HK1077435B - Resource allocation for mimo-ofdm communication systems - Google Patents

Description

Resource allocation for MIMO-OFDM communication systems

Background

FIELD

The present invention relates generally to data communication, and more specifically to techniques for allocating resources within a multiple-input multiple-output (MIMO) communication system using orthogonal frequency division multiplexing (i.e., a MIMO-OFDM system).

FIELD

Multiple Input Multiple Output (MIMO) communication systems using multiple (N)_T) Transmitting antenna and a plurality of (N)_R) The receive antennas are used for multiple independent data stream transmissions. In a MIMO system implementation, all data streams are used for communication between a multi-antenna base station and a single multi-antenna terminal at any given moment. However, in a multiple access communication system, the base station also concurrently communicates with a plurality of terminals. In this case, each terminal uses a sufficient number of antennas so that it can transmit and/or receive one or more data streams.

Multiple base stationsThe RF channel between the antenna array and the multiple antenna arrays at a given terminal is referred to as a MIMO channel. N is a radical of_TA transmission sum N_RThe MIMO channel formed by the receiving antennas can be decomposed into N_SA separate channel of which N_S≤min{N_T，N_R}。N_SEach of the individual channels is also referred to as a spatial subchannel of the MIMO channel and corresponds to a dimension. MIMO systems may provide improved performance (e.g., increased transmission capacity) if the additional dimensionalities created by the multiple transmit and receive antennas are utilized.

Wideband MIMO systems typically experience frequency selective fading, which is characterized by different amounts of fading over the system bandwidth. This frequency selective fading causes inter-symbol interference (ISI), a phenomenon in which each symbol in a received signal interferes with a subsequent symbol in the received signal. This distortion degrades performance by affecting the ability to correctly detect the received symbols.

Orthogonal Frequency Division Multiplexing (OFDM) may be used for ISI resistance and/or other purposes. An OFDM system effectively partitions the overall system bandwidth into multiple (N)_F) Frequency subchannels, which may be referred to as frequency sub-bands or frequency bins. Each frequency subchannel is associated with a respective subcarrier on which data is modulated. The frequency subchannels of an OFDM system may also experience frequency selective fading, depending on the characteristics of the propagation path (e.g., multipath characteristics) between the transmitting and receiving antennas. In OFDM, ISI due to frequency selective fading can overcome its effect by repeating a portion of each OFDM symbol (i.e., adding a cyclic prefix to each OFDM symbol), as is known in the art.

For MIMO systems using OFDM (i.e., MIMO-OFDM systems), for N of the MIMO channels_SEach of the spatial subchannels has N_FAnd available frequency subchannels. Each frequency subchannel of each spatial subchannel may be referred to as a transmission channel. Up to N_F·N_SMultiple transmission channels may be used at any time for communication between a multi-antenna base station and a multi-antenna terminal.

Between the base station and each terminalTypically experience different link characteristics and are therefore associated with different transmission capacities. Moreover, each spatial subchannel may further be subject to frequency selective fading, wherein the frequency subchannels may also be associated with different transmission capacities. Thus, the transmission channels available to each terminal may have different effective capacities. If N is present_F·N_SEfficient use of available resources and higher throughput may be achieved by efficiently allocating the available transmission channels such that they are used by a "suitable" set of one or more terminals in the MIMO-OFDM system.

Accordingly, there is a need in the art for techniques to allocate resources in a MIMO-OFDM system to provide high system performance.

SUMMARY

Techniques are provided herein for scheduling data transmissions for a terminal on the downlink and/or uplink based on the spatial and/or frequency "signature" of the terminal. In a MIMO-OFDM system, each "active" terminal desiring data transmission within an incoming time interval may be associated with a transmission channel with different capacity due to different link conditions experienced by the terminal. Various scheduling schemes are provided herein to select a "suitable" set of one or more terminals for data transmission on each frequency band and to allocate available transmission channels to the selected terminals to achieve system goals (e.g., high throughput, fairness, etc.).

The scheduler may be designed to form one or more sets of terminals for possible (downlink or uplink) data transmission on each of a plurality of frequency bands. Each set includes one or more active terminals and corresponds to a hypothesis to be evaluated. Each frequency band corresponds to one or more groups of frequency subchannels within the MIMO-OFDM system. The scheduler then further forms one or more sub-hypotheses for each hypothesis. For the downlink, each sub-hypothesis may correspond to a particular assignment of multiple transmit antennas at the base station to one or more terminals within the hypothesis. And for the uplink, each sub-hypothesis may correspond to a particular order in which to process uplink data transmissions from one or more terminals within the hypothesis. The performance of each sub-hypothesis is then evaluated (e.g., based on one or more performance metrics, such as a performance metric indicative of the overall throughput of the terminals within the hypothesis). A sub-hypothesis is then selected for each frequency band based on the evaluated performance, and one or more terminals within each selected sub-hypothesis are then scheduled for data transmission on the corresponding frequency band.

The set of one or more terminals scheduled for data transmission (downlink or uplink) on each frequency band may include multiple SIMO terminals, a single MIMO terminal, multiple MISO terminals, or a combination of SIMO, MISO, and MIMO terminals. A SIMO terminal is a terminal scheduled for data transmission over a single spatial subchannel in a MIMO-OFDM system and which uses multiple receive antennas and a single transmit antenna, a MISO terminal is a terminal receiving transmission using a single receive antenna using a single spatial subchannel, and a MIMO terminal is a terminal scheduled for data transmission over two or more spatial subchannels. Each SIMO, MISO, or MIMO terminal may be allocated one or more frequency bands for data transmission. The available transport channels may be assigned to the terminals to achieve the system goal.

Various aspects, embodiments, and features of the invention are described in further detail below. The present invention also provides methods, computer products, altimeters, base stations, terminals, systems and apparatuses implementing various aspects, embodiments and features of the invention, as will be described in detail below.

Brief description of the drawings

The features, nature, and advantages of the present invention will become more apparent from the detailed description set forth below when taken in conjunction with the drawings in which like reference characters identify correspondingly throughout and wherein:

FIG. 1 is a diagram of a MIMO-OFDM system;

fig. 2 is a process flow diagram for terminal scheduling for downlink data transmission;

fig. 3 is a flow diagram of a process for assigning transmit antennas to terminals using the "max-max" criterion;

FIG. 4 is a flow chart of a priority-based downlink scheduling scheme in which N is considered_TThe highest priority terminal set is used for scheduling;

fig. 5 is a process flow diagram for terminal scheduling for uplink transmissions;

FIGS. 6A and 6B are flow diagrams of two successive cancellation receiver processing schemes, wherein the processing order is (1) applied by the ordered set of terminals and (2) determined based on the processed SNR;

FIG. 7 is a flow chart for a priority based uplink scheduling scheme, where N is considered_TThe highest priority terminal set is used for scheduling;

fig. 8A and 8B are block diagrams of a base station and a terminal for downlink and uplink data transmission, respectively;

FIG. 9 is a block diagram of an embodiment of a transmitter unit; and

FIGS. 10A and 10B are block diagrams of two embodiments of receiver processing without and with successive cancellation, respectively;

Detailed Description

Fig. 1 is a diagram of a multiple-input multiple-output communication system (i.e., MIMO-OFDM system) 100 that uses orthogonal frequency division multiplexing. MIMO-OFDM system 100 uses multiple (N)_T) A plurality of transmitting antennas and a plurality of (N)_R) And the receiving antenna is used for data transmission. MIMO-OFDM system 100 may be a multiple-access communication system having one or more Base Stations (BSs) (only shown in fig. 1 for simplicity) that may concurrently communicate with one or more terminals (T) 106. A base station may also be called an access point, UTRAN, or some other terminology, and a terminal may also be called a handset, mobile station, or some other terminology,A remote station, user equipment, or some other terminology.

Each base station 104 uses multiple antennas and represents the Multiple Input (MI) for downlink transmissions from the base station to the terminal and the Multiple Output (MO) for uplink transmissions from the terminal to the base station. The set of one or more "communicating" terminals 106 together represent multiple outputs for downlink transmissions and multiple inputs for uplink transmissions. As used herein, a communication terminal is a terminal that transmits and/or receives user-specific data to/from a base station, and an "active" terminal refers to a terminal that desires downlink and/or uplink data transmission in an upcoming or future time slot. The active terminals may include terminals that are currently communicating.

As with the example shown in fig. 1, the base station 104 concurrently communicates with the terminals 106 a-106 d via multiple antennas available at the base station and one or more antennas available at each communication terminal (as represented by solid lines). Terminals 106e through 106h may receive pilot and/or other signaling information (as represented by the dashed lines) from base station 104, but not transmit or receive user-specific data to or from the base station.

For the downlink, the base station uses N_TOne antenna and one or N for each communication terminal_RMultiple antennas are used to receive one or more data streams from a base station. General formula N_RAny integer of two or more may be used. N is a radical of_TA transmitting antenna and N_RThe MIMO channel formed by the receiving antennas can be decomposed into N_SA separate channel of which N_S≤min{N_T，N_R}. Each such independent channel is referred to as a spatial subchannel of the MIMO channel.

For the downlink, the number of receive antennas at the communication terminal may be equal to or greater than the number of transmit antennas at the base station (i.e., N)_R≥N_T). For such terminals, the number of spatial subchannels is limited by the number of transmit antennas at the base station. Each multi-antenna terminal and base station pass through N of base station_TA transmitting antenna and its own N_RRespective MIMO formed by individual receiving antennasA channel communication. However, even if multiple multi-antenna terminals are selected for downlink data transmission, only N, no matter how many terminals are receiving the downlink transmission_SOne spatial subchannel is available. The terminals considered for downlink data transmission need not be equipped with an equal number of receive antennas.

For the downlink, the number of receive antennas at the communication terminal may also be less than the number of transmit antennas at the base station (i.e., N)_R＜N_T). In particular, MISO terminals use a single receive antenna (N) for downlink transmission_R1). The base station may also communicate with multiple MISO terminals simultaneously using beam steering and Spatial Division Multiple Access (SDMA), as described below.

For the uplink, each communication channel may use a single antenna or multiple antennas for the uplink. Each terminal also utilizes all or a subset of its antennas available for uplink transmission. N of the uplink at any given moment_TThe plurality of transmitting antennas is formed by all antennas used by one or more communication terminals. The MIMO channel is then formed by N from all communication terminals_TN of one transmitting antenna and base station_RA plurality of receiving antennas. The number of spatial subchannels is limited by the number of transmit antennas, which in turn is limited by the number of receive antennas at the base station (i.e., N)_S≤min(N_T，N_R))。

In SDMA, "spatial signatures" associated with different terminals are used to allow multiple terminals to operate simultaneously on the same channel, which may be a time slot, frequency band, code channel, etc. The spatial signature consists of the complete RF signature of the propagation path between each transmit-receive antenna pair used for data transmission. In the downlink, the spatial signature may be derived at the terminal and reported to the base station. The base station may then process the spatial signatures to select terminals for data transmission on the same channel and manually derive "orthogonal" steering vectors for each independent data stream to be transmitted to the selected terminals. On the uplink, the base station may also derive spatial signatures for different terminals. The base station may then process the signatures to schedule terminals for data transmission and further process transmissions from the scheduled terminals to demodulate each transmission separately.

If the terminal is equipped with multiple received antennas, so that N_R≥N_TThen the base station does not need the spatial signature of the terminal to obtain the benefits of SDMA. All that is needed at the base station is information from each terminal indicating the "processed" SNR that is demodulated at the terminal in relation to the signal from each base station transmit antenna. The SNR estimation process may be facilitated by periodically transmitting pilots from each base station transmit antenna, as described below.

As used herein, a SIMO terminal is a terminal designated (or scheduled) to transmit and/or receive data over a single spatial subchannel and which uses multiple receive antennas for data transmission, a MISO terminal is a terminal designated to receive data transmission over a single spatial subchannel and which uses a single receive antenna, and a MIMO terminal is a terminal designated to transmit and/or receive data over multiple spatial subchannels. For the downlink, SIMO terminals may receive data transmissions from a single transmit antenna at a base station, and MISO terminals may transmit over N at the base station_TThe beams formed by the transmit antennas receive data transmissions. And for the uplink, an SIMO terminal may transmit data from one antenna at the terminal.

For MIMO-OFDM systems, each spatial subchannel is further divided into N_FA frequency subchannel. Each frequency subchannel of each spatial subchannel may be referred to as a transmission channel. For downlink and uplink, N_TOne transmit antenna can thus be used at N_F·N_STransmitting up to N on one transmission channel_F·N_SAn independent data stream. Each independent data stream is associated with a particular "rate" that indicates a respective transmission parameter value, such as, for example, a particular data rate, a particular coding scheme, a particular modulation scheme, etc. for the data stream. The rate is typically determined by the capacity of one or more transmission channels used to transmit the data stream.

Multi-user OFDM system

For multiple-access OFDM systems without MIMO capacity, the total system bandwidth W is divided into N_FA plurality of orthogonal frequency sub-channels, each of the sub-channels having a bandwidth of W/N_F. For this system, multiple terminals share the available spectrum through Time Division Multiplexing (TDM). In a "pure" TDM scheme, a single terminal may be allocated the entire system bandwidth W in each fixed time interval, referred to as a timeslot. The terminal may perform data transmission scheduling by allocating time slots on a request basis. Alternatively, for the OFDM system, only N may be allocated within a given time slot_FA portion N of frequency sub-channels_AAssigned to a given terminal with the remainder (N) in the same time slot_F-N_A) One frequency subchannel may be used for other terminals. In this way, the TDM access scheme is converted into a hybrid TDM/FDM access scheme.

Allocating different frequency subchannels to different terminals may provide improved performance of the frequency selective channel. In a pure TDM scheme, where all N are_FWhere multiple frequency subchannels are allocated to a single terminal in a given time slot, some of the frequency subchannels associated with the terminal may experience fading, thus resulting in low SNR and poor throughput for these faded subchannels. However, these same frequency subchannels may have a high SNR for another terminal within the system, since the RF channels may not be correlated for each terminal. If the scheduler knows every active terminal and all N_FSNR of each frequency sub-channel is determined by comparing N_FEach of the frequency subchannels is assigned to the terminal that obtains the best SNR for that subchannel, it is possible to maximize system throughput. In fact, it is generally necessary to meet a certain minimum performance requirement for all terminals, so that the scheduler needs to meet fairness criteria to ensure that the terminals in the best position do not continuously "grab" resources.

The pure TDM scheduling scheme described above may allocate time slots to terminals with the best fading conditions. To improve performance, the scheduler may further consider allocating frequency subchannels to terminals and possibly allocating per-subchannel transmit power in each time slot. The ability to allocate transmit power provides additional scheduling flexibility that can be used to improve performance (e.g., increase throughput).

Single-user MIMO-OFDM system

For MIMO-OFDM systems, N may be used_FFrequency sub-channel at N_STransmitting up to N on each of a plurality of spatial subchannels_FAn independent data stream. The total number of transmission channels is thus N_C＝N_F·N_S. For pure TDM scheme, N_CEach transmission channel is allocated to a single terminal in each time slot.

N_CThe individual transmission channels may be associated with different SNRs and may have different transmission capacities. A portion of the transmission channel may obtain a poor SNR. In one scheme, additional redundancy (e.g., lower code rate codes) may be used for transmission channels with poor SNR to achieve a target Packet Error Rate (PER). The additional redundancy effectively reduces throughput. In another scheme, some or all of the transmission channels with poor SNR may be removed from use and only a subset of the available frequency subchannels is selected for use for each spatial subchannel.

The total available transmit power may be evenly or unevenly distributed over the transmission channel to improve throughput. For example, the total available transmit power for each transmit antenna may be evenly or unevenly distributed over the frequency subchannels selected for use for that transmit antenna. Thus, transmit power is not wasted on some transmission channels that provide little or no information to assist the receiver in recovering the transmitted data. Frequency subchannel selection and power allocation may be implemented on a per transmit antenna basis, where (1) N for each transmit antenna may be selected_FAll or a subset of the frequency subchannels, and (2) the transmit power for each transmit antenna may be evenly or unevenly distributed over the selected frequency subchannels.

The techniques used to process the received signal at the receiver may have an impact on which transmission channels are selected for use. If successive equalization and interference cancellation (or "successive cancellation") receiver processing techniques (described below) are used at the receiver, then it may be desirable to disable certain transmit antennas to increase throughput on the link. In this case, the receiver may determine which subset of transmit antennas should be used for data transmission and provide this information to the transmitter over the feedback channel. If the RF channel experiences frequency selective fading, the set of transmit antennas for one frequency subchannel may not be the optimal set for another frequency subchannel. In this case, the scheduler may select a suitable set of transmit antennas on a per-frequency subchannel basis to improve throughput.

Multi-user MIMO-OFDM system

The various techniques described above are used to: (1) different frequency subchannels are assigned to different terminals in a multi-user OFDM system, and (2) a transmission channel is assigned to a single terminal in a single-user MIMO-OFDM system. The techniques may also be used to allocate resources (e.g., transmission channels and transmit power) to multiple terminals in a multiple-access MIMO-OFDM system. Various scheduling schemes may be designed to achieve high system throughput through these and possibly other multi-user environment techniques.

System resources may be allocated by selecting the "best" set of terminals for data transmission to achieve high throughput and/or some other criteria. In the presence of frequency selective fading, resource allocation may be implemented for one or more frequency subchannel groups. The resource allocation per portion of the overall system bandwidth may provide additional gain in a scheme that attempts to maximize throughput on an overall system bandwidth basis (i.e., as is the case for single carrier MIMO systems).

If the entire system bandwidth is treated as a single frequency channel (e.g., as in a single carrier MIMO system), then the maximum number of terminals that can be scheduled for simultaneous transmission is equal to the number of spatial subchannels, i.e., N_S≤min(N_R，N_T). If the system bandwidth is divided into N_FA number of frequency channels (e.g., as in a MIMO-OFDM system), the maximum number of terminals that can be scheduled for simultaneous transmission is N_F·N_SSince each transmission channel (i.e., each frequency subchannel for each spatial subchannel) may be assigned to a different terminal. And if the system bandwidth is divided into N_GA number of frequency subchannel groups, the maximum number of terminals that can be scheduled for simultaneous transmission is N_G·N_SSince each frequency subchannel group of each spatial subchannel may be assigned to a different terminal. If the number of terminals is less than the maximum number allowed, multiple transmission channels may be allocated to a given terminal.

MIMO-OFDM systems may support various modes of operation. In the MIMO mode, all spatial subchannels of a particular frequency subchannel set are allocated to a single MIMO terminal. Can still pass through N_GMultiple frequency subchannel sets concurrently support multiple MIMO terminals. In N-SIMO mode, N of a particular frequency subchannel group_SThe spatial subchannels are assigned to a plurality of different SIMO terminals, one for each SIMO terminal. A given SIMO terminal may be assigned one or more frequency subchannel groups of a particular spatial subchannel. In the N-MISO mode (which may also be referred to as a multi-user beam steering mode), then N of a particular frequency subchannel group_SThe spatial subchannels are assigned to a plurality of different MISO terminals, each MISO terminal being assigned one spatial subchannel. A full characterization of the transmit-receive antenna paths can be used to derive the different beams of data transmission to these MISO terminals. Similarly, a given MISO terminal may be assigned one or more frequency subchannel groups of a particular spatial subchannel. And in mixed mode, N of a particular group of frequency subchannels_SThe spatial subchannels may be allocated to a combination of SIMO, MISO, and MIMO terminals, and concurrently support multiple types of terminals. Any combination of operating modes may be supported at any particular time slot. For example, a MIMO mode may be supported for the first frequency subchannel group, an N-SIMO mode may be supported for the second frequency subchannel group, an N-MISO mode may be supported for the third frequency subchannel group, a mixed mode may be supported for the fourth frequency subchannel group, and so on. By simultaneousCommunicating with multiple SIMO terminals, multiple MISO terminals, one or more MIMO terminals, or a combination of SIMO, MISO, and MIMO terminals increases the throughput of the system.

If the propagation environment is sufficiently spread, MIMO receiver processing techniques may be used to efficiently utilize the spatial dimension of the MIMO channel to increase transmission capacity. MIMO receiver processing techniques may be used regardless of whether a base station is in simultaneous communication with one or more terminals. For the downlink, the same receiver processing technique may be used to process different signals to the terminal (if a MIMO terminal), or just N, from the terminal's perspective_TOne of the signals (in case of a SIMO terminal). Certain restrictions may apply if successive cancellation receiver processing is used at the terminals, since the data stream assigned to one terminal may not be detected error-free for another terminal. And for the uplink, from the base station's perspective, is processing N from a single MIMO terminal_TA different channel and process from N_TThere is no significant difference between one signal for each of the different SIMO terminals.

As shown in fig. 1, terminals may be randomly distributed within a coverage area (or "cell") of a base station or may be relocated. For wireless communication systems, the link characteristics typically change over time due to a number of factors such as fading and multipath. At a particular moment in time, N_TA transmitting antenna and N_RMIMO channel corresponding usable matrix formed by receiving antennasH(k) Description, its elements consist of independent gaussian random variables, as follows:

for the downlink, N_TThe array of transmit antennas is at a base station and N may be formed at a single SIMO or MIMO terminal (for N-SIMO or MIMO mode) or at multiple MISO terminals (for N-SIMO mode)_RA receiving antenna. And for the uplink, the transmit antenna array may be formed by the antennas used by all communication terminals, and the receive antenna array is at the base station. In the case of the equation (1),H(k) channel response of a MIMO channel being a k-th frequency subchannel setShould be matrix, and h_i，j(k) Is the coupling (i.e., complex gain) between the jth transmit antenna and the ith receive antenna of the kth frequency subchannel set.

Each frequency subchannel group may include one or more frequency subchannels and correspond to a particular frequency band of the overall system bandwidth. Depending on the particular system design, there may be (1) all N present_FOnly one group of frequency subchannels, or (2) N_FGroups, each group having a single frequency subchannel, or (3) at 1 and N_FAny number of groups in between. Number of frequency subchannel groups N_GCan be between 1 and N_FBetween, including 1 and N_F(i.e., 1. ltoreq. N)_G ≤N_F). Each group may include any number of frequency subchannels, and N_GA group may include the same or different number of frequency subchannels. Moreover, each group may include any combination of frequency subchannels (e.g., the frequency subchannels of a group need not be adjacent to each other).

As shown in equation (1), the MIMO channel response for each frequency subchannel set may be represented by a corresponding matrixH(k) Indicating that the matrix has N corresponding to the number of receiving antennas and the number of transmitting antennas_R×N_TAnd (4) each element. Matrix arrayH(k) Each element of (a) describes a response of a corresponding transmit-receive antenna pair of the kth frequency subchannel set. For flat fading channels (i.e. when N is_G1), the overall system bandwidth for each transmit-receive antenna pair (i.e., for all N) may be provided_FOne frequency subchannel) uses a complex value.

In a practical operating environment, the channel response typically varies across the system bandwidth, and a more detailed channel description may be used for the MIMO channel. Thus, for a frequency selective fading channel, a channel response matrix may be provided for each frequency subchannel groupH(k) In that respect Alternatively, a channel impulse response matrix may be provided for a MIMO channelEach element of the matrix corresponds to a sampled impulse response of a transmit-receive antenna pair indicative of the responseThe corresponding sequence of values.

The receiver may periodically estimate the channel response for each transmit-receive antenna pair. Channel estimation may be facilitated in a number of ways, for example, using pilot and/or data-in-the-field decision-directed techniques. The channel estimates may include complex-valued channel response estimates (e.g., gain and phase) for each frequency subchannel set for each transmit-receive antenna pair, as shown in equation (1). The channel estimates provide information of the transmission characteristics (e.g., which data rates are supported) of each spatial subchannel for each frequency subchannel group.

The information given by the channel estimate may also be refined to obtain (1) an estimate of the processed signal-to-noise-and-interference ratio (SNR) for each spatial subchannel of each frequency subchannel group (as described below), and/or (2) some other statistic that enables the transmitter to select the appropriate rate for each individual data stream. This process of deriving critical statistics may reduce the amount of data required to describe the MIMO channel. The complex channel gain and the processed SNR represent different forms of Channel State Information (CSI) reported by the receiver to the transmitter. For a Time Division Duplex (TDD) system, the transmitter can derive or derive some channel state information based on transmissions (e.g., pilots) from the receiver, since there may be sufficient correlation between the downlink and uplink for such a system, as described below. Other forms of CSI may also be derived and reported, as described below.

The overall CSI received from the receiver may also be used to achieve high throughput by allocating a suitable set of one or more terminals to the available transmission channels so that they can be allowed to communicate with the base station at the same time. The scheduler may evaluate which particular combinations of terminals provide the best system performance (e.g., highest throughput) under any system constraints and requirements.

By using the spatial and frequency "signatures" (i.e., estimates of their channel responses, which may be a function of frequency) of a single terminal, the average throughput may be increased relative to that obtained by a single terminal. In addition, by using multi-user diversity, the scheduler can identify combinations of "mutually compatible" terminals that can be allowed to communicate at the same time on the same channel, effectively enhancing system capacity relative to single-user scheduling and random scheduling of multiple users.

Terminal scheduling for data transmission may be based on various factors. A factor set may relate to system constraints and requirements such as desired quality of service (QoS), maximum latency, average throughput, etc. Some or all of these factors may need to be satisfied on a per-terminal (e.g., for each terminal) basis within a multiple-access communication system. Another set of factors may relate to system performance, which may be quantified by an average system throughput or some other indication of performance. These various factors are described in detail below.

For the downlink, the scheduler may (1) select a "best" set of one or more terminals for data transmission, (2) allocate available transmission channels to the selected terminals, (3) allocate transmit power to the allocated transmission channels uniformly or non-uniformly, and (4) determine an appropriate rate for each independent data stream to be transmitted to the selected terminals. For the uplink, the scheduler may (1) select the best set of one or more terminals for data transmission, (2) assign the available transmission channels to the selected terminals, (3) determine the appropriate order in which to process the data streams from these selected terminals (if successive cancellation receiver processing techniques are used at the base station), and (4) determine the rate for each individual data stream from the selected terminals. Various details of resource allocation for the uplink and downlink are described in detail below.

To simplify scheduling, the terminals may be assigned transmission channels (and possibly transmit power) based on their priorities. Initially, the active terminals may be ordered by their priorities, which may be determined based on various factors, as described below. Then consider N within each scheduling interval_XThe highest priority terminal. This enables the scheduler to allocate the available transport channels to only N_XIndividual terminals rather than all active terminals. Resource allocation can be further simplified by (1) selectingSelecting N_X＝N_SAnd all frequency subchannels of one spatial subchannel are allocated to each terminal, or (2) N is selected_X＝N_GAnd all spatial subchannels of a frequency subchannel group are assigned to each terminal, or (3) some other simplification. Even with some simplification, the gain in throughput is important compared to a pure TDM scheduling scheme that allocates all transmission channels to each time slot, especially if N is considered in the resource allocation_XIndependent frequency selective fading for each terminal.

For the sake of brevity, some assumptions are made in the following description. First, it is assumed that the average received power per independent data stream can be adjusted to obtain a specific target energy per bit to total noise plus interference ratio (E) after channel processing by the receiver (terminal for downlink transmission and base station for uplink)_b/N_t). The object E_b/N_tCommonly referred to as a power control set point (or simply set point) and is selected to provide a particular level of performance (e.g., a particular PER). The set point may be obtained by a closed loop power control mechanism that adjusts the transmit power of each data stream (e.g., based on a power control signal from the receiver). For simplicity, a common set point may be used for all data streams received at the receiver. Alternatively, a different set point may be set for each data stream and the techniques described herein may be generalized to cover this mode of operation. Also for the uplink, it is assumed that simultaneous transmissions from different terminals are synchronized such that the transmissions arrive at the base station within a predetermined time window.

For simplicity, for the following description of N-SIMO and MIMO modes, it is assumed that the number of receive antennas is assumed to be equal to the transmit antenna data (N)_R＝N_T). This is not a requirement as the analysis can be applied to N_R≥N_TThe case (1). For the N-SIMO mode, the number of receive antennas per MISO terminal is assumed to be equal to one (i.e., N_R1). And for simplicity, the number of spatial subchannels is assumed to be equal to the number of transmit antennas (i.e., N)_S＝N_T)。

Downlink resource allocation

The resource allocation for the downlink includes (1) selecting one or more terminal sets to evaluate, (2) allocating available transport channels to the terminals in each set and evaluating performance, and (3) identifying the best terminal set and its transport channel allocation. Each set may include multiple SIMO terminals, multiple MISO terminals, one or more MIMO terminals, or a combination of SIMO, MISO, and MIMO terminals. All or a subset of the active terminals may be considered for evaluation, and these terminals may be selected to form one or more sets to be evaluated. Each terminal set corresponds to a hypothesis. For each hypothesis, the available transport channels are assigned to terminals within the hypothesis based on any one of a plurality of channel assignment schemes. The terminals within the best hypothesis may then be scheduled for data transmission in the next upcoming time slot. The flexibility in selecting the best set of terminals for data transmission and assigning transmission channels to the selected terminals enables the scheduler to exploit multi-user diversity environments to achieve high performance in flat-fading and frequency-selective fading channels.

To determine the "optimal" transmission for a set of terminals, the SNR or some other sufficient statistic may be provided for each terminal. For N-SIMO and MIMO modes, where (N)_R≥N_T) Spatial processing may be implemented at SIMO and MIMO terminals to separate the transmitted signals, and the base station does not need the spatial signature of the terminal to transmit multiple data streams simultaneously on the available spatial subchannels. What may be needed at the base station is the processed SNR associated with the signal from each base station transmit antenna. For simplicity, downlink scheduling for SIMO and MIMO terminals is described first, followed by downlink scheduling for MISO terminals.

Downlink scheduling for SIMO and MIMO terminals

Scheduling of SIMO and MIMO terminals may be achieved based on different types of channel state information, including full CSI (e.g., complex channel gain) and partial CSI (e.g., SNR). If it is usedIf the statistic for the scheduled terminals is SNR, then for each set of one or more terminals to be evaluated for data transmission in the incoming time slot, the hypothetical matrix of processed SNRs for that set of terminals for the k-th frequency subchannel groupΓ(k) Can be expressed as:

wherein gamma is_i，j(k) Is the (hypothetical) data stream transmitted from the jth transmit antenna to the ith terminal for the kth frequency subchannel set. N is a radical of_GA matrix of this kindΓ(k) Set of (1) k is not less than N_GThe entire frequency and spatial dimension of the terminal set may be described.

At each terminal within the set to be evaluated, N_TThe data streams may be (hypothetically) transmitted from N of each frequency subchannel group_TOne transmitting antenna is used for transmitting and is N of the terminal_RAnd receiving by the receiving antenna. N is a radical of_RThe received signals at the terminal may be processed using space or space-time equalization to separate the N of each frequency subchannel group_TThe data streams are transmitted as follows. The SNR of the processed data stream (i.e., after equalization) may be estimated and include the processed SNR of the data stream. For each terminal, may be N_TProviding N for each data stream_TPost-processing SNR, N_TThe data stream may be N_GThe terminals of each of the frequency subchannel groups receive.

In N-SIMO mode, a matrix is assumedΓ(k) N of (A)_TThe rows correspond to N of the k-th frequency subchannel group_TN of different terminals_TA SNR vector. In this mode, a matrix is assumedΓ(k) Each row of (a) gives N of the k-th frequency subchannel group from one SIMO terminal_TN of transmitting antenna_TThe SNR of each of the data streams is (assumed). In MIMO mode, the matrix is assumedΓ(k) N of (A)_TA row corresponds to a single SNR vector for a single MIMO terminal. The SNR vector comprises N of the k-th frequency subchannel group_TSNR of each data stream, and N can be replicated_TTo form a matrixΓ(k) In that respect In hybrid mode, for a particular MIMO terminal that may potentially be assigned two or more spatial subchannels of the kth frequency subchannel group, the SNR vector for that terminal may be duplicated such that the SNR vector appears in the hypothesis matrixΓ(k) The number of rows in is equal to the number of spatial subchannels to be allocated to the terminal (i.e. one row per spatial subchannel).

Alternatively, for all modes of operation, a matrix is assumedΓ(k) One row within may be used for each SIMO or MIMO terminal and the scheduler may be designed to flag and evaluate these different types of terminals accordingly. For the following description, it is assumed that the matrixΓ(k) Is assumed to include N_TSNR vector for individual terminals, where SIMO terminals are represented as a single terminal in a matrix and MIMO terminals can be represented as N in a matrix_TTwo or more of the terminals.

If successive cancellation receiver processing techniques are used at the terminal to process the received signal, then for each transmitted data stream for a particular frequency subchannel set, the received processed SNR obtained at the terminal depends on the order in which the transmitted data streams are detected (i.e., demodulated and decoded) to recover the transmitted data, as described below. In this case, multiple sets of SNRs may be provided for each terminal for multiple possible detection orders. Multiple hypothesis matrices may be formed for each frequency subchannel group for each set of terminalsΓ(k) And these matrices can be evaluated to determine which terminal-specific combinations and detection orders provide the best system performance.

In any case, each hypothesis matrixΓ(k) Including the processed SNRs for a given set of frequency subchannels for a particular set of terminals (i.e., hypothesis) to be evaluated. These processed SNRs represent the SNRs available to the terminal and are used to evaluate these hypotheses.

For N-SIMO and MIMO modes, each transmit antenna in the base station's antenna array may use channel state information (e.g., SNR or some other statistic) derived by the terminals in the coverage area for transmitting a different data stream on each frequency subchannel set. High performance can be achieved on the basis of CSI, which is used to schedule terminals and process data.

Various downlink scheduling schemes may be used to allocate resources (e.g., transport channels) to active terminals. These various schemes include (1) an "exhaustive" scheduling scheme that can assign each transport channel to a particular terminal for "optimal" performance as determined by some metric, (2) a priority-based scheduling scheme that assigns transport channels based on the priority of the active terminals, (3) an FDM-TDM scheduling scheme that assigns all spatial subchannels of each frequency subchannel group to a particular terminal, and (4) an SDMA-TDM scheduling scheme that assigns all frequency subchannels of each spatial subchannel to a particular terminal. These various downlink scheduling schemes are described in detail below. Other scheduling schemes may also provide better or near optimal performance and may require less processing and/or statistics, which may also be used and are within the scope of the invention.

Fig. 2 is a flow diagram of a process 200 for terminal scheduling for downlink data transmission. Process 200 may be used to implement various downlink scheduling schemes, as described below. For clarity, the overall process is described first, followed by some step details in the process.

In one embodiment, the transmission channels are assigned to the active terminals by evaluating one frequency subchannel set at a time. The first frequency subchannel set is considered by setting the frequency index k to 1 at step 210. The spatial subchannels of the k-th frequency subchannel group are then allocated to the terminals conducting the downlink transmission starting from step 212. For N-SIMO and MIMO modes for the downlink, allocating spatial subchannels to terminals is equivalent to allocating the transmit antennas of the base station to the terminals, since N is assumed_S＝N_T。

Initially, one or more performance metrics for selecting the best set of terminals for downlink transmission are initialized at step 212. Various performance metrics may be used to evaluate the set of terminals, some of which are described in further detail below. For example, a performance metric that maximizes system throughput may be used.

A new set of one or more active terminals is then selected from all active terminals and considered for transmit antenna assignment, step 214. Various techniques may be used to limit the number of active terminals considered for scheduling, which reduces the number of hypotheses to evaluate, as described below. For each terminal within the hypothesis, the SNR vector is retrieved in step 216γ _i (k)＝[γ_i，1(k)，γ_i，2(k)，...，γ_i，NT(k)]Indicating N in the k-th frequency subchannel group_TThe processed SNRs for the individual transmit antennas. For MIMO mode, a single MIMO terminal is selected for evaluation of the k-th frequency subchannel set, and one SNR vector for that terminal is retrieved. For N-SIMO mode, N_TIndividual SIMO terminals are selected for evaluation and the N for these terminals is retrieved_TA SNR vector. And for mixed mode, SNR vectors are retrieved for SIMO and MIMO terminals within the selected set. For each MIMO terminal within the MIMO and mixed modes, the SNR vectors may be replicated (or substantially marked) such that the number of SNR vectors for the terminal is equal to the number of transmit antennas allocated to the terminal. The SNR vectors for all selected terminals within a hypothesis are used to form a hypothesis matrix shown in equation (2)Γ(k)。

For N_TA transmitting antenna and N_TEach hypothesis matrix of each terminalΓ(k) Having N of_T| A Possible allocation combinations (i.e. N) for allocating transmit antennas to terminals_T| A Sub-hypotheses). Since the MIMO terminal is represented as a matrixΓ(k) A plurality of terminals in the matrix, if assumedΓ(k) There are fewer sub-hypotheses to include one or more MIMO terminals. In any case, a new combination of specific antenna/terminal assignments to be evaluated is selected at step 218. The combination includes assigning to N_TOne antenna for each of the terminals. The antenna assignment may be such that all possible antenna/terminal assignment combinations are ultimately evaluated. Alternatively, a particular scheme may be used for assigning antennas to terminals, as described below. The new combination of antenna/terminal assignments forms the sub-hypothesis to be evaluated.

The sub-hypothesis is then evaluated at step 220 and a performance metric (e.g., system throughput) corresponding to the sub-hypothesis is determined (e.g., based on the SNR of the sub-hypothesis). At step 222, the performance metric corresponding to the best sub-hypothesis is updated to reflect the performance metric of the current sub-hypothesis. In particular, if the performance metric of the current sub-hypothesis is better than the performance metric of the best sub-hypothesis, the current sub-hypothesis becomes the new best sub-hypothesis, and the performance metric, terminal metric, and antenna/terminal assignment corresponding to the sub-hypothesis are saved. The performance and terminal metrics are described below.

It is then determined whether all sub-hypotheses for the current hypothesis have been evaluated at step 224. If all sub-hypotheses have not been evaluated, the process returns to step 218 and another different and unevaluated antenna/terminal assignment combination is selected. Steps 218 through 224 are repeated for each sub-hypothesis to be evaluated.

If all sub-hypotheses for a particular hypothesis are evaluated at step 224, a determination is made at step 226 whether all hypotheses for the current frequency subchannel group are considered. If all hypotheses are not considered, the process returns to step 214 and a different but not considered set of terminals is selected for evaluation. Steps 214 to 226 are repeated for each hypothesis to be considered for the current frequency subchannel set.

If at step 226 all hypotheses for the current frequency subchannel group have been considered, then the best sub-hypothesis for that frequency subchannel group is saved at step 228. The best sub-hypothesis corresponds to a particular set of one or more active terminals that provides the best performance of the frequency subchannel set.

If the scheduling scheme requires maintenance of other system and terminal metrics (e.g., N in the past)_PAverage data rate over a time slot, latency of data transmission, etc.) then these metrics are updated and possibly saved in step 230. Terminal metrics may also be used to evaluate the performance of individual terminals and are described below.

It is then determined whether all frequency subchannels have been allocated for downlink transmission in step 232. If all frequency subchannels have not been allocated, the next frequency subchannel group is considered by incrementing the index k (i.e., k-k +1) at step 234. The process then returns to step 212 to allocate spatial subchannels of the new frequency subchannel set to terminals for downlink transmission. Steps 212 to 234 are repeated for each frequency subchannel to be allocated.

If all frequency subchannel groups are assigned in step 232, the data rate and code modulation scheme for the terminal within the best sub-hypothesis for each frequency subchannel group is determined in step 236 (e.g., based on its post-processing SNR). A schedule may be formed at step 236 and communicated to the terminals (e.g., via a control channel) prior to the scheduled time slot, the schedule indicating the particular active terminal selected for downlink data transmission, its assigned transmission channel, the scheduled time slot, the data rate, the coding and modulation scheme, other information, or any combination thereof. Alternatively, the active terminal may implement a "blind" detection and attempt to detect all transmitted data streams to determine which, if any, data streams are directed to them. Downlink scheduling is typically implemented for each scheduling interval, which corresponds to one or more time slots.

The process illustrated in fig. 2 may be used to implement the various downlink scheduling schemes described above. For an exhaustive scheduling scheme, each available transport channel may be assigned to any active terminal. This can be achieved by considering the following: (1) all possible terminal sets (i.e., all possible hypotheses) for each frequency subchannel group and (2) all possible antenna assignments (i.e., all possible sub-hypotheses) for each terminal set. This scheme may provide the best performance and the maximum flexibility, but requires the most processing for terminal scheduling of downlink data transmission.

For a priority-based scheduling scheme, the active terminals to be considered for transmission channel allocation may be selected based on their priorities, and the performance metric may be a function of the terminal priorities, as described below. This scheme may reduce the number of terminals for which transmission channel allocation is to be considered, which may reduce scheduling complexity. For FDM-TDM scheme, oneThe MIMO channel is allocated all spatial subchannels of each frequency subchannel group. In this case, a matrix is assumedΓ(k) Comprising a single processed SNR vector for one MIMO terminal and only one sub-hypothesis per hypothesis. And for SDMA-TDM scheduling schemes all frequency subchannels of each spatial subchannel are allocated to a single terminal, which may be a SIMO or MIMO terminal. For this scenario, steps 210, 212, 232, and 234 in FIG. 2 may be omitted.

For a given hypothesis matrixΓ(k) The scheduler evaluates various combinations of transmit antennas and terminal pairs (i.e., sub-hypotheses) to determine a hypothesized best antenna/terminal assignment. Various schemes may be used to allocate transmit antennas to terminals to achieve various system goals, such as fairness, high performance, and so on.

In an antenna allocation scheme, all possible sub-hypotheses are evaluated based on a particular performance metric, and the sub-hypothesis with the best performance metric is selected. For each hypothesis matrixΓ(k) There is N that can be evaluated_TFactorial (i.e. N)_T| A ) A possible sub-hypothesis. Each sub-hypothesis corresponds to a particular assignment of each transmit antenna to a particular terminal. Each sub-hypothesis can thus be represented by a processed SNR vector, which can be expressed as:

γ _sub-hyp(k)＝{γ_a，1(k)，γ_b，2(k)，...，γ_r，NT(k)}，(3)

wherein gamma is_i，j(k) Is the post-processing SNR of the data stream for the jth transmit antenna to the ith terminal for the kth frequency subchannel group, and the subscripts { a, b,.. and r } identify the particular terminal within the sub-hypothesized transmit antenna/terminal pair.

Each sub-hypothesis is further associated with a performance metric R_sub-hyp(k) Correlation, which may be a function of various factors. For example, the performance metric based on the processed SNR may be expressed as:

R_sub-hyp(k)＝f(γ _sub-hyp(k))， (4)

where f (.) is a specific true function of the parameter in parentheses.

Various functions may be used to form the performance metrics. In one embodiment, all N of the sub-hypotheses_TThe throughput function achievable for each transmit antenna can be used as a performance metric, which can be expressed as:

wherein gamma is_j(k) Is the throughput associated with the jth transmit antenna within the sub-hypothesis for the kth frequency subchannel group, which can be expressed as:

r_j(k)＝c_j·log₂(1+γ_j(k))， (6)

wherein c is_jIs a portion of theoretical capacity obtained reflecting the coding and modulation scheme selected for the data stream transmitted on the jth transmit antenna, and is_j(k) Is the kth frequency sub-channelThe processed SNR for the jth data stream on the lane group.

To simplify scheduling, resource allocation may be based on multiple frequency subchannel groups rather than a single frequency subchannel. Even if a given group comprises a plurality of frequency subchannels, the frequency-selective nature of the channel response may be taken into account when allocating resources to terminals. This may be achieved by evaluating a performance metric based on the response of the group of frequency subchannels. For example, it may be based on N_kResource allocation is achieved for a group of frequency subchannels, where N_kNot less than 2. In N_kThe channel responses on the individual frequency subchannels may then be used to evaluate the performance metric. If the performance metric is throughput, the sum of the available rates in equation (5) can be implemented over the transmit antennas and frequency subchannels as follows:

wherein gamma is_j(i) Is associated with the jth transmit antenna within the sub-hypothesis for the ith frequency subchannelThroughput of (1), and N_kIs the number of frequency subchannels of the k-th frequency subchannel set. Therefore, even if scheduling and resource allocation are implemented for a plurality of frequency subchannel groups, the performance of individual frequency subchannels within each group may be considered in the scheduling.

The first antenna allocation scheme described above for fig. 2 represents a specific scheme for evaluating all possible allocation combinations of transmit antennas to terminals. The number of potential sub-hypotheses per hypothesis to be evaluated by the scheduler may be up to N_T| A This results in a large number of total sub-hypotheses to be evaluated, since a large number of hypotheses need to be considered.

The scheduling scheme shown in fig. 2 performs an exhaustive search to determine sub-hypotheses that provide "optimal" system performance, which is quantified by a performance metric used to select the best sub-hypothesis. Various techniques may be used to reduce the processing complexity of assigning transmit antennas to terminals. One of these techniques is described below, and others may also be used and are within the scope of the invention. These techniques may also provide high system performance while reducing the amount of processing required to allocate antennas to terminals.

In a second antenna allocation scheme, the antennas are allocated to terminals within the hypothesis to be evaluated using a maximum-maximum (max-max) criterion. Using the max-max criterion, each transmit antenna is assigned to the terminal that achieves the best SNR for that transmit antenna. Antenna allocation may be implemented for each frequency subchannel and one transmit antenna at a time.

Fig. 3 is a flow diagram of a process 218a for allocating transmit antennas to terminals of a particular frequency subchannel group using the max-max criterion. The process 218a is implemented for a particular hypothesis, which corresponds to a particular set of one or more active terminals to be evaluated. Process 218a may be used for step 218 of fig. 2, in which case only one sub-hypothesis is evaluated for each hypothesis in process 200.

Initially, a hypothesis matrix is determined at step 312Γ(k) The maximum SNR of the inner. The maximum SNR corresponds to a particular transmit antenna/terminal pair, and in step314 assign transmit antennas to the terminals. The transmit antennas and terminals are then slaved to the matrix in step 316Γ(k) Is removed and the matrix reduces the dimensionality to (N) by removing the columns corresponding to the transmit antennas and the rows of the terminal just allocated_T-1)×(N_T-1)。

At step 318, it is determined whether all transmit antennas within the hypothesis are assigned. If all transmit antennas are assigned, then an antenna assignment is provided at step 320 and the process terminates. Otherwise, the process returns to step 312 and other transmit antennas are assigned in a similar manner.

Table 1 shows a matrix of terminal-derived SNRs in a 4 × 4MIMO systemΓ(k) An example, where the base station includes four transmit antennas and each terminal includes four receive antennas. For the antenna allocation scheme based on the max-max criterion, the best SNR (16dB) within the original 4x 4 matrix is obtained by the transmit antenna 3 and allocated to terminal 1 as indicated by the shaded box within the third row of the fourth column in the table. The transmit antennas 3 and the terminal 1 are then removed from the matrix. The best SNR (14dB) within the reduced 3 x 3 matrix can be obtained by transmit antennas 1 and 4, which are assigned to terminals 3 and 2, respectively. The remaining transmit antennas 2 are then allocated to terminals 4.

Table 1

Table 2 shows an example matrix within Table 1Γ(k) Antenna allocation using the max-max criterion. For the terminal 1, the best SNR (16dB) is obtained when the processed signal is transmitted from the transmission antenna 3. The best transmit antennas for the other terminals are indicated in table 2. The scheduler then uses this information to select the appropriate coding and modulation scheme to use for data transmission.

Table 2

Terminal device	Transmitting antenna	SNR(dB)
			1	3	16
2	4	14
			3	1	14
4	2	10

Once the max-max criterion is used for a particular hypothesis matrixΓ(k) Having made the antenna assignments, a performance metric (e.g., system throughput) corresponding to the hypothesis can be determined (e.g., based on the SNR corresponding to the antenna assignments), as shown in equations (4) through (6). The performance metric is updated for each hypothesis within the particular frequency subchannel group. When all hypotheses within a frequency subchannel are evaluated, the best set of terminal and antenna assignments is selected for downlink data transmission on the set of frequency subchannels within the incoming time slot. May be N_GEach of the frequency subchannel groups implements scheduling.

The downlink scheduling schemes depicted in fig. 2 and 3 represent a particular scheme that evaluates various assumptions corresponding to various possible sets of active terminals (which may include SIMO and/or MIMO terminals) desiring downlink data transmission in an incoming time slot. The total number of hypotheses evaluated by the scheduler may be large, even for a small number of active terminals. In practice, assume a total number N_hypCan be expressed as:

wherein N is_UIs an active terminal to consider schedulingNumber of the cells. For example, if N_G＝16，N_U8 and N_TIf 4, then N_hyp1120. An exhaustive search may be used to determine the specific hypotheses and specific antenna assignments that provide the best system performance, which is quantified by the performance metric used to select the best hypotheses and antenna assignments.

As described above, other downlink scheduling schemes of lesser complexity may also be implemented. These scheduling schemes may also provide high system performance while reducing the amount of processing required by terminals scheduling downlink data transmissions.

In a priority-based scheduling scheme, active terminals are scheduled for data transmission based on their priorities. The priority of each active terminal may be derived based on one or more metrics (e.g., average throughput), system constraints and requirements (e.g., maximum latency), other factors, or a combination thereof, as described below. A list may be maintained for all active terminals desiring data transmission in an incoming time slot. When the terminal desires a downlink transmission, it is added to the list and its metric is initialized (e.g., to zero). The metric for each terminal in the list is thereafter updated at each time slot. Once the terminal no longer desires data transmission, it is removed from the list.

For each frequency subchannel in each time slot, all or a subset of the terminals in the list may be considered for scheduling. The particular number of terminals to be considered may be selected based on various factors. In one embodiment, only N is selected_TThe highest priority terminal. In another embodiment, consider N within the list_XScheduling by the highest priority terminal, where N_X＞N_T. When N is selected_TOr N_XWhen scheduling is performed for the highest priority terminal, the MIMO terminal may be represented as multiple terminals. For example, if N_T4 and 4 independent data streams are transmitted from the base station for a given frequency subchannel group, one SIMO terminal may be selected along with the MIMO terminal to be assigned three spatial subchannels (in which case the MIMO terminal actually represents the selection of the four most significant data streamsThree of the high priority terminals. )

Fig. 4 illustrates a flow diagram of a priority-based downlink scheduling scheme 400 in which N is considered for each frequency subchannel group_TThe highest priority terminal set. Initially, the first frequency subchannel set is considered by setting the frequency index k to 1 at step 410. The spatial subchannels of the k-th frequency subchannel group may then be allocated to terminals of the downlink transmission starting at step 412.

The scheduler checks the priority of all active terminals in the list and selects N in step 412_TThe highest priority terminal set. The scheduling of the group of frequency subchannels in the scheduling interval is performed irrespective of the remaining active terminals in the list. The channel estimate for each selected terminal is then retrieved, step 414. For example, N may be retrieved_TThe processed SNRs for selected terminals are used to form a hypothesis matrixΓ(k)。

Based on the channel estimates and using any of a plurality of antenna allocation schemes, at step 416N_TA transmitting antenna is allocated to N_TAnd (c) a selected terminal. For example, antenna allocation may be based on the exhaustive search or maximum-maximum criteria described above. In another antenna allocation scheme, transmit antennas are allocated to terminals such that their priorities are normalized as much as possible after updating the terminal metrics.

The data rate and coding and modulation scheme for the terminal are determined at step 418 based on the antenna assignments. The metrics of the scheduled (unscheduled) terminals in the list are updated to reflect the scheduled data transmissions (and corresponding non-transmissions), and then the system metrics are updated at step 420.

It is then determined whether all frequency subchannels have been allocated for downlink transmission in step 422. If all frequency subchannels have not been allocated, the next frequency subchannel group is considered by incrementing the index k (i.e., k-k +1) at step 424. The process then returns to step 412 to assign the spatial subchannels of the new frequency subchannel group to the same or different sets of active terminals. Steps 412 to 424 are repeated for each frequency subchannel to be allocated.

If all frequency subchannel sets are allocated at step 422, a schedule indicating the particular active terminals selected for downlink data transmission, their allocated transmission channels, scheduled time slots, data rates, coding and modulation schemes, etc., and any combination thereof, is formed and communicated to the terminals at step 426. The process then terminates for that scheduling interval.

As described above, the transmit antennas may be assigned to selected terminals for each frequency subchannel group based on various schemes. In an antenna allocation scheme, transmit antennas are allocated to obtain high system performance, and this is based on the priority of the terminal.

Table 3 shows an example of the post-processing SNR derived for a terminal within each hypothesis considered for a particular frequency subchannel group. For terminal 1, the best SNR is obtained when detecting the data stream from transmit antenna 3, as indicated by the shaded box within column 4, column 3, row 3 of the table. The best transmit antennas of the other terminals within the hypothesis are also indicated by the shading within the table.

Table 3

If each terminal identifies a different transmit antenna from which the best processed SNR is detected, that transmit antenna may be assigned to the terminal based on its best processed SNR. For the example shown in table 3, terminal 1 may be assigned to transmit antenna 3 and terminal 2 may be assigned to transmit antenna 2.

If more than one terminal requires the same transmit antenna, the scheduler may determine the antenna allocation based on various criteria (e.g., fairness, performance metrics, etc.). For example, table 3 indicates that there are the best processed SNRs for terminals 3 and 4 for the data streams transmitted from the same transmit antenna 1. If the goal is to maximize throughput, the scheduler may assign transmit antenna 1 to terminal 3 and antenna 2 to terminal 4. However, if antennas are allocated to obtain fairness, transmit antennas may be allocated to terminal 4 if terminal 4 has a higher priority than terminal 3.

Scheduling of MIMO terminals may also be based on full CSI implementation. In this case, the statistics for scheduling the terminal are the complex channel gains between the transmit antennas of the base station and the receive antennas of the terminal, which are used to form the channel response matrix shown in equation (1)H(k) In that respect Scheduling is then implemented such that a mutually compatible set of spatial signatures is selected for each frequency channel group. Based on the channel response matrixH(k) Is described in detail below.

Downlink scheduling for MISO terminals

For N-SIMO mode, where N is_R＜N_TThe complex channel gain between the transmit antennas at the base station and the receive antennas at the terminals may be used to form a channel response matrixH(k) As shown in equation (1) for each set of MISO terminals to be evaluated. The selection of MISO terminals for downlink transmission is then implemented on the active terminals, with the goal of making the spatial signatures compatible with each other across the frequency band.

For downlink in multi-user N-MISO mode, the base station uses N_TN transmit antennas and (for simplicity) considered for downlink scheduling_UEach of the MISO terminals uses a single receiving antenna (i.e., N)_R1). In this case, it is possible to use any given frequency subchannel set (N)_U≤N_T) Up to N simultaneously served by a base station_TAnd (4) a terminal. The MISO channel model for terminal i can be expressed as:

y_i(k)＝H _i(k)x(k)+n_i(k)， (8)

wherein y is_i(k) Is the code received by the ith terminal on the kth frequency subchannel groupElement, to

i∈{1，...，N_U}；

x(k) Is a transmit vector (i.e.x＝[x₁ x₂ ... x_NT]^T) Where { x_jIs the item sent by the jth transmitting antenna, j belongs to {1_TAnd for any matrixM，M ^TTo representMTransposing;

H _i(k) is 1 XN of MISO channels of the ith terminal of the kth frequency subchannel group_TChannel response matrix of which element h_i，jIs the coupling (i.e., complex gain) between the jth transmit antenna and the receive antenna of the kth terminal, where i e {1_UAnd j ∈ { 1., N }_T}; and

n_i(k) additive Gaussian self-noise (AWGN) of the k-th frequency sub-channel group of the ith terminal, with mean 0 and variance σ_i ²；

For simplicity, each frequency subchannel group is assumed to be a flat-fading narrowband channel, which may be represented by a constant complex value. Thus, the elements of the channel response vectorH _i(k) Is a scalar quantity, where i ∈ { 1.,. N_U}. It is further assumed that there is a maximum power limit on each transmit antenna, which is denoted by P_max，jRepresents, j ∈ { 1.,. N_T}. The transmit power on antenna j at any given time is denoted as P_jIn which P is_j≤P_max，j。

N per frequency subchannel_TN transmitted from one transmitting antenna_TEach data stream is based on the channel response vectorH _i(k) Interfering with each other at the receive antennas at each terminal. Without any pre-processing at the base station, different data streams to different MISO terminals may suffer interference, which is called Multiple Access Interference (MAI). Since only one receive antenna is used per MISO terminal, all channel countermeasures need to be implemented at the transmitterAnd spatial processing of MAIs.

If the base station knows the channel response vector of each MISO terminal considered for downlink schedulingH _i(k) (i.e., full channel state information), one technique for removing or reducing MAI is to use Channel Correlation Matrix Inversion (CCMI).

The transmit vector at the base station isx(k)＝[x₁(k) x₂(k) ... x_NT(k)]^TWhere { x_j(k) Is the jth transmit antenna of the kth frequency subchannel set. By d_i(k) Representing the data stream for terminal i, the actual data vector beingd(K)＝[d₁(k) d₂(k) ... d_NU(k)]^TWherein the relationship between the data vector and the transmitted vector can be expressed as:

x(k)＝A(k)S(k)d(k)， (9)

whereinA(k) Is N_T×N_UCCMI matrix of, andS(k) is N_U×N_UThe scaling matrix of (1). The CCMI matrix can be viewed as comprising a plurality of steering vectors, one for each MISO terminal, each steering vector being used to generate a beam for a respective MISO terminal. The CCMI technique decorrelates the data streams of the MISO terminals, anA(k) The solution of (a) can be expressed as:

A(k)＝H ^T(k)(H(k)H ^T(k))^-1， (10)

whereinIs N_U×N_TA matrix containing N of downlink schedules considered within the current hypothesis_UA set of MISO terminals.

A(k) Need not solveH(k) Is a square matrix, which means N_U≠N_T. However, ifH(k) Is a square matrix, the solution in equation (10) can be rewritten asA(k)＝H ^-1(k) WhereinH ^-1(k) Is thatH(k) Is reversed, such thatH ^-1(k)H(k)＝H(k)H ^-1(k)＝IWhereinIIs a unit square matrix with one on the diagonal and the rest zero.

Due to P on each transmitting antenna_max，jIs determined, j e 1,...，N_Tthen may need to be pairedA(k) Is scaled to ensure the power P used on the transmit antenna j_jNot exceeding P_max，j. However, in order to maintainH(k) Rows and columnsA(k) Orthogonality between columns of (1), thenA(k) All entries within each column of (a) need to be scaled with the same value. Scaling is represented by the scaling matrix in equation (9)S(k) This was done in the form:

wherein the scalar value S_i(k) Multiplying by the data stream d_i(k) In that respect Scalar value set { S_i(k) Can be obtained by solving the following set of equations

diag((A(k)S(k))(A(k)S(k))^T)≤P _max(k)， (12)

WhereinP _max，j(k)＝[P_max，1(k)P_max，2(k)ΛP_max，NT(k)]^TAnd P is_max，j(k) Is the maximum power of the kth frequency subchannel group assigned to the jth transmit antenna. Value S_i(k) Can be solved from equation (12) and guarantees P to be used on each transmit antenna on the k-th frequency subchannel group_max，j(k) Not exceeding P_max，j。

Total transmission power P on each transmitting antenna_max，jMay be assigned to N in various ways_GA group of frequency subchannels. In one embodiment, at N_GUniformly distributing the total transmit power over a group of frequency subchannels such that P_j(k)＝P_max，j/N_G. In another embodiment, may be at N_GUnequally allocating total transmit power among frequency subchannel groups while maintainingTotal transmission power P_max，jMay be allocated based on various techniques, including "water filling" or "water-filling" techniques that allocate transmit power such that throughput is maximized. Irrigation techniques are described by Robert G.Gallager in "information therapy and replaceable Communication", John Wiley and Sons, 1968, incorporated herein by reference. A specific algorithm for implementing the basic watering process for a MIMO System is described in U.S. patent application Ser. No. 09/978337, entitled "Method and Apparatus for Determining Power Allocation in a MIMO Communication System", filed on 10/15 of 2001, assigned to the assignee of the present invention and incorporated herein by reference. Transmit power allocation techniques are also described in U.S. patent application No. 10/017308 entitled "Time-Domain Transmit and receive processing with Channel Eigen-mode components for MIMO systems", filed on 12/7/2001, assigned to the assignee of the present invention and incorporated herein by reference. Total transmission power P_max，jFor N, for the optimal allocation of_GN between groups of frequency subchannels_TEach of the individual transmit antennas is typically complex and iterative techniques may be used to solve for the optimal power allocation.

Substituting equation (9) into equation (8), the symbol received by terminal i can be represented as:

y_i(k)＝H _i(k)A(k)S(k)d(k)+n_i(k)， (13)

this can be simplified as:

y_i(k)＝S_i(k)d_i(k)+n_i(k)， (14)

because of the fact thatH _i(k) Except thatA(k) Is orthogonal to all others.

The SNR produced by the ith terminal for the kth frequency subchannel group may be expressed as:

in selecting a set of MISO terminals that have mutually compatible spatial signatures for downlink data transmission on a given set of frequency subchannels, the above analysis may be performed for each set of MISO terminals (i.e., each hypothesis) to be evaluated. The SNR for each terminal in the set may be determined as shown in equation (15). This SNR may be used for performance metrics such as SNR based on throughput shown in equations (5) and (6). Mutual compatibility may thus be defined based on throughput or some other criteria (e.g., the most mutually compatible MISO terminal may be the terminal that achieves the highest overall throughput.)

MISO terminals may also be scheduled for downlink data transmission based on their priority. At the position ofIn this case, the above-described SIMO and MIMO terminal scheduling based on priority may also be applied to the scheduling of the MISO terminal. For example, N may be considered_TThe highest priority MISO terminals perform scheduling for each frequency subchannel group.

Other techniques for generating multiple beams for multiple terminals may also be used and are within the scope of the invention. For example, beam steering may be implemented based on Minimum Mean Square Error (MMSE) techniques. CCMI and MMSE techniques are described in detail in U.S. patent application Ser. Nos. 09/826481 and 09/956449, both entitled "Method and Apparatus for Utilizing Channel State Information in an aWireless Communication System", filed on 3/23 and 9/18, 2001, respectively, assigned to the assignee of the present invention and incorporated herein by reference.

The transmission of data bursts to Multiple terminals based on Spatial signatures is also described in U.S. patent application No. 5515378, entitled "Spatial Division Multiple Access Wireless communication system," filed 5/7 1996, which is incorporated herein by reference.

The beam steering techniques described above may also be used for MIMO terminals.

The ability to schedule MISO terminals on a per-frequency subchannel group basis may improve system performance because the frequency signatures of MISO terminals may be used to select a set of terminals that are compatible with each other for each frequency subchannel group.

The techniques described above may also be generalized to handle the case of SIMO, MISO, and MIMO terminal combinations. For example, if four transmit antennas are available at the base station, there may be four independent data streams sent to a single 4x 4MIMO terminal, two 2 x 4MIMO terminals, four 1 x 4SIMO terminals, four 4x 1MISO terminals, one 2 x 4MIMO terminal and four 1 x 4SIMO terminals, or any other combination of terminals designated to receive a total of four data streams for each frequency subchannel group. The scheduler may be designed to select the best combination of terminals based on the post-processing SNRs for a set of terminals for various hypotheses, where each set of hypotheses may include a mix of SIMO, MISO, and MIMO terminals.

Various metrics and factors may be used to determine the priority of the active terminals. In an embodiment, a "score" may be maintained for each active terminal and each metric used for scheduling. In an embodiment, a score is maintained for each active terminal, the score indicating an average throughput over a particular average time interval. In an implementation, the fraction Φ of terminal i at slot n_i(N) can be represented by_PThe linear average throughput obtained over the previous time slot is calculated and can be expressed as:

wherein r is_i(n) is the "achieved" data rate (in bits per slot) for terminal i at slot n, and may be calculated based on the post-processing SNR shown in equation (6). General r_i(n) subject to a certain maximum obtainable data rate r_maxAnd a limit of a bound for a particular minimum data rate (e.g., zero). In another implementationFraction of terminal i at slot n_i(n) is the exponential average throughput obtained over some time interval and can be expressed as:

φ_i(n)＝(1-α)·φ_i(n-1)+α·r_i(n)/r_max， (17)

where alpha is an exponential averaging time constant, with larger alpha values corresponding to shorter averaging time intervals.

When a terminal desires data transmission, it is added to the active terminal list and its score is initialized to zero. The score for each active terminal in the list may then be updated at each time slot. Whenever an active terminal is not scheduled for transmission in a given time slot, its data rate for that time slot is set to zero (i.e., r)_i(n) ═ 0), and their scores are updated accordingly. If a data packet transmitted in a scheduled time slot is received by a terminal in error, the valid data of the terminal of the time slot may be set to zero. Packet errors may not be immediately known (e.g., because of round-trip latency of an acknowledgement/negative-acknowledgement (Ack/Nak) scheme for data transmission), but once this information is available, the score may be adjusted accordingly.

The priority of the active terminal may also be determined based in part on system constraints and needs. For example, a terminal may be boosted to a higher priority if the maximum latency for a particular terminal exceeds a threshold.

Other factors may also be considered in determining the priority of the active terminals. One such factor may be related to the type of data to be sent to the terminal. Delay sensitive data may be associated with a higher priority and delay insensitive data may be associated with a lower priority. The retransmitted data due to decoding errors within the previous transmission may also be associated with a higher priority because other processes may wait for the retransmitted data at the terminal. Another factor relates to the type of data service provided to the terminal. Other factors may also be considered in determining priority and this is within the scope of the invention.

The priority of an active terminal may thus be any combination of (1) the score maintained for the terminal for each metric to be considered, (2) other parameter values maintained for system constraints and requirements, and (3) other factors. In one embodiment, the system limits and requirements represent "hard" values (i.e., high or low priority depending on whether the constraints and requirements are violated) and the scores represent "soft" values. For this embodiment, terminals for which system constraints and requirements have not been met, and other terminals based on their scores, must be considered immediately.

The priority-based scheduling scheme may be designed such that all active terminals in the list obtain an equal average throughput (e.g., equal quality of service QoS). In this case, the active terminals are prioritized based on the average throughput they obtain, which may be determined as shown in equations (16) and (17). In this priority-based scheduling scheme, the scheduler uses the scores to prioritize the terminals assigned to the available transport channels. The score of the terminal is updated based on its allocation or non-allocation of the transmission channel and may further be adjusted for packet errors. The active terminals in the list may be prioritized such that the terminal with the lowest score is given the lowest priority. Other methods for ordering terminals may also be used. Prioritization may also assign a non-uniform weighting factor to the terminal score.

For a downlink scheduling scheme, in which terminals are selected and scheduled for data transmission based on their priorities, poor terminal grouping may sometimes occur. The "poor" set of terminals results in a similar channel response matrixH(k) The channel response matrix causes a poor SRN for all terminals on all transmitted data streams. This then results in a lower throughput for each terminal in the set and a lower overall system throughput. When this happens, the priority of the terminal may not change over several slots. The scheduler may then keep the particular set of terminals unchanged until there is a sufficient change in terminal priority to cause a change in membership in the set.

To avoid the above-mentioned "clustering" effect, the scheduler may be designed to recognize this situation before assigning the terminals to the available transmission channels, and/or to detect it when it occurs. Multiple schemes may be used to determine the channel response matrixH(k) The degree of linear dependence of. One approach to detecting clustering is to pair hypothesis matricesΓ(k) A specific threshold is applied. If matrixΓ(k) If all or most of the SNRs within the cluster are below the threshold, then the cluster condition is considered to exist. In the event that a cluster is detected, the scheduler may note that the terminal (in a random manner) is attempting to reduce linear dependencies within the hypothesis matrix. The use of a switching scheme is designed to force the scheduler to select the set of terminals that results in a "good" hypothetical matrix (i.e., the set of terminals with the least linear dependence).

Scheduling of terminals for Downlink data transmission and scheduling of terminals based on priority is also described in U.S. patent application nos. 09/859345, 09/539157 and 09/675706, the first of which is entitled "Method and Apparatus for Allocating Downlink Resources in a Multiple-Input Multiple-out (mimo) Communication System" filed on day 5/16 of 2001; a second patent application entitled "Method and Apparatus for controlling Transmission of a Communication System" was filed on 30/3 of 2000, and a third patent application entitled "Method and Apparatus for Determining available Transmission Power in a Wireless Communication System" was filed on 29/9 of 2000, all of which are assigned to the assignee of the present invention and are incorporated herein by reference.

Some downlink scheduling schemes described herein use techniques to reduce the amount of processing required to select terminals to be evaluated and to assign transmission channels to the selected terminals. These and other techniques may also be combined to derive other scheduling schemes, and this is within the scope of the invention. For example, N, which may be considered for scheduling using any of the schemes described above_XThe highest priority terminal.

For the downlink scheduling scheme described above, the total available transmit power for each transmit antenna may be assumed to be allocated evenly across all frequency subchannels selected for downlink data transmission. However, this uniform transmit power allocation is not necessary. Other downlink scheduling schemes may also be devised for selecting terminals for data transmission, allocating transmission channels to the selected terminals and further allocating transmit power to the allocated transmission channels. Some such scheduling techniques are described below.

In a downlink scheduling scheme with non-uniform transmit power allocation, only transmission channels that achieve SNRs above a certain threshold SNR are selected for use, and transmission channels that achieve SNRs below the threshold SNR are not used. The scheme can also be used to remove poor transmission channels with limited transmission capacity by not allocating transmit power to these transmission channels. The total available transmit power may then be distributed evenly or unevenly over the selected transmission channel.

In another downlink scheduling scheme, the transmit power is allocated such that all transmission channels used to transmit each data stream achieve approximately equal SNR. A particular data stream may be transmitted over multiple transmission channels (i.e., over multiple spatial subchannels and/or multiple frequency subchannels) and the transmission channels may achieve different SNRs if they are assigned equal transmit powers. Approximately equal SNRs may be achieved by assigning different amounts of transmit power to the transport channels, which may enable the data streams transmitted on the transport channels to use a single common coding and modulation scheme. In effect, unequal power allocation effects channel inversion on the transmission channel such that they appear similar at the receiver. Channel Inversion for all Transmission channels, and for only selected Transmission channels, is described in U.S. patent application Ser. No. 09/860274 filed on.5/17/2001, U.S. patent application No. 09/881610 filed on.6/14/2001, and U.S. patent application No. 09/892379 filed on.6/26/2001, all entitled "Method and Apparatus for Processing Data for Transmission in a multiple-Channel Communication System Using selected Channel Inversion", assigned to the assignee of the present invention and incorporated herein by reference.

In another downlink scheduling scheme, the allocation of transmit power may be such that each scheduled terminal obtains a desired data rate. For example, terminals with higher priority may be assigned more transmit power and terminals with lower priority may be assigned less transmit power.

In another downlink scheduling scheme, transmit power may be unevenly allocated to achieve high throughput. High system throughput can be achieved by allocating more transmit power to better transport channels and less transmit power to worse transport channels. The "optimal" allocation of the transmission power to the variable-capacity transmission channels can be achieved on the basis of a flooding technique. The scheme of allocating transmit power based on water injection is described in the aforementioned U.S. patent application serial No. 09/978337.

Other downlink schemes may also be implemented to distribute the transmit power in an uneven manner to achieve the desired result and this is within the scope of the invention.

In general, the terminal determines its post-processing SNR from some "hypothesized" power allocation, which may be a fixed power for the pilot transmitted from the base station. Therefore, if the power used for data transmission deviates from the assumed power, the processed SNRs may differ. Since the data rate for the data transmission is based largely on the processed SNR, the actual data rate may be transmitted to the terminal (e.g., within the preamble of the data packet). The terminal may also perform "blind" rate detection and attempt to process data transmissions received at various possible data rates until the data transmission is either correctly received or cannot be recovered error-free for all possible rates. Varying the transmit power within a given spatial subchannel may affect the post-processing SNR of another spatial subchannel within the same frequency subchannel group, and this effect may be taken into account when selecting a terminal for data transmission.

"water-pouring" power allocation can also be used to allocate the available transmit power among the transmission channels to maximize throughput. The water-pouring process may be implemented in various ways, such as (1) on all frequency subchannel groups per spatial subchannel, (2) on all spatial subchannels per frequency subchannel group, (3) on all frequency subchannels of all spatial subchannels, or (4) on some defined set of transmission channels. For example, flooding may be implemented on a set of transport channels for a single data stream to a particular terminal.

If partial CSI is used (e.g., a scheme using post-processing SNR), then there is a per-antenna constraint on transmit power allocation. For the multi-user case, transmit power may be allocated/reallocated (1) among multiple terminals scheduled on the same transmit antenna, (2) among multiple transmission channels allocated to each scheduled terminal (the total power allocated to each terminal is fixed), or (3) based on some other allocation scheme. For a full CSI scheme (e.g., a channel gain based scheme), there is additional flexibility in that transmit power may be reallocated on the transmit antennas (i.e., eigenmodes) as well as on the frequency subchannel sets. The allocation/reallocation of the transmit power over multiple terminals is then done over additional dimensions.

Thus, more complex downlink scheduling schemes can be designed to achieve near optimal throughput. These scheduling schemes may evaluate a number of hypotheses and antenna assignments (and possibly different transmit power assignments) to determine the best set of terminals and the best antenna assignments. Other downlink scheduling schemes may also be designed to take advantage of the statistical allocation of data rates obtained by each terminal. This information may be useful in reducing the number of hypotheses to evaluate. In addition, for some applications, it may be known by analyzing the performance over time which terminal group (hypothesis) works well. This information may then be stored and used by the scheduler for future scheduling intervals.

The techniques described above may be used for scheduling terminals for data transmission in MIMO mode, N-SIMO mode, and mixed mode. Other considerations may also apply to each of these modes of operation, as described below.

In MIMO mode, N from each frequency subchannel group may be simultaneously transmitted by the base station_TOne transmitting antenna transmitting (up to) N_TA separate data stream, the object of which is to carry N_RSingle MIMO terminal with multiple receive antennas (i.e., N)_R×N_TMIMO). The MIMO terminal may use spatial equalization (non-dispersive MIMO channel for flat fading) and space-time equalization (dispersive MIMO channel for flat selective fading) to process and separate N for each frequency subchannel group_TA transmitted data stream. The SNR for each processed data stream may be estimated (i.e., after equalization) and sent back to the base station as channel state information. The base station may then use this information to select an appropriate rate for each data stream so that the MIMO terminal can detect the transmitted data stream at a desired performance level (e.g., target (PER)).

If all data streams are sent to one terminal, as is the case in MIMO mode, successive cancellation receiver processing techniques can be used at that terminal to process N_RReceiving signals to recover N of each frequency sub-channel group_TA transmitted data stream. This technique processes N multiple times (iterations) in succession_RThe received signals are processed to recover the signals transmitted from the base station, one transmitted signal at each iteration. For each iteration, the technique pairs N_RSpatial or space-time equalization is performed on each received signal. A transmitted signal is then recovered, interference due to the recovered signal is estimated, and cancelled from the received signal to derive a "modified" signal with the interference component removed.

The modified signal is then processed by the next iteration to recover another transmitted signal. By removing the interference caused by each recovered signal from the received signal, the SNR of the transmitted signal that is included in the modified signal but is not recovered is improved. The improved SNR results in improved performance of the terminal and the system.

The successive cancellation receiver processing technique is further described in detail in U.S. patent application No. 09/854235 entitled "Method and Apparatus for processing Data in a Multiple-Input Multiple-output (mimo) Communication System using the Channel State Information" filed on day 11/5 2001, and in U.S. patent application serial No. 09/993087 entitled "Multiple-Access Multiple-Input Multiple-output (mimo) Communication System" filed on day 6/11/2001, both assigned to the assignee of the present invention and incorporated herein by reference.

In an embodiment, each MIMO system within the system estimates and transmits N for each frequency subchannel group that may be separately allocated to a terminal_TN of transmitting antenna_TThe post-processing SNR. The SNR from the active terminals can be evaluated by a scheduler to determine which terminals transmit data and when, and the appropriate rate for each data stream transmitted to the selected terminal. MIMO terminals for data transmission may be selected based on specific performance metrics developed to achieve desired system goals. The performance metric may be based on one or more functions and any number of parameters. Various functions may be used to form the performance metric, such as the throughput function available to the MIMO terminal, which is shown in equations (5) and (6) above.

In N-SIMO mode (up to) N_TN independent data streams may be simultaneously transmitted by the base station from each frequency subchannel group_TMultiple transmit antennas transmitting to (up to) N_TAnd different SIMO terminals. To achieve high performance, the scheduler may consider a large set of possible terminals for data transmission. The scheduler then determines the best N_TA set of SIMO terminals transmit simultaneously for each frequency subchannel group. In multiple access communication systems, there are typically limitations in meeting certain requirements on a per-terminal basis, such as maximum latency or average data rate. In this case, the scheduler may be designed to select the best set of terminals that meet these constraints.

In an implementation of an N-SIMO mode, the terminal uses spatial equalization to process the received signal and the post-processing SNR for each data stream is provided to the base station. The scheduler then uses this information to select the active terminals for data transmission and to assign transmission channels to the selected terminals.

In another implementation of the N-SIMO mode, the terminal uses successive cancellation receiver processing to process the received signal to obtain a higher post-processing SNR. With successive cancellation receiver processing, the post-processing SNR of the transmitted data stream depends on the order in which the data streams are detected (i.e., demodulated and decoded). In some cases, a particular SIMO terminal may not be able to cancel interference from a particular data stream to another terminal because the selection of the coding and modulation schemes for the data stream is based on the post-processing SNRs of the other terminals. For example, the transmitted data stream may be targeted to terminal u_xAnd is at the target terminal u_xIs coded and modulated with a certain post-processing SNR (e.g. 10dB), but another terminal u_yThe same transmission data stream can be received but its post-processing SNR is poor and thus the data stream cannot be properly detected. If a data stream to another terminal cannot be detected error-free, the interference caused by the data stream is unlikely to cancel. Successive cancellation receiver processing is only true if the post-processing SNR corresponding to the transmitted data stream can be reliably detected.

The terminal may attempt to use successive cancellation receiver processing on other transmitted data streams than the one to it before processing its own data stream to improve the reliability of the detection. However, in order for the system to take full advantage of this improvement, the base station needs to know the assumed post-processing SNR, in case the interference from the other antennas has been successfully cancelled. Independent constraints on the scheduler may result in data rate assignments for other antennas precluding successful cancellation by the terminal. Thus, there is no guarantee that the base station can select the data rate based on the post-processing SNR derived by the successful cancellation of the receiver processing. However, the base station may use successive cancellation receiver processing on the uplink because it is the intended recipient of all data streams transmitted on the uplink.

In order to enable the scheduler to exploit SIMO terminal usage continuationWith the benefit of improved post-processing SNRs obtained by receiver processing, each such terminal can derive a post-processing SNR corresponding to a different possible detection order for the transmitted data stream. N for each frequency subchannel group_TThe transmitted data streams may be based on N at the SIMO terminal_TFactorial (i.e. N)_T| A ) Possible ranks are detected, and each rank is associated with N_TIndividual processed SNR values are associated. Thus each active terminal may report N to the base station for each subchannel group_TN_T| A SNR value (e.g., if N_TEach SIMO terminal reports 96 SNR values for each frequency subchannel group ═ 4). The scheduler may then use this information to select terminals for data transmission and further assign transmit antennas to the selected terminals.

The scheduler may also take into account the possible detection order of each terminal if successive cancellation receiver processing is used at the terminal. However, a large number of such orderings are generally ineffective because a particular terminal can properly detect data streams transmitted to other terminals because of the lower post-processing SNR achieved at that terminal for the undetectable data streams.

In hybrid mode, the (e.g. MIMO) terminal imposes additional constraints on the scheduler using successive cancellation receiver processing due to the introduced dependencies. These limitations may lead to evaluating more sets of assumptions, because in addition to considering different sets of terminals, the scheduler needs to consider the various orders in which each terminal in a given set demodulates the data stream. The allocation of transmit antennas and the selection of coding and modulation schemes may take these dependencies into account to achieve high performance.

The set of transmit antennas at the base station may be a physically distinct set of "apertures," each of which may be used to directly transmit a respective data stream. Each aperture may be formed by a set of one or more antenna elements, which are distributed in space (e.g., physically located at a single location or distributed over multiple locations). Alternatively, the antenna apertures may precede one or more (fixed) beamforming matrices, each matrix being used to synthesize a different set of antenna beams from the set of apertures. In this case, the above description of the transmit antennas may be equally applied to the switched antenna beams.

For the downlink, multiple fixed beamforming matrices may be defined in advance, and the terminal may evaluate the processed SNR for each possible matrix (or set of antenna beams) and send the SNR vector back to the base station. Different switched antenna beams typically achieve different performance (i.e., post-processing SNR) and this is reflected in the reported SNR vector. The base station may then implement scheduling and antenna allocation (using the reported SNR vectors) for each possible beamforming matrix, and select a particular beamforming matrix and the set of terminals and their antenna allocations that result in the best use of the available resources.

The use of a beamforming matrix allows additional flexibility in terminal scheduling and provides improved performance. As an example, the following applies to beamforming switching:

● the correlation within the MIMO channel is high so that the best performance can be obtained with a smaller number of data streams. However, transmitting with only a subset of the available transmit antennas (and using their associated transmit amplifiers) results in less total transmit power. The switching may be selected to use all or most of the transmit antennas (and amplifiers thereof) of the data stream to be transmitted. In this case, higher transmit power may be obtained for the transmitted data stream.

● the ends of the physical dispersion may be isolated by their location. In this case, the terminal may be transformed by a standard FFT type of horizontally spaced aperture to serve a set of beams pointing at different azimuth angles.

Uplink resource allocation

On the uplink, since the base station is the intended recipient of data transmissions from the scheduling terminal, successive cancellation receiver processing techniques may be used at the base station to process transmissions from multiple terminals. This technique continuously processes N a plurality of times_RRecovering the signals transmitted from the terminals on a received signal basis, one transmitted at a time for each iterationA signal.

When using successive cancellation receiver processing techniques for processing the received signals, the SNR associated with each received data stream is a function of the particular order in which the transmitted signals are processed at the base station. The scheduling scheme may take into account the best set of terminals for uplink data transmission when selecting it.

Fig. 5 is a flow diagram of a process 500 for scheduling terminals for uplink transmissions. In this embodiment, the transmission channels are assigned to the active terminals by evaluating one frequency subchannel group at a time. The first frequency subchannel group is considered by setting the frequency index k to 1 at step 510. The best set of terminals for uplink transmission for the k-th frequency subchannel group is then determined starting at step 512.

Initially, one or more performance metrics used to select the best set of terminals for uplink transmission on the current frequency subchannel set are initialized at step 512. Various performance metrics may be used, such as a performance metric that maximizes system throughput as described above. Also, terminal metrics such as the post-processing SNR of the signal from the terminal, average throughput, etc. may also be used within the evaluation.

In step 514, one or more active terminal sets are then selected from all active terminals desiring to transmit data in the incoming time slot. As described above, the set of active terminals considered for scheduling may be limited (e.g., based on their priorities). The selected set of terminals forms the hypothesis to be evaluated. For each selected terminal, the channel estimate for each transmit antenna used for uplink data transmission is retrieved in step 516. For MIMO mode, a single MIMO terminal is selected for evaluation of the kth frequency subchannel set, and N for that terminal is retrieved_TA transmitting antenna N_TA channel estimate vector. For N-SIMO mode, N is selected_TOne SIMO terminal is used for evaluation and N is retrieved_TN of one transmitting antenna at each of the terminals_TA channel estimate vector. And for mixed mode, combined retrieval for SIMO and MIMO terminals within the setN_TA channel estimate vector. In any case, N is used_TThe channel estimate vectors to form a channel response matrix as shown in equation (1)H(k) Each channel estimation vector corresponds to a matrixH(k) One column of (c). Collectionu(k) Identifying terminals whose channel estimation vectors are included in the channel response matrixH(k) Therein, whereinu(k)＝{u_a(k)，u_b(k)，...，u_NT(k) And MIMO terminals can be represented as setsu(k) A plurality of terminals therein.

When using the successive cancellation receiver processing technique at the base station, the order in which the terminals are processed directly affects their performance. Thus, a processing set is selected at step 518u(k) A specific new order of the internal terminals. This particular order forms the sub-hypotheses to be evaluated.

The sub-hypothesis is then evaluated and terminal metrics for the sub-hypothesis are provided, step 520. The terminal metrics may be from the setu(k) The inner terminal(s) (assuming) the processed SNR of the signal sent to the base station. Step 520 may be obtained based on a successive cancellation receiver processing technique, which is described in fig. 6A and 6B. A performance metric (e.g., system throughput) corresponding to the sub-hypothesis is then determined (e.g., based on the post-processing SNR of the terminal) at step 522. The performance metrics are then used to update the performance metrics of the best sub-hypothesis, also at step 522. In particular, if the performance metric of the current sub-hypothesis is better than the performance metric of the best sub-hypothesis, the current sub-hypothesis becomes the new best sub-hypothesis, and the performance and terminal metrics corresponding to the sub-hypothesis are saved.

It is then determined whether all sub-hypotheses for the current hypothesis have been evaluated at step 524. If all sub-hypotheses have not been evaluated, the process returns to step 518 and a set is selectedu(k) And another different and unevaluated terminal is evaluated in sequence. Steps 518 through 524 are repeated for each sub-hypothesis to be evaluated.

If all sub-hypotheses for the current hypothesis have been evaluated at step 524, a next determination is made at step 526 as to whether all hypotheses have been considered. If all assumptions are not considered, the process returns to step 514 and another different but not considered set of terminals is selected for evaluation. Steps 514 through 526 are repeated for each hypothesis to be considered.

If all hypotheses for the current frequency subchannel set are evaluated at step 526, the best sub-hypothesis results for that frequency subchannel set are saved at step 528. The best sub-hypothesis corresponds to a particular set of one or more active terminals that provide the best performance metric for the group of frequency subchannels. The best sub-hypothesis is further related to the particular receiver processing order at the base station if successive cancellation receiver processing is used at the base station. The stored results may thus include the SNR available to the terminal and the selected processing order.

If the scheduling scheme requires maintenance of other system and terminal metrics (e.g., at the previous N)_PAverage throughput over a time slot, latency for data transmission, etc.) then these metrics are updated for the current frequency subchannel group in step 530. Terminal and system metrics may also be saved.

It is then determined whether all frequency subchannels have been allocated for uplink transmission in step 532. If all frequency subchannels have not been allocated, the next frequency subchannel group is considered by incrementing the index k (i.e., k-k +1) at step 534. The process then returns to step 512 to select the best set of terminals for uplink transmission on the new frequency subchannel group. Steps 512 through 534 are repeated for each frequency subchannel to be allocated.

If all frequency subchannel groups are assigned at step 532, the data rates and coding and modulation schemes for the terminals within the best sub-hypothesis for each frequency subchannel group are determined at step 536 (e.g., based on their post-processing SNRs). A schedule indicating the particular selected terminals and their associated transmission channels and rates may be formed and communicated to the terminals in advance of the scheduled time slots at step 536. Uplink scheduling is typically implemented for each scheduling interval.

FIG. 6A is a flow diagram of a processing scheme 520a for a successive cancellation receiver, where the processing isThe order is applied in a sorted set of terminals. This flowchart may be used for step 520 in fig. 5. The process shown in fig. 6A is implemented for a particular sub-hypothesis, which corresponds to ordering the set of terminals,u(k)＝{u_a(k)，u_b(k)，...，u_NT(k) and (4) dividing. Initially, at step 612, the first terminal in the sorted set is selected as the current terminal to be processed (i.e., u)_i＝u_a(k))。

For the successive cancellation receiver processing technique, the base station first implements at step 614 at N_RSpatial or space-time equalisation over received signals in an attempt to separate setsu(k) A single signal transmitted by the internal terminal. Spatial or space-time equalization may be performed as described below. The amount of available signal separation depends on the amount of correlation between the transmitted signals, and greater signal separation can be achieved if the signal correlations are reduced. Step 614 provides a slave N_RN derived from received signal_TA processed signal and corresponding setsu(k) N sent by internal terminal_TA signal. As part of spatial or space-time processing, a corresponding current terminal u is also determined_iSNR of the processed signal.

In step 616, terminal u_iIs further processed (i.e., "detected") to obtain a decoded data stream for the terminal. The detection may include demodulating, deinterleaving, and decoding the processed signal to obtain a decoded data stream.

At step 618, a determination is made as to whether the collection was processedu(k) All terminals in the network. If all terminals are processed, the SNR for the terminals is provided at step 926 and receiver processing for the ordered set terminates. Otherwise, it is estimated that u is due to the terminal in step 620_iThe transmitted signal causes interference to each received signal. Interference may be based on aggregationu(k) Channel response matrix of inner terminalH(k) Is estimated (e.g., as described below). In step 622, u is terminated_iThe resulting estimated interference may then be subtracted (i.e., cancelled) from the received signal to derive a modified signal. If terminal u_iNot yet transmitted, these modifications areThe signal represents an estimate of the received signal (i.e., assuming that interference cancellation is effectively implemented). The modified signal is used in the next iteration to process the signal from the setu(k) Including the signal transmitted by the next terminal. At step 624, aggregateu(k) The next terminal in the list is then selected as the (new) current terminal u_i. In particular, for ordered setsu(k)＝{u_a(k)，u_b(k)，...，u_NT(k) U for the second iteration_i＝u_b(k) For the third iteration u_i＝u_c(k) Etc., for the last iteration u_i＝u_NT(k)。

Set of processing pairs implemented by steps 614 to 616u(k) The modified signal (rather than the received signal) of each successive terminal within is repeated. Steps 620 to 624 are performed for each iteration, except for the last iteration.

Using successive cancellation receiver processing techniques for N_TEach hypothesis for a terminal, having N_TPossible ordering of factorial (e.g. if N_TIf 4, then N_T| A 24). For each terminal ordering within a particular hypothesis (i.e., for each sub-hypothesis), the successive cancellation receiver processing (step 520) provides a set of SNRs for the processed signals for those terminals, which can be expressed as:

γ _hyp，order＝{γ₁(k)，γ₂(k)，...γ_NT(k)}， (18)

wherein gamma is_i(k) Is the SNR of the k-th frequency subchannel set after processing by the receiver at the ith terminal within the sub-hypothesis.

Each sub-hypothesis is further associated with a performance metric R_hyp，order(k) This may be a function of various factors. For example, the performance metric based on the terminal SNR can be represented as shown in equation (4). In one embodiment, the performance metrics for the sub-hypotheses are setsu(k) All N in_TA function of the achievable throughput of the individual terminals, which can be expressed as shown in equation (5), where the sub-hypothesis is satisfiedOf the ith terminal-related throughput γ_i(k) Can be expressed as shown in equation (6).

The downlink scheduling schemes described in fig. 5 and 6A may be used to evaluate all possible orderings of all possible sets of each active terminal that desires to transmit data on the uplink. The total number of potential sub-hypotheses to be evaluated by the uplink scheduler may be large, even for a small number of active terminals. In practice, the total number of sub-hypotheses may be expressed as:

wherein N is_UIs to consider the number of terminals performing scheduling (again, a MIMO terminal may be represented as a terminal in multiple schedules). For example, if N_G＝16，N_U8 and N_TIf 4, then N_sub-hyp26880. An exhaustive search may be used to determine sub-hypotheses that provide the best system performance for each frequency sub-channel group, as quantified by a performance metric used to select the sub-hypotheses.

Similar to the downlink, a number of techniques may be used to reduce the complexity of terminal scheduling for handling uplink transmissions. Some scheduling schemes based on some of these techniques are described below. Other scheduling schemes may also be implemented and are within the scope of the present invention. These scheduling schemes can provide high system performance while reducing the amount of processing required for scheduling of terminals for uplink data transmission.

In the second uplink scheduling scheme, terminals included in each hypothesis are processed in a specific order determined based on a specific defined criterion. In one embodiment, the scheme relies on successive cancellation receiver techniques to determine the specific order in which the terminals within a hypothesis are processed. For example and as described below, for each iteration, the successive cancellation receiver processing scheme may recover the transmitted signal with the best SNR after equalization. In this case, the processing order is determined based on the post-processing SNR of the intra-hypothesis terminal.

Fig. 6B is a flow diagram of a successive cancellation receiver processing scheme 520B, where the order of processing is determined based on the post-processing SNR. The flowchart may also be used for step 520 of fig. 5. However, since the processing order is determined based on the post-processing SNRs obtained by the successive cancellation receiver processing, only one sub-hypothesis is actually evaluated for each hypothesis, steps 518 and 524 in fig. 5 may be omitted.

Initially, at step 614, spatial or space-time equalization may be performed on the received signal in an attempt to separate the individually transmitted signals. The SNR of the equalized transmitted signal is estimated at step 615. In one embodiment, the corresponding transmitted signal with the best SNR is selected and further processed (i.e., demodulated and decoded) at step 616 to obtain the corresponding decoded data stream. At step 618, a determination is made as to whether all transmitted signals (i.e., all terminals within the hypothesis) have been processed. If all terminals have been processed, the processing order for the terminals and their SNRs are provided at step 628, and the receiver processing for the set of terminals terminates. Otherwise, the interference caused by the just-processed transmitted signal is estimated at step 620 and subtracted (i.e., canceled) from the received signal at step 622 to derive a modified signal. Steps 614, 616, 618, 620 and 622 in fig. 6B correspond to the same numbered steps in fig. 6A.

In the third uplink scheduling scheme, terminals included in each hypothesis are processed based on a specific order. With successive cancellation receiver processing, each iteration improves the SNR of the unprocessed terminals because the interference from each processed terminal is removed. Thus, on average, the first terminal to be processed has the lowest SNR, the second terminal to be processed has the next lowest SNR, etc. Knowing this, the processing order of the terminal can be defined for the hypothesis. The processing order represents another degree of freedom that may be used by the scheduler to achieve system goals and requirements.

In the third uplink scheduling scheme embodiment, the processing order of each hypothesis is selected based on the terminal priority within the hypothesis. For example, the lowest priority terminal within the hypothesis may be processed first, then the next lowest priority terminal, etc., and the highest priority terminal may be processed last. This embodiment allows the highest priority terminal to achieve the highest SNR that is assumed to be possible, which in turn supports the highest possible data rate. In this way, terminals may be assigned transmission channels in a particular order based on their priority such that the highest priority terminal is assigned the highest possible data rate. In another embodiment of the third uplink scheduling scheme, each hypothesized processing order is selected based on user payload, latency requirements, emergency service priority, etc.

In a fourth uplink scheduling scheme, the terminals are scheduled based on their priorities, which may be determined based on one or more metrics (e.g., average throughput), system limitations and requirements (e.g., maximum latency), other factors, or a combination thereof, as described above. For each scheduling interval, multiple highest priority terminals may be considered for scheduling.

Fig. 7 is a flow diagram of a priority-based uplink scheduling scheme 700 in which the N to be scheduled is considered for each frequency subchannel group_TThe highest priority terminal set. Initially, at step 710, a first frequency subchannel group is considered by setting a frequency index k to 1. The spatial subchannels of the k-th frequency subchannel group may then be allocated to the terminals for uplink transmission beginning in step 712.

The scheduler checks the priority of all active terminals in the list and selects N in step 712_TThe highest priority terminal set. The remaining active terminals in the list are not considered for scheduling for the group of frequency subchannels in the scheduling interval. In step 714, the channel estimate for each selected terminal may be retrieved and used to form a channel response matrixH(k)。

Then, at step 716, each sub-hypothesis is derived, and N is then evaluated_TEach sub-hypothesis of the hypothesis formed by the selected terminals, and the corresponding processed SNR,γ _hyp，order(k) and (5) vector quantity. The best sub-hypothesis is selected and the data rates and coding and modulation schemes (e.g., SNRs based thereon) for the terminals within the best sub-hypothesis are determined at step 718. The metrics and system metrics for the active terminals in the list are then updated at step 720.

It is then determined whether all frequency subchannels for the uplink are allocated in step 722. If all frequency subchannels are not allocated, the next frequency subchannel group is considered by incrementing the index k (i.e., k-k +1) in step 724. Processing then returns to step 712 to assign the spatial subchannels of the new frequency subchannel group to the same or a different set of terminals. Steps 712 through 724 are repeated for each frequency subchannel set to be allocated.

If all frequency subchannel groups are assigned in step 722, a schedule may be formed and transmitted to the terminals in step 726, the schedule indicating the selected terminals and their assigned transmission channels and rates. The process is then terminated for the scheduling interval.

Priority-based uplink Scheduling for terminals is described in U.S. patent application Ser. Nos. 09/859346 and 5923650, entitled "Method and Apparatus for Allocating uplink resources in a Multiple-Input Multiple-output (MIMO) communication System", filed on 16/5 2001, and "Method and Apparatus for reverse Link Rate Scheduling", filed on 13/6 1999. These patents and patent applications are assigned to the assignee of the present invention and are incorporated herein by reference.

The same target set point may be used for all data streams received at the base station. However, a common set point for all received data streams is not necessary. Other uplink scheduling schemes may also be designed to select terminals for data transmission, assign a transport channel to the selected terminal, and further select a setpoint for the assigned transport channel. A particular setpoint may be obtained for a data stream through a power control mechanism that directs the terminal to adjust its transmit power of the data stream such that the SNR received by the data stream is approximately equal to the setpoint.

Various uplink scheduling schemes can be devised for non-uniform set points of the data stream transmitted by the scheduled terminals. In one embodiment, higher set points may be used for higher priority terminals and lower set points may be used for lower priority terminals. In another embodiment, the selection of the set point may be such that a desired data rate is obtained for each scheduled terminal. In another embodiment, the set point may be selected to achieve high system throughput by using a higher set point for better transport channels and a lower set point for poorer transport channels. Other schemes for selecting different set points for different transport channels to achieve the desired results may also be implemented and are within the scope of the invention.

Similar to the downlink, it is not necessarily required to use all available transport channels for uplink data transmission. In one embodiment, only transmission channels having an achieved SNR above a certain threshold SNR are selected for use, and transmission channels having an SNR below the threshold are not used.

For many of the above uplink scheduling schemes, successive cancellation receiver processing techniques are used to process the signal received at the base station, which may provide improved SNR for higher throughput. Uplink scheduling may also be achieved without using successive cancellation receiver processing at the base station. For example, the base station may simply use spatial or space-time equalization to process the received signal to recover the transmitted signal. It can be shown that a large gain (i.e., not dependent on successive cancellation receiver processing at the base station) can be obtained by using the multi-user diversity environment and/or the frequency signature of the terminal in the uplink data transmission scheduling.

Other uplink scheduling schemes may also be used and are within the scope of the invention. For the FDM-TDM uplink scheduling scheme, a MIMO terminal may be assigned all spatial subchannels of each frequency subchannel group, and the frequency signature of the terminal may be considered in uplink scheduling to achieve high performance. For SDMA-TDM uplink scheduling schemes, all frequency subchannels of each spatial subchannel may be allocated to a single terminal, which may be a SIMO or MIMO terminal.

Other scheduling considerations

For the downlink and uplink, if partial CSI (e.g., post-processing SNR) is used to schedule terminals for data transmission, a common coding and modulation scheme may be used for all of the transmission channels assigned to a given terminal, or different coding and modulation schemes may be used for each assigned transmission channel. Using the same coding and modulation scheme for all allocated transport channels may simplify processing at the terminal and base station. The scheduler may be designed to take this point into account when scheduling terminals for data transmission on the available transmission channels. For example, transmission channels with similar transmission capacities (e.g., similar SNRs) are preferably assigned to the same terminal, such that a common coding and modulation scheme may be used for the data transmission on the multiple transmission channels assigned to the terminal.

For downlink and uplink, the scheduling scheme may be designed to consider a set of terminals with similar link margins. Terminals may be grouped according to their link margin characteristics. The scheduler may consider combinations of terminals within the same "link margin" group when searching for mutually compatible spatial signatures. Grouping of terminals according to the link margin may improve the overall frequency efficiency of the scheduling scheme compared to the overall spectral efficiency obtained ignoring the link margin. Moreover, by scheduling terminals with similar link margins to transmit concurrently, power control (i.e., over the entire set of terminals) can be simply implemented to improve overall spectral reuse. This can be viewed as a combination of adaptive reuse scheduling and SIMO/MIMO (which relies on spatial processing at the receiver to separate multiple transmitted data streams) or MISO (which relies on beam steering by the transmitter to separate multiple transmitted data streams). Also, a scheduling scheme that evaluates a mixture of the two (beam and margin) may be implemented and is within the scope of the invention.

Scheduling based on link margin and adaptive reuse is further described in detail in U.S. patent application Ser. Nos. 09/532492 and 09/848937, entitled "High Efficiency, High Performance Communications System Employing Multi-Carrier Modulation", filed 30/3/2000, and "Method and Apparatus for controlling uplink Transmissions of a Wireless Communication System", filed 3/5/2001, both assigned to the assignee of the present invention and incorporated herein by reference.

To simplify the description of various scheduling schemes, where (1) N is selected for downlink or uplink transmission for a given frequency subchannel group_TA set of terminals (where MIMO terminals can represent these N_TA plurality of terminals) each assigned to a spatial subchannel, (2) the number of transmit antennas is equal to the number of receive antennas (i.e., N)_T＝N_R) And (3) transmitting an independent data stream on each spatial subchannel on each frequency subchannel group. In this case, the number of data streams per frequency subchannel is equal to the number of spatial subchannels, and N within the set_TEach of the terminals is actually assigned to a respective spatial subchannel.

For the downlink, each scheduled terminal may be equipped with more receive antennas than the total number of data streams. Also, multiple scheduled terminals may share a particular transmit antenna at the base station. Sharing may be by time division multiplexing (e.g., assigning different portions of a time slot to terminals), frequency division multiplexing (assigning different frequency subchannels within each frequency subchannel group to different terminals), code division multiple access (e.g., assigning different orthogonal codes to different terminals), some other multiplexing scheme, or a combination of multiplexing schemes.

For the uplink, scheduled terminals may also share a multiplexed array of receive antennas at the base station. In this case, the total number of transmit antennas for the scheduled terminals may be greater than the number of receive antennas at the base station, and the terminals may share the available transmission channels using other multiple access techniques (e.g., time, frequency, and/or code division multiplexing).

The scheduling scheme described herein selects a terminal and allocates a transmission channel to the selected terminal based on channel state information, which includes a post-processing SNR. The post-processing SNR for the terminal depends on the particular transmit power level used for the data stream. For simplicity, the same transmit power level is assumed for all data streams (i.e., no power control of the transmit power).

However, the achievable SNR can be adjusted by allocating different amounts of transmit power to different data streams and/or by controlling the transmit power of each data stream. For the downlink, by reducing the transmit power for a particular data stream by power control, the SNR associated with that data stream may be reduced, the data stream may not cause a reduction in interference to other data streams, and the other data streams may achieve better SNR. For the uplink, by reducing the transmit power of a particular terminal by power control, the SNR for that terminal is reduced, the interference due to that terminal is also reduced, and other terminals can also obtain better SNR. Power control (and power allocation among them) for multiple terminals sharing non-orthogonal spatial channels simultaneously may be achieved by imposing various constraints to ensure system stability, as described above. Accordingly, transmit power allocation and/or power control may also be used in conjunction with the scheduling schemes described herein and this is within the scope of the invention.

The downlink and uplink scheduling schemes described herein may be designed to support multiple characteristics. First, the scheduling scheme may support mixed mode operation, where any combination of SIMO and MIMO terminals for data transmission may be scheduled on a "channel," which may be a time slot, frequency band, code channel, etc. Second, the scheduling scheme provides scheduling within each scheduling gap based on the spatial and frequency signatures of a group of terminals that includes a set of "mutually compatible" terminals. Considering mutual compatibility means that transmissions coexist on the same channel and at the same time, given the specific constraints of terminal data rate requirements, transmit power, link margin, capacity between SIMO and MIMO terminals, and possibly other factors. Third, the scheduling scheme supports variable data rate adaptation based on the SNR of the processed signal of the terminal. Each scheduled terminal is informed of when to communicate, which data rate to use (e.g., on a per data stream basis), and a specific mode (SIMO, MIMO).

MIMO-OFDM system

Fig. 8A is a block diagram of a base station 104 and two terminals 106 within a MIMO-OFDM system 100 for downlink data transmission. At the base station 104, a data source 808 provides data (i.e., information bits) to a Transmit (TX) data processor 810. For each independent data stream, TX data processor 810(1) encodes data based on a particular coding scheme, (2) interleaves (i.e., reorders) the coded bits based on a particular interleaving scheme, and (3) maps the interleaved bits into modulation symbols for the one or more transport channels selected for use for that data stream. The encoding increases the reliability of the data transmission. Interleaving provides time diversity of the coded bits, enables data to be transmitted over the transmission channel at an average SNR, removes correlation of the coded bits used to form each modulation symbol, and may further provide frequency diversity if the coded bits are transmitted over multiple frequency subchannels. The encoding and modulation (i.e., symbol mapping) may be performed based on control signals provided by a controller 830.

A TX MIMO processor 820 receives and demultiplexes the modulation symbols from TX data processor 810 and provides a stream of symbol vectors, one symbol vector for each symbol period, for each transmit antenna used for the data transmission. Each symbol vector comprising N of the transmit antennas_FUp to N frequency sub-channels_FAnd a modulation symbol. If full CSI processing is implemented (e.g., if the channel response matrix isH(k) Available), TX MIMO processor 820 may further pre-condition the modulation symbols. MIMO and full CSI processing are described in detail in U.S. patent application No. 09/993087. Each vector of symbols is then received and modulated by a respective Modulator (MOD)822 and transmitted via an associated antenna 824.

At each terminal 106 to which a data transmission is directed, the transmitted signal is received by an antenna 852 and the received signal from each antenna is provided to a respective demodulator (DEMOD) 854. Each demodulator (or front end unit) 854 performs processing complementary to that performed at modulator 822. The received modulation symbols from all demodulators 854 are then provided to a Receive (RX) MIMO/data processor 860 and processed to recover one or more data streams transmitted to the terminals. RX MIMO/data processor 860 performs processing complementary to that performed by TX data processor 810 and TX MIMO processor 820 and provides decoded data to a data sink 862. The processing of the terminal 106 is described in detail below.

At each active terminal 106, RX MIMO/data processor 860 further estimates the channel conditions for the downlink and provides Channel State Information (CSI) indicative of the estimated channel conditions. The CSI may include post-processing SNR, channel gain estimates, etc. Controller 870 receives and may further convert downlink csi (dl csi) into some other form (e.g., rate). The downlink CSI is processed (e.g., coded and symbol mapped) by a TX data processor 880, further processed by a TX MIMO processor 882, modulated by one or more modulators 854, and transmitted back to base station 104 over an uplink (feedback) channel. The downlink CSI may be reported by the terminal using various signaling techniques, as described below.

At base station 104, the transmitted feedback signals are received by antennas 824, demodulated by a demodulator 822, and processed by a RX MIMO/data processor 840 in a manner complementary to that performed by TX data processor 880 and TX MIMO processor 882. The reported downlink CSI is then provided to controller 830 and scheduler 834.

Scheduler 834 uses the reported downlink CSI to perform a number of functions such as (1) selecting the best set of terminals for downlink data transmission, and (2) allocating available transmission channels to the selected terminals. Scheduler 834 or controller 830 may further use the reported downlink CSI to determine the coding and modulation schemes to be used for each data stream. Scheduler 834 may schedule terminals for high throughput based on some other performance criteria or metric.

Fig. 8B is a block diagram of a base station 104 and two terminals 106 for uplink data transmission. At each terminal, which is scheduled for data transmission on the uplink, a data source 878 provides data to a TX data processor 880, which encodes, interleaves, and maps the data into modulation symbols. If multiple transmit antennas are used for uplink data transmission, a TX MIMO processor 882 receives and further processes the modulation symbols to provide a vector stream of modulation symbols for each antenna used for data transmission. Each vector stream of symbols is then received and modulated by a respective modulator 854 and transmitted via an associated antenna 852.

At base station 104, the transmitted signals are received by antennas 824, and the received signals from each antenna are provided to a respective demodulator 822. Each demodulator 822 performs processing complementary to that performed at modulator 854. The modulation symbols from all demodulators 822 are then provided to an RX MIMO/data processor 840 and processed to recover the data streams transmitted by the scheduled terminals. An RX MIMO/data processor 840 performs processing complementary to the TX data processor 880 and TX MIMO processor 882 and provides decoded data to a data sink 842.

For each terminal 106 (or just N) desiring to transmit data on the uplink within the coming scheduling interval_TOr N_XThe highest priority terminals), RX MIMO/data processor 840 further estimates the channel condition for the uplink and derives uplink CSI (ul CSI), which is provided to controller 830. Scheduler 834 may also receive and use uplink CSI to perform a number of functions, such as (1) selecting the best set of terminals for data transmission on the uplink, (2) determining a particular processing order for the data streams from the selected terminals, and (3) determining a rate for each data stream. For each scheduling interval, scheduler 834 provides uplink scheduling, which indicates which terminals are selected for data transmission, and their assigned transmission channels and rates. The rate for each data stream may include the data rate and coding and modulation schemes used for the data stream.

A TX data processor 810 receives and processes the uplink schedule and provides processed data to indicate scheduling to one or more modulators 822. A scheduler 822 further conditions the processed data and transmits uplink scheduling to the terminals over the radio link. Uplink scheduling may be transmitted to the terminals using various signaling and messaging techniques.

At each active terminal 106, the transmitted signals are received by antennas 852, demodulated by a demodulator 854, and provided to a RX MIMO/data processor 860. Processor 860 performs processing complementary to that performed by TX MIMO processor 820 and TX data processor 810 and recovers the uplink scheduling for the terminal (if any), which is then provided to a controller 870 and used to control the terminal's uplink transmissions.

In fig. 8A and 8B, scheduler 834 is shown implemented within base station 104. In other embodiments, scheduler 834 may be implemented within other elements of MIMO-OFDM system 100 (e.g., a base station controller coupled to and interacting with multiple base stations).

Fig. 9 is a block diagram of an embodiment of a transmitter unit 900. For clarity, transmitter unit 900 is depicted as the transmitter portion of base station 104 in fig. 8A and 8B. However, transmitter unit 900 may also be used for the transmitter part of each terminal for uplink transmissions.

Transmitter unit 900 may process multiple data streams for one or more terminals based on the CSI (e.g., as reported by the terminals). Transmitter unit 900 includes (1) a TX data processor 814x that receives and processes the information bits to provide modulation symbols, and (2) a pair N_TA TX MIMO processor 820x that demultiplexes the modulation symbols for the transmit antennas.

In the particular embodiment shown in fig. 9, TX data processor 814x includes a demultiplexer 908, N coupled to a plurality of channel data processors 910_DOne for each of the independent data streams to be transmitted to the terminal. Demultiplexer 908 receives and demultiplexes the total information bits into N_DMultiple data streams, each of which may be transmitted over one or more transport channels. Each data stream is provided to a respective channel data processor 910.

In the embodiment shown in fig. 9, each channel data processor 910 includes an encoder 912, a channel interleaver 914, and a symbol mapping element 916. Encoder 912 encodes information bits within the received data stream based on a particular coding scheme to provide coded bits. Channel interleaver 914 interleaves the coded bits based on a particular interleaving scheme to provide diversity. And symbol mapping element 916 maps the interleaved bits to modulation symbols for one or more transmission channels used to transmit the data stream.

Pilot data (e.g., data of a known pattern) may also be encoded and multiplexed with the processed information bits. The processed pilot data may be transmitted on all or a subset of the transmission channels used to transmit the information bits (e.g., in a Time Division Multiplexed (TDM) or Code Division Multiplexed (CDM) manner). The pilot data may be used at the receiver system to perform channel estimation.

As shown in fig. 9, data coding, interleaving, and modulation (or a combination thereof) may be adjusted based on available CSI (e.g., reported by a receiver system). In one coding and modulation scheme, adaptive coding may be achieved using a fixed base code (e.g., a turbo code of rate 1/3) and adjusting puncturing to achieve a desired code rate that is supported by the SNR of the transmission channel used to transmit the data. For this scheme, puncturing may be implemented after channel interleaving. In another coding and modulation scheme, different coding schemes may be used based on the reported CSI. For example, each data stream may be encoded using independent encoding. In this scheme, successive cancellation receiver processing techniques may be used at the receiver to detect and decode the data stream to derive a more reliable estimate of the transmitted data stream, as will be described in detail below.

Symbol mapping element 916 may be designed to combine the sets of interleaved bits to form non-binary symbols and map each non-binary symbol to a point on a signal constellation corresponding to a particular modulation scheme (e.g., QPSK, M-PSK, M-QAM, or some other scheme). Each mapped signal point corresponds to a modulation symbol. The number of information bits transmitted PER modulation symbol for a particular performance level (e.g., one percent PER) depends on the SNR of the transmission channel used to transmit the data stream. Thus, the coding and modulation schemes for each data stream may be selected based on the available CSI. The channel interleaving may also be adjusted based on the available CSI.

The modulation symbols from TX data processor 814x are provided to a TX MIMO processor 820 x. TX MIMO processor 820x from N_DN channel data processor 910 receives_DA stream of modulation symbols and demultiplexing the received modulation symbols into N_TStream of symbol vectors V₁To V_NtOne symbol vector stream for each antenna used to transmit data. Each symbol vector stream is provided to a respective modulator 822. In the embodiment illustrated in fig. 9, each modulator 822 comprises an inverse fourier transform (IFFT) processor 940, a cyclic prefix generator 942, and a transmitter (TMTR) 944.

IFFT processor 940 converts each received symbol into its time domain representation (this is referred to as an OFDM symbol) using an IFFT. IFFT processor 940 may be designed to pair any number of frequency subchannels (e.g., 8, 16, 32_F..) implement an IFFT. In an embodiment, cyclic prefix generator 942 repeats a portion of the time domain representation of the OFDM symbol for each symbol vector converted to an OFDM symbol to form a "transmission symbol" for a particular transmit antenna. The cyclic prefix ensures the orthogonal characteristic of the transmission code element in multipath delay spread, thereby improving the performance of resisting the influence of a degraded path. The IFFT processors 940 and the implementation of cyclic prefixes are known in the art and are not described in detail herein.

A transmitter 944 then converts the time-domain transmission symbols from the associated cyclic prefix generator 942 to an analog signal that is further amplified, filtered, quadrature modulated, and frequency upconverted to provide a modulated signal suitable for transmission over the wireless link. The modulated signal from transmitter 944 is then transmitted from an antenna 824 to the terminals.

An example MIMO-OFDM system is described in the aforementioned U.S. patent application No. 09/532492. OFDM Modulation is also described in the title "Multicarrier Modulation for Data Transmission: an ideaweose Time Has Come, "by the authors John a.c. bingham, ieee communication Magazine, 1990, month 5, which is incorporated herein by reference.

Fig. 9 illustrates an example coding and modulation scheme that may be used with full or partial CSI to provide improved performance (e.g., high throughput). Some other Coding and modulation schemes are further described in the above-mentioned U.S. patent application nos. 09/854235, 09/826481, and 09/956449 and U.S. patent application No. 09/776075, the last patent application entitled "Coding Scheme for a Wireless communication system", filed on 2/1/2001, assigned to the assignee of the present invention, and incorporated herein by reference. Other coding and modulation schemes may also be used and are within the scope of the invention.

Fig. 10A is a block diagram of an embodiment of a receiver unit 1000A. For clarity, receiver unit 1000a is depicted as the receiver portion of one terminal 106 in fig. 8A and 8B. However, receiver unit 1000a may also be used for the receiver portion of base station 104 for uplink transmissions.

From N_TThe transmission signal of each transmitting antenna is N_REach of the antennas 852a through 852r receives and the received signal from each antenna is routed to a respective demodulator 854 (which is also referred to as a front end processor). Each demodulator 854 conditions (e.g., filters and amplifies) a respective received signal, downconverts the conditioned signal to an intermediate frequency or baseband signal, and digitizes the downconverted signal to provide data samples. Each demodulator 854 can further demodulate the data samples with recovered pilots.

Each demodulator 854 also implements a complementary process to that implemented by modulator 822 shown in fig. 9. For OFDM, each demodulator 854 includes an FFT processor and a demultiplexer (not shown in fig. 10A for simplicity). An FFT processor generates a transformed representation of the data samples and provides a stream of symbol vectors. Each symbol vector comprising N_FN of one frequency subchannel_FOne symbol and one vector is provided for each symbol period. From all N_RN of FFT processor of demodulator_RThe symbol vector streams are provided to a demultiplexer which demultiplexes each symbol vector stream into N_GN of a group of frequency subchannels_GA stream of received symbol vectors. Each received symbol vector includes N within the kth frequency subchannel group_KN of one frequency subchannel_KA received symbol, where 1 ≦ N_K≤N_F. The demultiplexer may then be N_RN within a received channel_GMultiple frequency subchannel sets providing up to N_G·N_RThe received symbol vector stream.

Within RX MIMO/data processor 860a, a space/space-time processor 1010 is used to perform MIMO processing on the received symbols for each frequency subchannel set used for the data transmission. One spatial/space-time processor may be used to implement MIMO processing for each frequency subchannel set, or one spatial/space-time processor may be used to implement MIMO processing for all frequency subchannel sets (e.g., in a time-division multiplexed manner).

Space/space-time processor 1010 may be designed to perform spatial or space-time processing on the received symbols to provide estimates of the transmitted adjustment symbols. Space-time processing may be used for non-dispersive channels (i.e., flat fading channels) to cancel undesired signals and/or to maximize the received SNR of each constituent signal in the presence of interference and noise from other signals. Spatial processing may be implemented based on a Channel Correlation Matrix Inversion (CCMI) technique, a Minimum Mean Square Error (MMSE) technique, a full CSI technique, or some other technique. Space-time processing may be used for dispersive channels (i.e., frequency selective fading channels) to improve crosstalk from other transmitted signals and inter-symbol interference (ISI) from all transmitted signals caused by dispersion of the channel. The space-time processing may be implemented based on an MMSE linear equalizer (MMSE-LE), a Decision Feedback Equalizer (DFE), a Maximum Likelihood Sequence Estimator (MLSE), or some other technique. Spatial and space-time processing is further detailed in the aforementioned U.S. patent application serial No. 09/993087.

For a particular group of frequency subchannels, a space/space-time processor 1010 receives and processes N_RA stream of received symbol vectors and providing N_TA stream of recovered symbol vectors. Each recovered symbol vector includes up to N_KA recovered symbol, which is N for the k frequency subchannel group in one symbol period_KN transmitted on frequency subchannels_KAn estimate of the individual modulation symbols. Space/space time processor 1010 may further process each received numberThe processed SNR is estimated from the stream. The SNR estimates may be derived as described in U.S. patent application serial nos. 09/956449 and 09/854235 and 09/993087.

Selector 1012 receives N from space/space-time processor 1010_TA stream of recovered symbol vectors and extracts recovered symbols corresponding to one or more data streams to be recovered. Alternatively, the desired recovered symbols are decimated in space/space-time processor 1010. In any event, the desired recovered symbols are extracted and provided to an RX data processor 1020.

Within RX data processor 1020, a demodulation element 1022 demodulates each recovered symbol in accordance with a demodulation scheme (e.g., M-PSK, M-QAM) used for the symbol at the transmitter unit. The demodulated data is then deinterleaved by a deinterleaver 1024 and the deinterleaved data is further implemented in a manner complementary to the modulation, interleaving, and coding implemented at the transmitter unit. For example, a Turbo decoder or a Viterbi decoder may be used for decoder 1026 if Turbo or convolutional coding, respectively, is implemented at the transmitter unit. The decoded data stream from decoder 1026 represents an estimate of the transmitted data stream.

Fig. 10B is a block diagram of a receiver unit 1000B that can implement a successive cancellation receiver processing technique. Receiver unit 1000b may also be used for a receiver portion of a base station 104 or a terminal 106. The transmitted signal is composed of N_REach of the antennas 852 receives and the received signal from each antenna is routed to a respective demodulator 854. Each demodulator 854 processes a respective received signal and provides a stream of received symbols to an RXMIMO/data processor 860 b. RX MIMO/data processor 860b may also be used to process N from each frequency subchannel group used for data transmission_RN of one receiving antenna_RA stream of received symbol vectors, wherein each received symbol vector comprises N within a k-th frequency subchannel group_KN of one frequency subchannel_KThe received symbols.

In the embodiment illustrated in fig. 10B, RX MIMO/data processor 860B includes multiple sequential (i.e., cascaded) receiver processing stages 1050, one for each transmitted signal to be recovered. In one transmit processing scheme, an independent data stream is transmitted on each spatial subchannel of each frequency subchannel group. For this transmit processing scheme, the number of data streams per frequency subchannel is equal to the number of transmit signals, which is also equal to the number of transmit antennas used for data transmission (which may be all or a subset of the available transmit antennas). For clarity, RX MIMO/data processor 860b is described for the transmit processing scheme.

Each receive processing stage 1050 (except for the last stage 1050n) includes a channel MIMO/data processor 1060 coupled to an interference canceller 1070, and the last stage 1050n includes only a channel MIMO/data processor 1060 n. For the first receiver processing stage 1050a, a channel MIMO/data processor 1060a receives and processes N from demodulators 854a through 854r_RThe received symbol vector stream to provide a decoded data stream of the first transmitted signal. And for each of the second through last stages 1050b through 1050N, the channel MIMO/data processor 1060 of that stage receives and processes N from the interference canceller of the previous stage_RA stream of modified symbol vectors to derive a decoded data stream of the transmitted signal being recovered by the stage. Each channel MIMO/data processor 1060 further provides CSI (e.g., SNR) associated with the transmission channel in question.

For the first receiver processing stage 1050a, interference canceller 1070a proceeds from all N_RA demodulator 854 receives N_RA stream of received symbol vectors. And for each of the second through penultimate stages, interference canceller 1070 receives N from the interference canceller of the previous stage_RA stream of modified symbol vectors. Each interference canceller 1070 also receives the decoded data stream from the channel MIMO/data processor 1060 at the same stage and performs processing (e.g., coding, interleaving, and modulation) to derive N_TA stream of remodulated symbol vectors, which are N of a group of frequency subchannels_TAn estimate of a vector stream of transmit modulation symbols.

N_TA stream of remodulated symbol vectors (forNth iteration) is further processed with the estimated channel response to provide an estimate of interference due to the decoded data stream. EstimatingComprising N_RA number of vectors, each vector being N due to the decoded data stream_RAn estimate of a component in one of the received signals. These components are pairs of_RThe remaining (not yet detected) transmitted signals within the received signal. Thus, interference estimationFrom a received stream of symbol vectorsr ⁿIs subtracted (i.e., canceled) to provide N_RA stream r of modified symbol vectorsⁿ⁺¹Some of which are removed from the decoded data stream. Modified symbol vector stream rⁿ⁺¹Is provided to the next receiver processing stage as shown in fig. 10B. Each interference canceller 1070 thus provides a signal comprising all N except the cancelled interference component_RA stream of modified symbol vectors. The controller 870 may also be utilized to direct various steps in the successive cancellation receiver processing.

The successive cancellation receiver processing techniques are further detailed in the aforementioned U.S. patent application serial numbers 09/854235 and 09/993087 in the following paper, authored by p.w.wolfiansky et al, entitled "V-BLAST: an Architecture for the engineering of the extreme High Data Rates over the Rich-Scattering Wireless Channel ", Proc. ISSSE-98, pizza, Italy, which is incorporated herein by reference.

Fig. 10B shows a receiver structure that may be used directly when one independent data stream is transmitted on each transmit antenna of each frequency subchannel group. In this case, each receiver processing stage 1050 can be used to recover one of the transmitted data streams and provide a decoded data stream corresponding to the recovered data stream.

For some other transmit processing schemes, the data streams may be transmitted over multiple transmit antennas, frequency subchannels, and/or time intervals to provide corresponding spatial, frequency, and/or time diversity. For these schemes, receiver processing initially derives a received symbol stream for each transmit antenna for each frequency subchannel. The modulation symbols for multiple transmit antennas, frequency subchannels, and/or time intervals may then be combined in a manner complementary to the demultiplexing implemented at the transmitter unit. The combined symbol streams are then processed to recover the transmitted data streams.

For simplicity, the receiver architecture shown in fig. 10B provides streams of symbol vectors (received or modified) to each receiver processing stage 1050, with the streams having interference components due to the previously removed (i.e., canceled) decoded data streams. In the embodiment shown in fig. 10B, each stage removes interference components due to the data stream decoded by the stage. In some other designs, the received symbol vector streams may be provided to all stages, and each stage may implement cancellation of interference components from all previously decoded data streams (which may be provided from previous stages). Interference cancellation may also be skipped in one or more stages (e.g., if the SNR of the data stream is high). Various modifications may be made to the receiver structure shown in fig. 10B and are within the scope of the invention.

Fig. 10A and 10B illustrate two embodiments of a receiver unit that can process a data transmission, determine characteristics of the transmission channel (e.g., post-processing SNR), and report CSI back to the transmitter unit. Other designs and other receiver processing techniques based on the techniques presented herein are also contemplated and are within the scope of the present invention.

Channel State Information (CSI)

The CSI, which may comprise any type of information indicative of the characteristics of the communication link, is used to select the appropriate data rate and coding and modulation scheme for each individual data stream. CSI may be categorized as "full CSI" or "partial CSI". Various types of information may be provided as full or partial CSI, and some examples are described below.

In an embodiment, the partial CSI includes SNR, which may be derived as a ratio of signal power to noise and interference power. The SNR is typically estimated and provided for each transmission channel used for data transmission (e.g., each transmitted data stream), although the overall SNR may be provided for multiple transmission channels. The SNR estimate may be quantized to a value with a certain number of bits. In one embodiment, the SNR estimate is mapped to an SNR index, for example, using a look-up table.

In another embodiment, the partial CSI includes signal power, noise power, and interference power. These three terms may be derived and provided for each transport channel or each set of transport channels used for data transmission.

In another embodiment, the partial CSI includes a signal-to-noise ratio and an interference power list of observable interference terms. This information may be derived and provided for each transport channel or each set of transport channels used for data transmission.

In another embodiment, the partial CSI comprises signal components in the form of a matrix (e.g., N for all transmit-receive antenna pairs)_R×N_TComplex terms) and matrix-form noise plus interference terms (e.g., N)_R×N_TA plurality of terms). The transmitter unit may then suitably combine the signal components and the noise and interference components for the appropriate transmit-receive antenna pairs to derive a quality for each transmission channel used for the data transmission (e.g., a post-processing SNR for each transmitted data stream, as received at the receiver unit).

In another embodiment, the partial CSI includes a data rate indicator for each transmitted data stream. The transmission channel quality for the data transmission may be initially determined (e.g., based on an estimated SNR for the transmission channel) and a data rate corresponding to the determined channel quality may then be identified for each transmission channel or each group of transmission channels (e.g., based on a look-up table). The identified data rate indicates a maximum data rate that can be transmitted on the transport channel for a desired level of performance. The data rate may be mapped and represented by a Data Rate Indicator (DRI), which may be efficiently encoded. For example, if the transmitter unit of each transmit antenna supports (up to) seven possible data rates, the DRI may be represented using a 3-bit value, where, for example, zero may indicate a data rate of zero (i.e., no transmit antenna is used), and 1 to 7 may be used to indicate seven different data rates. In a general implementation, channel quality measurements (e.g., SNR estimates) are directly mapped to the DRI based on, for example, a look-up table.

In another embodiment, the partial CSI includes a rate used at the transmitter unit for each data stream. In this embodiment, the rate may identify a particular coding and modulation scheme to use for the data stream to achieve a desired level of performance.

In another embodiment, the partial CSI comprises different indicators for specific quality measurements of the transmission channel. Initially, the SNR or DRI or some other quality measure of the transmission channel is determined and reported as a reference measurement value. Thereafter, link quality monitoring continues and the difference between the last reported measurement and the current measurement is determined. The difference may then be quantized into one or more bits, and the quantized difference mapped to and identified with a different indicator, which is then reported. The discrepancy indicator may indicate that the last reported measurement is incremented or decremented by a particular step size (or the last reported measurement is maintained). For example, the discrepancy indicator may include (1) the SNR observed for a particular transmission channel is incremented or decremented by a particular step size, or (2) the data rate should be adjusted by a particular amount or some other change. The reference measurements may be sent periodically to ensure that errors in the difference indicators and/or erroneous receptions in these indicators do not accumulate.

The full CSI includes N over the entire system bandwidth (i.e., each frequency subchannel)_R×N_TChannel response matrixH(k) Sufficient characteristics (e.g., complex gain) of the propagation path between each transmit-receive antenna pair within the antenna array.

In an embodiment, the full CSI includes eigenmodes plus any other information indicative of or equivalent to SNR. For example, the SNR-related information may be an indication of the data rate per eigenmode, an indication of the coding and modulation scheme used per eigenmode, the signal and interference power per eigenmode, the signal-to-noise ratio per eigenmode, etc. The above-mentioned information of the partial CSI may also be provided as SNR-related information.

In another embodiment, the full CSI comprises a matrixA＝H ^H H. The matrixAIt is sufficient to determine the eigenmodes and the eigenvalues of the channel and may be a more efficient representation of the channel (e.g., fewer bits may be needed to send the full CSI of the representation).

Different update techniques may also be used for all full CSI data types. For example, different updates of the full CSI characteristics may be sent periodically when the channel changes by some amount, etc.

Other forms of full or partial CSI may also be used and are within the scope of the invention. In general, the full or partial CSI data comprises sufficient information in any form for adjusting the processing at the transmitter unit such that a desired level of performance is obtained for the transmitted data stream.

Deriving and reporting CSI

The CSI may be derived based on signals transmitted by the transmitter unit and received at the receiver unit. In one embodiment, the CSI is derived based on pilots included in the transmitted signal. Alternatively or additionally, CSI is derived based on data included in the transmitted signal.

In another embodiment, the CSI includes one or more signals transmitted on the reverse link from the receiver unit to the transmitter unit. In some systems, there may be a degree of correlation on the downlink and uplink (e.g., for Time Division Duplex (TDD) systems where the uplink and downlink share the same system bandwidth in a time division multiplexed manner). In these systems, uplink may be basedThe path quality estimates the downlink quality (with some necessary accuracy), which may be estimated based on the signal (e.g., pilot) transmitted from the receiver unit. The pilot signal sent on the uplink may then represent the manner in which the transmitter unit may estimate the CSI observed at the receiver unit. In a TDD system, a transmitter unit may derive a channel response matrixH(k) (e.g., based on pilots sent on the uplink), the difference between the transmit and receive array replicas is taken into account and a noise variance estimate is received at the receiver unit. The array replica delta (arraymanifold delta) can be accounted for by a periodic calibration procedure that may involve feedback between the receiver unit and the transmitter unit.

The signal quality may be estimated at the receiver unit based on various techniques. Some such techniques are described in the following patents, which are assigned to the assignee of the present invention and are incorporated herein by reference:

● U.S. Pat. No. 5799005 entitled "System and Method for determining received Pilot Power and Path in a CDMA communication System" filed on 25/8 of 1998;

● U.S. Pat. No. 5903554 entitled "Method and Apparatus for measuring Link Quality in a Spread Spectrum Communication System", filed 5/11.1999;

● U.S. Pat. Nos. 5056109 and 5265119, both entitled "Method and Apparatus for controlling Transmission Power in a CDMA Cellular Mobile telephone System", filed on 1991, 8.10 and 23.11.1993, respectively; and

● U.S. Pat. No. 6097972 entitled "Method and Apparatus for processing Power Control Signals in CDMA Mobile Telephone System", filed on 8/1/2000;

the CSI may be reported back to the transmitter unit using various CSI transmission schemes. For example, the CSI may be transmitted fully, differentially, or in a combination thereof. In thatIn an embodiment, full or partial CSI is reported periodically and differential updates are sent based on previously sent CSI. As an example of full CSI, the update may be a correction to the reported eigenmodes (based on the error signal). Eigenvalues generally do not change as fast as eigenmodes, so they can be updated at a lower rate. In another embodiment, the CSI is sent only when there is a change (e.g., if the change exceeds a certain threshold), which may reduce the effective rate of the feedback channel. As an example of partial CSI, the SNR may be sent back (e.g., differentially) when it changes. For OFDM systems, correlation in the frequency domain may be used to reduce the amount of CSI to be fed back. As an example of an OFDM system using partial CSI, if corresponding to N_MThe SNR of a particular spatial subchannel for each frequency subchannel is similar, then the condition may be reported as a true SNR and the first and last frequency subchannels. Other compression and feedback channel error recovery techniques that reduce the amount of data fed back for CSI may also be used and are within the scope of the invention.

Various types of information for CSI and various CSI reporting mechanisms are described in U.S. patent application Ser. No. 08/963386, entitled "Method and Apparatus for High Rate Packet Data Transmission", filed on 1997, month 11, 3, assigned to the assignee of the present invention, and also "TIE/EIA/IS-856 cdma2000 High Rate Packet Data Air interface Specification", incorporated herein by reference.

For clarity, various aspects and embodiments of resource allocation are described on the downlink and uplink, among others. The various techniques described herein may be used in an "ad hoc" or peer-to-peer network, and this is within the scope of the invention.

The MIMO-OFDM system described herein may also be designed to implement any standard and design of CDMA, TDMA, FDMA and other multiple access techniques. The CDMA standards include IS-95, CDMA2000, and W-CDMA standards, and the TDMA standards include Global System for Mobile communications (GSM) standards. These standards are known in the art and are incorporated herein by reference.

The implementation of the elements of the base station and the terminal may employ: one or more Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), processors, controllers, microcontrollers, microprocessors, Field Programmable Gate Arrays (FPGAs), programmable logic devices, and other electronic units designed to perform the functions described herein, or a combination thereof. Some of the functions and processes described herein may also be implemented in software executing on a processor.

Aspects of the invention may also be implemented in a combination of software and hardware. For example, a terminal scheduling process for downlink and/or uplink data transmissions can be implemented based on program code executing on a processor (e.g., scheduler 834 of fig. 8).

Headings are included herein for reference and to aid in locating sections. These headings are not intended to limit the concepts described herein, and these concepts may have application in other sections throughout the specification.

The previous description of the preferred embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without the use of the inventive faculty. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (63)

1. A method for scheduling data transmission on a plurality of frequency subchannel groups for a plurality of terminals in a wireless communication system, comprising:

forming at least one set of terminals for possible data transmission on the plurality of frequency subchannel groups, wherein each set comprises one or more terminals and corresponds to a hypothesis to be evaluated;

evaluating the performance of each hypothesis;

selecting a hypothesis for each frequency subchannel group based on the evaluated performance; and

one or more terminals within each selected hypothesis are scheduled for data transmission on the corresponding set of frequency subchannels.

2. The method of claim 1, wherein scheduling of the plurality of terminals is performed for downlink data transmission.

3. The method of claim 2, wherein the step of evaluating the performance of each hypothesis further comprises:

forming one or more sub-hypotheses for each hypothesis, wherein each sub-hypothesis corresponds to a particular allocation of multiple transmit antennas to one or more terminals within the hypothesis, and wherein performance of each sub-hypothesis is evaluated, and selecting one hypothesis for each frequency subchannel group further comprises selecting one sub-hypothesis for each frequency subchannel based on the evaluated sub-hypothesis performance.

4. The method of claim 2, wherein the step of scheduling the one or more terminals further comprises:

multiple transmit antennas are assigned to one or more terminals within each hypothesis, and performance of each hypothesis is evaluated based in part on the antenna assignments for the hypothesis.

5. The method of claim 4, wherein the assigning for each hypothesis comprises:

identifying the transmit antenna and the terminal pair with the best metric among all unassigned transmit antennas;

assigning the transmit antennas within the pair to the terminals within the pair; and

the assigned transmit antennas and terminals are removed from the hypothesis.

6. The method of claim 4, wherein multiple transmit antennas are assigned to the one or more terminals within each hypothesis based on a priority of each terminal.

7. The method of claim 6, wherein the highest priority terminal within each hypothesis is assigned a transmit antenna associated with the highest throughput, and the lowest priority terminal within the hypothesis is assigned a transmit antenna associated with the lowest throughput.

8. The method of claim 2, wherein: the step of evaluating the performance of each hypothesis further comprises:

a channel response matrix is formed for a plurality of terminals within the hypothesis, and wherein performance of the hypothesis is evaluated based on the channel response matrix.

9. The method of claim 8, wherein: the step of evaluating the performance of each hypothesis further comprises:

a steering vector matrix is derived for generating a plurality of beams for a plurality of terminals within the hypothesis.

10. The method of claim 9, wherein: the step of evaluating the performance of each hypothesis further comprises:

a scaling matrix is derived for adjusting the transmit power of each terminal within the hypothesis.

11. The method of claim 1, wherein a plurality of terminals are scheduled for uplink data transmission.

12. The method of claim 11, wherein: the step of evaluating the performance of each hypothesis further comprises: forming one or more sub-hypotheses for each hypothesis, wherein each sub-hypothesis corresponds to a ranking of characteristics of one or more terminals within the hypothesis, and wherein performance of each sub-hypothesis is evaluated, and selecting one hypothesis for each frequency subchannel group further comprises selecting one sub-hypothesis for each frequency subchannel group based on the evaluated sub-hypothesis performance.

13. The method of claim 12, wherein: the step of forming one or more sub-hypotheses for each hypothesis includes forming a terminal ordering for each hypothesis based on the priority of each terminal within the hypothesis.

14. The method of claim 12, wherein each sub-hypothesis is evaluated by:

processing signals transmitted from one or more terminals within a hypothesis to provide processed signals; and

the signal-to-noise ratio SNR is estimated for the processed signal.

15. The method of claim 14, wherein the SNR of the processed signals is dependent upon a particular order in which the signals hypothetically transmitted are processed, and wherein the signals hypothetically transmitted are processed in the particular order determined by the terminal ordering of the sub-hypothesis being evaluated.

16. The method of claim 14, wherein: the step of forming one or more sub-hypotheses for each hypothesis includes: a sub-hypothesis is formed for each hypothesis, and the terminal ordering of the sub-hypotheses is determined based on the SNR of the processed signal.

17. The method of claim 14, wherein: the step of forming one or more sub-hypotheses for each hypothesis includes: a sub-hypothesis is formed for each hypothesis, and where the transmitted signal from the lowest priority terminal within the hypothesis is processed first and the transmitted signal from the highest priority terminal is processed last.

18. The method of claim 11, wherein the performance of each hypothesis is evaluated based on successive cancellation receiver processing.

19. The method of claim 18, wherein the successively canceling receiver processing implements a plurality of iterations to recover a plurality of signals hypothesized to be transmitted from the one or more terminals within each hypothesis, wherein one iteration is performed for each hypothesized transmitted signal to be recovered.

20. The method of claim 19, wherein each iteration comprises:

processing the plurality of input signals according to a particular equalization scheme to provide a plurality of processed signals;

detecting a processed signal corresponding to the hypothesized transmitted signal being restored within the iteration to provide a decoded data stream; and

selectively deriving a plurality of modified signals based on the input signal and substantially removing interference components generated by the decoded data stream, an

Where the input signal for the first iteration is a signal received from one or more terminals within the hypothesis being evaluated, and the input signal for each successive iteration is a modified signal from the previous iteration.

21. The method of claim 1, wherein each hypothesis is evaluated based in part on Channel State Information (CSI) for each terminal within the hypothesis.

22. The method of claim 21, wherein the channel state information comprises a signal-to-noise-plus-interference ratio (SNR).

23. The method of claim 22, wherein each set of one or more terminals to be evaluated for a frequency subchannel group is associated with a respective SNR matrix obtained for the one or more terminals in the set of terminals for the frequency subchannel group.

24. The method of claim 21, wherein the channel state information includes a channel gain for each transmit-receive antenna pair used for data transmission.

25. The method of claim 1, wherein: the step of scheduling one or more terminals further comprises:

a data rate is determined for each data stream transmitted by each scheduled terminal, and wherein a plurality of data streams are transmitted at the determined data rate.

26. The method of claim 25, wherein: the step of scheduling one or more terminals further comprises:

a coding and modulation scheme to be used for each data stream to be transmitted is determined, and wherein the plurality of data streams are processed based on the coding and modulation schemes determined prior to transmission.

27. The method of claim 1, wherein the plurality of terminals are scheduled for data transmission on a plurality of spatial subchannels.

28. The method of claim 27, wherein each selected hypothesis includes a plurality of SIMO terminals, and wherein each SIMO terminal is assigned one spatial subchannel.

29. The method of claim 27, wherein each selected hypothesis comprises a single MIMO terminal assigned all spatial subchannels.

30. The method of claim 27, wherein each selected hypothesis comprises a combination of SIMO and MIMO terminals, wherein each SIMO terminal is assigned one spatial subchannel and each MIMO terminal is assigned two or more spatial subchannels.

31. The method of claim 1, wherein at least one set comprises a plurality of MISO terminals, each terminal having a single antenna to receive downlink data transmissions.

32. The method of claim 1, wherein each set of multiple terminals includes terminals with similar link margins.

33. The method of claim 1, wherein evaluating each hypothesis comprises:

performance metrics are calculated for the hypotheses.

34. The method of claim 33, wherein the performance metric is a function of an overall throughput achievable by one or more terminals within a hypothesis for the particular group of frequency subchannels.

35. The method of claim 34, wherein the throughput for each terminal within a hypothesis is determined based on signal-to-noise-and-interference ratios, SNRs, obtained by the terminal for a particular group of frequency subchannels.

36. The method of claim 34, wherein the throughput for each terminal is determined based on a signal-to-noise-plus-interference ratio (SNR) obtained by the terminal for each of a plurality of frequency subchannels in a particular group of frequency subchannels.

37. The method of claim 33, wherein for each frequency subchannel group, the hypothesis with the best performance metric is selected for scheduling.

38. The method of claim 1, wherein: the step of scheduling one or more terminals further comprises:

a plurality of terminals to be scheduled for data transmission are prioritized.

39. The method of claim 38, wherein: the step of prioritizing a plurality of terminals to be scheduled for data transmission comprises:

the group of N highest priority terminals considered for scheduling for each frequency subchannel group is selected, where N is equal to or greater than 1.

40. The method of claim 38, wherein: the step of prioritizing a plurality of terminals to be scheduled for data transmission comprises:

one or more metrics are maintained for each terminal to be considered for scheduling, and wherein the priority for each terminal is determined based on the one or more metrics maintained for the terminal.

41. The method of claim 40, wherein one metric maintained for each terminal relates to an average throughput achieved by the terminal.

42. The method of claim 38, wherein the priority for each terminal is determined based on one or more factors maintained for the terminal and related to quality of service, QoS.

43. A method for downlink data transmission scheduling on a plurality of frequency subchannel groups for a plurality of terminals in a multiple-input multiple-output, MIMO, communication system utilizing orthogonal frequency division multiplexing, OFDM, comprising:

forming at least one set of terminals for possible data transmission for each of a plurality of frequency subchannel groups, wherein each set includes one or more terminals and corresponds to a hypothesis to be evaluated

Forming one or more sub-hypotheses for each hypothesis, wherein each sub-hypothesis corresponds to a particular assignment of multiple transmit antennas to one or more terminals within the hypothesis;

evaluating the performance of each sub-hypothesis;

selecting a sub-hypothesis for each frequency subchannel group based on the evaluated performance; and

one or more terminals within each selected sub-hypothesis are scheduled for downlink data transmission on the corresponding frequency subchannel group.

44. The method of claim 43, wherein evaluating for each sub-hypothesis comprises:

determining an overall throughput for one or more terminals within a sub-hypothesis based on the particular antenna assignment; and

wherein for each frequency subchannel group, the sub-hypothesis with the highest throughput is selected.

45. The method of claim 43, wherein: the step of forming at least one terminal set comprises: one or more terminals within the set of terminals are selected based on the priority.

46. A method for scheduling downlink data transmissions for a plurality of terminals on a plurality of frequency subchannel groups in a multiple-input multiple-output, MIMO, communication system using orthogonal frequency division multiplexing, OFDM, comprising:

forming at least one set of terminals for possible data transmission for each of a plurality of frequency subchannel sets, wherein each set includes a plurality of terminals and corresponds to a hypothesis to be evaluated;

forming a channel response matrix for the plurality of terminals within each hypothesis;

evaluating the performance of each hypothesis based on the channel response matrix;

selecting a hypothesis for each frequency subchannel group based on the evaluated performance;

one or more terminals within each selected hypothesis are scheduled for downlink data transmission on the corresponding frequency subchannel group.

47. A method for scheduling uplink data transmission for a plurality of terminals on a plurality of frequency subchannel groups in a multiple-input multiple-output, MIMO, communication system utilizing orthogonal frequency division multiplexing, OFDM, comprising:

forming at least one set of terminals for possible data transmission on the plurality of frequency subchannel groups, wherein each set comprises one or more terminals and corresponds to a hypothesis to be evaluated;

forming one or more sub-hypotheses for each hypothesis, wherein each sub-hypothesis corresponds to a particular order of one or more terminals within the hypothesis;

evaluating the performance of each sub-hypothesis;

selecting a sub-hypothesis for each frequency subchannel group based on the evaluated performance; and

one or more terminals within each selected sub-hypothesis are scheduled for uplink data transmission on the corresponding frequency subchannel group.

48. The method of claim 47, wherein signals transmitted by one or more scheduled terminals within a selected sub-hypothesis for each frequency subchannel group are processed in a particular order determined by the ordering of the sub-hypotheses.

49. The method of claim 47, wherein the evaluation of each sub-hypothesis comprises:

processing each signal hypothesized to be transmitted from each terminal within the sub-hypothesis to provide a corresponding processed signal;

a signal-to-noise-and-interference ratio, SNR, is determined for each processed signal.

50. The method of claim 49, wherein: the step of forming one or more sub-hypotheses for each hypothesis includes: the ranking within the sub-hypotheses is selected to obtain the best performance of the hypothesis, as determined by one or more performance metrics.

51. A method for use in a memory communicatively coupled to a digital signal processing device, DSPD, capable of interpreting digital signals, the method comprising:

receiving channel state information, CSI, indicating channel estimates for a plurality of terminals within a wireless communication system;

forming at least one set of terminals for possible data transmission for each of a plurality of frequency subchannel groups, wherein each set includes one or more terminals and corresponds to a hypothesis to be evaluated;

evaluating performance of each hypothesis based in part on channel state information for one or more terminals within the hypothesis;

selecting a hypothesis for each frequency subchannel group based on the evaluated performance; and

one or more terminals within each selected hypothesis are scheduled for data transmission on the corresponding set of frequency subchannels.

52. A scheduler for use in a multiple-input multiple-output, MIMO, communication system utilizing orthogonal frequency division multiplexing, OFDM, comprising:

means for receiving channel state information, CSI, indicative of channel estimates for a plurality of terminals within a communication system;

means for forming at least one set of terminals for possible data transmission on a plurality of frequency subchannel groups, wherein each set comprises one or more terminals and corresponds to a hypothesis to be evaluated;

means for evaluating performance of each hypothesis based in part on channel state information for one or more terminals within the hypothesis;

means for selecting a hypothesis for each frequency subchannel group based on the evaluated performance; and

means for scheduling one or more terminals within each selected hypothesis for data transmission on the corresponding set of frequency subchannels.

53. The scheduler of claim 52, wherein: the means for evaluating the performance of each hypothesis further comprises: means for forming one or more sub-hypotheses for each hypothesis, wherein each sub-hypothesis corresponds to a particular allocation of multiple transmit antennas to terminals for downlink data transmission within the one or more hypotheses, wherein performance within each sub-hypothesis is evaluated, and

the means for selecting a hypothesis for each frequency subchannel group further comprises: one sub-hypothesis is selected for each frequency subchannel group based on the evaluated sub-hypothesis performance.

54. The scheduler of claim 52, wherein: the means for evaluating the performance of each hypothesis further comprises: means for forming one or more sub-hypotheses for each hypothesis, wherein each sub-hypothesis corresponds to a particular order for processing uplink data transmissions from one or more terminals within the hypothesis, wherein performance of each sub-hypothesis is evaluated, and

the means for selecting a hypothesis for each frequency subchannel group further comprises: one sub-hypothesis is selected for each frequency subchannel group based on the evaluated sub-hypothesis performance.

55. The scheduler of claim 52, wherein: the apparatus for scheduling one or more terminals further comprises:

means for prioritizing a plurality of terminals scheduled for data transmission.

56. A base station for use in a multiple-input multiple-output, MIMO, communication system utilizing orthogonal frequency division multiplexing, OFDM, comprising:

a scheduler for receiving channel state information, CSI, indicative of channel estimates for a plurality of terminals in a communication system, selecting one or more sets of terminals for data transmission on a plurality of frequency subchannel groups, and allocating a plurality of spatial subchannels in a corresponding frequency band to the one or more terminals in each selected set;

a transmit data processor for receiving and processing data to provide a plurality of data streams for transmission to one or more scheduled terminals, wherein the data is processed based on channel state information for the one or more scheduled terminals;

at least one modulator for processing a plurality of data streams to provide a plurality of modulated signals; and

a plurality of antennas for receiving and transmitting a plurality of modulated signals to one or more scheduled terminals.

57. The base station of claim 56, wherein the scheduler is further configured to select a data rate for each data stream.

58. The base station of claim 56, wherein the scheduler is further configured to select a coding and modulation scheme for each data stream, and wherein the transmit data processor is further configured to process the data for each data stream based on the coding and modulation scheme selected for the data stream.

59. The base station of claim 56, further comprising:

at least one demodulator for processing a plurality of signals received via a plurality of antennas to provide a plurality of received signals; and

a receive data processor for processing the plurality of received signals to derive channel state information for a plurality of terminals in the communication system.

60. A transmitter apparatus for use in a multiple-input multiple-output, MIMO, communication system utilizing orthogonal frequency division multiplexing, OFDM, comprising:

means for receiving channel state information, CSI, indicative of channel estimates for a plurality of terminals within a communication system;

means for selecting a set of one or more terminals for data transmission on a plurality of frequency subchannel groups;

means for assigning a plurality of spatial subchannels in a corresponding frequency subchannel group to one or more terminals in each selected set;

means for processing data to provide a plurality of data streams for transmission to one or more scheduled terminals, wherein the data is processed based on channel state information for the one or more scheduled terminals;

means for processing a plurality of data streams to provide a plurality of modulated signals; and

means for transmitting the plurality of modulated signals to one or more scheduled terminals.

61. A terminal in a multiple-input multiple-output, MIMO, communication system, comprising:

a plurality of antennas, each antenna for receiving a plurality of transmitted signals and providing a corresponding received signal;

a plurality of front end units, each front end unit for processing a respective received signal to provide a corresponding sample stream and deriving channel state information, CSI, for the plurality of sample streams;

a receive processor for processing the plurality of sample streams from the plurality of front end units to provide one or more decoded data streams; and

a transmit data processor for processing the transmitted channel state information; and

wherein a terminal is one or more terminals included in a set scheduled for data transmission over one or more of a plurality of frequency subchannel sets for a particular time interval.

62. The terminal of claim 61, comprising:

at least one demodulator may be configured to process the multiple sample streams to provide one or more received symbol streams for one or more spatial subchannels of the one or more frequency subchannels allocated for downlink transmission by the terminal.

63. A multiple-input multiple-output, MIMO, communication system utilizing orthogonal frequency division multiplexing, OFDM, comprising:

a scheduler for receiving channel state information, CSI, indicative of channel estimates for a plurality of terminals in a communication system, selecting one or more sets of terminals on a plurality of frequency subchannel groups for data transmission, and allocating a plurality of spatial subchannels in a corresponding frequency subchannel group to the one or more terminals in each selected set;

a base station for processing transmissions for one or more terminals scheduled for data transmission on a plurality of spatial subchannels of a plurality of frequency subchannel groups; and

a plurality of terminals, each terminal for communicating with the base station through one or more spatial subchannels of the one or more frequency subchannel groups assigned to the terminal when scheduled for data transmission by the scheduler.