HK1209521B - System and methods for coping with doppler effects in distributed-input distributed-output wireless systems - Google Patents

Description

System and method for handling the Doppler effect in a distributed-input-distributed-output wireless system

Related patent applications

This patent application is a continuation-in-part of the following co-pending U.S. patent application:

U.S. patent application serial number 12/917,257, filed on November 1, 2010, entitled “Systems and Methods To Coordinate Transmissions In Distributed Wireless Systems Via User Clustering”; U.S. patent application serial number 12/802,988, filed on June 16, 2010, entitled “Interference Management, Handoff, Power Control And Link Adaptation In Distributed-Input Distributed-Output (DIDO) Communication Systems”; and U.S. patent application serial number 12/802,988, filed on June 16, 2010, entitled “System And Method For Adjusting DIDO Interference Cancellation Based On Signal No. 12/802,976, entitled “System and Method for Adjusting DIDO Interference Cancellation Based on Signal Strength Measurements,” now U.S. Granted Patent No. 8,170,081, published May 1, 2012; No. 12/802,974, entitled “System and Method for Managing Inter-Cluster Handoff Of Clients Which Traverse Multiple DIDO Clusters,” filed June 16, 2010; and No. 12/802,975, entitled “System and Method for Managing Handoff Of A Client Between Different Distributed-Input-Distributed-Output (DIDO) Networks Based On Detected Velocity Of The No. 12/802,989, filed on June 16, 2010, entitled “System and Method for Managing Handoff of Clients Between Distributed-Input-Distributed-Output (DIDO) Networks Based on Detected Client Speed”; No. 12/802,958, filed on June 16, 2010, entitled “System and Method For Power Control And Antenna Grouping In A Distributed-Input-Distributed-Output (DIDO) Network”; No. 12/802,975, filed on June 16, 2010, entitled “System and Method For Link Adaptation In DIDO Multicarrier Systems”; and No. 12/802,975, filed on June 16, 2010, entitled “System and Method For DIDO Precoding Interpolation In Multicarrier Systems”. No. 12/802,938, entitled “System and Method for DIDO Precoding Interpolation in Multicarrier Systems”, filed December 3, 2009; No. 12/630,627, entitled “System and Method For Distributed Antenna Wireless Communications”, filed December 3, 2009; No. 12/143,503, entitled “System and Method For Distributed Input-Distributed Output Wireless Communications”, filed June 20, 2008, now U.S. Granted Patent No. 8,160,121, issued April 17, 2009; No. 12/163,503, entitled “System and Method For Distributed Input-Distributed Output Wireless Communications”, filed August 20, 2007; and No. 12/163,503, entitled “System and Method For Distributed Input-Distributed Output Wireless Communications”, filed June 20, 2008. No. 11/894,394, entitled “System and method for Distributed Input-Distributed Wireless Communications,” filed on August 20, 2007, now U.S. Grant Patent No. 7,599,420, published on October 6, 2009; No. 11/894,362, entitled “System and method for Distributed Input-Distributed Wireless Communications,” filed on August 20, 2007, now U.S. Grant Patent No. 7,633,994, published on December 15, 2009; No. 11/894,540, entitled “System and Method For Distributed Input-Distributed Output Wireless Communications,” filed on August 20, 2007, now U.S. Grant Patent No. 7,636,381, published on December 22, 2009; and No. 11/894,570, entitled “System and Method For Distributed Input-Distributed Output Wireless Communications,” filed on October 21, 2005. No. 11/256,478, entitled “Spatial-Multiplexed Tropospheric Scatter Communications,” now U.S. Grant Patent No. 7,711,030, published May 4, 2010; and No. 10/817,731, entitled “System and Method For Enhancing Near Vertical Incidence Skywave (“NVIS”) Communication Using Space-Time Coding,” filed April 2, 2004, now U.S. Grant Patent No. 7,885,354, published February 28, 2011.

Background Art

Prior art multi-user wireless systems may include only a single base station or several base stations.

A single WiFi base station (e.g., utilizing the 2.4 GHz 802.11b, g, or n protocol) connected to a broadband wired internet connection in an area where no other WiFi access points exist (e.g., a WiFi access point connected to a DSL connection in a rural user's home) is an example of a relatively simple multi-user wireless system with a single base station shared by one or more users within its transmission range. If a user is in the same room as the wireless access point, the user will generally experience a high-speed link with few transmission interruptions (e.g., packet loss may occur due to a 2.4 GHz interferer (e.g., a microwave oven), but not due to spectrum sharing with other WiFi devices). If the user is moderately far away or there are several obstacles in the path between the user and the WiFi access point, the user will likely experience a medium-speed link. If the user is approaching the edge of the WiFi access point's range, the user will likely experience a low-speed link and may experience periodic interruptions if channel variations cause the signal SNR to drop below a usable level. Finally, if the user is out of range of the WiFi base station, the user will have no link at all.

When multiple users access a WiFi base station simultaneously, the available data throughput is shared among them. Different users will typically place different throughput demands on the WiFi base station at a given time, but sometimes when the aggregate throughput demands exceed the available throughput from the WiFi base station to the users, some or all users will receive less data throughput than they are seeking. In extreme cases where a WiFi access point is shared among a very large number of users, the throughput to each user can slow to a crawl, or worse, the data throughput to each user may arrive in short bursts separated by long periods of no data throughput at all, during which time other users are being served. This "intermittent" data transmission can harm certain applications, such as media streaming.

Adding additional WiFi base stations in scenarios with a large number of users will only help to a certain extent. Within the 2.4 GHz ISM band in the United States, there are three non-interfering channels available for WiFi, and if three WiFi base stations in the same coverage area are configured to each use a different non-interfering channel, the aggregate throughput of the coverage area across multiple users will increase by up to three times. Beyond that, however, adding more WiFi base stations in the same coverage area will not increase aggregate throughput because they will begin to share the same available spectrum among themselves, effectively utilizing time division multiple access (TDMA) by “taking turns” using that spectrum. This scenario is common in coverage areas with high population density (such as in multi-dwelling units). For example, a user in a large apartment building with a WiFi adapter may experience significantly poor throughput due to the numerous interfering WiFi networks (e.g., in other apartments) serving other users in the same coverage area, even if the user's access point is in the same room as the client device accessing the base station. While the link quality may be good in this scenario, the user will receive interference from neighboring WiFi adapters operating in the same frequency band, reducing the user's effective throughput.

Current multi-user wireless systems (including both unlicensed spectrum (such as WiFi) and licensed spectrum) suffer from several limitations. These limitations include coverage area, downlink (DL) data rate, and uplink (UL) data rate. A key goal of next-generation wireless systems (such as WiMAX and LTE) is to improve coverage area and DL and UL data rates via multiple-input multiple-output (MIMO) technology. MIMO uses multiple antennas on the transmit and receive sides of a wireless link to improve link quality (producing wider coverage) or data rate (by creating multiple non-interfering spatial channels to each user). However, if sufficient data rate is available for each user (note that the terms "user" and "client" are used interchangeably in this article), it may be necessary to create non-interfering channels to multiple users (rather than a single user) using channel spatial diversity according to multi-user MIMO (MU-MIMO) technology. See, for example, the following references:

G. Caire and S. Shamai, “On the achievable throughput of a multiantenna Gaussian broadcast channel,” IEEE Trans. Info. Th., vol. 49, pp. 1691–1706, July 2003.

P.Viswanath and D.Tse, “Sum capacity of the vector Gaussian broadcast channel and uplink-downlink duality,” IEEE Trans.Info.Th., vol.49, pp.1912–1921, Aug.2003.

S. Vishwanath, N. Jindal, and A. Goldsmith, “Duality, achievable rates, and sum-rate capacity of Gaussian MIMO broadcast channels,” IEEE Trans. Info. Th., vol. 49, pp. 2658–2668, Oct. 2003.

W.Yu and J.Cioffi, “Sum capacity of Gaussian vector broadcast channels,” IEEE Trans.Info.Th., vol.50, pp.1875–1892, Sep.2004.

M. Costa, “Writing on dirty paper,” IEEE Transactions on Information Theory, vol. 29, pp. 439–441, May 1983.

M. Bengtsson, “A pragmatic approach to multi-user spatial multiplexing,” Proc. of Sensor Array and Multichannel Sign. Proc. Workshop, pp. 130–134, Aug. 2002.

K.-K. Wong, R.D. Murch, and K.B. Letaief, “Performance enhancement of multiuser MIMO wireless communication systems,” IEEE Trans. Comm., vol. 50, pp. 1960–1970, Dec. 2002.

M. Sharif and B. Hassibi, “On the capacity of MIMO broadcast channel with partial side information,” IEEE Trans. Info. Th., vol. 51, pp. 506–522, Feb. 2005.

For example, in a MIMO 4×4 system (i.e., four transmit antennas and four receive antennas) with a 10 MHz bandwidth, 16-QAM modulation, and forward error correction (FEC) coding at a rate of 3/4 (yielding a spectral efficiency of 3 bps/Hz), the ideal peak data rate achievable at the physical layer for each user is 4×30 Mbps = 120 Mbps, which is much higher than the rate required to transmit high-definition video content (which may only require about 10 Mbps). In a MU-MIMO system with four transmit antennas, four users, and a single antenna per user, in an ideal case (i.e., independent and identically distributed (i.i.d.) channels), the downlink data rate can be shared among the four users and channel spatial diversity can be exploited to create four parallel 30 Mbps data links to the users.

Different MU-MIMO schemes have been proposed as part of the LTE standard, as described, for example, in 3GPP, “Multiple Input Multiple Output in UTRA,” 3GPP TR 25.876 V7.0.0, Mar. 2007; 3GPP, “Base Physical channels and modulation,” TS 36.211, V8.7.0, May 2009; and 3GPP, “Multiplexing and channel coding,” TS 36.212, V8.7.0, May 2009. 36.212, V8.7.0, May 2009. However, these schemes can only provide up to a 2x improvement in DL data rate with four transmit antennas. Practical implementations of MU-MIMO technology in standard and proprietary cellular systems by companies like ArrayComm (see, for example, ArrayComm, “Field-proven "(Field Verified Results), http://www.arraycomm.com/serve.php?page=proo) has produced up to a 3x increase in DL data rates (via four transmit antennas) from spatial division multiple access (SDMA). A key limitation of MU-MIMO schemes in cellular networks is the lack of spatial diversity on the transmit side. Spatial diversity is a function of antenna spacing and multipath angular spread in the wireless link. In cellular systems using MU-MIMO technology, the transmit antennas at the base station are typically clustered together and placed only one or two wavelengths apart due to the limited footprint on the antenna support structure (referred to herein as a "tower," whether physically tall or not) and due to restrictions on where the tower can be located. Furthermore, because cell towers are typically placed high above obstructions (10 meters or more) to produce wide coverage, the multipath angular spread is low.

Other practical issues with cellular system deployment include the excessive cost and limited availability of cellular antenna locations (e.g., due to municipal restrictions on antenna placement, real estate costs, physical obstructions, etc.), as well as the cost and/or availability of network connectivity to transmitters (referred to herein as "backhaul"). Furthermore, cellular systems often struggle to reach clients located deep within buildings due to losses through walls, ceilings, floors, furniture, and other obstructions. Indeed, the entire concept of a cellular architecture for wide-area wireless networks presupposes relatively fixed placement of cellular towers, alternation of frequencies between adjacent cells, and frequent sectorization to avoid interference between transmitters (base stations or users) using the same frequency. Consequently, a given sector of a given cell ultimately becomes a shared block of DL and UL spectrum among all users in that cell sector, which is then primarily shared between these users only in the time domain. For example, cellular systems based on time division multiple access (TDMA) and code division multiple access (CDMA) both share spectrum between users in the time domain. By overlaying such cellular systems with sectorization, perhaps a 2-3x spatial domain benefit can be achieved. And, then, by overlaying such cellular systems with MU-MIMO systems (such as those described previously), perhaps another 2-3x space-time domain benefit could be achieved. However, given that the cells and sectors of a cellular system are typically in fixed locations (often dictated by where towers can be placed), even these limited benefits are difficult to exploit if the user density (or data rate demand) at a given time doesn't match the tower/sector placement well. Cellular smartphone users often experience the following: one day, the user may be talking on the phone or downloading a web page with no problems, and then, after driving (or even walking) to a new location, suddenly find that the voice quality degrades, the web page slows to a crawl, or even loses the connection completely. However, on different days, the user may encounter completely opposite situations in each location. Assuming the same environmental conditions, the situation the user may be experiencing is that the user density (or data rate demand) is highly variable, but the total available spectrum shared among users at a given location (and therefore the total data rate, using existing technology) is largely fixed.

Furthermore, prior art cellular systems rely on using different frequencies in different adjacent cells, typically 3 different frequencies. For a given amount of spectrum, this reduces the available data rate by one-third.

So, in summary, a prior art cellular system can lose perhaps 3x spectrum utilization due to cellularization, and can improve spectrum utilization by perhaps 3x through sectorization and perhaps another 3x through MU-MIMO technology, resulting in a net 3*3/3 = 3x possible spectrum utilization. The bandwidth is then typically divided among users in the time domain based on which sector of which cell a user belongs to at a given time. Even further inefficiencies arise due to the fact that a given user's data rate requirements are generally independent of the user's location, but the available data rate varies depending on the link quality between the user and the base station. For example, users farther from a cellular base station will generally have a lower available data rate than users closer to the base station. Because the data rate is typically shared among all users in a given cellular sector, the consequence of this is that all users are impacted by high data rate demands from distant users with poor link quality (e.g., at the cell's edge), as these users will still demand the same amount of data rate, but will consume more of the shared spectrum to achieve it.

Other proposed spectrum sharing systems, such as those used by WiFi (e.g., 802.11b, g, and n) and those proposed by the White Spaces Coalition, share spectrum very inefficiently because simultaneous transmissions by base stations within range of users cause interference, and thus the systems utilize collision avoidance and sharing protocols. These spectrum sharing protocols are in the time domain, and therefore, when there are a large number of interfering base stations and users, regardless of how efficient each base station is in spectrum utilization, the base stations are collectively limited to time-domain sharing of spectrum with each other. Other prior art spectrum sharing systems similarly rely on similar methods to mitigate interference between base stations (whether cellular base stations with tower-mounted antennas or small-scale base stations such as WiFi access points (APs)). These methods include: limiting the transmit power from the base stations to limit the range of interference; beamforming (via synthesis or physical means) to narrow the area of interference; time-domain multiplexing of spectrum; and/or MU-MIMO technology with multiple clustered antennas on user devices, base stations, or both. And, with advanced cellular networks deployed today or in planning, many of these technologies are often used simultaneously.

However, as is evident from the fact that even advanced cellular systems can only achieve a roughly three-fold increase in spectrum utilization compared to single-user spectrum utilization, all of these techniques are ineffective in increasing the aggregate data rate among shared users within a given coverage area. Specifically, as a given coverage area scales in terms of users, it becomes increasingly difficult to scale the available data rate within a given amount of spectrum to keep pace with the increase in users. For example, in the case of cellular systems, to increase the aggregate data rate within a given area, cells are typically subdivided into smaller cells (often referred to as nano-cells or femto-cells). These small cells can be extremely expensive given the restrictions on where towers can be placed, and the requirement that towers be placed in a suitably structured pattern to provide coverage with minimal "dead zones" while avoiding interference between neighboring cells using the same frequency. Essentially, the coverage area must be mapped, available locations for tower or base station placement must be identified, and then, given these constraints, designers of cellular systems must do their best to accomplish this. And, of course, if user data rate demands increase over time, cellular system designers must once again redraw the coverage area, try to locate towers or base stations, and once again work within the constraints of the environment. And often, there simply is no good solution, resulting in dead zones in the coverage area or insufficient aggregate data rate capacity. In other words, the strict physical placement requirements of cellular systems to avoid interference between towers or base stations utilizing the same frequency lead to significant difficulties and constraints in cellular system design, and user data rate and coverage requirements are often not met.

So-called prior art "cooperative" and "cognitive" radio systems seek to increase spectrum utilization in a given area by using intelligent algorithms within the radios to enable the radios to minimize interference between each other and/or to enable the radios to potentially "listen" to other spectrum usage in order to wait until the channel is free of interference. Such systems are proposed to be used particularly in unlicensed spectrum in order to increase spectrum utilization of this spectrum.

Mobile ad hoc networks (MANETs) (see http://en.wikipedia.org/wiki/Mobile_ad_hoc_network) are examples of cooperative, self-configuring networks designed to provide peer-to-peer communications and can be used to create communications between radios without cellular infrastructure and, with sufficiently low-power communications, can potentially mitigate interference between simultaneous transmissions that are out of range of each other. A large number of routing protocols have been proposed and implemented for MANET systems (see http://en.wikipedia.org/wiki/List_of_ad-hoc_routing_protocols for a list of many routing protocols in various categories), but the common theme among them is that they are all techniques for routing (e.g., repeating) transmissions so as to minimize transmitter interference within the available spectrum with the goal of achieving a specific efficiency or reliability paradigm.

All prior art multi-user wireless systems seek to improve spectrum utilization within a given coverage area by utilizing techniques that allow simultaneous spectrum utilization between a base station and multiple users. Note that in all of these cases, the techniques for simultaneous spectrum utilization between a base station and multiple users enable simultaneous spectrum use by multiple users by mitigating interference between the waveforms to the multiple users. For example, in the case of three base stations each using a different frequency to transmit to one of three users, interference is mitigated because the three transmissions are at three different frequencies. In the case of sectorization from a base station to three different users (each 180 degrees apart relative to the base station), interference is mitigated because beamforming prevents the three transmissions from overlapping at any one user.

When such technology is enhanced with MU-MIMO, and (for example) each base station has 4 antennas, this has the potential to increase downlink throughput by a factor of 4 by creating four non-interfering spatial channels to users in a given coverage area. However, some techniques must still be utilized to mitigate interference between multiple simultaneous transmissions to multiple users in different coverage areas.

Furthermore, as previously mentioned, these prior art techniques (e.g., cellularization, sectorization) not only typically suffer from increasing the cost and/or deployment flexibility of multi-user wireless systems, but they also typically encounter physical or practical limitations on the aggregate throughput in a given coverage area. For example, in a cellular system, there may not be enough available locations to install more base stations to create smaller cells. Furthermore, in a MU-MIMO system, given the clustered antenna spacing at each base station location, limited spatial diversity leads to diminishing returns in throughput as more antennas are added to the base station.

Furthermore, in the case of multi-user wireless systems where user locations and densities are unpredictable, this results in unpredictable throughput (with frequent and dramatic changes), which is inconvenient for users and renders some applications (e.g., delivery of services requiring predictable throughput) impractical or of low quality. Therefore, prior art multi-user wireless systems still leave much to be desired in terms of their ability to provide predictable and/or high-quality services to users.

While prior art multi-user wireless systems have become very sophisticated and complex over time, a common theme exists: distributing transmissions among different base stations (or ad hoc transceivers) and structuring and/or controlling the transmissions so as to avoid RF waveform transmissions from different base stations and/or different ad hoc transceivers interfering with each other at a given user's receiver.

Or, in other words, it is considered a known fact that if a user happens to receive transmissions from more than one base station or ad hoc transceiver simultaneously, the interference from the multiple simultaneous transmissions will result in a reduction in the SNR and/or bandwidth of the signal to the user, which (if severe enough) will result in the loss of all or some of the potential data (or analog information) that would otherwise be received by the user.

Therefore, in a multi-user wireless system, one or more spectrum sharing methods or another method must be utilized to avoid or mitigate such interference to users from multiple base stations or ad hoc transceivers transmitting simultaneously at the same frequency. Numerous prior art methods exist for avoiding such interference, including controlling the physical location of base stations (e.g., cellularization), limiting the power output of base stations and/or ad hoc transceivers (e.g., limiting the transmission range), beamforming/sectorization, and time domain multiplexing. In short, all of these spectrum sharing systems address a limitation of multi-user wireless systems: when multiple base stations and/or ad hoc transceivers transmitting simultaneously at the same frequency are received by the same user, the resulting interference reduces or destroys the data throughput to the affected users. If most or all of the users in a multi-user wireless system experience interference from multiple base stations and/or ad hoc transceivers (e.g., in the event of a component failure in the multi-user wireless system), this can result in a situation where the aggregate throughput of the multi-user wireless system is drastically reduced or even loses functionality.

Prior art multi-user wireless systems add complexity and introduce limitations to wireless networks, frequently leading to situations where a given user's experience (e.g., available bandwidth, latency, predictability, reliability) is impacted by the spectrum utilization of other users in the area. Given the increasing demand for aggregate bandwidth within a wireless spectrum shared by multiple users, and the growing number of applications that can rely on the reliability, predictability, and low latency of a multi-user wireless network for a given user, it is clear that prior art multi-user wireless technologies suffer from numerous limitations. Indeed, due to the limited availability of spectrum suitable for certain types of wireless communications (e.g., at wavelengths that can effectively penetrate building walls), it is likely that prior art wireless technologies will be insufficient to meet the increasing demand for reliable, predictable, and low-latency bandwidth.

Prior art related to the present invention describes beamforming systems and methods for null steering in multi-user scenarios. Beamforming was originally conceived to maximize the received signal-to-noise ratio (SNR) by dynamically adjusting the phase and/or amplitude (i.e., beamforming weights) of the signal fed to the antennas of the array, thereby concentrating energy in the direction of the user. In multi-user scenarios, beamforming can be used to suppress interference sources and maximize the signal-to-interference-plus-noise ratio (SINR). For example, when beamforming is used at the receiver of a wireless link, weights are calculated to create a null in the direction of the interference source. When beamforming is used at the transmitter in a multi-user downlink scenario, weights are calculated to pre-eliminate inter-user interference and maximize the SINR to each user. Alternative techniques for multi-user systems, such as BD precoding, calculate precoding weights to maximize throughput in the downlink broadcast channel. The co-pending patent application incorporated herein by reference describes the above technology (see the co-pending patent application for specific citations).

BRIEF DESCRIPTION OF THE DRAWINGS

A better understanding of the present invention may be obtained from the following detailed description taken in conjunction with the accompanying drawings, in which:

FIG1 illustrates a main DIDO cluster surrounded by neighboring DIDO clusters in one embodiment of the present invention.

FIG2 illustrates a frequency division multiple access (FDMA) technique used in one embodiment of the present invention.

FIG3 illustrates a time division multiple access (TDMA) technique used in one embodiment of the present invention.

FIG4 illustrates different types of interference zones addressed in one embodiment of the present invention.

FIG5 shows a framework used in one embodiment of the present invention.

FIG6 shows a graph illustrating the variation of SER with SNR, assuming SIR=10 dB for a target client in an interference zone.

FIG7 shows a graph illustrating the SER obtained by two IDCI-precoding techniques.

FIG8 shows an exemplary scenario where a target client moves from a primary DIDO cluster to an interfering cluster.

FIG9 shows the relationship between the signal to interference plus noise ratio (SINR) and the distance (D).

FIG10 shows the symbol error rate (SER) performance for three cases of 4-QAM modulation in a flat fading narrowband channel.

FIG11 shows a method for IDCI precoding according to an embodiment of the present invention.

FIG. 12 shows the SINR variation as a function of the distance of the client from the center of the primary DIDO cluster, in one embodiment.

FIG13 shows an embodiment where the SER is derived for 4-QAM modulation.

FIG. 14 illustrates an embodiment of the present invention in which a finite state machine implements a handoff algorithm.

FIG. 15 illustrates one embodiment of a handoff strategy in the presence of shadowing.

FIG. 16 illustrates the hysteresis loop mechanism when switching between any two states of FIG. 15 .

FIG17 illustrates one embodiment of a DIDO system with power control.

FIG18 shows the SER versus SNR relationship in different scenarios assuming four DIDO transmit antennas and four clients.

FIG. 19 shows the relationship between the MPE power density and the distance from the RF radiation source for different transmit power values according to one embodiment of the present invention.

20a-20b illustrate different distributions of low-power and high-power DIDO distributed antennas.

21a-21b show two power distributions corresponding to the configurations in FIG. 20a and FIG. 20b , respectively.

22a-22b show the rate distributions for the two scenarios shown in FIG. 21a and FIG. 21b , respectively.

FIG23 illustrates one embodiment of a DIDO system with power control.

FIG. 24 illustrates one embodiment of a method that is repeated across all antenna groups according to a cyclic scheduling strategy for transmitting data.

FIG. 25 shows a comparison of the uncoded SER performance of power control with antenna grouping and conventional eigenmode selection in U.S. Patent No. 7,636,381.

26a-26c illustrate three scenarios where BD precoding dynamically adjusts the precoding weights to account for different power levels on the wireless link between the DIDO antennas and the clients.

27 shows the amplitude of a low frequency selective channel (assuming β=1) for a DIDO 2×2 system in the delay domain or instantaneous PDP (upper curve) and the frequency domain (lower curve).

FIG. 28 shows one embodiment of the channel matrix frequency response for DIDO 2×2 with a single antenna per client.

FIG. 29 shows one embodiment of the channel matrix frequency response for DIDO 2×2 with a single antenna per client for a channel characterized by high frequency selectivity (eg, with β=1).

FIG30 shows exemplary SERs for different QAM schemes (ie, 4-QAM, 16-QAM, 64-QAM).

FIG31 illustrates one embodiment of a method for implementing link adaptation (LA) technology.

FIG32 illustrates the SER performance of one embodiment of the Link Adaptation (LA) technique.

FIG33 shows how the entries of the matrix in Equation (28) vary with OFDM tone index for a DIDO 2×2 system with _NFFT = 64 and _L0 = 8.

FIG34 shows the SER versus SNR for L ₀ =8, M=N _t =2 transmit antennas, and a variable number of P. ...

FIG35 shows the SER performance of one embodiment of the interpolation method for different DIDO orders and L ₀ =16.

FIG36 illustrates one embodiment of a system using super clusters, DIDO-clusters, and user clusters.

FIG37 illustrates a system with user clustering according to one embodiment of the present invention.

38a-38b illustrate link quality metric thresholds used in one embodiment of the present invention.

39-41 show examples of link quality matrices used to create user clusters.

FIG42 illustrates an embodiment of a client moving across different DIDO clusters.

43-46 show the relationship between the resolution of a spherical array and its area A in one embodiment of the present invention.

Figure 47 shows the degrees of freedom of a MIMO system in practical indoor and outdoor propagation scenarios.

FIG48 shows the degrees of freedom in a DIDO system as a function of array diameter.

FIG49 illustrates an embodiment comprising multiple centralized processors (CPs) and distributed nodes (DNs) communicating via wired or wireless connections.

FIG50 shows an embodiment in which the CP exchanges control information with unlicensed DNs and reconfigures them to shut down the frequency band for licensed use.

Figure 51 shows an embodiment where the entire spectrum is allocated to the new service and the CP uses control information to shut down all unlicensed DNs to avoid interfering with the licensed DNs.

FIG52 illustrates an embodiment of a cloud wireless system including a plurality of CPs, distributed nodes, and a network interconnecting the CPs and DNs.

53-59 illustrate embodiments of a multi-user (MU) multi-antenna system (MAS) that adaptively reconfigures parameters to compensate for the Doppler effect due to changes in user mobility or propagation environment.

FIG. 60 shows multiple BTSs, some of which have good SNR and some of which have low Doppler relative to the UE.

FIG61 shows one embodiment of a matrix containing the values of SNR and Doppler for multiple BTS-UE links recorded by the CP.

FIG62 shows channel gains (or CSI) at different times according to one embodiment of the present invention.

DETAILED DESCRIPTION

One solution to overcome many of the limitations of the prior art described above is an embodiment of Distributed Input Distributed Output (DIDO) technology. DIDO technology is described in the following patents and patent applications, all of which are assigned to the assignee of this patent and incorporated by reference. This patent application is a continuation-in-part (CIP) of these patent applications. These patents and patent applications are sometimes collectively referred to herein as "related patents and patent applications."

U.S. Patent Application Serial No. 13/232,996, filed September 14, 2011, entitled “Systems and Methods To Exploit Areas of Coherence in Wirless Systems”

U.S. patent application serial number 13/233,006, filed September 14, 2011, entitled “Systems and Methods for Planned Evoluation and Obsolescence of Multiuser Spectrum.”

U.S. Patent Application Serial No. 12/917,257, filed November 1, 2010, entitled “Systems and Methods To Coordinate Transmissions In Distributed Wireless Systems Via User Clustering”

U.S. patent application Ser. No. 12/802,988, filed on June 16, 2010, entitled “Interference Management, Handoff, Power Control And Link Adaptation In Distributed-Input Distributed-Output (DIDO) Communication Systems”

U.S. Patent Application Serial No. 12/802,976, filed June 16, 2010, entitled “System And Method For Adjusting DIDO Interference Cancellation Based On Signal Strength Measurements”

U.S. Patent Application Serial No. 12/802,974, filed June 16, 2010, entitled “System And Method For Managing Inter-Cluster Handoff Of Clients Which Traverse Multiple DIDO Clusters”

U.S. Patent Application Serial No. 12/802,989, filed June 16, 2010, entitled “System And Method For Managing Handoff Of A Client Between Different Distributed-Input-Distributed-Output (DIDO) Networks Based On Detected Velocity Of The Client”

U.S. patent application serial number 12/802,958, entitled “System And Method For Power Control And Antenna Grouping In A Distributed-Input-Distributed-Output (DIDO) Network,” filed June 16, 2010

U.S. Patent Application Serial No. 12/802,975, filed June 16, 2010, entitled “System And Method For Link Adaptation In DIDO Multicarrier Systems”

U.S. Patent Application Serial No. 12/802,938, filed June 16, 2010, entitled “System And Method For DIDO Precoding Interpolation In Multicarrier Systems”

U.S. Patent Application Serial No. 12/630,627, filed December 2, 2009, entitled “System and Method For Distributed Antenna Wireless Communications”

U.S. Patent No. 7,599,420, entitled “System and Method for Distributed Input Distributed Output Wireless Communication,” filed on August 20, 2007 and published on October 6, 2009;

U.S. Patent No. 7,633,994, “System and Method for Distributed Input Distributed Output Wireless Communication,” filed on August 20, 2007 and published on December 15, 2009;

U.S. Patent No. 7,636,381, “System and Method for Distributed Input Distributed Output Wireless Communication,” filed on August 20, 2007 and published on December 22, 2009;

U.S. patent application serial number 12/143,503, entitled “System and Method For Distributed Input-Distributed Output Wireless Communications,” filed on June 20, 2008;

U.S. patent application serial number 11/256,478, entitled “System and Method For Spatial-Multiplexed Tropospheric Scatter Communications,” filed on October 21, 2005;

U.S. Patent No. 7,418,053, “System and Method for Distributed Input Distributed Output Wireless Communication,” filed on July 30, 2004 and published on August 26, 2008;

U.S. patent application serial number 10/817,731, filed April 2, 2004, entitled “System and Method For Enhancing Near Vertical Incidence Skywave (“NVIS”) Communication Using Space-Time Coding”.

In order to reduce the length and complexity of this patent application, the disclosures of some of the related patents and patent applications are not explicitly listed below. Please refer to the related patents and patent applications for a complete and detailed description of the present disclosure.

It should be noted that the following Section I (from the disclosure of related patent application Ser. No. 12/802,988) uses its own set of endnotes that refer to prior art references and prior patent applications assigned to the assignee of the present patent application. The endnote references are listed at the end of Section I (just before the Section II header). The references used in Section II can have numerical labels for their references that overlap with those used in Section I, and even identify different references (listed at the end of Section II) by these numerical labels. Thus, the references identified by a particular numerical label can be identified in the section that uses that numerical label.

I. Disclosure from Related Patent Application Serial No. 12/802,988

1. Methods for Removing Inter-Cluster Interference

The following describes a wireless radio frequency (RF) communication system and method for creating locations with zero RF energy in space using multiple distributed transmit antennas. When M transmit antennas are used, up to (M-1) zero RF energy points can be created in predefined locations. In one embodiment of the present invention, the zero RF energy point is a wireless device, and the transmit antenna is aware of the channel state information (CSI) between the transmitter and the receiver. In one embodiment, the CSI is calculated at the receiver and fed back to the transmitter. In another embodiment, assuming channel reciprocity, the CSI is calculated at the transmitter via training from the receiver. The transmitter can use the CSI to determine the interfering signals that will be transmitted simultaneously. In one embodiment, block diagonalization (BD) precoding is used at the transmit antenna to generate the zero RF energy point.

The systems and methods described herein differ from the conventional receive/transmit beamforming techniques described above. While receive beamforming computes weights to suppress interference on the receive side (via null steering), some embodiments of the invention described herein apply weights on the transmit side to create interference patterns that result in one or more locations in space with “zero RF energy.” Unlike conventional transmit beamforming or BD precoding, which are designed to maximize signal quality (or SINR) or downlink throughput to each user, respectively, the systems and methods described herein minimize signal quality under certain conditions and/or from certain transmitters, thereby creating zero RF energy points at client devices (sometimes referred to herein as “users”). Furthermore, in the context of Distributed Input Distributed Output (DIDO) systems (described in our related patents and patent applications), the transmit antennas distributed in space provide higher degrees of freedom (i.e., higher channel spatial diversity) for creating multiple zero RF energy points and/or maximum SINR to different users. For example, with M transmit antennas, up to (M-1) RF energy points can be created. In contrast, practical beamforming or BD multi-user systems are typically designed with densely packed antennas at the transmit side, limiting the number of simultaneous users that can be served on the wireless link for any number M of transmit antennas.

Considering a system with M transmit antennas and K users, where K < M, we assume that the transmitter knows the CSI between the M transmit antennas and the K users (H∈C ^KxM ). For simplicity, each user is assumed to be equipped with a single antenna, but the same approach can be extended to multiple receive antennas per user. The precoding weights (w∈C ^Mx1 ) that create zero RF energy at the K user locations are calculated to satisfy the following condition:

Hw＝0 ^Kx1

where 0 ^Kx1 is a vector with all zero entries, and H is the channel matrix obtained by combining the channel vectors (h _k ∈ ^{C 1xM} ) from M transmit antennas to K users as follows

In one embodiment, the singular value decomposition (SVD) of the channel matrix H is computed, and the precoding weights w are defined as the right singular vectors corresponding to the null subspace of H (identified by zero singular values).

The transmit antenna transmits RF energy using the weight vector defined above, while creating K zero RF energy points at the locations of K users, so that the signal received at the kth user is given by

r _k = h _k ws _k + n _k = 0 + n _k

where n _k ∈ ^{C 1x1} is the additive white Gaussian noise (AWGN) at the k-th user.

In one embodiment, the singular value decomposition (SVD) of the channel matrix H is computed, and the precoding weights w are defined as the right singular vectors corresponding to the null subspace of H (identified by zero singular values).

In another embodiment, the wireless system is a DIDO system and creates zero RF energy points to preemptively eliminate interference to clients between different DIDO coverage areas. In U.S. patent application serial number 12/630,627, a DIDO system is described that includes:

DIDO client

DIDO distributed antenna

DIDO base transceiver station (BTS)

DIDO Base Station Network (BSN)

Each BTS is connected via the BSN to multiple distributed antennas that serve a given coverage area, known as a DIDO cluster. In this patent application, we describe systems and methods for removing interference between adjacent DIDO clusters. As shown in Figure 1, we assume that the primary DIDO cluster hosts clients (i.e., user devices served by a multi-user DIDO system) that are affected by interference (or target clients) from a neighboring cluster.

In one embodiment, adjacent clusters operate at different frequencies using frequency division multiple access (FDMA) techniques, similar to conventional cellular systems. For example, with a frequency reuse factor of 3, every third DIDO cluster reuses the same carrier frequency, as shown in Figure 2. In Figure 2, the different carrier frequencies are identified as _F1 , _F2 , and _F3 . While this embodiment may be used in some implementations, the solution incurs a loss in spectral efficiency because the available spectrum is divided into multiple sub-bands and only a subset of DIDO clusters operate in the same sub-band. Furthermore, it requires complex cell planning to associate different DIDO clusters with different frequencies to prevent interference. Similar to prior art cellular systems, this cellular planning requires specific placement of antennas and limitations on transmit power to avoid interference between clusters using the same frequency.

In another embodiment, adjacent clusters operate in the same frequency band but at different time slots according to a time division multiple access (TDMA) technique. For example, as shown in FIG3 , DIDO transmissions in time slots T ₁ , T ₂ , and T ₃ are allowed only for certain clusters, as shown. Time slots can be equally allocated to different clusters, such that different clusters are scheduled according to a round-robin policy. If different clusters are characterized by different data rate requirements (i.e., clusters in a crowded urban environment versus clusters in a rural area with a smaller number of clients per coverage area), different priorities are assigned to different clusters, such that more time slots are allocated to clusters with greater data rate requirements. While TDMA as described above can be used for one embodiment of the present invention, the TDMA approach may require time synchronization across different clusters and may result in lower spectral efficiency because interfering clusters cannot use the same frequency simultaneously.

In one embodiment, all adjacent clusters transmit simultaneously in the same frequency band and use spatial processing across clusters to avoid interference. In this embodiment, the multi-cluster DIDO system: (i) uses conventional DIDO precoding within the primary cluster to transmit synchronized, non-interfering data streams to multiple clients within the same frequency band (as described in related patents and patent applications, including 7,599,420; 7,633,994; 7,636,381; and patent application serial number 12/143,503); and (ii) uses DIDO precoding with interference cancellation within the adjacent clusters to avoid interference with clients located in interference zone 8010 in FIG. 4 by creating a point of zero radio frequency (RF) energy at the location of the target client. If the target client is in interference zone 410, it will receive the sum of the RF containing the data stream from primary cluster 411 and the zero RF energy from interfering clusters 412-413, which will be the RF containing the data stream from the primary cluster. Thus, adjacent clusters can use the same frequency simultaneously without causing interference to the target client in the interference zone.

In practical systems, the performance of DIDO precoding can be affected by various factors, such as channel estimation errors or the Doppler effect (which produces outdated channel state information at the DIDO distributed antennas); intermodulation distortion (IMD) in multicarrier DIDO systems; and time or frequency offsets. Due to these effects, achieving the zero RF energy point can be impractical. However, as long as the RF energy from the interfering cluster at the target client is negligible compared to the RF energy from the primary cluster, the link performance at the target client will not be affected by the interference. For example, let's assume that the client requires a 20dB signal-to-noise ratio (SNR) to demodulate a 4-QAM constellation using forward error correction (FEC) coding to achieve a target bit error rate (BER) of ^10-6 . If the RF energy received from the interfering cluster at the target client is 20dB lower than the RF energy received from the primary cluster, the interference is negligible and the client can successfully demodulate the data within the predefined BER target. Therefore, as used herein, the term "zero RF energy" does not necessarily mean that the RF energy from the interfering RF signal is zero. Rather, it means that the RF energy is sufficiently low relative to the RF energy of the desired RF signal so that the desired RF signal can be received at the receiver. Furthermore, while certain desired thresholds of interfering RF energy relative to desired RF energy are described, the underlying principles of the present invention are not limited to any particular threshold.

As shown in Figure 4, there are different types of interference zones 8010. For example, a "Type A" zone (denoted by the letter "A" in Figure 4) is affected by interference from only one adjacent cluster, while a "Type B" zone (denoted by the letter "B") illustrates interference from two or more adjacent clusters.

Figure 5 illustrates a framework used in one embodiment of the present invention. The dots represent DIDO distributed antennas, the crosses represent DIDO clients, and the arrows indicate the direction of RF energy propagation. The DIDO antennas in the primary cluster transmit precoded data signals to client MC 501 in that cluster. Similarly, the DIDO antennas in the interfering cluster serve client IC 502 within that cluster via conventional DIDO precoding. The green cross 503 represents the target client TC 503 in the interference zone. The DIDO antennas in the primary cluster 511 transmit precoded data signals to the target client (black arrows) via conventional DIDO precoding. The DIDO antennas in the interfering cluster 512 use precoding to create an RF null in the direction of the target client 503 (green arrows).

The signal received at the target client k in any interference area 410A, 410B in FIG4 is given by

where k=1, ..., K, where K is the number of clients in the interfering zones 8010A, 8010B, U is the number of clients in the primary DIDO cluster, C is the number of interfering DIDO clusters 412-413, and _Ic is the number of clients in the interfering cluster c. In addition, r _k ∈ ^CN×M is a vector containing the received data streams at client k, assuming there are M transmit DIDO antennas and N receive antennas at the client device; _sk ∈ ^CN×1 is a vector of transmit data streams to client k in the primary DIDO cluster; _su ∈ ^CN×1 is a vector of transmit data streams to client u in the primary DIDO cluster; sc _,i ∈ ^CN×1 is a vector of transmit data streams to client i in the c-th interfering DIDO cluster; n _k ∈ ^CN×1 is a vector of additive white Gaussian noise (AWGN) at the N receive antennas of client k; H _k ∈ ^CN×M is the DIDO channel matrix from the M transmit DIDO antennas to the N receive antennas at client k in the primary DIDO cluster; H _c,k ∈ ^CN×M is the DIDO channel matrix from the M transmit DIDO antennas to the N receive antennas at client k in the c-th interfering DIDO cluster; W _k ∈ C ^M×N is the matrix of DIDO precoding weights to client k in the primary DIDO cluster; _Wk∈C ^M×N is the matrix of DIDO precoding weights to client u in the primary DIDO cluster; _Wc,i∈C ^M×N is the matrix of DIDO precoding weights to client i in the c-th interfering DIDO cluster.

For simplicity of notation and without loss of generality, we assume that all clients are equipped with N receive antennas and that there are M DIDO distributed antennas in each DIDO cluster, where M ≥ (N·U) and if M is greater than the total number of receive antennas in the cluster, then additional transmit antennas are used to preemptively cancel interference to target clients in the interference zone or to improve link robustness to clients within the same cluster through diversity schemes described in related patents and patent applications including 7,599,420; 7,633,994; 7,636,381; and patent application serial number 12/143,503.

DIDO precoding weights are calculated to preemptively eliminate inter-client interference within the same DIDO cluster. For example, block diagonalization (BD) precoding as described in related patents and patent applications (including 7,599,420, 7,633,994, 7,636,381, and patent application serial number 12/143,503 and [7]) can be used to remove inter-client interference such that the following condition is satisfied in the primary cluster:

The precoding weight matrices in adjacent DIDO clusters are designed such that the following conditions are satisfied

To calculate the precoding matrix W _c,i , the downlink channels from the M transmit antennas to the Ic clients in the interfering cluster and to the client k in the interference zone are estimated, and the precoding matrix is calculated by the DIDO BTS in the interfering cluster. If the BD method is used to calculate the precoding matrix in the interfering cluster, the following effective channel matrix is constructed to calculate the weight to the i-th client in the neighboring cluster:

where is the matrix obtained from the channel matrix for interfering cluster c with the row corresponding to the i-th client removed.

Substituting conditions (2) and (3) into (1), we obtain the data stream received by target client k, where intra-cluster and inter-cluster interference are removed

r _k ＝H _k W _k s _k +n _k . (5)

The precoding weights Wc _,i in (1) calculated in the neighboring clusters are designed to transmit the precoded data stream to all clients in those clusters while pre-cancelling the interference to the target client in the interference zone. The target client receives the precoded data only from its primary cluster. In a different embodiment, the same data stream is sent to the target client from both the primary cluster and the neighboring clusters to obtain diversity gain. In this case, the signal model in (5) is expressed as

where W _c,k is the DIDO precoding matrix from the DIDO transmitter in the cth cluster to the target client k in the interference zone. Note that the approach in (6) requires time synchronization across neighboring clusters, which can be complex to implement in large systems, but nonetheless it is quite feasible if the diversity gain benefits justify the implementation cost.

We begin by evaluating the performance of the proposed method in terms of the symbol error rate (SER) versus the signal-to-noise ratio (SNR). Without loss of generality, we define the following signal model assuming each client has a single antenna and reformulate (1) as

Where INR is the interference-to-noise ratio, defined as INR=SNR/SIR, and SIR is the signal-to-interference ratio.

Figure 6 shows the SER as a function of SNR, assuming SIR = 10 dB for the target client in the interference zone. Without loss of generality, we measure the SER for 4-QAM and 16-QAM without forward error correction (FEC) coding. For the uncoded system, we fix the target SER to 1%. This target corresponds to different values of SNR depending on the modulation order (i.e., SNR = 20 dB for 4-QAM and SNR = 28 dB for 16-QAM). When FEC coding is used, due to the coding gain, a lower SER target can be met for the same SNR value. We consider the case of two clusters (one primary cluster and one interfering cluster), each with two DIDO antennas and two clients (each equipped with a single antenna). One of the clients in the primary cluster is located in the interference zone. We assume a flat-fading narrowband channel, but the following results can be extended to frequency selective multicarrier (OFDM) systems, where each subcarrier experiences flat fading. We consider two scenarios: (i) one with DIDO inter-cluster interference (IDCI), where the precoding weights w _c,i are calculated without considering the target client in the interference zone; and (ii) another scenario where the IDCI is removed by calculating the weights w _c,i to cancel the IDCI for the target client. We observe that in the presence of IDCI, the SER is high and above the predefined target. By using IDCI-precoding at neighboring clusters, the interference for the target client is removed, and the SER target is achieved for SNR > 20 dB.

The results in Figure 6 assume IDCI-precoding as in (5). If the IDCI-precoding at the neighboring clusters is also used to precode the data stream to the target client in the interference zone as in (6), an additional diversity gain is obtained. Figure 7 compares the SER obtained by the following two techniques: (i) "Method 1" using IDCI-precoding as in (5); (ii) "Method 2" using IDCI-precoding as in (6), where the neighboring clusters also transmit the precoded data stream to the target client. Compared to conventional IDCI-precoding, Method 2 produces about 3dB gain due to the additional array gain provided by the DIDO antennas in the neighboring clusters used to transmit the precoded data stream to the target client. More generally, the array gain of Method 2 relative to Method 1 is proportional to 10*log10(C+1), where C is the number of neighboring clusters and the factor "1" refers to the master cluster.

We then evaluate the performance of the above approach as the target client's location relative to the interference zone varies. We consider a simple scenario where the target client 8401 moves from the primary DIDO cluster 802 to the interfering cluster 803, as shown in Figure 8. We assume that all DIDO antennas 812 within the primary cluster 802 use BD precoding to eliminate inter-cluster interference to satisfy condition (2). We assume a single interfering DIDO cluster, a single receiver antenna at the client device 801, and equal path loss from all DIDO antennas in the primary or interfering cluster (i.e., DIDO antennas placed in a circle around the client) to the client. We use a simplified path loss model [11] with a path loss exponent of 4 (as in a typical urban environment).

The following analysis is based on the following simplified signal model that extends (7) to take path loss into account:

where the signal-to-interference ratio (SIR) is derived as SIR=((1-D)/D) ⁴ . In modeling IDCI, we consider three cases: i) the ideal case without IDCI; ii) pre-cancelling IDCI in the interfering cluster via BD precoding to satisfy condition (3); iii) with IDCI not pre-cancelled by the neighboring cluster.

FIG9 shows the signal-to-interference-plus-noise ratio (SINR) as a function of D (i.e., when the target client moves from the primary cluster 802 toward the DIDO antenna 813 in the interfering cluster 8403). SINR is derived as the ratio of signal power to interference-plus-noise power using the signal model in (8). We assume that for D=D _o , D _o =0.1 and SNR=50 dB. In the absence of IDCI, the radio link performance is only affected by noise, and SINR is reduced due to path loss. In the presence of IDCI (i.e., no IDCI-precoding), interference from DIDO antennas in neighboring clusters helps to reduce SINR.

Figure 10 shows the symbol error rate (SER) performance of the above three cases for 4-QAM modulation in a flat fading narrowband channel. These SER results correspond to the SINR in Figure 9. We assume that a SER threshold of 1% for an uncoded system (i.e., without FEC) corresponds to the SINR threshold SINR _T = 20 dB in Figure 9. The SINR threshold depends on the modulation order used for data transmission. Higher modulation orders are typically characterized by higher SINR _T to achieve the same target error rate. With FEC, due to coding gain, a lower target SER can be achieved for the same SINR value. In the case of IDCI without precoding, the target SER is only achieved within the range D < 0.25. With IDCI-precoding at adjacent clusters, the range of meeting the target SER is extended to up to D < 0.6. Outside the range, SINR increases due to path loss and the SER target is not met.

FIG11 shows an embodiment of a method for IDCI precoding, which includes the following steps:

SIR estimation 1101: The client estimates the signal power from the primary DIDO cluster (i.e., based on the received precoded data) and the interference plus noise signal power from the neighboring DIDO clusters. In a single-carrier DIDO system, the framework structure can be designed with short silent periods. For example, the silent period can be defined between the transmission of precoded data during training for channel estimation and channel state information (CSI) feedback. In one embodiment, the interference plus noise signal power from the neighboring clusters is measured by the DIDO antennas in the primary cluster during the silent period. In practical DIDO multi-carrier (OFDM) systems, null tones are typically used to prevent direct current (DC) offsets and attenuation at the band edges due to filtering at the transmit and receive sides. In another embodiment using a multi-carrier system, the interference plus noise signal power is estimated by the null tones. A correction factor can be used to compensate for the transmit/receive filter attenuation at the band edges. Once the signal plus interference and noise power ( _PS ) from the primary cluster and the interference plus noise power ( _PIN ) from the neighboring cluster are estimated, the client calculates the SINR as

Alternatively, the SINR estimate is derived from the Received Signal Strength Indicator (RSSI) used in typical wireless communication systems to measure radio signal power.

We observe that the metric in (9) cannot distinguish between noise and interference power levels. For example, clients that are affected by shadowing in an interference-free environment (i.e., behind an obstacle that attenuates the signal power from all DIDO distributed antennas in the primary cluster) may estimate a low SINR even though they are not affected by inter-cluster interference.

A more reliable measure of the proposed method is SIR, which is calculated as

where _PN is the noise power. In a practical multi-carrier OFDM system, the noise power _PN in (10) is estimated from the null tones, assuming that all DIDO antennas from the primary and neighboring clusters use the same set of null tones. The interference plus noise power ( _PIN ) is estimated from the silent periods as described above. Finally, the signal plus interference and noise power ( _Ps ) is derived from the data tones. From these estimates, the client calculates the SIR in (10).

Channel estimation at neighboring clusters 1102-1103: If it is determined at 8702 of FIG11 that the estimated SIR in (10) is below a predefined threshold (SIR _T ), then the client starts listening for training signals from the neighboring cluster. Note that SIR _T depends on the modulation and FEC coding scheme (MCS) used for data transmission. Different SIR targets are defined based on the client's MCS. When the DIDO distributed antennas from different clusters are time synchronized (i.e., locked to the same pulse per second (PPS) time reference), the client delivers its channel estimates to the DIDO antennas in the neighboring cluster using a training sequence at 8703. The training sequences used for channel estimation in the neighboring cluster are designed to be orthogonal to the training from the master cluster. Alternatively, when the DIDO antennas in different clusters are not time synchronized, orthogonal sequences (with good cross-correlation properties) are used for time synchronization in the different DIDO clusters. Once the client is locked to the time/frequency reference of the neighboring cluster, channel estimation is performed at 1103.

IDCI Precoding 1104: Once channel estimates are available at the DIDO BTSs in neighboring clusters, IDCI-precoding is calculated to satisfy the condition in (3). The DIDO antennas in the neighboring clusters transmit only the precoded data stream to the clients in their cluster while pre-cancelling interference to the clients in the interference zone 410 in Figure 4. We observe that if a client is located in the Type B interference zone 410 in Figure 4, then the interference to the client is generated by multiple clusters and IDCI-precoding is performed simultaneously by all neighboring clusters.

Handoff method

In the following, we describe different handoff methods for clients moving across a DIDO cluster populated by distributed antennas located in separate areas or providing different types of services (i.e., low or high mobility services).

a. Handoff between adjacent DIDO clusters

In one embodiment, the IDCI precoder used to remove inter-cluster interference described above is used as the baseline for handoff methods in DIDO systems. Conventional handoff in cellular systems is envisioned as a client seamlessly switching across cells served by different base stations. In DIDO systems, handoff allows a client to move from one cluster to another without losing connectivity.

To illustrate one embodiment of a handoff strategy for a DIDO system, let's again consider the example of Figure 8 with only two clusters 802 and 803. When client 801 moves from primary cluster (C1) 802 to neighboring cluster (C2) 803, one embodiment of a handoff method dynamically calculates the signal quality in the different clusters and selects the cluster that yields the lowest error rate performance for the client.

Figure 12 shows the SINR variation as a function of the distance of the client from the center of cluster C1. For 4-QAM modulation without FEC coding, we consider a target SINR = 20dB. The line marked with a circle represents the SINR of the target client served by the DIDO antenna in C1 when both C1 and C2 use DIDO precoding without interference cancellation. The SINR decreases with D due to path loss and interference from neighboring clusters. When IDCI-precoding is implemented at a neighboring cluster, the SINR loss is only due to path loss (as shown by the line with triangles) because the interference is completely removed. When the client is served by a neighboring cluster, a symmetric behavior is experienced. One embodiment of the handover strategy is defined so that when the client moves from C1 to C2, the algorithm switches between different DIDO schemes to keep the SINR above the predefined target.

From the graph in Figure 12, we derive the SER for 4-QAM modulation in Figure 13. We observe that by switching between different precoding strategies, the SER is kept within the predefined target.

One embodiment of a handoff strategy is as follows.

C1-DIDO and C2-DIDO precoding: When the client is located in C1 and away from the interference area, clusters C1 and C2 both work independently with conventional DIDO precoding.

C1-DIDO and C2-IDCI precoding: As a client moves toward an interference zone, its SIR or SINR decreases. When the target SINR _T1 is reached, the target client begins estimating the channel from all DIDO antennas in C2 and provides CSI to the BTS in C2. The BTS in C2 calculates the IDCI precoding and transmits it to all clients in C2 while preventing interference to the target client. As long as the target client remains within the interference zone, it continues to provide its CSI to both C1 and C2.

· C1-IDCI and C2-DIDO precoding: As the client moves towards C2, its SIR or SINR continues to decrease until it reaches the target again. At this point, the client decides to switch to a neighboring cluster. In this case, C1 starts using the CSI from the target client through IDCI-precoding to create zero interference in its direction, while the neighboring cluster uses CSI for conventional DIDO-precoding. In one embodiment, when the SIR estimate approaches the target, clusters C1 and C2 both try both DIDO-precoding schemes and IDCI-precoding schemes alternately to allow the client to estimate the SIR in both cases. The client then selects the best scheme to maximize some error rate performance metric. When this method is applied, the intersection point for the handover strategy appears at the intersection of the curves with triangles and diamonds in Figure 12. One embodiment uses the modified IDCI-precoding method described in (6), where the neighboring cluster also transmits the precoded data stream to the target client to provide array gain. With this method, the handover strategy is simplified because the client does not need to estimate the SINR of the two strategies at the intersection point.

· C1-DIDO and C2-DIDO precoding: When the client moves out of the interference zone towards C2, the main cluster C1 stops pre-cancelling interference towards the target client via IDCI-precoding and switches back to conventional DIDO-precoding for all clients remaining in C1. This final intersection in our handover strategy can be used to avoid unnecessary CSI feedback from the target client to C1, thereby reducing overhead on the feedback channel. In one embodiment, a second target SINR _T2 is defined. When the SINR (or SIR) increases above this target, the strategy is switched to C1-DIDO and C2-DIDO. In one embodiment, cluster C1 keeps alternating between DIDO-precoding and IDCI-precoding to allow the client to estimate the SINR. The client then selects the method for C1 that more closely approaches the target SINR _T1 from above.

The method described above calculates SINR or SIR estimates for different schemes in real time and uses them to select the optimal scheme. In one embodiment, a handoff algorithm is designed based on the finite state machine shown in FIG14. When the SINR or SIR drops below or above the predefined threshold shown in FIG12, the client tracks its current state and switches to the next state. As described above, in state 1201, clusters C1 and C2 both independently operate with conventional DIDO precoding, and the client is served by cluster C1; in state 1202, the client is served by cluster C1, the BTS in C2 calculates IDCI-precoding, and cluster C1 operates with conventional DIDO precoding; in state 1203, the client is served by cluster C2, the BTS in C1 calculates IDCI-precoding, and cluster C2 operates with conventional DIDO precoding; and in state 1204, the client is served by cluster C2, and clusters C1 and C2 both independently operate with conventional DIDO precoding.

In the presence of shadowing, the signal quality or SIR may fluctuate around a threshold as shown in FIG15 , causing iterative switching between the successive states in FIG14 . Iteratively changing states is an undesirable effect because it results in significant overhead on the control channel between the client and the BTS to implement switching between transmission schemes. FIG15 shows an example of a handoff strategy in the presence of shadowing. In one embodiment, the shadowing coefficient is modeled according to a lognormal distribution with variance 3 [3]. In the following, we define some methods to prevent the iterative switching effect during DIDO handoff.

One embodiment of the present invention uses a hysteresis loop to address state switching effects. For example, when switching between the "C1-DIDO, C2-IDCI" 9302 and "C1-IDCI, C2-DIDO" 9303 states in Figure 14 (or vice versa), the threshold SINR _T1 can be adjusted to be within the range A _1. This method avoids repeated switching between states when the signal quality oscillates around SINR _T1 . For example, Figure 16 shows the hysteresis loop mechanism when switching between any two states in Figure 14. In order to switch from state B to state A, the SIR must be greater than (SIR _T1 +A ₁ /2), but in order to switch back from A to B, the SIR must drop below (SIR _T1 -A ₁ /2).

In various embodiments, the threshold SINR _T2 is adjusted to avoid repeated switching between the first and second states (or the third and fourth states) of the finite state machine in Figure 14. For example, a range of values _A2 can be defined such that the threshold SINR _T2 is selected within the range according to channel conditions and shadowing effects.

In one embodiment, the SINR threshold is dynamically adjusted within the range [SINR _T2 , SINR _T2 + A ₂ ] based on the variance of expected shadowing on the wireless link. The variance of the log-normal distribution can be estimated based on the variance of the received signal strength (or RSSI) as the client moves from its current cluster to a neighboring cluster.

The above method assumes that the client triggers the handoff strategy.In one embodiment, assuming that communication across multiple BTSs is enabled, the handoff decision is deferred to the DIDO BTS.

For simplicity, the above method is derived assuming no FEC coding and 4-QAM. More generally, SINR or SIR thresholds are derived for different modulation and coding schemes (MCSs), and handoff strategies are designed in conjunction with link adaptation (see, e.g., U.S. Patent No. 7,636,381) to optimize the downlink data rate for each client in the interference zone.

b. Handoff between Low-Doppler DIDO Network and High-Doppler DIDO Network

The DIDO system uses a closed-loop transmission scheme to precode the data stream on the downlink channel. The closed-loop scheme is inherently constrained by the delay on the feedback channel. In an actual DIDO system, when delivering CSI and baseband precoding data from the BTS to the distributed antenna, the calculation time can be shortened by a transceiver with high processing power, and most of the delay is expected to be introduced by the DIDO BSN. The BSN may include various network technologies, including but not limited to digital subscriber lines (DSL), cable modems, fiber rings, T1 lines, hybrid fiber-coaxial (HFC) networks, and/or fixed wireless (e.g., WiFi). Dedicated optical fiber typically has very large bandwidth and low latency, which may be less than 1 millisecond in local areas, but its deployment range is not as wide as DSL and cable modems. Today, DSL and cable modem connections in the United States typically have last-mile delays between 10-25ms, but their deployment is very widespread.

The maximum delay at the BSN determines the maximum Doppler frequency that can be tolerated on the DIDO wireless link without degrading DIDO precoding performance. For example, in [1], we showed that at a carrier frequency of 400 MHz, a network with a delay of about 10 milliseconds (i.e., DSL) can tolerate client speeds of up to 8 mph (running speed), while a network with a delay of 1 millisecond (i.e., a fiber ring) can support speeds of up to 70 mph (i.e., highway traffic).

We define two or more DIDO subnetworks based on the maximum Doppler frequency that can be tolerated on the BSN. For example, a BSN with high-latency DSL connections between the DIDO BTS and distributed antennas can only provide low-mobility or fixed wireless services (i.e., a low-Doppler network), while a low-latency BSN on a low-latency fiber ring can tolerate high mobility (i.e., a high-Doppler network). We observe that most broadband users are immobile when using broadband, and furthermore, most are unlikely to be located near areas with many high-speed moving objects (e.g., near highways) because such locations are generally less desirable residential or business locations. However, there are also broadband users who use broadband at high speeds (e.g., while in a car traveling on a highway) or near high-speed objects (e.g., in a store located near a highway). To address these two different user Doppler scenarios, in one embodiment, a low-Doppler DIDO network consists of a typically large number of DIDO antennas with relatively low power (i.e., 1W to 100W for indoor or rooftop installations) spread over a wide area, while a high-Doppler network consists of a typically smaller number of DIDO antennas transmitting at high power (i.e., 100W for rooftop or tower installations). A low-Doppler DIDO network serves a typically larger number of low-Doppler users and can use inexpensive high-latency broadband connections (such as DSL and cable modems) with a typically lower connection cost. A high-Doppler DIDO network serves a typically smaller number of high-Doppler users and can use more expensive low-latency broadband connections (such as fiber) with a typically higher connection cost.

To avoid interference between different types of DIDO networks (e.g., low-Doppler and high-Doppler), different multiple access technologies can be used, such as time division multiple access (TDMA), frequency division multiple access (FDMA), or code division multiple access (CDMA).

In the following, we propose a method to assign clients to different types of DIDO networks and allow handoff between them. Network selection is based on the mobility type of each client. The client's velocity (v) is proportional to the maximum Doppler shift according to the following equation [6],

where _fd is the maximum Doppler shift, λ is the wavelength corresponding to the carrier frequency, and θ is the angle between the vector indicating the direction transmitter-client and the velocity vector.

In one embodiment, the Doppler shift of each client is calculated by a blind estimation technique. For example, similar to a Doppler radar system, the Doppler shift can be estimated by sending RF energy to the client and analyzing the reflected signal.

In another embodiment, one or more DIDO antennas send training signals to the client. Based on those training signals, the client estimates the Doppler shift using techniques such as counting the zero crossing rate of the channel gain or performing spectral analysis. We observe that for a fixed velocity v and the trajectory of the client, the angular velocity in vsinθ(11) can depend on the relative distance of the client from each DIDO antenna. For example, a DIDO antenna close to a mobile client produces a larger angular velocity and Doppler shift than an antenna farther away. In one embodiment, the Doppler velocity is estimated by multiple DIDO antennas at different distances from the client, and the average, weighted average, or standard deviation is used as an indicator of the client's mobility. Based on the estimated Doppler indicator, the DIDO BTS decides whether to assign the client to a low-Doppler network or a high-Doppler network.

Doppler indicators are periodically monitored for all clients and sent back to the BTS. When one or more clients change their Doppler speed (i.e., a client riding a bus versus a client walking or sitting), those clients are dynamically reallocated to a different DIDO network that can accommodate their level of mobility.

Although the Doppler of a low-speed client may be affected by being near high-speed objects (such as near a highway), the Doppler is generally much lower than the Doppler of a client that is itself in motion. Therefore, in one embodiment, the client's speed is estimated (e.g., by using methods such as monitoring the client's position using GPS), and if the speed is low, the client is assigned to a low-Doppler network, and if the speed is high, the client is assigned to a high-Doppler network.

Methods for power control and antenna grouping

Figure 17 shows a block diagram of a DIDO system with power control. First, one or more data streams ( _sk ) for each client (1, ..., U) are multiplied by weights generated by the DIDO precoding unit. The precoded data streams are multiplied by a power scaling factor calculated by the power control unit based on input channel quality information (CQI). The CQI is fed back by the client to the DIDO BTS or derived from the uplink channel assuming uplink-downlink channel reciprocity. The U precoded streams from different clients are then combined and multiplexed into M data streams ( _tm ), one for each of the M transmit antennas. Finally, the streams _tm are sent to the digital-to-analog converter (DAC) unit, the radio frequency (RF) unit, the power amplifier (PA) unit, and finally to the antenna.

The power control unit measures the CQI for all clients. In one embodiment, the CQI is the average SNR or RSSI. Depending on path loss or shadowing, the CQI may vary for different clients. Our power control method adjusts the transmit power scaling factors P _k for each client and multiplies them by the precoded data streams generated for each client. Note that one or more data streams can be generated for each client, depending on the number of receive antennas on the client.

To evaluate the performance of the proposed method, we define the following signal model including path loss and power control parameters based on (5):

Where k = 1, ..., U, U is the number of clients, SNR = P _o /N _o , where P _o is the average transmit power, N _o is the noise power, and α _k is the path loss/shadowing factor. To model path loss/shadowing, we use the following simplified model

where a=4 is the path loss exponent, and we assume that the path loss increases with client index (ie, clients are located at increasing distances from the DIDO antennas).

Figure 18 shows the SER vs. SNR for different scenarios assuming four DIDO transmit antennas and four clients. The ideal case assumes that all clients have the same path loss (i.e., a = 0), resulting in P _k = 1 for all clients. The curve with squares refers to the case where clients have different path loss coefficients and there is no power control. The curve with dots is based on the same case (with path loss) with the power control coefficient chosen so that P _k = 1/α _k . With the power control method, more power is allocated to the data stream intended for clients with higher path loss/shadowing, resulting in a 9dB SNR gain (for this particular case) compared to the case without power control.

The Federal Communications Commission (FCC) of the United States (and other international regulatory bodies) define constraints on the maximum power that can be transmitted from wireless devices to limit human exposure to electromagnetic (EM) radiation. There are two types of limits [2]: i) "occupational/controlled" limits, where people are fully aware of radio frequency (RF) sources through fencing, warnings, or signage; and ii) "general population/uncontrolled" limits, where there are no controls on exposure.

Different emission classes are defined for different types of wireless devices. Generally speaking, DIDO distributed antennas for indoor/outdoor applications meet the requirements of the FCC "mobile" device category, which is defined as [2]:

“Transmitting equipment designed for use in locations other than fixed locations, normally where the radiating structure is maintained at a distance of 20 cm or more from the body of the user or nearby persons.”

EM emissions from "mobile" devices are measured in terms of Maximum Permissible Exposure (MPE) expressed in mW/ ^cm² . Figure 19 shows the MPE power density as a function of distance from the RF radiation source for various values of transmit power at a 700 MHz carrier frequency. The maximum permissible transmit power required to meet the FCC's "unregulated" limits for devices typically operating at distances greater than 20 cm from the human body is 1 W.

Less restrictive power emission limits are defined for transmitters mounted on rooftops or buildings away from the “general public.” For these “rooftop transmitters,” the FCC defines a looser emission limit of 1000W measured in terms of effective radiated power (ERP).

Based on the above FCC constraints, in one embodiment, we define two types of DIDO distributed antennas for practical systems:

Low-power (LP) transmitter: Located anywhere at any altitude (i.e., indoors or outdoors) with a maximum transmit power of 1W and a 5Mbps consumer-grade broadband (e.g., DSL, cable modem, fiber-to-the-home (FTTH)) backhaul connection.

High-power (HP) transmitters: Rooftop or building-mounted antennas at approximately 10 meters in height, with 100W of transmit power and commercial-grade broadband (e.g., fiber ring) backhaul (with effectively “unlimited” data rates compared to the throughput available over the DIDO wireless link).

Note that LP transmitters connected using DSL or cable modems are good candidates for low-Doppler DIDO networks (as described in the previous section) because their clients are mostly stationary or have low mobility. HP transmitters connected using commercial fiber can allow for higher client mobility and can be used for high-Doppler DIDO networks.

To get a realistic sense of the performance of a DIDO system with different types of LP/HP transmitters, we consider a real-world DIDO antenna installation in downtown Palo Alto, CA. Figure 20a shows a random distribution of N _LP = 100 low-power DIDO distributed antennas in Palo Alto. In Figure 20b, the 50 LP antennas are replaced by N _HP = 50 high-power transmitters.

Based on the DIDO antenna distribution in Figures 20a-20b, we derive coverage maps for a system using DIDO technology in Palo Alto. Figures 21a and 21b show two power distributions corresponding to the configurations in Figures 20a and 20b, respectively. The received power distribution (in dBm) is derived assuming the path loss/shadowing model defined by the 3GPP standard [3] for urban environments at a 700 MHz carrier frequency. We observe that using 50% HP transmitters results in better coverage for the selected area.

Figures 22a-22b show the rate distribution for the two scenarios above. The throughput (in Mbps) is derived based on the power thresholds for different modulation and coding schemes defined in the 3GPP Long Term Evolution (LTE) standard in [4,5]. At a 700 MHz carrier frequency, the total available bandwidth is fixed to 10 MHz. Two different frequency allocation plans are considered: i) only 5 MHz of spectrum is allocated to LP stations; ii) 9 MHz is allocated to HP transmitters and 1 MHz is allocated to LP transmitters. Note that the lower bandwidth is generally allocated to LP stations due to their DSL backhaul connections with limited throughput. Figures 22a-22b show that the rate distribution can be significantly improved when 50% of the HP transmitters are used, increasing the average per-client data rate from 2.4 Mbps in Figure 22a to 38 Mbps in Figure 22b.

We then define an algorithm to control the power transmission of the LP stations such that higher power is allowed at any given time, thereby increasing the throughput on the downlink channel of the DIDO system in Figure 22b. We observe that the FCC limit on power density is defined based on time averaging as [2]

Where is the MPE averaging time and _tn is the time period of exposure to radiation with power density _Sn . For "controlled" exposure, the averaging time is 6 minutes, while for "uncontrolled" exposure, it increases to a maximum of 30 minutes. Then, any power source is allowed to transmit at a power level greater than the MPE limit, as long as the average power density in (14) meets the FCC limit of 30 minutes for "uncontrolled" exposure.

Based on this analysis, we define an adaptive power control method to increase the instantaneous per-antenna transmit power while keeping the average power per DIDO antenna below the MPE limit. We consider a DIDO system with more transmit antennas than active clients. This is a reasonable assumption considering that DIDO antennas can be conceived as inexpensive wireless devices (similar to WiFi access points) and can be placed anywhere there is a DSL, cable modem, fiber, or other Internet connection.

Figure 23 shows the framework of a DIDO system with adaptive per-antenna power control. Before the digital signal generated by the multiplexer 234 is sent to the DAC unit 235, its amplitude is dynamically adjusted using power scaling factors S ₁ ,…, _SM . The power scaling factors are calculated by the power control unit 232 based on the CQI 233.

In one embodiment, N _g DIDO antenna groups are defined. Each group contains at least as many DIDO antennas as the number of active clients (K). At any given time, only one group has _Na > K active DIDO antennas transmitting to clients at a power level (So) greater than the MPE limit. A method according to the round-robin scheduling strategy shown in Figure 24 is repeated across all antenna groups. In another embodiment, a different scheduling technique (i.e., proportional fair scheduling [8]) is used for cluster selection to optimize error rate or throughput performance.

Assuming cyclic power allocation, we derive the average transmit power of each DIDO antenna from (14) as

where _t is the time period during which the antenna group is active, and T _MPE = 30 min is the averaging time defined by FCC guidelines [2]. The ratio in (15) is the duty cycle (DF) of the group, which is defined such that the average transmit power from each DIDO antenna meets the MPE limit. The duty cycle depends on the number of active clients, the number of groups, and the number of active antennas per group according to the following definition:

The SNR gain (in dB) achieved in a DIDO system with power control and antenna grouping is expressed as a function of duty cycle as follows

We observe that the gain in (17) is achieved at the cost of G _dB additional transmit power across all DIDO antennas.

In general, the total transmit power of all _Na from all _Ng groups is defined as

Where _Pij is the average daily antenna transmit power, which is given by the following formula

and S _ij (t) is the power spectral density of the i-th transmit antenna in the j-th group. In one embodiment, the power spectral density in (19) is designed for each antenna to optimize error rate or throughput performance.

To get some intuition about the performance of the proposed approach, consider 400 DIDO distributed antennas in a given coverage area and 400 clients subscribing to wireless internet service provided via the DIDO system. It is unlikely that every internet connection will be fully utilized at all times. We assume that 10% of the clients will actively use the wireless internet connection at any given time. The 400 DIDO antennas can then be divided into _Ng = 10 groups, each with _Na = 40 antennas, with each group serving K = 40 active clients at any given time with a duty cycle DF = 0.1. The resulting SNR gain from this transmission scheme is _GdB = _10log10 (1/DF) = 10dB, provided by the 10dB additional transmit power from all DIDO antennas. However, we observe that the average per-antenna transmit power is constant and within the MPE limit.

Figure 25 compares the (uncoded) SER performance of the above power control with antenna grouping with conventional eigenmode selection in U.S. Patent No. 7,636,381. All schemes use BD precoding with four clients, each equipped with a single antenna. SNR refers to the ratio of power per transmit antenna to noise power (i.e., transmit SNR per antenna). The curves represented by DIDO 4×4 assume four transmit antennas and BD precoding. The curve with squares represents the SER performance with two additional transmit antennas and BD with eigenmode selection, resulting in a 10dB SNR gain relative to conventional BD precoding (at a 1% SER target). Power control with antenna grouping and DF=1/10 also produces a 10dB gain at the same SER target. We observe that eigenmode selection changes the slope of the SER curve due to diversity gain, while our power control method shifts the SER curve to the left (maintaining the same slope) due to the increased average transmit power. For comparison, the SER with a larger duty cycle DF=1/50 is shown, providing an additional 7dB gain compared to DF=1/10.

Note that our power control can be less complex than conventional eigenmode selection methods. In fact, the antenna IDs for each group can be pre-computed and shared between the DIDO antennas and clients via a lookup table, requiring only K channel estimates at any given time. For eigenmode selection, (K+2) channel estimates are computed, and additional computational processing is required to select the eigenmode that minimizes the SER for all clients at any given time.

We then describe another approach involving DIDO antenna grouping to reduce CSI feedback overhead in some special cases. Figure 26a shows a scenario where clients (points) are randomly scattered in an area covered by multiple DIDO distributed antennas (crosses). The average power on each transmit-receive radio link can be calculated as

A＝{|H| ² }. (20)

where H is the channel estimation matrix available at the DIDO BTS.

The matrix A in Figures 26a-26c was numerically obtained by averaging the channel matrix over 1000 instances. Figures 26b and 26c depict two alternative scenarios, respectively, where clients are grouped together around a subset of DIDO antennas and the clients receive negligible power from distant DIDO antennas. For example, Figure 26b shows two sets of antennas that produce a block diagonal matrix A. An extreme case is when each client is very close to only one transmitter and the transmitters are far enough away from each other that the power from all other DIDO antennas is negligible. In this case, the DIDO link degenerates into multiple SISO links and A is a diagonal matrix as in Figure 26c.

In all three cases described above, BD precoding dynamically adjusts the precoding weights to account for the varying power levels on the wireless link between the DIDO antennas and the clients. However, it is convenient to identify multiple groups within a DIDO cluster and operate DIDO precoding only within each group. Our proposed grouping approach yields the following advantages:

Computational Gain: DIDO precoding is computed only within each group in the cluster. For example, if BD precoding is used, the singular value decomposition (SVD) has complexity O( ⁿ³ ), where n is the minimum dimension of the channel matrix H. If H can be reduced to a block diagonal matrix, the SVD of each block is computed with reduced ^complexity . In fact, if the channel matrix is split into two block matrices with dimensions _n1 and _n2 , such that n = n1 _{+ n2} , the complexity of SVD is only O( _n13 ) + O( _n23 ) < O( ⁿ³ ). In the extreme case, if H is a diagonal matrix, the DIDO link is reduced to multiple SISO links ^and no SVD computation is required.

Reduced CSI feedback overhead: When DIDO antennas and clients are grouped, in one embodiment, CSI is computed from clients to antennas only within the same group. In TDD systems, assuming channel reciprocity, antenna grouping reduces the number of channel estimates used to compute the channel matrix H. In FDD systems, where CSI is fed back over the wireless link, antenna grouping further reduces CSI feedback overhead over the wireless link between DIDO antennas and clients.

Multiple access technology for DIDO uplink channels

In one embodiment of the present invention, different multiple access techniques are defined for the DIDO uplink channel. These techniques can be used to feed back CSI or transmit data streams from the client to the DIDO antennas on the uplink. Hereinafter, we refer to the fed back CSI and data streams as uplink streams.

Multiple Input Multiple Output (MIMO): The uplink streams are transmitted from the clients to the DIDO antennas via an open-loop MIMO multiplexing scheme. This approach assumes that all clients are time/frequency synchronized. In one embodiment, synchronization between the clients is achieved via training from the downlink and all DIDO antennas are assumed to be locked to the same time/frequency reference clock. Note that variations in delay spread at different clients can generate jitter between the clocks of different clients, which can affect the performance of the MIMO uplink scheme. After the client sends the uplink streams via the MIMO multiplexing scheme, the receiving DIDO antennas can use nonlinear (i.e., maximum likelihood, ML) or linear (i.e., zero-forcing minimum mean square error) receivers to cancel co-channel interference and individually demodulate the uplink streams.

Time Division Multiple Access (TDMA): Different clients are assigned to different time slots. Each client sends its uplink stream when its time slot is available.

Frequency Division Multiple Access (FDMA): Assigns different clients to different carrier frequencies. In multi-carrier (OFDM) systems, subsets of tones are assigned to different clients transmitting uplink streams simultaneously, thus reducing latency.

Code Division Multiple Access (CDMA): Assigns each client a different pseudo-random sequence and achieves orthogonality across clients in the code domain.

In one embodiment of the present invention, clients are wireless devices that transmit at significantly lower power than the DIDO antennas. In this case, the DIDO BTS defines client subsets based on uplink SNR information to minimize interference across the subsets. Within each subset, the aforementioned multiple access techniques are used to create orthogonal channels in the time, frequency, spatial, or code domains, thereby avoiding uplink interference across different clients.

In another embodiment, the uplink multiple access technique described above is used in conjunction with the antenna grouping method proposed in the previous section to define different client groups within a DIDO cluster.

Systems and methods for link adaptation in DIDO multi-carrier systems

A link adaptation method for a DIDO system that exploits the time, frequency, and spatial selectivity of a wireless channel is defined in U.S. Patent No. 7,636,381. The following describes embodiments of the present invention for link adaptation in a multi-carrier (OFDM) DIDO system that exploits the time/frequency selectivity of a wireless channel.

We model the Rayleigh fading channel according to the exponentially decaying power delay profile (PDP) or Saleh-Valenzuela model in [9]. For simplicity, we assume a single cluster channel with multipath PDP defined as

P _n =e ^-βn (21)

where n = 0, ..., L-1 is the index of the channel tap, L is the number of channel taps, and β = 1/σ _DS is the PDP index, which is an indicator of the channel coherence bandwidth and is inversely proportional to the channel delay spread (σ _DS ). Low values of β produce frequency-flat channels, while high values of β produce frequency-selective channels. The PDP in (21) is normalized so that the total average power of all L channel taps is unity

Figure 27 shows the amplitude of a low-frequency selective channel (assuming β = 1) for a DIDO 2×2 system in the delay domain or instantaneous PDP (upper curve) and the frequency domain (lower curve). The first subscript indicates the client and the second subscript indicates the transmit antenna. A high-frequency selective channel (where β = 0.1) is shown in Figure 28.

Next, we study the performance of DIDO precoding in frequency selective channels. Assuming that the signal model in (1) satisfies the conditions in (2), we calculate the DIDO precoding weights via BD. We reformulate the DIDO received signal model in (5) using the conditions in (2) as

r _k =H _ek s _k +n _k · (23)

where _Hek = _{Hk Wk} is the effective channel matrix for user k. For DIDO 2×2 with a single antenna per client, the effective channel matrix reduces to a value with the frequency response shown in FIG29 and for a channel characterized by high frequency selectivity (e.g., with β = 0.1) in FIG28. The solid line in FIG29 refers to client 1, while the dotted line refers to client 2. Based on the channel quality metrics in FIG29, we define a time/frequency domain link adaptation (LA) method that dynamically adjusts the MCS according to changing channel conditions.

We begin by evaluating the performance of different MCSs in AWGN and Rayleigh fading SISO channels. For simplicity, we assume no FEC coding, but the following LA method can be extended to systems including FEC.

FIG30 shows the SER for different QAM schemes (i.e., 4-QAM, 16-QAM, 64-QAM). Without loss of generality, we assume a target SER of 1% for the uncoded system. The SNR thresholds used to meet the target SER in an AWGN channel are 8 dB, 15.5 dB, and 22 dB for the three modulation schemes, respectively. In Rayleigh fading channels, it is well known that the SER performance of the above modulation schemes is worse than that of AWGN [13], and the SNR thresholds are 18.6 dB, 27.3 dB, and 34.1 dB, respectively. We observe that DIDO precoding transforms the multi-user downlink channel into a set of parallel SISO links. Therefore, on a per-client basis, the same SNR thresholds used for the SISO system as in FIG30 apply to the DIDO system. Furthermore, if instantaneous LA is performed, the thresholds in the AWGN channel are used.

The key idea of the proposed LA method for DIDO systems is to use a low MCS order to provide link robustness when the channel experiences deep fading in the time or frequency domain (shown in FIG28 ). Conversely, when the channel is characterized by large gain, the LA method switches to a higher MCS order to increase spectral efficiency. One contribution of this patent application compared to U.S. Patent No. 7,636,381 is the use of (23) and the effective channel matrix in FIG29 as a metric to allow for adaptation.

The overall framework of the LA method is shown in Figure 31 and defined as follows:

• CSI estimation: The DIDO BTS calculates the CSI from all users at 3171. Users can be equipped with single or multiple receive antennas.

DIDO Precoding: At 3172, the BTS calculates the DIDO precoding weights for all users. In one embodiment, BD is used to calculate these weights. The precoding weights are calculated on a tone-by-tone basis.

· Link quality metric calculation: At 3173, the BTS calculates the frequency domain link quality metric. In an OFDM system, this metric is calculated based on the CSI and the DIDO precoding weights for each tone. In one embodiment of the present invention, the link quality metric is the average SNR over all OFDM tones. We define this method as LA1 (based on average SNR performance). In another embodiment, the link quality metric is the frequency response of the effective channel in (23). We define this method as LA2 (based on per-tone performance to exploit frequency diversity). If each client has a single antenna, the frequency domain effective channel is shown in Figure 29. If the client has multiple receive antennas, the link quality metric is defined as the Frobenius norm of the effective channel matrix for each tone. Alternatively, multiple link quality metrics are defined for each client as the singular values of the effective channel matrix in (23).

Bitloading Algorithm: Based on the link quality metric, the BTS determines the MCS for different clients and different OFDM tones at 3174. For the LA1 method, the same MCS is used for all clients and all OFDM tones based on the SNR threshold for the Rayleigh fading channel in Figure 30. For LA2, different MCSs are assigned to different OFDM tones to exploit channel frequency diversity.

· Precoded data transmission: At 3175, the BTS transmits the precoded data stream from the DIDO distributed antenna to the client using the MCS derived by the bit loading algorithm. A header is attached to the precoded data to convey the MCS for the different tones to the client. For example, if eight MCSs are available and the OFDM symbol is defined with N=64 tones, then _log2 (8)*N=192 bits are required to convey the current MCS to each client. Assuming that those bits are mapped into symbols with 4-QAM (2 bits/symbol spectral efficiency), only 192/2/N=1.5 OFDM symbols are required to map the MCS information. In another embodiment, multiple subcarriers (or OFDM tones) are grouped into subbands and the same MCS is assigned to all tones in the same subband to reduce the overhead due to control information. In addition, the MCS is adjusted based on the time variation of the channel gain (proportional to the coherence time). In fixed wireless channels (characterized by low Doppler effects), the MCS is recalculated every fraction of the channel coherence time, thereby reducing the overhead required for control information.

Figure 32 shows the SER performance of the LA method described above. For comparison, the SER performance in a Rayleigh fading channel is plotted for each of the three QAM schemes used. The LA2 method adapts the MCS to the fluctuations of the effective channel in the frequency domain, providing a 1.8 bps/Hz gain in spectral efficiency for low SNRs (i.e., SNR = 20 dB) and a 15 dB gain in SNR (for SNR > 35 dB) compared to LA1.

Systems and methods for DIDO precoding interpolation in multi-carrier systems

The computational complexity of a DIDO system is primarily confined to the centralized processor or BTS. The most computationally expensive operation is the calculation of precoding weights for all clients based on their CSI. When using BD precoding, the BTS must perform as many singular value decomposition (SVD) operations as there are clients in the system. One way to reduce complexity is through parallel processing, where the SVD is calculated on a separate processor for each client.

In a multi-carrier DIDO system, each subcarrier experiences a flat fading channel, and SVD is performed on each subcarrier for each client. Obviously, the complexity of the system increases linearly with the number of subcarriers. For example, in an OFDM system with a 1 MHz signal bandwidth, the cyclic prefix (L ₀ ) must have at least eight channel taps (i.e., a duration of 8 microseconds) to avoid inter-symbol interference in an outdoor urban macrocell environment with large delay spread [3]. The size of the fast Fourier transform (FFT) used to generate OFDM symbols ( _NFFT ) is typically set to a multiple of _L0 to reduce the loss of data rate. If _NFFT = 64, the effective spectral efficiency of the system is limited by the factor _NFFT / ( _NFFT + L ₀ ) = 89%. Larger values of _NFFT produce higher spectral efficiency at the expense of higher computational complexity at the DIDO precoder.

One way to reduce the computational complexity at the DIDO precoder is to perform an SVD operation on a subset of tones (which we call pilot tones) and derive the precoding weights for the remaining tones via interpolation. Weight interpolation is a source of error that leads to inter-client interference. In one embodiment, an optimal weight interpolation technique is used to reduce inter-client interference, resulting in improved error rate performance and lower computational complexity in multi-carrier systems. In a DIDO system with M transmit antennas, U clients, and N receive antennas per client, the condition for the precoding weight ( _Wk ) of the kth client to guarantee zero interference to other clients u is derived from (2) as

where _Hu is the channel matrix corresponding to other DIDO clients in the system.

In one embodiment of the present invention, the objective function of the weight interpolation method is defined as

where _θk is the set of parameters to be optimized for user k, is the weight interpolation matrix and ||·|| _F represents the Frobenius norm of the matrix. The optimization problem is formulated as

Where _Θk is the feasible set of the optimization problem and θk _,opt is the optimal solution.

The objective function in (25) is defined for one OFDM tone. In another embodiment of the present invention, the objective function is defined as a linear combination of the Frobenius norms in (25) of the matrices of all OFDM tones to be interpolated. In another embodiment, the OFDM spectrum is divided into subsets of tones and the optimal solution is given by

where n is the OFDM tone index and A is the subset of tones.

The weight interpolation matrix _Wk ( _θk ) in (25) is expressed as a function of the set of parameters _θk . Once the best set is determined according to (26) or (27), the optimal weight matrix can be calculated. In one embodiment of the present invention, the weight interpolation matrix for a given OFDM tone n is defined as a linear combination of the weight matrices of the pilot tones. An example of a weight interpolation function for a beamforming system with a single client is defined in [11]. In a DIDO multi-client system, we write the weight interpolation matrix as

where 0≤l≤( _L0-1 ), _L0 is the number of pilot tones and _cn =(n-1)/ _N0 , where _N0 = _NFFT / _L0 . The weight matrix in (28) is then normalized so that to ensure uniform power transmission from each antenna. If N=1 (single receive antenna per client), the matrix in (28) becomes a vector normalized with respect to its norm. In one embodiment of the present invention, the pilot tones are picked uniformly over the range of OFDM tones. In another embodiment, the pilot tones are picked adaptively based on the CSI to minimize interpolation error.

We observe that a key difference between the system and method in [11] and the system and method proposed in this patent application is the objective function. Specifically, the system in [11] assumes multiple transmit antennas and a single client, and thus the related method is designed to maximize the product of the precoding weight and the channel to maximize the client's received SNR. However, this method does not work in the multi-client scenario because it generates inter-client interference due to interpolation errors. In contrast, our method is designed to minimize inter-client interference, thereby improving the error rate performance for all clients.

FIG33 shows how the entries of the matrix in (28) vary with the OFDM tone index for a DIDO 2×2 system with _NFFT = 64 and _L0 = 8. The channel PDP is generated according to the model in (21) with β = 1, and the channel consists of only eight channel taps. We observe that _L0 must be chosen to be larger than the number of channel taps. The solid line in FIG33 represents the ideal function, while the dashed line is the interpolated function. According to the definition in (28), for the pilot tone, the interpolated weights match the ideal function. The weights computed on the remaining tones only approximate the ideal case due to estimation error.

One way to implement the weight interpolation method is via an exhaustive search of the feasible set _Θk in (26). To reduce the complexity of the search, we quantize the feasible set to P values uniformly in the range [0, 2π]. Figure 34 shows the SER versus SNR for _L0 = 8, M = _Nt = 2 transmit antennas, and a variable number of P. As the number of quantization levels increases, the SER performance improves. We observe that the case of P = 10 approaches the performance of P = 100, due to the much lower computational complexity due to the reduced number of searches.

Figure 35 shows the SER performance for different DIDO orders and the interpolation method with L ₀ = 16. We assume that the number of clients is the same as the number of transmit antennas and each client is equipped with a single antenna. When the number of clients increases, the SER performance decreases due to the increase in inter-client interference caused by weight interpolation errors.

In another embodiment of the present invention, weight interpolation functions different from those in (28) are used. For example, a linear predictive autoregressive model [12] can be used to interpolate weights across different OFDM tones based on an estimate of the channel frequency correlation.

References

[1] A. Forenza and S. G. Perlman, “System and method for distributed antenna wireless communications,” U.S. patent application Ser. No. 12/630,627, filed Dec. 2, 2009, entitled “System and Method For Distributed Antenna Wireless Communications.”

[2]FCC, “Evaluating compliance with FCC guidelines for human exposure to radiofrequency electromagnetic fields,” OET Bulletin 65, Ed. 97-01, Aug. 1997

[3] 3GPP, “Spatial Channel Model AHG (Combined ad-hoc from 3GPP&3GPP2)”, SCM Text V6.0, April 22, 2003

[4] 3GPP TR 25.912, “Feasibility Study for Evolved UTRA and UTRAN”, V9.0.0 (2009-10)

[5] 3GPP TR 25.913, “Requirements for Evolved UTRA (E-UTRA) and Evolved UTRAN (E-UTRAN)”, V8.0.0 (2009-01)

[6]W.C.Jakes, Microwave Mobile Communications, IEEE Press, 1974

[7] K.K.Wong, et al., “A joint channel diagonalization for multiuser MIMO antenna systems,” IEEE Trans. Wireless Comm., vol. 2, pp. 773-786, July 2003.

[8] P.Viswanath, et al., “Opportunistic beamforming using dump antennas,” IEEE Trans.On Inform.Theory, vol.48, pp.1277–1294, June 2002.

[9]A.A.M.Saleh,et al.,“A statistical model for indoor multipath propagation,”IEEE Jour.Select.Areas in Comm.,vol.195SAC-5,no.2,pp.128–137,Feb.1987.

[10]A.Paulraj,et al.,Introduction to Space-Time Wireless Communications,Cambridge University Press,40 West 20th Street,New York,NY,USA,2003(A.Paulraj et al.,Introduction to Space-Time Wireless Communications,Cambridge University Press,40 West 20th Street,New York,NY,USA,2003).

[11] J. Choi, et al., “Interpolation Based Transmit Beamforming for MIMO-OFDM with Limited Feedback,” IEEE Trans. on Signal Processing, vol. 53, no. 11, pp. 4125-4135, Nov. 2005.

[12] I.Wong, et al., “Long Range Channel Prediction for Adaptive OFDM Systems,” Proc. of the IEEE Asilomar Conf. on Signals, Systems, and Computers, vol. 1, pp. 723-736, Pacific Grove, CA, USA, Nov. 7-10, 2004.

[13] J.G.Proakis, Communication System Engineering, Prentice Hall, 1994

[14] B.D. Van Veen, et al., “Beamforming: a versatile approach to spatial filtering,” IEEE ASSP Magazine, Apr. 1988.

[15] R.G.Vaughan, “On optimum combining at the mobile,” IEEE Trans.On Vehic.Tech., vol. 37, n. 4, pp. 181-188, Nov. 1988

[16] F. Qian, “Partially adaptive beamforming for correlated interference rejection,” IEEE Trans. On Sign. Proc., vol. 43, n. 2, pp. 506-515, Feb. 1995

[17] H. Krim, et al., “Two decades of array signal processing research,” IEEE Signal Proc. Magazine, pp. 67-94, July 1996

[19]W.R.Remley, “Digital beamforming system”, US Patent N.4,003,016, Jan. 1977

[18] R.J.Masak, “Beamforming/null-steering adaptive array”, US Patent No. 4,771,289, Sep. 1988

[20] K.-B.Yu, et al., “Adaptive digital Beamforming architecture and algorithm for nulling mainlobe and multiple sidelobe radar jammers while preserving monopulse ratio angle estimation accuracy”, US Patent 5,600,326, Feb. 1997

[21] H. Boche, et al., “Analysis of different precoding/decoding strategies for multiuser Beamforming”, IEEE Vehic. Tech. Conf., vol. 1, Apr. 2003

[22] M. Schubert, et al., “Joint’dirty paper’ pre-coding and downlink Beamforming” vol. 2, pp. 536-540, Dec. 2002 (M. Schubert et al., “Joint’dirty paper’ pre-coding and downlink beamforming”, vol. 2, pp. 536-540, Dec. 2002)

[23] H. Boche, et al. “A general duality theory for uplink and downlink Beamforming”, vol. 1, pp. 87-91, Dec. 2002

[24] K.K.Wong, R.D.Murch, and K.B.Letaief, “A joint channel diagonalization for multiuser MIMO antenna systems,” IEEE Trans.Wireless Comm., vol. 2, pp. 773–786, Jul 2003.

[25] Q.H.Spencer, A.L.Swindlehurst, and M.Haardt, “Zero forcing methods for downlink spatial multiplexing in multiuser MIMO channels,” IEEE Trans.Sig.Proc., vol.52, pp.461–471, Feb.2004.

II. Disclosure from Related Patent Application Serial No. 12/917,257

The following describes a wireless radio frequency (RF) communication system and method using multiple distributed transmit antennas that operate cooperatively to create a wireless link to a given user while suppressing interference to other users. Coordination across different transmit antennas is enabled through user clustering. A user cluster is a subset of transmit antennas whose signals can be reliably detected by a given user (i.e., the received signal strength is above the noise or interference level). Each user in the system defines its own user cluster (user-cluter). The waveforms transmitted by the transmit antennas within the same user cluster coherently combine to create RF energy at the location of the target user and create a zero RF interference point at the location of any other users that can be reached by those antennas.

Consider a system with M transmit antennas and K users reachable by those M antennas within a user cluster, where K ≤ M. We assume that the transmitter knows the CSI between the M transmit antennas and the K users (H∈C ^K×M ). For simplicity, each user is assumed to be equipped with a single antenna, but the same approach can be extended to multiple receive antennas per user. Consider the following channel matrix H obtained by combining the channel vectors from the M transmit antennas to the K users (h _k ∈C ^1×M )

Calculate the precoding weights (w _k ∈ ^{C M×1} ) that create RF energy to user k and zero RF energy to all other K-1 users, such that the following conditions are satisfied:

where is the effective channel matrix for user k obtained by removing the kth row of matrix H, and 0 ^K×1 is a vector with all zero entries.

In one embodiment, the wireless system is a DIDO system and uses user clustering to create a wireless communication link to a target user while preemptively eliminating interference to any other users reachable by antennas within the user cluster. In U.S. patent application Ser. No. 12/630,627, a DIDO system is described that includes:

DIDO client: A user terminal equipped with one or more antennas;

DIDO distributed antennas: base transceiver stations that operate cooperatively to transmit precoded data streams to multiple users, thereby suppressing inter-user interference;

DIDO Base Transceiver Station (BTS): A centralized processor that generates precoded waveforms to the DIDO distributed antennas;

DIDO Base Station Network (BSN): The wired backhaul that connects the BTS to the DIDO distributed antennas or other BTSs.

DIDO distributed antennas are grouped into different subsets based on their spatial distribution relative to the BTS or DIDO client locations. We define three types of clusters, as shown in Figure 36:

Supercluster 3640: A DIDO distributed antenna group connected to one or more BTSs such that the round-trip delay between all BTSs and the corresponding users is within the constraints of the DIDO precoding loop.

DIDO cluster 3641: A group of DIDO distributed antennas connected to the same BTS. When a supercluster contains only one BTS, its definition is the same as that of a DIDO cluster.

User cluster 3642: is a group of DIDO distributed antennas that cooperatively transmit precoded data to a given user.

For example, a BTS is a local hub connected to other BTSs and DIDO distributed antennas via a BSN. The BSN can include various network technologies, including but not limited to digital subscriber lines (DSL), ADSL, VDSL[6], cable modems, fiber rings, T1 lines, hybrid fiber-coaxial (HFC) networks, and/or fixed wireless (e.g., WiFi). All BTSs within the same super cluster share information about DIDO precoding via the BSN, so that the round-trip delay is within the DIDO precoding loop.

In Figure 37, dots represent DIDO distributed antennas, crosses represent users, and dashed lines indicate user clusters for users U1 and U8, respectively. The method described below is designed to create a communication link to the target user U1 while creating zero RF energy points for any other users (U2 to U8) inside or outside the user cluster.

We propose a similar approach in [5], where zero RF energy points are created to remove interference in the overlapping areas between DIDO clusters. Additional antennas are required to transmit signals to clients within a DIDO cluster while suppressing inter-cluster interference. One embodiment of the method proposed in this patent application does not attempt to remove inter-DIDO cluster interference; rather, it assumes that clusters are bound to clients (i.e., user-clusters) and guarantees that no interference (or negligible interference) is generated to any other clients in the neighborhood.

One concept associated with the proposed method is that users far enough away from a user cluster are not affected by radiation from the transmit antenna due to large path loss. Users close to or within the user cluster receive interference-free signals due to precoding. Furthermore, additional transmit antennas can be added to the user cluster (as shown in FIG37 ) such that the condition K ≤ M is satisfied.

One embodiment of a method using user clustering consists of the following steps:

a. Link quality measurement: The link quality between each DIDO distributed antenna and each user is reported to the BTS. The link quality metric consists of the signal-to-noise ratio (SNR) or the signal-to-interference-plus-noise ratio (SINR).

In one embodiment, DIDO distributed antennas transmit training signals, and users estimate the received signal quality based on the training. The training signals are designed to be orthogonal in the time, frequency, or code domains so that users can distinguish between different transmitters. Alternatively, DIDO antennas transmit narrowband signals (i.e., a single tone) at a specific frequency (i.e., the beacon channel), and users estimate the link quality based on the beacon signal. A threshold is defined as the minimum signal amplitude (or power) above the noise level for successful data demodulation, as shown in Figure 38a. Any link quality metric value below this threshold is assumed to be zero. The link quality metric is quantized by a finite number of bits and fed back to the transmitter.

In various embodiments, training signals or beacons are sent from users and link quality is estimated at the DIDO transmit antennas (as shown in FIG38b ), assuming reciprocity between uplink (UL) path loss and downlink (DL) path loss. Note that path loss reciprocity is a realistic assumption in time division duplex (TDD) systems (with UL and DL channels at the same frequency) and frequency division duplex (FDD) systems when the UL and DL frequency bands are relatively close.

As shown in FIG. 37 , information about link quality metrics is shared across different BTSs via the BSN so that all BTSs are aware of the link quality between each antenna/user coupling across different DIDO clusters.

b. Definition of User-Clusters: The link quality metrics of all wireless links in a DIDO cluster are entries of a link quality matrix shared across all BTSs via the BSN. An example of a link quality matrix for the scenario in FIG37 is shown in FIG39.

The link quality matrix is used to define user clusters. For example, FIG39 illustrates the selection of a user cluster for user U8. First, a subset of transmitters with non-zero link quality metrics for user U8 (i.e., active transmitters) is identified. These transmitters populate the user cluster for user U8. Then, a sub-matrix containing non-zero entries from transmitters within the user cluster to other users is selected. Note that because the link quality metric is only used to select the user cluster, it can be quantized using only two bits (i.e., to identify states above or below the threshold in FIG38 ), thereby reducing feedback overhead.

Another example is shown in Figure 40 for user U1. In this case, the number of active transmitters is lower than the number of users in the submatrix, violating the condition K ≤ M. Therefore, one or more columns are added to the submatrix to satisfy this condition. If the number of transmitters exceeds the number of users, additional antennas can be used for diversity schemes (i.e., antenna or eigenmode selection).

Yet another example for user U4 is shown in Figure 41. We observe that the sub-matrix can be obtained as a combination of two sub-matrices.

c. CSI reporting to BTSs: Once a user cluster is selected, CSI from all transmitters within the user cluster to each user reached by those transmitters is made available to all BTSs. CSI information is shared across all BTSs via the BSN. In TDD systems, UL/DL channel reciprocity can be exploited to derive CSI from training on the UL channel. In FDD systems, a feedback channel from all users to the BTS is required. To reduce the amount of feedback, only CSI corresponding to non-zero entries in the link quality matrix is fed back.

d. DIDO precoding: Finally, DIDO precoding is applied to each CSI sub-matrix corresponding to a different user cluster (eg, as described in related US patent applications).

In one embodiment, the singular value decomposition (SVD) of the effective channel matrix is computed and the precoding weights _wk for user k are defined as the right singular vectors in the null subspace corresponding to . Alternatively, if M>K and the SVD decomposes the effective channel matrix into , then the DIDO precoding weights for user k are given by

w _k ＝U _o (U _o ^H ·h _k ^T )

where U _o is the matrix whose columns are the singular vectors of the null subspace of .

From basic linear algebra considerations, we observe that the right singular vectors in the null subspace of a matrix are equal to the eigenvectors of C corresponding to the zero eigenvalues.

where the effective channel matrix is decomposed according to the SVD into Then, an alternative method to computing the SVD of is to compute the eigenvalue decomposition of C. There are several methods for computing the eigenvalue decomposition, such as the power method. Since we are only interested in the eigenvectors corresponding to the zero subspace of C, we use the inverse power method described by the following iteration

The first iterated vector (u _i ) is a random vector.

Considering that the eigenvalue (λ) of the null subspace is known (i.e., zero), the inverse power method requires only one iteration to converge, thus reducing the computational complexity. We then write the precoding weight vector as

w＝C ^-1 u ₁

where u ₁ is a vector whose real entries are equal to 1 (ie, the precoding weight vector is the sum of the columns of C ^-1 ).

The DIDO precoding computation requires a matrix inversion. Several numerical solutions exist to reduce the complexity of the matrix inversion, such as Strassen's algorithm [1] or the Coppersmith-Winograd algorithm [2,3]. Since C is a Hermitian matrix by definition, an alternative solution is to decompose C into its real and imaginary parts and compute the matrix inversion of the real matrix according to the method in [4, Section 11.4].

Another feature of the proposed method and system is its reconfigurability. As a client moves across different DIDO clusters, as shown in Figure 42, the user-cluster follows it. In other words, as the client changes its location, the subset of transmit antennas is continuously updated and the effective channel matrix (and corresponding precoding weights) is recalculated.

The proposed method works within the super cluster in Figure 36 because the links between BTSs via the BSN must be low latency. To suppress interference in the overlapping areas of different super clusters, our method in [5] can be used, which uses additional antennas to create zero RF energy points in the interference area between DIDO clusters.

It should be noted that the terms "user" and "client" are used interchangeably herein.

References

[1] S. Robinson, “Toward an Optimal Algorithm for Matrix Multiplication”, SIAM News, Volume 38, Number 9, November 2005.

[2] D.Coppersmith and S.Winograd, “Matrix Multiplication via Arithmetic Progression”, J.Symb.Comp.vol.9,pp.251-280,1990

[3] H. Cohn, R. Kleinberg, B. Szegedy, C. Umans, “Group-theoretic Algorithms for Matrix Multiplication”, p. 379-388, Nov. 2005.

[4] W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannery "Numerical Recipes in C: The Art of Scientific Computing", Cambridge University Press, 1992

[5] A. Forenza and S. G. Perlman, “Interference Management, Handoff, Power Control and Link Adaptation in Distributed-Input Distributed-Output (DIDO) Communication Systems,” patent application serial number 12/802,988, filed June 16, 2010.

[6] Per-Erik Eriksson and Odenhammar, “VDSL2: Next important broadband technology”, Ericsson Review No. 1, 2006.

III. Systems and Methods for Utilizing Coherence Regions in Wireless Systems

The capacity of a multiple antenna system (MAS) in a real propagation environment varies with the available spatial diversity on the radio link. Spatial diversity is determined by the distribution of scatterers in the radio channel and the geometry of the transmit and receive antenna arrays.

A popular model for MAS channels is the so-called clustered channel model, which defines groups of scatterers as clusters located around a transmitter and a receiver. In general, the more clusters there are and the larger their angular spread, the higher the spatial diversity and capacity that can be achieved on the wireless link. Clustered channel models have been validated by real-world measurements [1-2], and variations of those models have been adopted by different indoor (i.e., the IEEE 802.111n technical group for WLANs [3]) and outdoor (3GPP technical specification group for 3G cellular systems [4]) wireless standards.

Other factors that determine spatial diversity in wireless channels are the characteristics of the antenna array, including: antenna element spacing [5-7], number of antennas [8-9], array aperture [10-11], array geometry [5,12,13], polarization, and antenna pattern [14-28].

A unified model describing the effects of antenna array design and propagation channel characteristics on the spatial diversity (or degrees of freedom) of a wireless link is proposed in [29]. The received signal model in [29] is given by

y(q)＝∫C(q，p)x(p)dp+z(q)

Where x(p)∈C ³ is the polarization vector describing the transmitted signal, p,q∈R ³ are the polarization vector positions describing the transmit array and receive array respectively, and C(·,·)∈C ^3×3 is the matrix describing the system response between the transmit vector position and the receive vector position, which is given by

where _At (·,·), _Ar (·,·)∈C3 ^×3 are the transmit array response and receive array response, respectively, and is the channel response matrix, where the terms are the complex gains between the transmit and receive directions. In a DIDO system, a user device can have a single or multiple antennas. For simplicity, we assume a single-antenna receiver with an ideal isotropic pattern and rewrite the system response matrix as

Only the transmitting antenna pattern is considered

According to Maxwell's equations and the far-field term of Green's function, the array response can be approximated as [29]

where p∈P, P is the space defining the antenna array and where

where for an unpolarized antenna, studying the array response is equivalent to studying the integration kernel above. In the following, we show that the expressions for the integration kernel for different types of arrays are closed.

Unpolarized linear array

For an unpolarized linear array of length L (normalized by wavelength) and antenna elements oriented along the z-axis and centered at the origin, the integral kernel is given by [29]

a(cosθ, p _z )=exp(-j2πp _z cosθ).

Expanding the above equation into a series of shifted dyadic vectors, we obtain that the sine function has a resolution of 1/L and the dimension (i.e., degrees of freedom) of the array-finite and approximately wave-vector-finite subspace is

D _F =L|Ω _θ |

where Ω _θ = {cosθ:θ∈Θ}. We observe that for the broadside array |Ω _θ |= |Θ|, while for the endfire array |Ω _θ |≈ |Θ| ² /2.

Unpolarized spherical array

The integral kernel of a spherical array of radius R (normalized by the wavelength) is given by [29]

Decomposing the above function as a sum of spherical Bessel functions of the first kind, we obtain that the resolution of the spherical array is 1/(πR ² ), and the degrees of freedom are given by:

D _F ＝A|Ω|＝πR ² |Ω|

where A is the area of the spherical array, and

Coherence Regions in Wireless Channels

The relationship between the resolution of spherical arrays and their area, A, is shown in Figure 43. The center sphere is a spherical array of area A. The projection of the channel cluster onto the unit sphere defines distinct scattering regions whose sizes are proportional to the angular spread of the cluster. The region of size 1/A within each cluster (which we call the "coherence region") represents the projection of the basis functions of the array's radiation field and defines the array's resolution in the wavevector domain.

Comparing Figure 43 with Figure 44, we observe that the size of the region of coherence decreases with the inverse of the array size. In effect, a larger array concentrates energy into a smaller region, resulting in a larger number of degrees of freedom, _DF . Note that the total number of degrees of freedom also depends on the angular spread of the cluster, as shown in the definition above.

Figure 45 shows another example where the array size covers an even larger area compared to Figure 44, resulting in additional degrees of freedom. In a DIDO system, the array aperture can be approximated by the total area covered by all DIDO transmitters (assuming the antennas are spaced a fraction of the wavelength). Next, Figure 45 shows that a DIDO system can achieve an increased number of degrees of freedom by distributing the antennas in space, thereby reducing the size of the coherence region. Note that these figures are generated assuming an ideal spherical array. In reality, DIDO antennas are randomly scattered throughout a wide area, and the resulting coherence region may not be as regular in shape as shown.

Figure 46 shows that as the array size increases, more clusters are included in the wireless channel as radio waves are scattered by an increasing number of objects between the DIDO transmitters. Therefore, an increasing number of basis functions (spanning the radiation field) can be excited, resulting in additional degrees of freedom as defined above.

The multi-user (MU) multi-antenna system (MAS) described in this patent application exploits the coherence region of the wireless channel to create multiple, simultaneous, independent, non-interfering data streams to different users. Given channel conditions and user distribution, the basis functions of the radiation field are selected to create independent and simultaneous wireless links to different users, allowing each user to experience an interference-free link. Since the MU-MAS knows the channel between each transmitter and each user, it adjusts the precoded transmission based on this information to create individual coherence regions for different users.

In one embodiment of the present invention, MU-MAS employs nonlinear precoding, such as dirty paper coding (DPC) [30-31] or Tomlinson-Harashima (TH) [32-33] precoding. In another embodiment of the present invention, MU-MAS employs nonlinear precoding, such as block diagonalization (BD) or zero-forcing beamforming (ZF-BF) [34] as described in our previous patent applications described in the "Related Patent Applications" section of this specification.

To allow precoding to be implemented, MU-MAS requires knowledge of channel state information (CSI). CSI is available to MU-MAS via a feedback channel, or estimated on the uplink channel (assuming uplink/downlink channel reciprocity is possible in time division duplex (TDD) systems). One way to reduce the amount of feedback required for CSI is to use limited feedback techniques [35-37]. In one embodiment, MU-MAS uses limited feedback techniques to reduce the CSI overhead of the control channel. In limited feedback techniques, codebook design is key. One embodiment defines the codebook from a basis function across the transmit array radiation field.

As users move through space or the propagation environment changes over time due to moving objects (such as people or vehicles), the coherence region changes its position and shape. This is due to the Doppler effect, which is well known in wireless communications. As the environment changes due to the Doppler effect, the MU-MAS described in this patent application adjusts the precoding to continuously adapt the coherence region for each user. This adaptation of the coherence region is intended to create simultaneous non-interfering channels to different users.

Another embodiment of the present invention adaptively selects subsets of the MU-MAS system's antennas to create coherence zones of varying sizes. For example, if users are sparsely distributed across space (i.e., in rural areas or at times with low wireless resource usage), only a small subset of antennas is selected, and the size of the coherence zone is large relative to the size of the array in FIG43 . Alternatively, in densely populated areas (i.e., in urban areas or at times with peak wireless service usage), more antennas are selected to create smaller coherence zones for users that are directly adjacent to each other.

In one embodiment of the present invention, the MU-MAS is a DIDO system as described in the previous patent applications described in the "Related Patent Applications" section of this specification. The DIDO system uses linear or nonlinear precoding and/or limited feedback techniques to create coherence zones to different users.

Numerical results

We begin by calculating the number of degrees of freedom in a conventional multiple-input multiple-output (MIMO) system as a function of the array size. We consider unpolarized linear arrays and two types of channel models: an indoor model as in the IEEE 802.11n standard for WiFi systems and an outdoor model as in the 3GPP-LTE standard for cellular systems. The indoor channel model in [3] defines the number of clusters in the range [2, 6] and the angular spread in the range [15°, 40°]. The outdoor channel model for urban microcells defines approximately 6 clusters and an angular spread of approximately 20° at the base station.

Figure 47 illustrates the degrees of freedom of a MIMO system in practical indoor and outdoor propagation scenarios. For example, considering a linear array with 10 antennas spaced one wavelength apart, the maximum degrees of freedom (or number of spatial channels) available on the wireless link is limited to approximately 3 for the outdoor scenario and 7 for the indoor scenario. Of course, indoor channels offer more degrees of freedom due to their greater angular spread.

Next, we calculate the degrees of freedom in a DIDO system. We consider the case where antennas are distributed in 3D space, such as in a city center where DIDO access points may be located on different floors of adjacent buildings. Therefore, we model the DIDO transmit antennas (all connected to each other via fiber or a DSL backbone) as a spherical array. Furthermore, we assume that the clusters are uniformly distributed across the solid angle.

Figure 48 shows the degree of freedom in a DIDO system as a function of array diameter. We observe that for a diameter equal to 10 wavelengths, approximately 1,000 degrees of freedom are available in a DIDO system. Theoretically, up to 1,000 non-interfering channels to users can be created. The increased spatial diversity due to the spatially distributed antennas is key to the multiplexing gain DIDO offers over conventional MIMO systems.

For comparison, we show the degrees of freedom achievable with a DIDO system in a suburban environment. We assume that clusters are distributed within the elevation angle [α, π-α] and define the cluster solid angle as |Ω| = 4π cosα. For example, in a suburban scenario with a two-story building, the elevation angle of the scatterer can be α = 60°. In this case, the number of degrees of freedom varies with wavelength as shown in Figure 48.

References

[1]A.A.M.Saleh and R.A.Valenzuela, “A statistical model for indoor multipath propagation,” IEEE Jour.Select.Areas in Comm., vol.195SAC-5, no.2, pp.128–137, Feb.1987.

[2] J.W.Wallace and M.A.Jensen, “Statistical characteristics of measured MIMO wireless channel data and comparison to conventional models,” Proc.IEEE Veh.Technol.Conf., vol.2, no.7-11, pp.1078–1082, Oct.2001.

[3] V. Erceg et al., “TGn channel models,” IEEE 802.11-03/940r4, May 2004.

[4]3GPP Technical Specification Group, “Spatial channel model, SCM-134 text V6.0,” Spatial channel Model AHG (Combined ad-hoc from 3GPP and 3GPP2), Apr. 2003.

[5] D.-S.Shiu, G.J.Foschini, M.J.Gans, and J.M.Kahn, “Fading correlation and its effect on the capacity of multielement antenna systems,” IEEE Trans.Comm., vol.48, no.3, pp.502–513, Mar.2000.

[6] V.Pohl, V.Jungnickel, T.Haustein, and C.von Helmolt, “Antenna spacing in MIMO indoor channels,” Proc.IEEE Veh.Technol.Conf., vol.2, pp.749–753, May 2002.

[7] M. Stoytchev, H. Safar, A. L. Moustakas, and S. Simon, “Compact antenna arrays for MIMO applications,” Proc. IEEE Antennas and Prop. Symp., vol. 3, pp. 708–711, July 2001.

[8] K.Sulonen, P.Suvikunnas, L.Vuokko, J.Kivinen, and P.Vainikainen, “Comparison of MIMO antenna configurations in picocell and microcell environments,” IEEE Jour.Select.Areas in Comm., vol.21, pp.703–712, June 2003.

[9] Shuangqing Wei, D.L.Goeckel, and R.Janaswamy, “On the asymptotic capacity of MIMO systems with fixed length linear antenna arrays,” Proc.IEEE Int.Conf.on Comm., vol.4, pp.2633–2637, 2003.

[10] T.S.Pollock, T.D.Abhayapala, and R.A.Kennedy, “Antenna saturation effects on MIMO capacity,” Proc.IEEE Int.Conf.on Comm., 192 vol.4, pp.2301–2305, May 2003.

[11] M.L.Morris and M.A.Jensen, “The impact of array configuration on MIMO wireless channel capacity,” Proc.IEEE Antennas and Prop.Symp., vol.3, pp.214–217, June 2002.

[12] Liang Xiao, Lin Dal, Hairuo Zhuang, Shidong Zhou, and Yan Yao, “A comparative study of MIMO capacity with different antenna topologies,” IEEE ICCS’02, vol. 1, pp. 431–435, Nov. 2002.

[13] A. Forenza and R. W. Heath Jr., “Impact of antenna geometry on MIMO communication in indoor clustered channels,” Proc. IEEE Antennas and Prop. Symp., vol. 2, pp. 1700–1703, June 2004.

[14]M.R.Andrews,P.P.Mitra,and R.deCarvalho,“Tripling the capacity of wireless communications using electromagnetic polarization,”Nature,vol.409,pp.316–318,Jan.2001.

[15] D.D.Stancil,A.Berson,J.P.Van’t Hof,R.Negi,S.Sheth,and P.Patel,“Doubling wireless channel capacity using co-polarised,co-located electricand magnetic dipoles,” Electronics Letters,vol.38,pp.746–747,July 2002.

[16] T. Svantesson, “On capacity and correlation of multi-antennasystems employing multiple polarizations,” Proc. IEEE Antennas and Prop. Symp., vol. 3, pp. 202–205, June 2002.

[17] C. Degen and W. Keusgen, “Performance evaluation of MIMO systems using dual-polarized Antennas,” Proc. IEEE Int. Conf. on Telecommun., vol. 2, pp. 1520–1525, Feb. 2003.

[18] R. Vaughan, “Switched parasitic elements for antenna diversity,” IEEE Trans. Antennas Propagat., vol. 47, pp. 399–405, Feb. 1999.

[19] P.Mattheijssen, M.H.A.J.Herben, G.Dolmans, and L.Leyten, “Antenna-pattern diversity versus space diversity for use at handhelds,” IEEE Trans.onVeh.Technol., vol.53, pp.1035–1042, July 2004.

[20] L. Dong, H. Ling, and R. W. Heath Jr., “Multiple-input multiple-output wireless communication systems using antenna pattern diversity,” Proc. IEEE Glob. Telecom. Conf., vol. 1, pp. 997–1001, Nov. 2002.

[21] J.B.Andersen and B.N.Getu, “The MIMO cube-a compact MIMO antenna,” IEEE Proc.of Wireless Personal Multimedia Communications Int.Symp., vol.1, pp.112–114, Oct.2002.

[22] C.Waldschmidt, C.Kuhnert, S.Schulteis, and W.Wiesbeck, “Compact MIMO-arrays based on polarisation-diversity,” Proc.IEEE Antennas and Prop.Symp., vol.2, pp.499–502, June 2003.

[23] C.B.Dietrich Jr, K.Dietze, J.R.Nealy, and W.L.Stutzman, “Spatial, polarization, and pattern diversity for wireless handheld terminals,” Proc.IEEE Antennas and Prop. Symp., vol.49, pp.1271–1281, Sep.2001.

[24] S.Visuri and D.T.Slock, “Colocated antenna arrays: design desiderata for wireless communications,” Proc. of Sensor Array and Multichannel Sign. Proc. Workshop, pp. 580–584, Aug. 2002.

[25]A.Forenza and R.W.Heath Jr., “Benefit of pattern diversity via 2-element array of circular patch Antennas in indoor clustered MIMO channels,” IEEE Trans.on Communications, vol.54, no.5, pp.943-954, May 2006.

[26]A.Forenza and R.W.Heath,Jr., “Optimization Methodology forDesigning 2-CPAs Exploiting Pattern Diversity in clustered MIMO channels”, IEEE Trans.on Communications, Vol.56, no.10, pp.1748-1759, Oct.2008.

[27] D.Piazza, N.J.Kirsch, A.Forenza, R.W.Heath, Jr., and K.R.Dandekar, “Design and Evaluation of a Reconfigurable Antenna Array for MIMO Systems,” IEEE Transactions on Antennas and Propagation, vol.56, no.3, pp.869-881, March 2008.

[28] R. Bhagavatula, R. W. Heath, Jr., A. Forenza, and S. Vishwanath, “Sizing up MIMO Arrays,” IEEE Vehicular Technology Magazine, vol. 3, no. 4, pp. 31-38, Dec. 2008.

[29]Ada Poon, R.Brodersen and D.Tse, “Degrees of Freedom in Multiple Antenna channels: A Signal Space Approach”, IEEE Transactions on Information Theory, vol. 51(2), Feb. 2005, pp. 523-536.

[30]M.Costa, “Writing on dirty paper,” IEEE Transactions on Information Theory, Vol. 29, No. 3, Page(s): 439-441, May 1983.

[31] U.Erez, S.Shamai(Shitz), and R.Zamir, “Capacity and lattice-strategies for cancelling known interference,” Proceedings of International Symposium on Information Theory, Honolulu, Hawaii, Nov. 2000.

[32]M.Tomlinson, “New automatic equalizer employing moduloarithmetic,” Electronics Letters, Page(s): 138-139, March 1971.

[33]H.Miyakawa and H.Harashima, “A method of code conversion for digital communication channels with intersymbol interference,” Transactions of the Institute of Electronic

[34] R.A.Monziano and T.W.Miller, Introduction to Adaptive Arrays, New York: Wiley, 1980.

[35]T.Yoo,N.Jindal,and A.Goldsmith,“Multi-antenna broadcast channelswith limited feedback and user selection,”IEEE Journal on Sel.Areas inCommunications,vol.25,pp.1478-91,July 2007.

[36] P. Ding, D. J. Love, and M. D. Zoltowski, “On the sum rate of channelsubspace feedback for multi-antenna broadcast channels,” in Proc., IEEE Globecom, vol. 5, pp. 2699-2703, November 2005.

[37] N. Jindal, “MIMO broadcast channels with finite-rate feedback,” IEEE Trans. on Info. Theory, vol. 52, pp. 5045-60, November 2006.

IV. Systems and Methods for Planned Evolution and Obsolescence of Multi-User Spectrum

The growing demand for high-speed wireless services and the increasing number of cellular phone users have led to a fundamental technological revolution in the wireless industry over the past three decades, evolving from the initial analog voice services (AMPS[1-2]) to standards that support digital voice (GSM[3-4], IS-95CDMA[5]), data traffic (EDGE[6], EV-DO[7]), and Internet browsing (WiFi[8-9], WiMAX[10-11], 3G[12-13], 4G[14-15]). The advancement of wireless technology over the years has been made possible by two major efforts:

i) The Federal Communications Commission (FCC)[16] has been allocating new spectrum to support emerging standards. For example, in the first generation AMPS system, the number of channels allocated by the FCC increased from an initial 333 in 1983 to 416 in the late 1980s to support the growing number of cellular clients. More recently, the commercialization of technologies such as Wi-Fi, Bluetooth, and ZigBee has been made possible by the use of the unlicensed ISM bands[17] allocated by the FCC as early as 1985.

ii) The wireless industry has been generating new technologies that more efficiently utilize the limited available spectrum to support higher data rate links and an ever-increasing number of users. A major revolution in the wireless field was the migration from analog AMPS systems to digital D-AMPS and GSM in the 1990s, which allowed for a higher number of calls per given frequency band due to improved spectral efficiency. Another fundamental shift occurred in the early 21st century with spatial processing techniques such as Multiple Input Multiple Output (MIMO), which produced a 4x increase in data rates relative to previous wireless networks and were adopted by different standards (i.e., IEEE 802.11n for Wi-Fi, IEEE 802.16 for WiMAX, 3GPP for 4G-LTE).

Despite significant efforts to provide high-speed wireless connectivity solutions, the wireless industry is facing new challenges: delivering high-definition (HD) video streaming to meet the growing demand for gaming-like services, while also providing wireless coverage everywhere, including in rural areas where building a wired backbone is costly and impractical. Currently, the most advanced wireless standard systems (i.e., 4G-LTE) are unable to deliver the data rate requirements and latency constraints required to support HD streaming services, especially when the network is overloaded with a large number of concurrent links. Once again, the main drawbacks are limited spectrum availability and the lack of spectrally efficient technologies that can truly increase data rates and provide complete coverage.

In recent years, a new technology called Distributed Input Distributed Output (DIDO) [18-21] has emerged, which is described in our previous patent applications listed in the "Related Patent Applications" section of this specification. DIDO technology promises orders of magnitude increases in spectral efficiency, making HD wireless streaming services possible in overloaded networks.

Meanwhile, the US government has been addressing the spectrum shortage problem by implementing a plan to release 500 MHz of spectrum over the next 10 years. The plan was announced on June 28, 2010, with the goal of allowing emerging wireless technologies to operate in new frequency bands and provide high-speed wireless coverage in urban and rural areas [22]. As part of this plan, on September 23, 2010, the FCC opened up approximately 200 MHz of VHF and UHF spectrum for unlicensed use, which is called "white space" [23]. One of the restrictions on operating in those frequency bands is that harmful interference must not be caused to existing wireless microphone devices operating in the same frequency bands. Therefore, on July 22, 2011, the IEEE 802.22 working group finalized the standard for a new wireless system that uses cognitive radio technology (or spectrum sensing) with the key feature of dynamically monitoring the spectrum and operating in available frequency bands, thereby avoiding harmful interference to coexisting wireless devices [24]. Only recently has there been a debate about allocating part of the white space for licensed use and opening it up for spectrum auctions [25].

The coexistence of unlicensed devices in the same frequency band and the competition between unlicensed and licensed spectrum have been two major issues in the FCC's spectrum allocation plan over the years. For example, cognitive radio technology has enabled the coexistence of wireless microphones and wireless communication devices in white spaces. However, cognitive radio can only provide a fraction of the spectral efficiency of other technologies using spatial processing like DIDO. Similarly, the performance of Wi-Fi systems has been significantly degraded over the past decade due to the increase in the number of access points and the use of Bluetooth/ZigBee devices operating in the same unlicensed ISM bands and generating uncontrolled interference. A disadvantage of unlicensed spectrum is the unregulated use of RF devices, which will continue to pollute the spectrum for years to come. RF pollution also hinders the use of unlicensed spectrum for future licensed operations, thereby limiting the important market opportunities for commercial wireless broadband services and spectrum auctions.

We propose a new system and method that allows for dynamic allocation of wireless spectrum to allow different services and standards to coexist and evolve. One embodiment of our method dynamically allocates rights to RF transceivers to operate in certain portions of the spectrum and allows for the obsolescence of identical RF devices to provide:

i) Spectrum reconfigurability to allow new types of wireless operation (i.e., licensed vs. unlicensed) and/or to comply with new RF power emission limits. This feature allows spectrum auctions to be conducted whenever necessary, without requiring advance planning for the use of licensed spectrum versus unlicensed spectrum. It also allows transmit power levels to be adjusted to meet new power emission levels mandated by the FCC.

ii) Coexistence of different technologies (i.e., white spaces and wireless microphones, WiFi and Bluetooth/ZigBee) operating in the same frequency band, allowing the band to be dynamically reallocated as new technologies are created while avoiding interference with existing technologies.

iii) Enables seamless evolution of wireless infrastructure as systems migrate to more advanced technologies that offer higher spectral efficiency, better coverage, and improved performance to support new services requiring higher QoS (ie, HD video streaming).

In the following, we describe a system and method for planned evolution and obsolescence of multi-user spectrum. One embodiment of the system includes one or more centralized processors (CPs) 4901-4904 and one or more distributed nodes (DNs) 4911-4913, which communicate via wired or wireless connections as shown in Figure 49. For example, in the context of a 4G-LTE network [26], the centralized processor is an access core gateway (ACGW) connected to several Node B transceivers. In the context of Wi-Fi, the centralized processor is an Internet Service Provider (ISP) and the distributed nodes are Wi-Fi access points connected to the ISP via a modem or directly to cable or DSL. In another embodiment of the present invention, the system is a distributed input distributed output (DIDO) system having a centralized processor (or BTS) and distributed nodes that are DIDO access points (or DIDO distributed antennas connected to the BTS via a BSN), such as the DIDO system described in the "Related Patent Applications" section of this specification.

DNs 4911-4913 communicate with CPs 4901-4904. Information exchanged between DNs and CPs is used to dynamically adapt node configurations to the evolving design of the network architecture. In one embodiment, DNs 4911-4913 share their identification numbers with the CPs. The CPs store the identification numbers of all DNs connected via the network in a lookup table or shared database. These lookup tables or databases can be shared with other CPs, and the information synchronized so that all CPs always have access to the latest information about all DNs on the network.

For example, the FCC may decide to allocate a portion of the spectrum for unlicensed use, and the proposed system may be designed to operate in that spectrum. Due to a lack of spectrum, the FCC may subsequently need to allocate a portion of the spectrum for authorized use for commercial operators (i.e., AT&T, Verizon, or Sprint), national defense, or public safety. In conventional wireless systems, this coexistence would be impossible because existing wireless devices operating in unlicensed bands would cause harmful interference to licensed RF transceivers. In our proposed system, distributed nodes exchange control information with CPs 4901-4903 to adapt their RF transmissions to the evolving band plan. In one embodiment, DNs 4911-4913 are initially designed to operate on different bands within the available spectrum. When the FCC allocates one or more portions of the spectrum for authorized operation, the CPs exchange control information with the unlicensed DNs and reconfigure the DNs to shut down the bands used for authorized use so that the unlicensed DNs do not interfere with the licensed DNs. This scenario is shown in Figure 50, where unlicensed nodes (e.g., 5002) are represented by solid circles and licensed nodes are represented by hollow circles (e.g., 5001). In another embodiment, the entire spectrum can be allocated to the new licensed service, and control information is used by the CP to shut down all unlicensed DNs to avoid interfering with the licensed DNs. This scenario is shown in Figure 51, where the obsolete unlicensed nodes are covered with crosses.

By way of another example, it may be necessary to limit the power transmission of certain devices operating in a given frequency band to meet FCC exposure limits [27]. For example, a wireless system may initially be designed for fixed wireless links, with DNs 4911-4913 connected to outdoor rooftop transceiver antennas. Later, the same system may be updated to support DNs with indoor portable antennas to provide better indoor coverage. Portable devices are subject to more stringent FCC exposure limits than rooftop transmitters because they may be closer to the human body. In this case, the old DNs designed for outdoor applications can be reused for indoor applications simply by adjusting the transmit power setting. In one embodiment of the present invention, the DNs are designed with a predefined set of transmit power levels, and when the system is upgraded the CPs 4901-4903 send control information to the DNs 4911-4913 to select the new power level to meet the FCC exposure limits. In another embodiment, the DNs are manufactured with only one power transmission setting, and those DNs that exceed the new power transmission level are remotely shut down by the CP.

In one embodiment, CPs 4901-4903 periodically monitor all DNs 4911-4913 in the network to define their authority to operate as RF transceivers according to certain criteria. Those DNs that are no longer up-to-date can be marked as obsolete and removed from the network. For example, DNs operating within the current power limits and frequency bands remain active in the network, while all other DNs are shut down. It is important to note that the DN parameters controlled by the CP are not limited to power transmission and frequency bands; they can be any parameters that define the wireless link between the DN and the client device.

In another embodiment of the present invention, DNs 4911-4913 can be reconfigured to allow different standard systems to coexist within the same spectrum. For example, the power transmission, frequency band, or other configuration parameters of certain DNs operating in the context of WLAN can be adjusted to accommodate the use of new DNs designed for WPAN applications while avoiding harmful interference.

As new wireless standards are developed to increase data rates and coverage in wireless networks, DNs 4911-4913 can be updated to support those standards. In one embodiment, the DN is a software-defined radio (SDR) equipped with programmable computing power, such as an FPGA, DSP, CPU, GPU, and/or GPGPU that executes algorithms for baseband signal processing. If the standard is upgraded, the new baseband algorithm can be remotely uploaded from the CP to the DN to reflect the new standard. For example, in one embodiment, the first standard is a CDMA-based standard and is subsequently replaced by OFDM technology to support different types of systems. Similarly, the sampling rate, power, and other parameters can be remotely updated to the DN. This SDR feature of the DN allows for continuous upgrades to the network as new technologies are developed to improve overall system performance.

In another embodiment, the system described herein is a cloud wireless system consisting of multiple CPs, distributed nodes, and a network interconnecting the CPs with the DNs. Figure 52 shows an example of a cloud wireless system, where all nodes identified by solid circles (e.g., 5203) communicate with CP 5206, nodes identified by hollow circles communicate with CP 5205, and CPs 5205-5206 communicate with each other via network 5201. In one embodiment of the present invention, the cloud wireless system is a DIDO system, and the DNs are connected to the CPs and exchange information to periodically or immediately reconfigure system parameters and dynamically adjust to changing conditions of the wireless architecture. In the DIDO system, the CPs are DIDO BTSs, the distributed nodes are DIDO distributed antennas, the network is a BSN, and multiple BTSs are interconnected via a DIDO centralized processor as described in our previous patent application described in the "Related Patent Applications" section of this specification.

All DNs 5202-5203 within the cloud wireless system can be grouped into different groups. These groups of DNs can simultaneously create non-interfering wireless links to many client devices, with each group supporting different multiple access technologies (e.g., TDMA, FDMA, CDMA, OFDMA, and/or SDMA), different modulations (e.g., QAM, OFDM), and/or coding schemes (e.g., convolutional coding, LDPC, enhanced codes). Similarly, each client can be served using a different multiple access technology and/or a different modulation/coding scheme. Based on the active clients in the system and the standards they employ for their wireless links, the CPs 5205-5206 dynamically select a subset of DNs that support those standards and are within range of the client devices.

References

[1]Wikipedia, "Advanced Mobile Phone System"

http://en.wikipedia.org/wiki/Advanced_Mobile_Phone_System (Wikipedia, “Advanced Mobile Phone System”, http://en.wikipedia.org/wiki/Advanced_Mobile_Phone_System)

[2] AT&T, “1946: First Mobile Telephone Call”

AT&T, "1946: The First Mobile Phone Call," http://www.corp.att.com/attlabs/reputation/timeline/46mobile.html

[3]GSMA, "GSM technology"

http://www.gsmworld.com/technology/index.htm (Global System for Mobile Communications Association, "GSM Technology," http://www.gsmworld.com/technology/index.htm)

[4]ETSI, "Mobile technologies GSM"

http://www.etsi.org/WebSite/Technologies/gsm.aspx (European Telecommunications Standards Institute, "GSM Mobile Technology," http://www.etsi.org/WebSite/Technologies/gsm.aspx)

[5] Wikipedia, “IS-95”

http://en.wikipedia.org/wiki/IS-95 (Wikipedia, "IS-95", http://en.wikipedia.org/wiki/IS-95)

[6]Ericsson, "The evolution of EDGE"

http://www.ericsson.com/res/docs/whitepapers/evolution_to_edge.pdf (Ericsson, "Evolution of EDGE," http://www.ericsson.com/res/docs/whitepapers/evolution_to_edge.pdf)

[7]Q.Bi(2004-03). "A Forward Link Performance Study of the 1xEV-DORel.0System Using Field Measurements and Simulations" (PDF).LucentTechnologies.

http://www.cdg.org/resources/white_papers/files/Lucent%201xEV-DO%20Rev%20O%20Mar%2004.pdf (Q. Bi, March 2004, "Forward Link Performance Study of 1xEV-DO Rel.0 Systems Using Field Measurements and Simulations" (PDF), Lucent Technologies, http://www.cdg.org/resources/white_papers/files/Lucent%201xEV-DO%20Rev%20O%20Mar%2004.pdf)

[8]Wi-Fi alliance, http://www.wi-fi.org/(Wi-Fi Alliance, http://www.wi-fi.org/)

[9]Wi-Fi alliance, “Wi-Fi certified makes it Wi-Fi”

http://www.wi-fi.org/files/WFA_Certification_Overview_WP_en.pdf (Wi-Fi Alliance, "Wi-Fi CERTIFIED Takes Wi-Fi Technology to a New Level of Performance," http://www.wi-fi.org/files/WFA_Certification_Overview_WP_en.pdf)

[10]WiMAX forum, http://www.wimaxforum.org/(WiMAX Forum, http://www.wimaxforum.org/)

[11] C.Eklund, RBMarks, KLStanwood and S.Wang, "IEEE Standard 802.16: A Technical Overview of the WirelessMAN ^TM Air Interface for Broadband WirelessAccess"

http://ieee802.org/16/docs/02/C80216-02_05.pdf (C. Eklund, R.B. Marks, K.L. Stanwood, and S. Wang, "IEEE Standard 802.16: Technical Overview of the WirelessMAN ^™ Air Interface for Broadband Wireless Access," http://ieee802.org/16/docs/02/C80216-02_05.pdf)

[12] 3GPP, "UMTS", http://www.3gpp.org/article/umts

[13] H.A.J.Karlsson, M.Meyer, S.Parkvall, J.Torsner, and M.Wahlqvist “Technical Solutions for the 3G Long-Term Evolution”, IEEE Communications Magazine, pp.38-45, Mar. 2006

[14] 3GPP, "LTE", http://www.3gpp.org/LTE

[15] Motorola, “Long Term Evolution (LTE): A Technical Overview”, http://business.motorola.com/experiencelte/pdf/LTETechnicalOverview.pdf

[16] Federal Communications Commission, “Authorization of Spread Spectrum Systems Under Parts 15 and 90 of the FCC Rules and Regulations”, June 1985.

[17]ITU, "ISM band" http://www.itu.int/ITU-R/terrestrial/faq/index.html#g013

[18] S.Perlman and A.Forenza "Distributed-input distributed-output (DIDO) wireless technology: a new approach to multiuser wireless", Aug. 2011

http://www.rearden.com/DIDO/DIDO_White_Paper_110727.pdf (S. Perlman and A. Forenza, "Distributed Input Distributed Output (DIDO) Wireless Technology: A New Approach to Multi-User Wireless," August 2011, http://www.rearden.com/DIDO/DIDO_White_Paper_110727.pdf)

[19] Bloomberg Businessweek, “Steve Perlman’s Wireless Fix,” July 27, 2011

http://www.businessweek.com/magazine/the-edison-of-silicon-valley-07272011.html (Bloomberg Businessweek, “Steve Perlman’s Wireless Location Method,” July 27, 2011, http://www.businessweek.com/magazine/the-edison-of-silicon-valley-07272011.html)

[20]Wired, “Has OnLive’s Steve Perlman Discovered Holy Grail of Wireless?”, June 30, 2011

http://www.wired.com/epicenter/2011/06/perlman-holy-grail-wireless/ (Wired, “Has OnLive’s Steve Perlman Found the Holy Grail of Wireless?”, June 30, 2011, http://www.wired.com/epicenter/2011/06/perlman-holy-grail-wireless/)

[21]The Wall Street Journal “Silicon Valley Inventor’s Radical Rewrite of Wireless”, July 28,2011

http://blogs.wsj.com/digits/2011/07/28/silicon-valley-inventors-radical-rewrite-of-wireless/ (The Wall Street Journal, “Silicon Valley Inventors Radically Rewrite Wireless Networking,” July 28, 2011, http://blogs.wsj.com/digits/2011/07/28/silicon-valley-inventors-radical-rewrite-of-wireless/)

[22]The White House, "Presidential Memorandum: Unleashing the WirelessBroadband Revolution", June 28, 2010

http://www.whitehouse.gov/the-press-office/presidential-memorandum-unleashing-wireless-broadband-revolution (The White House, “Presidential Memorandum: Unleashing the Wireless Broadband Revolution,” June 28, 2010, http://www.whitehouse.gov/the-press-office/presidential-memorandum-unleashing-wireless-broadband-revolution)

[23]FCC, "Open commission meeting", Sept.23 ^rd , 2010

http://reboot.fcc.gov/open-meetings/2010/september (FCC, “Open Commission Meetings,” September 23, 2010, http://reboot.fcc.gov/open-meetings/2010/september)

[24] IEEE 802.22, “IEEE 802.22 Working Group on Wireless Regional Area Networks”, http://www.ieee802.org/22/

[25] "A bill", 112th congress, 1 ^st session, July 12, 2011

http://republicans.energycommerce.house.gov/Media/file/Hearings/Telecom/071511/DiscussionDraft.pdf (112th Congress, First Session, “Bill,” July 12, 2011)

[26] H.A.J.Karlsson, M.Meyer, S.Parkvall, J.Torsner, and M.Wahlqvist “Technical Solutions for the 3G Long-Term Evolution”, IEEE Communications Magazine, pp.38-45, Mar. 2006

[27]FCC, “Evaluating compliance with FCC guidelines for human exposure to radiofrequency electromagnetic fields,” OET Bulletin 65, Edition 97-01, August 1997

V. System and Method for Compensating for Doppler Effect in a Distributed-Input-Distributed-Output Wireless System

In this section of the detailed description, we describe a multi-user (MU) multi-antenna system (MAS) for multi-user wireless transmission that adaptively reconfigures parameters to compensate for the Doppler effect due to changes in user mobility or the propagation environment. In one embodiment, the MAS is a distributed input-distributed output (DIDO) system, as described in the co-pending patent application described in the "Related Patent Applications" section of this specification and shown in Figure 53. The DIDO system of one embodiment includes the following components:

• User Equipment (UE): The UE 5301 of one embodiment includes an RF transceiver for a fixed or mobile client that receives data streams on downlink (DL) channels from the DIDO backhaul and transmits data to the DIDO backhaul via uplink (UL) channels.

Base Transceiver Station (BTS): The BTSs 5310-5314 of one embodiment interface the DIDO backhaul with the wireless channel. The BTSs 5310-5314 are access points that include a DAC/ADC and a radio frequency (RF) chain to convert baseband signals into RF signals. In some cases, the BTS is a simple RF transceiver equipped with a power amplifier/antenna. The RF signal is transmitted to the BTS via radio frequency fiber transmission technology, as described in our patent application.

Controller (CTR): In one embodiment, the CTR 5320 is a specific type of BTS designed for certain specific purposes, such as transmitting training signals for time/frequency synchronization of the BTS and/or UE, receiving control information from the UE or transmitting control information to the UE, and receiving channel state information (CSI) or channel quality information from the UE.

Centralized Processor (CP): The CP 5340 of one embodiment is a DIDO server that interfaces the Internet or other type of external network 5350 with the DIDO backhaul. The CP computes the DIDO baseband processing and sends the waveform to the distributed BTSs for DL transmission.

Base Station Network (BSN): The BSN 5330 of one embodiment is a network that connects CPs to distributed BTSs, which carry information for downlink or uplink channels. The BSN is a wired or wireless network, or a combination of both. For example, the BSN can be a DSL, cable, or fiber optic network, or a line-of-sight or non-line-of-sight wireless link. Alternatively, the BSN can be a private network, a local area network, or the internet.

As described in the co-pending patent application, a DIDO system creates independent channels for multiple users, allowing each user to receive an interference-free channel. In a DIDO system, this is achieved by using distributed antennas or BTSs to exploit spatial diversity. In one embodiment, a DIDO system employs spatial, polarization, and/or pattern diversity to increase the degrees of freedom within each channel. The increased degrees of freedom of the wireless link are used to transmit independent data streams to a greater number of UEs (i.e., multiplexing gain) and/or improve coverage (i.e., diversity gain).

BTSs 5310-5314 are deployed anywhere that is convenient for accessing the Internet or BSN. In one embodiment of the present invention, UEs 5301-5305 are arbitrarily deployed between, around, and/or surrounded by BTSs or distributed antennas, as shown in FIG54.

In one embodiment, BTSs 5310-5314 transmit training signals and/or independent data streams to UE 5301 via a DL channel, as shown in FIG55 . The training signals are used by the UE for various purposes, such as time/frequency synchronization, channel estimation, and/or estimation of channel state information (CSI). In one embodiment of the present invention, MU-MAS DL employs nonlinear precoding, such as dirty paper coding (DPC) [1-2] or Tomlinson-Harashima (TH) [3-4] precoding. In another embodiment of the present invention, MU-MAS DL employs nonlinear precoding, such as block diagonalization (BD) or zero-forcing beamforming (ZF-BF) [5] as described in the co-pending patent applications described in the “Related Patent Applications” section of this specification. If the number of BTSs is greater than the number of UEs, the additional BTSs are used to improve the link quality to each UE through a diversity scheme, such as antenna selection or eigenmode selection as described in the “Related Patent Applications” section of this specification. If the number of BTSs is less than the number of UEs, the additional UEs share the radio link with other UEs through conventional multiplexing techniques (eg, TDMA, FDMA, CDMA, OFDMA).

The UL channel is used to send data from the UE 5301 to the CSI (or channel quality information) used by the CP 5340 and/or the DIDO precoder. In one embodiment, the UL channel from the UE is multiplexed to the CTR as shown in Figure 56 or to the nearest BTS using conventional multiplexing techniques (e.g., TDMA, FDMA, CDMA, OFDMA). In another embodiment of the present invention, spatial processing techniques are used to separate the UL channels from the UE 5301 to the distributed BTSs 5310-5314, as shown in Figure 57. For example, the UL stream is transmitted from the client to the DIDO antenna via a multiple-input multiple-output (MIMO) multiplexing scheme. The MIMO multiplexing scheme includes transmitting independent data streams from the client and removing co-channel interference using linear or nonlinear receivers at the DIDO antennas. In another embodiment, downlink weights are used on the uplink to demodulate the uplink stream, assuming that UL/DL channel reciprocity is maintained and that the channel does not differ significantly between DL and UL transmissions due to the Doppler effect. In another embodiment, a maximum ratio combining (MRC) receiver is used on the UL channel to improve the signal quality from each client's DIDO antenna.

Data, control information, and CSI sent via the DL/UL channels are shared between the CP 5340 and the BTSs 5310-5314 via the BSN 5330. Known training signals for the DL channels can be stored in memory at the BTSs 5310-5314 to reduce overhead through the BSN 5330. Depending on the network type (i.e., wireless/wired, DSL/cable, or fiber), the data rate available on the BSN 5330 may not be sufficient to exchange information between the CP 5340 and the BTSs 5310-5314, especially when delivering baseband signals to the BTS. For example, we assume that the BTS transmits an independent data stream of 10 Mbps to each UE over a 5 MHz bandwidth (depending on the digital modulation and FEC coding scheme used on the wireless link). If 16 bits are quantized for the real part and 16 bits are quantized for the imaginary part, the baseband signal requires a data throughput of 160 Mbps from the CP to the BTS via the BSN. In one embodiment, the CP and BTS are equipped with encoders and decoders to compress and decompress the information sent via the BSN. In the forward link, the precoded baseband data sent from the CP to the BTS is compressed to reduce the number of bits and the overhead sent through the BSN. Similarly, in the reverse link, the CSI (sent from the UE to the BTS via the uplink channel) and the data are compressed before being transmitted from the BTS to the CP via the BSN. Different compression algorithms are used to reduce the number of bits and the overhead sent through the BSN, including but not limited to lossless and/or lossy techniques [6].

One feature of the DIDO system used in one embodiment is that the CSI or channel quality information between all BTSs 5310-5314 and the UE 5301 is made known to the CP 5340 to allow precoding. As described above, the performance of DIDO depends on the rate at which CSI is delivered to the CP relative to the rate of change of the wireless link. It is well known that changes in channel reuse gain are caused by changes in UE mobility and/or propagation environment, which cause the Doppler effect. The rate of change of the channel is measured based on the channel coherence time ( _Tc ), which is inversely proportional to the maximum Doppler shift. In order for DIDO transmission to proceed reliably, the delay caused by CSI feedback must be a fraction of the channel coherence time (e.g., 1/10 or less). In one embodiment, the delay on the CSI feedback loop is measured, that is, the time between the time when the CSI training is sent and the time when the precoded data is demodulated on the UE side, as shown in Figure 58.

In a frequency division duplex (FDD) DIDO system, BTSs 5310-5314 send CSI training to UE 5301, which estimates CSI and provides feedback to the BTS. The BTS then sends the CSI to CP 5340 via the BSN, which calculates the DIDO precoded data streams and sends them back to the BTS via BSN 5330. Finally, the BTS sends the precoded streams to the UE, which demodulates the data. Referring to Figure 58, the total delay of the DIDO feedback loop is given by

2*T _DL +T _UL +T _BSN +T _CP

Where T _DL and T _UL include the time to construct, send, and process the downlink and uplink frames, respectively, T _BSN is the round-trip delay on the BSN, and T _CP is the time it takes the CP to process the CSI, generate the precoded data stream for the UE, and schedule the different UEs for the current transmission. In this case, T _DL is multiplied by 2 to account for the training signal time (from the BTS to the UE) and the feedback signal time (from the UE to the BTS). In time division duplexing (TDD), if channel reciprocity can be exploited, the first step (i.e., transmitting the CSI training signal from the BTS to the UE) is skipped when the UE sends CSI training to the BTS, which calculates the CSI and sends it to the CP. Therefore, in this embodiment, the total delay of the DIDO feedback loop is T _DL + T _UL + T _BSN + T _CP

The latency, T _BSN , depends on whether the BSN is a dedicated cable, DSL, fiber-optic connection, or general internet. Typical values can range from 1 millisecond to 50 milliseconds. If DIDO processing is implemented at the CP on a dedicated processor (such as an ASIC, FPGA, DSP, CPU, GPU, and/or GPGPU), the computation time at the CP can be reduced. Furthermore, if the number of BTSs 5310-5314 exceeds the number of UEs 5301, all UEs can be served simultaneously, eliminating the latency caused by multi-user scheduling. Therefore, the latency, T _{CP ,} is negligible compared to T _BSN . Finally, the transmit and receive processing for both the DL and UL are typically implemented on an ASIC, FPGA, or DSP, where the computation time is negligible. Furthermore, if the signal bandwidth is relatively large (e.g., greater than 1 MHz), the frame duration can be very short (i.e., less than 1 millisecond). Therefore, T _DL and T _UL are also negligible compared to T _BSN .

In one embodiment of the present invention, CP 5340 tracks the Doppler speed of all UEs 5301 and dynamically assigns the BTS 5310-5314 with the lowest _BSN to the UE with higher Doppler. This adaptation is based on different criteria:

Type of BSN : For example, dedicated fiber links typically experience lower latency than cable modems or DSL. Lower-latency BSNs are used for high-mobility UEs (e.g., cars on highways, trains), while higher-latency BSNs are used for fixed wireless or low-mobility UEs (e.g., home devices, pedestrians, and cars in residential areas).

Type of QoS : For example, a BSN can support different types of DIDO or non-DIDO communications. Different priorities of Quality of Service (QoS) can be defined for different communication types. For example, the BSN can assign high priority to DIDO communications and low priority to non-DIDO communications. Alternatively, high-priority QoS can be assigned to communications for high-mobility UEs and low-priority QoS can be assigned to UEs with low mobility.

Long-term statistics : For example, traffic on a BSN can vary significantly depending on the time of day (e.g., home use at night, office use during the day). Higher traffic loads result in higher latency. Consequently, at different times of the day, a BSN with higher traffic loads results in higher latency for low-mobility UEs, while a BSN with lower traffic loads results in lower latency for high-mobility UEs.

Short-term statistics : For example, any BSN may be affected by temporary network congestion, resulting in higher latency. However, the CP can adaptively select BTSs from the congested BSNs (if the congestion results in higher latency) for low-mobility UEs and use the remaining BSNs (if their latency is lower) for high-mobility UEs.

In another embodiment of the present invention, the BTSs 5310-5314 are selected based on the Doppler experienced on each individual BTS-UE link. For example, in the line-of-sight (LOS) link B in FIG59 , the maximum Doppler shift is a function of the angle (φ) between the BTS-UE link and the vehicle speed (v) according to the well-known equation

Where λ is the wavelength corresponding to the carrier frequency. Therefore, in an LOS channel, the Doppler shift of link A in Figure 59 is maximum, and the Doppler shift of link C is close to zero. In non-LOS (NLOS), the maximum Doppler shift depends on the direction of the multipath around the UE, but generally speaking, due to the distributed nature of BTSs in a DIDO system, some BTSs will experience a higher Doppler for a given UE (e.g., BTS 5312), while other BTSs will experience a lower Doppler for a given UE (e.g., BTS 5314).

In one embodiment, the CP tracks the Doppler velocity on each BTS-UE link and selects only the link with the lowest Doppler effect for each UE. Similar to the above technique, the CP 5340 defines a "user cluster" for each UE 5301. A user cluster is a group of BTSs with good link quality for the UE (defined based on a certain signal-to-noise ratio, SNR, threshold) and low Doppler (e.g., defined based on a predefined Doppler threshold), as shown in Figure 60. In Figure 60, BTSs 5 to 10 all have good SNRs for UE 1, but only BTSs 6 to 9 experience low Doppler effect (e.g., below a specified threshold).

The CP of this embodiment records all SNR and Doppler values for each BTS-UE link into a matrix and selects a submatrix that meets the SNR and Doppler thresholds for each UE. In the example shown in Figure 61, the submatrix is marked with a green dashed line surrounding _C2,6 , _C2,7 , _C3,9 , _C4,7 , _C4,8 , _C4,9 , and _C5,6 . The DIDO precoding weights for the UE are calculated based on this submatrix. Note that BTS5 and 10 are reachable by UE2, 3, 4, 5, and 7, as shown in the table of Figure 61. However, to avoid interference with UE1 when transmitting to those other UEs, BTS5 and 10 must be turned off or assigned to different orthogonal channels based on conventional multiplexing techniques (such as TDMA, FDMA, CDMA, or OFDMA).

In another embodiment, the adverse effects of the Doppler effect on the performance of the DIDO precoding system are reduced by linear prediction, which is a technique for estimating future complex channel coefficients based on past channel estimates. By way of example and not limitation, different prediction algorithms for single-input single-output (SISO) and OFDM wireless systems are proposed in [7-11]. Knowing the future channel complex coefficients can reduce errors caused by outdated CSI. For example, Figure 62 shows the channel gain (or CSI) at different times: i) t _CTR is the time at which the CTR in Figure 58 receives the CSI from the UE in the FDD system (or equivalently, the BTS estimates the CSI from the UL channel using DL/UL reciprocity in the TDD system); ii) t _CP is the time at which the CSI is delivered to the CP by the BSN; iii) t _BTS is the time at which the CSI is used for precoding on the wireless link. In Figure 62, we observe that due to the delay T _BSN (also shown in Figure 58), the CSI estimated at time t _CTR will be outdated (i.e., the complex channel gain has changed) when used for wireless transmission on the DL channel at time t _BTS . One way to avoid this effect due to Doppler is to run the prediction method at the CP. The CSI estimate available at time t _CTR and CP is delayed by T _BSN /2 due to the CTR-CP delay and corresponds to the channel gain at time t ₀ in Figure 62. The CP then uses all or part of the CSI estimated and stored in memory before time t ₀ to predict the future channel coefficients at time t ₀ + T _BSN = t _CP . If the prediction algorithm has minimal error propagation, the CSI predicted at time t _CP reliably reproduces the channel gain in the future. The time difference between the predicted CSI and the current CSI is called the prediction time domain and, in SISO systems, is typically scaled by the channel coherence time.

In DIDO systems, the prediction algorithm is more complex because it estimates future channel coefficients in both the time and space domains. [12-13] describe linear prediction algorithms that exploit the space-time characteristics of MIMO wireless channels. In [13], it is shown that the performance of prediction algorithms in MIMO systems (measured in terms of mean squared error or MSE) improves for higher channel coherence times (i.e., reduced Doppler effects) and lower channel coherence distances (due to lower spatial correlation). Therefore, the prediction time domain (expressed in seconds) of the space-time approach is proportional to the channel coherence time and inversely proportional to the channel coherence distance. In DIDO systems, the low coherence distance is due to the high spatial selectivity provided by the distributed antennas.

This paper describes prediction techniques that exploit the temporal and spatial diversity of DIDO systems to predict future vector channels (i.e., CSI from the base station to the UE). These embodiments exploit the spatial diversity available in the wireless channel to achieve negligible CSI prediction errors and extend the prediction time domain of any existing SISO and MIMO prediction algorithms. A key feature of these techniques is the utilization of distributed antennas, as they receive uncorrelated complex channel coefficients from distributed UEs.

In one embodiment of the invention, the temporal and spatial predictors are combined with an estimator in the frequency domain to allow CSI prediction over all available subcarriers in a system (such as an OFDM system). In another embodiment of the invention, the DIDO precoding weights (instead of CSI) are predicted based on previous estimates of the DIDO weights.

References

[1]M.Costa, “Writing on dirty paper,” IEEE Transactions on Information Theory, Vol. 29, No. 3, Page(s): 439-441, May 1983.

[2] U.Erez, S.Shamai (Shitz), and R.Zamir, “Capacity and lattice-strategies for cancelling known interference,” Proceedings of International Symposium on Information Theory, Honolulu, Hawaii, Nov. 2000.

[3]M.Tomlinson, “New automatic equalizer employing modulo arithmetic,” Electronics Letters, Page(s): 138-139, March 1971.

[4]H.Miyakawa and H.Harashima, “A method of code conversion for digital communication channels with intersymbol interference,” Transactions of the Institute of Electronic

[5] R.A.Monziano and T.W.Miller, Introduction to Adaptive Arrays, New York: Wiley, 1980.

[6]Guy E. Blelloch, “Introduction to Data Compression”, Carnegie Mellon University Tech. Report Sept. 2010.

[7]A.Duel-Hallen,S.Hu,and H.Hallen,“Long-Range Prediction of FadingSignals,”IEEE Signal Processing Mag.,vol.17,no.3,pp.62–75,May 2000.

[8] A. Forenza and R. W. Heath, Jr., “Link Adaptation and channel Prediction in Wireless OFDM Systems,” in Proc. IEEE Midwest Symp. on Circuits and Sys., Aug 2002, pp. 211–214.

[9] M. Sternad and D. Aronsson, “channel estimation and prediction for adaptive OFDM downlinks [vehicular applications],” in Proc. IEEE Vehicular Technology Conference, vol. 2, Oct 2003, pp. 1283–1287.

[10] D.Schafhuber and G.Matz, “MMSE and Adaptive Prediction of Time-Varying channels for OFDM Systems,” IEEE Trans.Wireless Commun., vol.4, no.2, pp.593–602, Mar 2005.

[11] I.C.Wong and B.L.Evans, “Joint channel Estimation and Prediction for OFDM Systems,” in Proc.IEEE Global Telecommunications Conference, St.Louis, MO, Dec 2005.

[12] M.Guillaud and D.Slock, “A specular approach to MIMO frequency selective channel tracking and prediction,” in Proc.IEEE SignalProcessing Advances in Wireless Communications, July 2004, pp.59–63.

[13] Wong, I.C. Evans, B.L., “Exploiting Spatio-Temporal Correlations in MIMO Wireless channel Prediction”, IEEE Globecom Conf., pp. 1-5, Dec. 2006.

Embodiments of the present invention may include the various steps described above. These steps may be embodied as machine-executable instructions that cause a general-purpose or special-purpose processor to perform certain steps. For example, the various components within the aforementioned base station/AP and client device may be implemented as software executed on a general-purpose or special-purpose processor. To avoid obscuring relevant aspects of the present invention, various well-known personal computer components, such as computer memory, hard drives, input devices, etc., are not shown in the figure.

Alternatively, in one embodiment, the various functional modules and associated steps shown herein may be performed by specific hardware components that contain hard-wired logic for performing the steps, such as an application specific integrated circuit ("ASIC"), or by any combination of programmed computer components and custom hardware components.

In one embodiment, certain modules, such as the encoding, modulation, and signal processing logic unit 903 described above, may be implemented on a programmable digital signal processor ("DSP") (or DSP group), such as a DSP using the Texas Instruments TMS320x architecture (e.g., TMS320C6000, TMS320C5000, etc.). The DSP in this embodiment may be embedded in an add-in card (such as a PCI card) for a personal computer. Of course, a variety of different DSP architectures may be used while still complying with the basic principles of the present invention.

The elements of the present invention can also be provided as a machine-readable medium for storing machine-executable instructions. Machine-readable media may include, but are not limited to, flash memory, optical discs, CD-ROMs, DVD ROMs, RAMs, EPROMs, EEPROMs, magnetic or optical cards, propagation media, or other types of machine-readable media suitable for storing electronic instructions. For example, the present invention can be downloaded as a computer program that can be transferred from a remote computer (e.g., a server) to a requesting computer (e.g., a client) via a communication link (e.g., a modem or a network connection) in the form of a data signal, the data signal being embodied as a carrier wave or other propagation medium.

Throughout the foregoing description, for purposes of explanation, numerous specific details are set forth to provide a deeper understanding of the present systems and methods. However, it will be apparent to one skilled in the art that the systems and methods may be practiced without some of these specific details. Accordingly, the scope and spirit of the present invention should be determined by reference to the following claims.

Furthermore, throughout the foregoing description, numerous publications are cited to provide a more thorough understanding of the present invention. All of these cited references are incorporated herein by reference.

Claims (78)

1. A distributed antenna system, comprising: Multiple distributed antennas are communicatively coupled to one or more centralized processors via a network. The centralized processors select multiple subsets of the distributed antennas based on the Quality of Service (QoS) across the network to communicate wirelessly with multiple subsets of users, or use linear prediction to estimate Channel State Information (CSI) between the distributed antennas and the users to compensate for the Doppler effect. 2. The system according to claim 1, wherein the Doppler effect is caused by user movement or changes in the propagation environment. 3. In the system according to claim 1, the plurality of user subsets equals one user. 4. In the system according to claim 1, the plurality of user subsets are equal to all users. 5. The system of claim 1, wherein one or more subsets of the distributed antennas are equal to all of the distributed antennas. 6. The system of claim 1, wherein the first distributed antenna subset serving a first plurality of users includes a second distributed antenna subset serving a second plurality of users. 7. The system of claim 1, wherein the first distributed antenna subset serving the first plurality of users does not include the second distributed antenna subset serving the second plurality of users. 8. The system of claim 1, wherein a plurality of subsets of the distributed antennas communicate with the plurality of users at different times. 9. The system of claim 1, wherein a plurality of subsets of the distributed antennas communicate with the plurality of users at different frequencies. 10. The system of claim 1, wherein a plurality of subsets of the distributed antennas communicate with the plurality of users at different spatial locations. 11. The system of claim 1, wherein different subsets of the distributed antennas are assigned different quality of service metrics. 12. The system of claim 1, wherein different subsets of the distributed antennas are assigned different data rates, reliability, or delays. 13. The system of claim 1, wherein the distributed antenna system is reconfigured to compensate for the Doppler effect attributable to user movement or changes in the propagation environment. 14. The system of claim 1, employing a distributed antenna that utilizes spatial, polarization, and/or pattern diversity to improve data rate and/or coverage for one or more users in the wireless system. 15. The system of claim 1, wherein the user is located around, between, or surrounded by the distributed antennas. 16. The system of claim 1, wherein the distributed antenna system employs multiple composite weights at the receiver of the uplink channel to demodulate multiple independent data streams from the user, wherein the data streams include data or channel state information (CSI). 17. The system of claim 16, wherein the composite weight of the uplink receiver is derived from a plurality of downlink precoded weights or calculated via a maximum ratio combining receiver. 18. A multi-user multi-antenna system (MU-MAS), comprising: Multiple users; Multiple distributed transceiver stations or antennas are communicatively coupled to the user via multiple wireless links; One or more centralized units are communicatively coupled to the plurality of distributed transceiver stations or antennas via a network; The network includes wired links or wireless links or a combination of both used as backhaul communication channels. The one or more centralized units communicate with the distributed transceiver station or antenna via the network to adaptively reconfigure communication between the distributed transceiver station or antenna and the user to compensate for the Doppler effect attributable to changes in the user's movement or propagation environment. 19. The system of claim 18, wherein the centralized unit is a centralized processor (CP), the network is a base station network (BSN), and the distributed transceiver station is a distributed transceiver base station (BTS). 20. The system of claim 19, wherein the centralized processor CP and the distributed transceiver base station BTS are equipped with encoders/decoders to compress/decompress information exchanged between them via the base station network BSN. 21. The system of claim 19, wherein the centralized processor CP adaptively selects the distributed transceiver base station BTS for a low-mobility UE or a high-mobility UE based on latency on the base station network BSN. 22. The system of claim 21, wherein the adaptation is based on the type of high/low data rate base station network (BSN) or quality of service (QoS) or average or instantaneous communication statistics on the base station network (BSN). 23. The system of claim 22, wherein the average communication statistics are obtained from daytime or nighttime use of different networks, and the instantaneous communication statistics are obtained from temporary network congestion. 24. The system of claim 19, wherein the centralized processor CP adaptively selects the distributed transceiver base station BTS for low-mobility users or high-mobility users based on the Doppler velocity of the user link of the distributed transceiver base station BTS. 25. The system of claim 18, wherein communication between the distributed transceiver station and the user comprises multiple precoded data streams transmitted over the wireless link. 26. The system of claim 25, wherein the precoded data stream is obtained from precoded weights calculated based on channel state information (CSI). 27. The system of claim 26, wherein linear prediction is used to estimate the future channel state information (CSI) or the precoding weights, thereby reducing the adverse effects of the Doppler effect on the MU-MAS performance. 28. The system of claim 27, wherein the prediction is employed in the time domain, frequency domain, or spatial domain. 29. A multi-user multiple antenna system (MU-MAS) for compensating for the Doppler effect, comprising at least one centralized unit communicatively coupled to multiple distributed transceiver base stations (BTSs) via a network, the multi-user multiple antenna system comprising: Measuring the Doppler velocity of the first mobile user relative to the plurality of distributed transceiver base stations (BTS); and The centralized unit communicates with the distributed transceiver base station (BTS) via the network to adaptively reconfigure communication between the BTS and the user. The adaptive reconfiguration includes dynamically assigning the first mobile user to the first BTS or the first group of BTSs based on the measured Doppler velocity of the first BTS or the first group of BTSs relative to the other BTSs. 30. The system of claim 29, wherein if the first mobile user has a relatively higher measured Doppler velocity compared to the second mobile user, then dynamic allocation includes allocating the first mobile user to a first distributed transceiver base station (BTS) and allocating the second mobile user to a second distributed transceiver base station (BTS), wherein the first distributed transceiver base station (BTS) has a relatively lower associated latency compared to the second distributed transceiver base station (BTS). 31. The system of claim 30, wherein the delay includes (a) the time spent transmitting a first training signal from the first mobile user to a distributed transceiver base station (BTS), (b) the round-trip delay of connecting the BTS to the base station network (BSN) of a centralized processor (CP), and (c) the time spent by the centralized processor (CP) processing multiple channel state information (CSI) values of the radio channel between the BTS and the first mobile user, generating a precoded data stream for the first mobile user based on the CSI values, and scheduling transmissions to different mobile users, including the first mobile user for the current transmission. 32. The system of claim 31, wherein the delay further includes the time spent transmitting the second training signal from the distributed transceiver base station (BTS) to the first mobile user. 33. The system of claim 29, wherein dynamic allocation further comprises allocating based on a combination of the link quality of the communication channel between each distributed transceiver base station BTS and the first mobile user and the measured Doppler velocity associated with each distributed transceiver base station BTS relative to the first mobile user. 34. The system of claim 33, wherein for a given Doppler velocity, a distributed transceiver base station (BTS) with relatively high link quality is selected. 35. The system of claim 33, wherein, for a given link quality, a distributed transceiver base station (BTS) with a relatively low Doppler speed is selected. 36. The system of claim 31, wherein the centralized processor CP estimates multiple future complex channel coefficients based on multiple past complex channel coefficients to compensate for the adverse effects of the Doppler effect on communication between the distributed transceiver base station BTS and the first mobile user. 37. The system of claim 36, wherein linear prediction is used for estimation. 38. The system of claim 31, wherein the first mobile user is dynamically assigned to the first distributed transceiver base station BTS based on both the Doppler velocity and channel state information (CSI) that define the communication channel quality between the mobile user and each of the plurality of distributed transceiver base stations (BTSs). 39. The system of claim 38, wherein the centralized processor CP performs the following additional operations: Construct a matrix of Doppler velocity and link quality for each of the plurality of distributed transceiver base stations (BTSs) relative to the first mobile user; and Select a distributed transceiver base station (BTS) with a Doppler speed below a specified threshold and a link quality above a specified threshold. 40. A method performed in a distributed antenna system, the system comprising a plurality of distributed antennas communicatively coupled to one or more centralized processors via a network, the method comprising: Multiple subsets of the distributed antennas are selected based on the Quality of Service (QoS) across the network to communicate wirelessly with multiple subsets of users, or linear prediction is used to estimate the Channel State Information (CSI) between the distributed antennas and the users to compensate for the Doppler effect. 41. The method of claim 40, wherein the Doppler velocity is caused by user movement or changes in the propagation environment. 42. The method of claim 40, wherein the plurality of user subsets equals one user. 43. The method of claim 40, wherein the plurality of user subsets are equal to all users. 44. The method of claim 40, wherein one or more subsets of the distributed antennas are equal to all of the distributed antennas. 45. The method of claim 40, wherein the first distributed antenna subset serving the first plurality of users includes a second distributed antenna subset serving the second plurality of users. 46. The method of claim 40, wherein the first distributed antenna subset serving the first plurality of users does not include the second distributed antenna subset serving the second plurality of users. 47. The method of claim 40, wherein a plurality of subsets of the distributed antennas communicate with the plurality of users at different times. 48. The method of claim 40, wherein a plurality of subsets of the distributed antennas communicate with the plurality of users at different frequencies. 49. The method of claim 40, wherein a plurality of subsets of the distributed antennas communicate with the plurality of users at different spatial locations. 50. The method of claim 40, wherein different subsets of the distributed antennas are assigned different quality of service metrics. 51. The method of claim 40, wherein different subsets of the distributed antennas are assigned different data rates, reliability, or delays. 52. The method of claim 40, wherein the distributed antenna system is reconfigured to compensate for the Doppler effect attributable to user movement or changes in the propagation environment. 53. The method of claim 40, employing a distributed antenna that utilizes spatial, polarization, and/or pattern diversity to improve data rates and/or coverage for one or more users in a wireless system. 54. The method of claim 40, wherein the user is located around, between, or surrounded by the distributed antennas. 55. The method of claim 40, wherein the distributed antenna system employs multiple composite weights at the receiver of the uplink channel to demodulate multiple independent data streams from the user, wherein the data streams include data or channel state information (CSI). 56. The method of claim 55, wherein the composite weight of the uplink receiver is derived from a plurality of downlink precoding weights or calculated via a maximum ratio combining receiver. 57. A method performed in a multi-user multiple antenna system (MU-MAS), the MU-MAS comprising: Multiple users; Multiple distributed transceiver stations or antennas are communicatively coupled to the user via multiple wireless links; One or more centralized units are communicatively coupled to the plurality of distributed transceiver stations or antennas via a network; The network includes wired links or wireless links or a combination of both used as backhaul communication channels. The method includes: the one or more centralized units communicating with the distributed transceiver station or antenna to adaptively reconfigure communication between the distributed transceiver station or antenna and the user to compensate for the Doppler effect attributable to changes in the user's movement or propagation environment. 58. The method according to claim 57, wherein the centralized unit is a centralized processor (CP), the network is a base station network (BSN), and the distributed transceiver station is a distributed transceiver base station (BTS). 59. The method of claim 58, wherein the centralized processor CP and the distributed transceiver base station BTS are equipped with encoders/decoders to compress/decompress information exchanged between them via the base station network BSN. 60. The method of claim 58, wherein the centralized processor CP adaptively selects the distributed transceiver base station BTS for a low-mobility UE or a high-mobility UE based on latency on the base station network BSN. 61. The method of claim 60, wherein the adaptation is based on the type of high/low data rate base station network (BSN) or quality of service (QoS) or average or instantaneous communication statistics on the base station network (BSN). 62. The method of claim 61, wherein the average communication statistics are obtained from daytime or nighttime use of different networks, and the instantaneous communication statistics are obtained from temporary network congestion. 63. The method of claim 58, wherein the centralized processor CP adaptively selects the distributed transceiver base station BTS for low-mobility users or high-mobility users based on the Doppler velocity of the user link of the distributed transceiver base station BTS. 64. The method of claim 57, wherein the communication between the distributed transceiver station and the user comprises a plurality of precoded data streams transmitted over the wireless link. 65. The method of claim 64, wherein the precoded data stream is obtained from precoded weights calculated based on channel state information (CSI). 66. The method of claim 65, wherein linear prediction is used to estimate the future channel state information (CSI) or the precoding weights, thereby reducing the adverse effects of the Doppler effect on the MU-MAS performance. 67. The method of claim 66, wherein the prediction is employed in the time domain, frequency domain, or spatial domain. 68. A method for execution in a multi-user multiple antenna system (Mu-MAS) comprising at least one centralized unit communicatively coupled to a plurality of distributed transceiver base stations (BTSs) via a network, the method comprising: Measuring the Doppler velocity of the first mobile user relative to the plurality of distributed transceiver base stations (BTS); and The centralized unit communicates with the distributed transceiver base station (BTS) via a network to adaptively reconfigure communication between the BTS and the user. The adaptive reconfiguration includes dynamically assigning the first mobile user to the first BTS or the first group of BTSs based on the measured Doppler velocity of the first BTS or the first group of BTSs relative to the other BTSs. 69. The method of claim 68, wherein if the first mobile user has a relatively higher measured Doppler velocity compared to the second mobile user, then dynamic allocation includes allocating the first mobile user to a first distributed transceiver base station (BTS) and allocating the second mobile user to a second distributed transceiver base station (BTS), wherein the first distributed transceiver base station (BTS) has a relatively lower associated latency compared to the second distributed transceiver base station (BTS). 70. The method of claim 69, wherein the delay includes (a) the time taken to transmit the first training signal from the first mobile user to the distributed transceiver base station BTS, (b) the round-trip delay on the base station network (BSN) of the centralized processor (CP) connecting the distributed transceiver base station (BTS) to the base station network (BSN), and (c) the time spent by the centralized processor (CP) in processing multiple channel state information (CSI) information of the radio channel between the distributed transceiver base station (BTS) and the first mobile user, generating a precoded data stream for the first mobile user based on the channel state information (CSI), and scheduling transmissions to different mobile users, including the first mobile user for the current transmission. 71. The method of claim 70, wherein the delay further includes the time spent transmitting the second training signal from the distributed transceiver base station (BTS) to the first mobile user. 72. The method of claim 68, wherein dynamic allocation further comprises allocating based on a combination of the link quality of the communication channel between each distributed transceiver base station BTS and the first mobile user and the measured Doppler velocity associated with each distributed transceiver base station BTS relative to the first mobile user. 73. The method of claim 72, wherein for a given Doppler velocity, a distributed transceiver base station (BTS) with relatively high link quality is selected. 74. The method of claim 72, wherein, for a given link quality, a distributed transceiver base station (BTS) with a relatively low Doppler speed is selected. 75. The method of claim 70, wherein the centralized processor CP estimates future complex channel coefficients based on past complex channel coefficients to compensate for the adverse effects of the Doppler effect on communication between the distributed transceiver base station BTS and the first mobile user. 76. The method of claim 75, wherein linear prediction is used for estimation. 77. The method of claim 68, wherein the first mobile user is dynamically assigned to the first distributed transceiver base station BTS based on both the Doppler velocity and channel state information (CSI) that define the communication channel quality between the mobile user and each of the plurality of distributed transceiver base stations BTS. 78. The method of claim 70, wherein the centralized processor CP performs the following additional operations: Construct a matrix of Doppler velocity and link quality for each of the plurality of distributed transceiver base stations (BTSs) relative to the first mobile user; and Select a distributed transceiver base station (BTS) with a Doppler speed below a specified threshold and a link quality above a specified threshold.