UA86191C2 - Channel calibrate for communication system having duplex commutation and time-division channeling - Google Patents

Abstract

Techniques to calibrate the downlink and uplink channels to account for differences in the frequency responses of the transmit and receive chains at an access point and a user terminal. In one embodiment, pilots are transmitted on the downlink and uplink channels and used to derive estimates of the downlink and uplink channel responses, respectively. Two sets of correction factors are then determined based on the estimate^ of the downlink and uplink channel responses. A calibrated downlink channel is formed by using a first set of correction factorsi for the downlink channel, and a calibrated uplink channel is formed by using a I second set of correction factors for the uplink channel. The first and second sets of correction factors may be determined using a matrix-ratio computation or a minimum mean sqdare error (MMSE) computation. The calibration may be performed in real-time based on over-the-air transmission.

Description

interpolated based on the correction factors obtained for the "calibrated" sub-bands.

Various aspects of the embodiments of the present invention are described in more detail below.

Brief description of the drawings

The salient features, essence and advantages of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which like reference numerals denote like elements.

Figure 1 shows the transmission and reception circuits in the access point and user terminal of the MIMO system;

Fig. 2 illustrates the application of correction factors to account for differences in the transmission/reception circuits of the access point and the user terminal; Fig. 3 shows the calibration process of downlink and uplink echoes in THAT MIMO-BOTH system; Figure 4 shows the process of deriving correction vectors from estimates of downlink and uplink responses;

Fig. 5 is a block diagram of an access point and a user terminal; and

Fig. b6 is a block diagram of the TX spatial processor.

Implementation of the invention

The calibration methods described in this description can be used for different communication systems. In particular, these methods can be used in single-input, single-output (5150), multiple-input, single-output (MI5ZO) systems, single-input, multiple-output (5IMO) systems, and multiple-input, multiple-output (MIMO) systems. ).

A MIMO system uses multiple (Mt) transmit antennas and multiple (Me) receive antennas to transmit data. The MIMO channel, formed by Mk" transmitting and Mk receiving antennas, can be divided into M5 independent channels, and Mz" type(Mt,Ma). Each of the M5 independent channels is also called a spatial channel or an eigenmode MIMO channel and corresponds to the dimension. A MIMO system can provide improved performance (ie, increased transmission capacity) if the additional dimension created by multiple transmit and receive antennas is used.

Of course, this requires an accurate estimate of the channel response between the transmitter and the receiver.

Figure 1 shows a block diagram of the transmission and reception circuits at the access point 102 and the user terminal 104 in the MIMO system. For this system, the uplink and downlink share the same frequency band in a time-division duplex method.

For the downlink at the access point 102, the symbols (defined by the xHap "transmission" vector) are processed by the transmission circuit (TMTC) 114 and transmitted through the Mar antennas 116 over a wireless channel. At the user terminal 104, the downlink signals are received by the My antennas 152 and processed by the receiving circuit (ESURC) 154 to provide the received symbols (denoted by the "receive" gap vector). Processing in transmission circuit 114 typically includes digital-to-analog conversion, amplification, filtering, upscaling, and the like. Processing in the receiving circuit 154 typically includes downscaling, amplification, filtering, analog-to-digital conversion, and the like.

In the case of an uplink in the user terminal 104, the symbols (denoted by the transmission vector xXyr) are processed in the transmission circuit 164 and transmitted through the My antenna 152 over the wireless channel. At the access point 102, the uplink signals are received by the Mar antennas 116 and processed in the receiving circuit 124 to provide the received symbols (denoted by the vector of Reception mountains).

In the downlink case, the reception vector at the user terminal can be expressed as

Tap-Vsh NTar:khap equation (1) where Khap is a transmission vector with Mar elements for symbols transmitted through Mar antennas of the access point;

Iap is a reception vector with M elements for symbols received by M antennas in the user terminal;

Tar is a MarhMar diagonal matrix with elements representing complex amplifications connected to the transmission chain for Mear antennas at the access point;

Vsh is a Mih Mi diagonal matrix with elements representing complex amplifications connected to the receiving circuit for Mi antennas in the user terminal;

H is the MykhMar matrix of channel responses for the downlink.

The echoes of the transmit/receive circuits and the wireless channel are generally a function of frequency.

For simplicity, it is assumed that the channel is an amplitude fading channel (that is, with a uniform frequency response).

In the case of the uplink, the reception vector at the access point can be expressed as

Tor-- Nar-N"-Ti: Chor, equation (d) where Chor is the transmission vector for the symbols transmitted through We antennas of the user terminal;

Tyr is the reception vector for symbols received by Mar antennas at the access point;

Ti is a MiM diagonal matrix with elements in the form of complex amplifications connected to the transmission chain for Mi antennas in the user terminal;

Var is a MarhMar diagonal matrix with elements in the form of complex amplifications connected to the receiving circuit for Mear antennas at the access point;

N' represents the Marsh matrix of channel responses for the uplink.

For a TOO system, since the downlink and uplink share the same frequency range, there is usually a strong correlation between the downlink and uplink responses. Thus, the downlink and uplink response matrices can be viewed as reciprocals (i.e., transposed) of each other, and denoted as H, H", respectively, as shown in Equations (1) and (2) . However, the responses of the transmission/reception circuits at the access point, as a rule, do not coincide with the responses of the transmission/reception circuits in the user terminal. These differences lead to the inequality Var"NT-Ty 4(Vy:-NTar)".

From equations (1) and (2), the "effective" downlink and uplink responses, Nap and Hp, which include the responses of the transmit and receive circuits used, can be expressed as

Nap-Vy-NTar and Nor-Var"N"-Tya equations (3)

Combining these two equations and equation (3), the following ratio can be obtained:

V" Nap T"ar- (Var Nor Ta)! - (4) -Gd NTer V"zr equation

Transforming equation (4), we obtain the following:

NT -Ha V'sh Nap T"ar Var B Kish Nap Kar or

Nv - (Kya Nap Kar)! equation (5) where Ky -G "m Vy and Kar -T "ar Var Equation (5) can also be represented in the form:

Nar Ky "(Nap Kar)". equation (6)

The left side of equation (6) represents the response of the calibrated uplink channel, and the right side represents the transposed response of the calibrated downlink channel. The use of the diagonal matrices, Ki and Kar, in the effective responses of the downlink and uplink channels, as shown in equation (6), allows us to express the responses of the calibrated channels for the downlink and uplink as the result of transposing each other. (MarhMar) the diagonal Kar matrix for an access point is a relation

Var

Kar 5th t recall Var of the receiving chain to the recall Tar of the transmitting chain (ie, tar ), and the relation is an element-by-element relation. Similarly, the diagonal matrix Ky for the user terminal is the ratio of the response of the receiving circuit to the response of the transmitting circuit.

The Kar and Ki matrices include values that take into account the differences in the transmission/reception circuits of the access point and the user terminal. This allows the channel response of one line to be expressed in terms of the channel response of another line, as shown in equation (6).

Calibration can be performed to determine the Kar and Ky matrices. Typically, the true H-channel response and transmit/receive circuit responses are not known and cannot be easily and accurately obtained. In contrast, the effective downlink and uplink responses, Nap and Nor, can be estimated based on the downlink and uplink pilot signals, respectively, as described below. Then, estimates of the Kar and Ki matrices, called correction matrices "Kar and "Ki, can be derived based on the estimates of the downlink and uplink responses, "Nap" and "Nur", as described below.

The "Kar" and "Ky" matrices include correction factors that allow you to take into account the differences in the transmission/reception circuits of the access point and the user terminal.

Fig. 2 illustrates the use of correction matrices "Kar" and "Ki" to account for differences in the transmission/reception circuits of the access point and the user terminal. In the case of a downlink vector

The transmission chord is first multiplied by the matrix "Kar in block 112. Further processing in the transmission circuit 114 and the reception circuit 154 for the downlink is the same as shown in Fig. 1. Similarly, in the case of the uplink, the transmission chord vector is first multiplied by the matrix " And in block 162. Again, further processing in the transmit circuit 164 and the receive circuit 124 for the uplink is the same as shown in Fig.1. The "calibrated" downlink and uplink echoes visible in the user terminal and the access point, respectively, can be expressed as

Neap - Nap "Kar and Netsr - Nor "Ki equation (7) where N"sap and Netso are expressions for estimating the "true" responses of calibrated channels in equation (6).

Combining the two equations of the set of equations (7) using the expression from equation (6), it can be shown that

NetsreNsap. The accuracy of the NeoreN'sap ratio depends on the accuracy of the "Kar" and "Ki" matrices, which in turn, as a rule, depends on the quality of the estimates of the responses of the descending and ascending channels, "Nap" and "Nur.

As shown above, calibration can be performed in the LLP system to determine and account for differences in the responses of the transmit/receive circuits at the access point and the user terminal. After the transmit/receive circuits are calibrated, the calibrated channel response estimate obtained for one line (e.g., Neap) can be used to determine the calibrated channel response estimate for another line (e.g., Netsr).

The calibration methods described in this description can also be used for wireless communication systems that use ORBOM. With OROM, the entire frequency range of the system is effectively divided into several (Me) orthogonal subbands, which are also called frequency bins or subchannels. In the case of OBM, each subband is associated with a corresponding subcarrier, which can be modulated by data. For a MIMO system using OEBOM (ie MIMO-OROM system), each subband of each eigenmode can be considered as an independent transmission channel.

Calibration can be done in different ways. For clarity, a specific calibration scheme is described below for the TOU MIMO-VOLUME system. For such a system, each subband of the wireless link can be considered as reciprocal.

Figure 3 shows a block diagram of the sequence of operations of the process 300 of calibrating downlink and uplink echoes in the THAT MIMO-OROM system. First, the user terminal receives the timing and frequency of the access point using the reception procedures defined for this system (step 310). The user terminal may then send a message to initiate calibration by the access point, or the calibration may be initiated by the access point. Calibration can be performed in parallel with the registration/authentication of the user terminal by the access point (eg during call setup) and can also be performed as needed at any time.

Calibration can be performed for all subbands that can be used for data transmission (called "data" subbands). Subbands not used for data transmission (ie guard subbands) generally do not require calibration. However, since the frequency responses of the transmit/receive circuits at the access point and the user terminal are usually uniform over most of the frequency bands of interest, and since adjacent subbands are likely to be correlated, calibration can only be performed for a subset of the data subbands. If not all data sub-ranges are calibrated, then information about the sub-ranges intended for calibration (referred to as "designated" sub-ranges) may be sent to the access point (eg, in a message sent to initiate calibration).

For calibration, the user terminal transmits a MIMO pilot signal on the assigned subbands to the access point (step 312). MIMO pilot generation is described in more detail below. The duration of the MIMO transmission of the pilot signal on the uplink may depend on the number of assigned subbands. For example, 8 OROM symbols may be sufficient if the calibration is performed for four subbands, while more subbands may require a larger number (eg, 20) OEM symbols. As a rule, the total transmission power is fixed, so if the MIMO pilot signal is transmitted over a small number of subbands, then a higher level of transmission power can be used for each of these subbands and the SSS for each subband will be high. Conversely, if the MIMO pilot signal is transmitted over a large number of subbands, then a lower level of transmit power will be used for each subband, and the SSS for each subband will be worse. If the SSR for each subband is not high enough, then a larger number of OEOM symbols can be sent for the MIMO pilot, which are integrated in the receiver to obtain a higher overall SSR for that subband.

The access point receives the MIMO pilot on the uplink and outputs an uplink echo estimate, "Nof(K), for each of the assigned subbands, where K is the subband index. The channel estimate based on the MIMO pilot is described below. Estimates echoes of the uplink channels are diffracted and sent to the user terminal (step 314). The elements of each matrix "Nus(K) represent the complex channel gains between the My transmit and Mar receive antennas for the uplink for the K-th subband. Channel gains for all arrays can be scaled by a specific scaling factor that is common to all assigned subbands to obtain the required dynamic range. For example, channel gains in each matrix "Nur(k) can be equally scaled to the largest channel gain for all matrices "Nur(K) for the assigned subbands so that the value of the largest channel gain is unity. Since the purpose of the calibration is to normalize the gain/phase differences between the downlink and the uplink, absolute channel gains are not important. If 12-bit complex values (ie, with 12-bit in-phase (I) and 12-bit quadrature (0) components) are used to represent channel gains, then estimates of downlink responses can be sent to the user terminal in 3-Mui-Mar 'Mei in bytes, where "Z" is due to the fact that to represent | and O components use 24 bits and M is the number of assigned subbands.

The user terminal also receives the MIMO downlink pilot signal transmitted by the access point (step 316) and outputs an estimate of the downlink response, "Nap(K), for each of the assigned subbands based on the received pilot signal (step 318). The user terminal then determines the correction factors, "Kar(K) and "Kia(K), for each of the assigned subbands, based on estimates of the uplink and downlink responses, "Nur(K) and "Nap(K) (step 320).

To derive the correction factors, the downlink and uplink responses for each subband are assumed to be mutually inverse, with gain/phase corrections to account for differences in the transmit/receive circuits of the access point and the user terminal, given as "Na(yu Ka(yu - (National) Kar(k))", for (B)

KkeK equation where K is the set of all subranges of data. Since only estimates of the effective downlink and uplink responses for the assigned subbands are available during calibration, equation (8) can be rewritten as "Na(yu Ka(yu- (Nat) Kar(k))", for (9) kek of Eq. where K represents the set of all assigned subranges. The correction vector "Ksh(k) can be defined as including only My diagonal elements "Ka(K). Similarly, the correction vector "Kar(K) can be defined as which includes only Mar diagonal elements "Kar(K).

The correction factors "Kar(k) and "Ky(K) can be derived from the estimates of the channels "Natsk(K)" and "Nor(Yu) in various ways, including by calculating the ratio of matrices and calculation with the minimum root mean square error (MM5E) . Both of these calculation methods are described in more detail below. Other methods of calculation may also be used and are within the scope of this invention.

A. Calculation of matrix ratio

Fig. 4 is a block diagram of the sequence of operations of a variant of the implementation of the process 320a for deriving correction "Ka(K) and "Kar(K) from estimates of the responses of the downlink and uplink channels "Nuyr(K) and "Nap(K),

using matrix relation calculus. Process 320a can be used as step 320 of Fig.3.

First, for each assigned subrange, the (MikhMar) matrix of SCC) is calculated (step 412), as follows: "Noria or

In the (10) is for KeeEK equation where the ratio is calculated element by element. Each element of C(K) can thus be calculated as

Puri, juice Zv re! ,

I) for ich, Ma) 1 and )-1...Mar), equations where "Por IC) and "Pap i; (K) is the (i, Y)-th (row, column) element from "NTrt"N"ap, respectively, and su(K) is the (i, Y)-th element from C(K).

In one embodiment, the correction vector for the access point, "Kar(K), is determined to be equal to the average of the normalized lines C(K) and output in the steps of block 420. Each line C(K) is first normalized by scaling each of the Mar elements in a row to the first element in this row (step 422).

Thus, if c(K)--(Ci(K)...Ci,mar(K)| is the i-th line of C(K), then the normalized line syu) can be expressed as sik)

AIS... сКИсК)... Si,mar(K)/СДК)), (12) equation

The mean value of the normalized rows is then determined as the sum of My normalized rows divided by My (step 424). The correction vector "Kar(K) is determined to be equal to the specified average (step 426), which can be expressed as aolya 1 cha

Bad FIK), for KEKS, on, x (13) equation

As a result of normalization, the first element "Kar(K) is single.

In one embodiment, the correction vector for the user terminal, "Kj(K), is defined as equal to the average of the inverse values of the normalized columns of the SCC), and is determined at the steps of block 430. First, the ith column of C(K) is normalized by scaling each element in the column on the |Yth element of the vector "Kar(K), y Ky eDk) he Den, DYU Sh which is designated as Kar)(K) (step 432). Thus, if represents the |th column - SC -

C(K), then the normalized column syu) can be expressed as I am IK E o

B "Te OY Ko - BAKY KAK sia, MEM Kiy. equation (14)

Then, the average of the inverse values of the normalized columns is determined as the sum of the inverse values Mar of the normalized columns divided by Mar (step 434). The correction vector "Ksh(K) is determined to be equal to the specified average (step 436), which can be expressed as

Kk zu g for Kek,

Me BC) (15) equation. vk. . where the reciprocal values of normalized columns, si ) are obtained on an elemental basis.

B. MM5E calculations

For MM5E calculation, the correction factors "Kar(K) and "Kia(K) are derived from the estimates of the responses of the downlink and uplink channels "Nap(K) and "Nur(kK) in such a way that the root mean square error (M5E) between the response of the calibrated downlink channel and response of the calibrated upstream channel is minimal. This condition can be expressed as Ty schy EI SHA 2 is (YUK, (U YUK , for hek, (16) equation which can also be written as x fl Se y - 2 g tiv KN - VOK, for hek, where " Kar(K)-"Kar(K), since "Kar(K) is a diagonal matrix.

The restriction imposed on equation (16) is that the first element "Kad(K) is defined as equal to one (ie "Karod(K)-1). Without such a restriction, a trivial solution will be obtained in which all elements of the matrices "Kar(K) and "Ksh(kK) are equal to zero. In equation (16), the matrix X(Kk) is first obtained as

Zh(K)- "Kar(yu)"N"ats(k)-"Nuv(k) "Ka(Y). Then the square of the absolute value is obtained for each of the Mar-My elements of the matrix Zh(kK). The root mean square error ( or quadratic error, if the division by Mar"My is not carried out) at the same time is equal to the sum of all the squares of Mar-My values.

MM5E calculations are performed for each designated sub-range to obtain correction factors "Kar(K)" and "Ksh(K)" for this sub-range. MM5E calculation for one sub-range is described below.

For simplicity, the subrange index, K, is omitted in the following description. Also, for simplicity, the evaluation elements

AN"ap of the downlink response is denoted by (a), the elements of the "Nur" response of the uplink channel are denoted as (Bi), the diagonal elements of the "Kar" matrix are denoted by (y) and the diagonal elements of the "Ky" matrix are denoted by (m), where

I-411...Maru and ) 2 41...Mou.

The root-mean-square error can be rewritten, based on equation (16), as follows: me M, 2

ME "U vyh HB , zn food (17) of the equation and again taking into account the restriction shshch1-1. The minimum root mean square error can be obtained by calculating the partial derivatives of equation (17) with respect to and equating the partial derivatives to zero.

The result of these operations are the following sets of equations: im .

Ua ni aut t, for iz, v (Iva) and Eq

X,

UChA-V; 0, for Change (186) of Eq

In equation (18a) n1-1, therefore, for this case there is no partial derivative, and the index i changes from 2 to Mar.

The set of (Mar-My-1) equations in the sets of equations (18a) and (180) can be more conveniently expressed in matrix form as follows:

Au-7, equation (19) where

In Ky ah o -Kayu sh -beuyzhk, : o - a Hi -v nah -,, kh.bv, K. ! day dreams What and if ty rya ve: from henna: WE oh! as sh o U y - : seam, I sia, sh syak bu, a - 0 U !

Mr. A

In, 0 c o

Nh. 0 t k ko si - i

X M she Yai

In pov:

I fl BH che

Matrix A includes (Mar"My-1) rows, and the first Mar-1 rows correspond to Mar-1 equations from the set of equations (18a), and the last My rows correspond to My equations from the set of equations (186). In particular, the first row of matrix A is formed from the set of equations (18a) at i-2, the second row is formed at i-3, etc. The second row of matrix A is formed from the set of equations (180) at |-1, etc. And the last row is formed at |Й- We. As shown above, the elements of the matrix A and the elements of the vector 7 can be obtained based on the elements of the matrices "NTap" and "No.

Correction factors are included in the vector y, which can be obtained as y -A 72 of equation (20)

The result of the MM5E calculation is the correction matrices "Kar" and "Ki", which minimize the root mean square error of the responses of the calibrated downlink and uplink channels, as shown in equation (16). Since the "Kar" and "Ky" matrices are obtained based on the feedback estimates of the descending and ascending channels, "Nap" and "Ner", the quality of the "Kar" and "Ky" correction matrices thus depends on the quality of the estimates of the "Nap" and "Nr" channels.

MIMO pilot signal can be averaged in the receiver to obtain more accurate estimates for "Nap" and "No".

The correction matrices Kar and Ki obtained based on the MM5E calculation are generally better than the correction matrices obtained based on the matrix ratio calculation, especially when some of the channel gains are small and the measured noise can cause severe degradation channel amplification.

C Additional calculations

Regardless of the specific calculation method chosen to be used, after completing the calculation of the correction matrices, the user terminal sends to the access point the correction vector for the access point, "Kar(K), for all assigned sub-bands. If for each correction factor in "Kar(K) are used 12-bit complex values, then the correction vector "Kar(K) for all assigned subbands can be sent to the access point in 3-(Mar-1)Mer bytes, where the "3" results from the fact that for | and O components in 24 bits are used for the sum and (Mar-1) results from the fact that the first element in each vector "Kar(K) is one and therefore does not need to be transmitted. If the value 22-12:4511 is assigned to the first element, then a range of 12dB is available (since the maximum positive 12-bit value with a sign is 1211-1-:--2047), which makes it possible, using 12-bit values, to adjust discrepancies of up to 12dB in gains between the downlink and the uplink.

If the downlink and uplink match each other within 12dB, and the first element is normalized to 511, then the other elements must not exceed 511.4-2044 in absolute magnitude, and can be represented by 12 bits.

A pair of correction vectors "Kar(K)" and "Ka(k)" are obtained for each assigned subband. If calibration is not performed for all sub-ranges of the data, then correction factors for the "uncalibrated" sub-ranges can be obtained by interpolating the correction factors obtained for the designated sub-ranges. Interpolation can be performed at the access point to obtain correction vectors "Kar(K) for KeK. Similarly, interpolation can be performed at the user terminal to obtain correction vectors "Ka(K) for KeK.

Subsequently, the access point and the user terminal use their respective correction vectors "Kar(K) and "Kka(yu) or the corresponding correction matrices "Kar(K) and "Ka(K)" for KeK to scale the modulation symbols before transmission over the wireless channel , as described below. At the same time, the effective downstream channel visible from the user terminal is Neap(K) - Nap(K) Kar(K).

The calibration scheme described above, by which a vector of correction factors is obtained for both the access point and the user terminal, makes it possible to derive "compatible" correction vectors for the access point when calibration is performed by different user terminals. If the calibration at the access point has already been performed (for example, by one or more user terminals), then the current correction vectors can be updated using newly derived correction vectors.

For example, if two user terminals simultaneously perform a calibration procedure, then the calibration results from these user terminals can be averaged to improve performance. However, as a rule, calibration is performed for one user terminal only once.

Thus, the second user terminal sees a downlink for which the correction vector for the first user terminal has already been used. In this case, the product of the second correction vector and the old correction vector can be used as the new correction vector or "weighted averaging" (described below) can also be used. Typically, an access point uses one correction vector for all user terminals, rather than different correction vectors for different user terminals (although this option can also be implemented). Updates from multiple user terminals or sequential updates from a single user terminal can be handled in the same way, vector updates can be applied directly (using a multiplication operation). Alternatively, if some averaging is required to reduce measurement noise, weighted averaging can be used as described below.

Thus, if the access point uses correction vectors "Kari(K) to transmit the MIMO pilot signal from which the user terminal determines new correction vectors "Karg(K), then the updated correction vectors "Karz(K) are the result of multiplying the current and of new correction vectors. Correction vectors "Kari(K) and "Karg(K) can be displayed in one or different user terminals.

In one of the implementation variants, the updated correction vectors are defined as "Karz(K)-- "Kari(K)- Karg(K), and the multiplication is performed element by element. In another embodiment, the updated vector corrections can be redefined as "Karz(K) - "Kari(K) - K"arg(K), where a is the factor used to provide weighted averaging (ie, O«o«1) . If the calibration updates are infrequent, then a with a value close to one works better. If the calibration updates are frequent but noisy, then smaller values of a are preferred. Then the updated correction vector "Karz(K) can be used by the access point to their next update.

As indicated above, calibration may not be performed for all sub-ranges of data. For example, calibration may be performed for each nth subband, where n may be determined from the expected response of the transmit/receive circuits (eg, n may be 2, 4, 8, 16, etc.). Calibration can also be performed for non-uniformly distributed subbands. For example, since the filter characteristic may have a larger rolloff at the frequency band boundaries, which may create a greater mismatch in the transmit/receive circuits, more subbands may be calibrated at the frequency band boundaries. In general, any number of subbands distributed in any way can be calibrated, and it is within the scope of this invention.

In the above description, the correction vector "Kar(K) and Ki(K) for KeK" are output by the user terminal, and the vector "Kar(K) is sent to the access point. This scheme preferably distributes the calibration processing among the user terminals in the case of a multiple access system However, correction vectors "Kar(K) and "Ky(k) can also be output at the access point, which then sends the vector "Ksh(K) to the user terminal, and this is within the scope of this invention.

The calibration scheme described above allows each user terminal to calibrate its transmit/receive circuits in real time when transmitting over the radio channel. This allows user terminals with different frequency responses to provide high performance without strict frequency response requirements or manufacturing calibration. The access point can be calibrated by multiple user terminals to provide improved accuracy.

O. Strengthening

Calibration can be performed based on the normalized gains for the downstream and upstream channels, which represent the gain relative to the noise level in the receiver. After the downlink and uplink calibrations are performed, the use of normalized gains allows you to obtain the characteristics of one line (including the channel gain and SSS for each eigenmode) based on the gain measurements for the other line.

The access point and the user terminal can first balance the input levels of their receivers so that the noise levels in the receiving circuits of the access point and the user terminal are approximately the same. Balancing can be performed by estimating the noise level, i.e., determining the section of a received TOO frame (ie, a downlink/uplink transmission unit) that has the minimum average power over a specific time interval (eg, one or two symbol periods). In general, the time slot immediately before the start of each TOO frame is free of transmissions, since any uplink data must be received by the access point, and then a time-consuming receive/transmit switching must be performed before the access point starts transmitting on downline Depending on the circumstances of the obstacle, the noise level can be determined based on several TOO frames. Then the downlink and uplink responses are measured relative to this noise level. More precisely, the channel gain for a given subband of a given pair of transmit/receive antennas can be obtained, for example, as a ratio of received pilot symbols to transmitted pilot symbols for that subband of a given pair of transmit/receive antennas. In this case, the normalized gain is the measured gain divided by the noise level.

A large difference between the normalized gains for the access point and the normalized gains for the user terminal can cause the correction factors for the user terminal to be very different from unity. The correction factors for the access point are close to unity because the first element of the Kar matrix is set to 1.

If the correction factors for the user terminal are significantly different from unity, the user terminal may not be able to use the calculated correction factors. This may be due to the fact that the user terminal has a limit on its maximum transmission power and may not be able to increase the transmission power for large correction factors. In addition, reducing the transmission power for small correction factors is generally not desirable, as it may reduce the data rate.

Thus, the user terminal can transmit using a scaled version of the calculated correction factors. Scaled calibration factors can be obtained by scaling the calculated correction factors to a specific scale, which can be set equal to the difference in gains (as a difference or ratio) between the downlink and uplink responses. Such a difference in gains can be calculated as the average of the difference (or differences) between the normalized gains for the downlink and the uplink. The scale (or gain difference) used for the correction factors at the user terminal can be sent to the access point along with the calculated correction factors for the access point.

With correction factors and scale or gain differences, the downlink characteristics can be determined from the measured uplink response and vice versa. If the noise level at either the access point or the user terminal changes, the difference in gains may be updated, and the updated difference in gains may be sent in a message to the other entity.

In the above description, the calibration resulted in two sets (vectors or matrices) of correction factors for each subband, with one set used at the access point for downlink data transmission and the other set used at the user terminal for uplink data transmission. The calibration can also be performed in such a way that two sets of correction factors are provided for each sub-band, with one set used at the access point for uplink data reception and the second set used at the user terminal for downlink data reception.

The calibration can also be performed in such a way that one set of correction factors is obtained for each sub-band, and this set can be used either at the access point or at the user terminal. In general, the calibration is performed in such a way that the responses of the calibrated downstream and upstream channels are mutually inverse regardless of where the correction factors are applied. 2. MIMO pilot signal

For MIMO calibration, a pilot signal is transmitted by the user terminal on the uplink to enable the access point to estimate the uplink response, and a MIMO pilot signal is transmitted by the access point on the downlink to enable the user terminal to estimate the downlink response. The same or different MIMO pilots may be used for downlink and uplink, and the MIMO pilots used are known to both the access point and the user terminal.

In one embodiment, the MIMO pilot signal contains an OROM-defined symbol (denoted "P") that is transmitted through each of the Mt transmit antennas, where Mt-Mear is for the downlink and Mt-My is for the uplink. For each transmitting antenna, the same ORBOM symbol P is transmitted in each symbol period intended for MIMO transmission of the pilot signal. However, the OROM symbols P for each antenna are covered by different Walsh sequences with M elementary signals assigned to that antenna, where

M2Mar for the downline and MM for the upline. Walsh coverage maintains orthogonality between Mt transmit antennas and allows the receiver to distinguish between individual transmit antennas.

The OROM symbol P includes one modulation symbol for each of the Mz assigned subbands.

The OROM symbol P thus contains a certain "word" of the Modulation symbols which can be selected to facilitate the evaluation of the channel by the receiver. This word can also be defined to minimize changes in the peak-to-average ratio during MIMO transmission of the pilot signal. This reduces the amount of distortion and nonlinearity generated by the transmit/receive circuits, which in turn leads to improved channel estimation accuracy.

For understanding, the MIMO defined pilot signal for the defined MIMO-OROM system is described below. For this system, both the access point and the user terminal are equipped with four receive/transmit antennas. The frequency band of the system is divided into 64 orthogonal subbands (that is, Me-64), which are assigned indices from "31 to -32. Of these 64 subranges, 48 subranges (for example, with indices ш1,..., 6, 8,..., 20, 22,..., 263) are used for data, 4 subranges (for example, 7, 21) are used for pilot and possibly for signaling, the OS subband (with index 0) is unused, and the remaining subbands are also unused and serve as guard subbands. This structure of OEOM subbands is described in more detail in the document of the IEEE 802.11a standard, entitled "Raii 11: M/geie55 GAM Medisht Aszez5 Sopigo! (MAS) apa Rpuzis! Gaueg (RNU) zresiiysayope: Nidp-zreey Rpuzisa! And aueg ip She 5 SN? Vapa", September 1999, which is in the public domain and is incorporated herein by reference in its entirety.

ORBROM symbol P includes a set of 52 ORBC modulation symbols for 48 data subbands and 4 pilot subbands. The indicated OROM symbol P can have the following form:

RI valid a (0,0,0,0,0,0,-1,-1,-1,-1,1,1,3,-1, -1,1,-1,1,1,1 ,1,-1,-1,1,-1,1,-1,-1,-1,-1,1,-1, 0,1,-1,-1,-1,-1,1 ,-1,-1,-1,-1,1,1,-1,-1,1,-3,-1,1,1, -1,1,-1,1,-1,1, -1,0,0,0,0,0),

P(IMAGINARY;)ao(0,0,0,0,0,0,-1,1,1,1,-1,-1,1,-1,1,1,1,-1,1, -1,-1,-1,-1,-1,-1,1,1,-1,1,1,-1,1, 0,-1,-1,-1,-1,1, 1,-1,1,-1,-1,1,-1,1,-1,3,1,1, AND

B 1-1,0,0,0,0, 0), where d is the gain for the pilot signal. Values inside parentheses 4) are given for subranges with indices from -32 to -1 (for the first line) and from 0 to "31 (for the second line). Thus, the first line for

P(real) and P(imaginary) indicates that the symbol (-1-)) is transmitted in the -26 subrange, the symbol (-1--)) is transmitted in the -25 subrange, and so on. The second line for P(real) and P(imaginary) indicates that the symbol (1-)) is transmitted in subrange 1, the symbol (-1st) is transmitted in subrange 2, and so on. Other OROM symbols can also be used for the MIMO pilot signal.

In one of the implementation options, the four transmission antennas for the MIMO pilot signal are assigned Walsh sequences M/l-1111, M/2-1010, M/3-1100 M/4-1001. For a given Walsh sequence, a value of "1" indicates that a P symbol is transmitted by OROM, and a value of "0" indicates that a -P symbol is transmitted by OEB (ie, each of the 52 modulation symbols in P is inverted).

Table 1 lists the OEOM symbols transmitted through each of the four transmitting antennas during MIMO transmission of a pilot signal with a duration of 4 symbol periods.

Table 1

The symbol BOTH Antenna 2 Antenna 3 Antenna 4 up and down

For longer MIMO pilot transmissions, the Walsh sequences for each transmit antenna are repeated. For such a set of Walsh sequences, MIMO transmission of the pilot signal is performed for a number of symbol periods multiples of 4 symbol periods to guarantee orthogonality between the four transmit antennas.

The receiver can derive an estimate of the channel recall based on the received MIMO pilot signal by performing complementary processing. In particular, to recover the pilot signal transmitted through the transmitting antenna and received by the receiving antenna |, the pilot signal received by the receiving antenna ) is first processed with the Walsh sequence assigned to the transmitting antenna and the complementary Walsh coating method performed in the transmitter.

The stripped OROM symbols for all Mr symbol periods for the MIMO pilot are then summed, with the summing performed individually for each of the 52 subbands used for the MIMO pilot transmission. The summation result is "P(K) for Ksh1,...,26) which is an estimate of the effective channel response from the transmitting antenna to the receiving antenna and (i.e., including the responses of the transmit/receive circuits) for 52 data subbands and signal pilot.

The same processing can be performed to restore the pilot signal from each transmitting antenna to each receiving antenna. The processing of the pilot signal provides the Mar-My values, which are the elements of the evaluation of the effective channel response, "Nur(KU)" and "Nat(K)" for each of the 52 subbands.

The channel estimation described above can be performed by both the access point and the user terminal during calibration to obtain an estimate of the effective uplink response, "Nur(K), and an estimate of the effective downlink response, "NacK), respectively, which is then used to derive correction factors as described above. 3. Spatial processing

To simplify channel estimation and spatial processing at the access point and user terminal for TOU

In MIMO and MIMO-OROM systems, the correlation between the downlink and uplink responses can be used. Such a simplification is possible after performing a calibration to account for differences in the transmit/receive circuits. As indicated above, the responses of the calibrated channels are:

Neapts(k) - Nap(K) Kaf(K), equation (21a) for the descending line and

Netsb(k)-(Nats(K) Kar(k)) Nyr(K)Kiv(K), (216) equations for the ascending line.

The last equality in equation (210) appears due to the relationship between the responses of the effective downstream and upstream channels,

Nyr(k)-(K"ap(K) Nap(k) Kar)"

The channel response NI(K) matrix for each subband can be "diagonalized" to obtain the M5 eigenmodes for that subband. This can be achieved either by decomposing the channel response matrix H(K) by singular values, or by decomposing the correlation matrix by eigenvectors for the NIK), which is B(K) - NA(UNIK).

The singular value expansion of the response matrix of the calibrated upstream channel Ner(K) can be expressed as

Neo(k)-Oar(yu)x(Y MNK), for KEK (215) the equation where Oar(K) is a unitary matrix of left eigenvectors for Neor(K);

X(K) is a (Mikh Mar) diagonal matrix of singular values for Neor(K); and

Ma(K) is a unitary matrix of right eigenvectors for Netsr(K);

The unitary matrix M is characterized by the property МУМ - И, where И is a unit matrix.

Accordingly, the singular value expansion of the response matrix of the calibrated downstream channel, Neap(K), can be expressed as

Neak)- SOFT (KYUSH Tar(K), for KeK (23) equation

Thus, the matrices Mya(K) and OtTar(K) are also matrices of left and right eigenvectors, respectively, for

NaapYu). Matrices Ma(Kk), Mya(K), MyaZhK) and MNd(kK) are different forms of the matrix Msh(K) and the matrix Ozr(K),

Шар(К), ОТар(К) and ОНар(К) are also different forms of the matrix О2р(К). For ease of reference, the matrices О2р(К) and

Ma(K) in the description below may also be a reference to their various forms. Matrices Ozr(K) and

Ma(K) are used in the access point and the user terminal, respectively, for spatial processing, and are defined as such by their subscripts.

Decomposition by singular values is described in more detail in the work of biiregi bigapo, entitled "I ipeag AIdebga apa 5 Arriisayop5", second edition, Asadetis Rgez5, 1980.

The user terminal may perform an estimate of the calibrated downlink recall based on

BYPASSING the pilot signal transmitted by the access point. After that, the user terminal can perform a singular value decomposition of the calibrated downlink response estimate, Neap(K), for KeK, to obtain the diagonal matrices "Х(К) and matrices M"x(K) of the left eigenvectors for NeacK). Such a singular value expansion can be described as Neap(k)-M'"(K) "x(K) "OtTar(K), where the sign ("7") above each matrix indicates that it is an estimate of a real matrix .

Similarly, the access point can perform an estimate of the calibrated uplink callback based on

MIMO pilot signal transmitted by the user terminal. Then the access point can perform the decomposition on the singular values of the response estimate of the calibrated uplink channel, "Neor(K), for KeK, to obtain the diagonal matrices "Х(К) and the matrices "Ц2р(К) of the left eigenvectors for "Нед(К), for KeK Such a decomposition by singular values can be described as "Y

Due to the fact that the channel and calibration are mutually inverse, the singular value decomposition can be performed either only at the user terminal or only at the access point to obtain both the matrices ku(k) and the matrices "Ozr(K). In the case of execution in the user terminal, the matrices "Msh(k) are used for spatial processing in the user terminal, and the matrices "Ozr(k) can be transmitted to the access point.

The access point may also be able to obtain the matrices "Shzr(K) and "X(K) based on the directional reference signal transmitted by the user terminal. Similarly, the user terminal may also be able to obtain the matrices "Mj(k) and "X(K) based on the directional reference signal transmitted by the access point. A directional reference signal is described in detail in the above-mentioned preliminary US patent application Meb0/421309.

Matrices "Ozr(k) and "7X(K) can be used to transmit independent data streams on M5 own modes of the MIMO channel, where M5 is the type of Mar,My). Spatial processing for the transmission of multiple downlink and uplink data streams is described below.

A. Spatial processing for the uplink

Spatial processing at the user terminal for uplink transmission can be expressed as

Khor(k)-"Ky(k MK) Bir(Kyu), for KEK (24) equation where khor(K) is the transmission vector for the uplink for the K-th subband; and 5er(K) is the "data" vector » with non-zero elements, the number up to M5, for modulation symbols intended for transmission via M5 of the K-th subband's own modes.

Also, additional processing of modulation symbols can be done before transmission. For example, for subbands of data (for example, for each eigenmode), channel inversion can be applied so that the received SSW is approximately the same for all subbands of data. At the same time, spatial processing can be expressed as

Khir(K)- "Kak MKM or(K)Bor(K), FOR

KeEK (25) of the equation where M/ts(K) is the weight matrix for (optional) inversion of the upstream channel.

Channel inversion can also be performed by assigning transmission power to each subband before performing modulation, in which case the vector z5ur(K) includes the channel inversion coefficients, and the matrix M/d(K) in equation (25) can be omitted. In the description below, the use of the matrix M/ur(K) indicates that the channel inversion coefficients are not included in the vector zuf(K). The absence of the matrix M/ur(K) in the equation may indicate that (1) or the channel inversion is not performed, (2) or the channel inversion is performed and taken into account in the vector zur(K).

Channel inversion can be performed as described in the aforementioned prior US patent application No. 60/421309, and in US patent application No. 10/229209, entitled "Sodey MIMO Zuzietv5 m/yp ZeiIesiime Spappei!

Impegsiop Arriei Reg Eideptode'', filed August 27, 2002, the rights to which belong to the assignee of this patent application and which is incorporated herein by reference in its entirety.

The received uplink transmission at the access point can be expressed as

Tor(K)-Nov(K)Khir(K)-"p(K), for KeK (2c) I Mar(Y)X(K)Boo(K) p(K) equation where hor(K) is received vector for the uplink for the K-th subband; n(K) is additive white Gaussian noise (AUMOM) for the K-th subband; and

Khir(Kk) is given by equation (24).

Spatial processing (or matched filtering) at the access point for a received uplink transmission can be expressed as

Vyr(k)A xx (KU OChar(K)hor(K), -y

Chkulonar(KYuar(KU IC) Vir(K)-p(K)) (26) -Avu(K) o (Y) for KeK equation where "5ur(K) is an estimate of the vector zur(K) transmitted by the user terminal upstream, and n(u) represents the post-processing noise Equation (27) assumes that no channel inversion is performed at the transmitter and that the received vector gur(kK) has the form represented by Equation (26).

B. Spatial processing for the downlink

Spatial processing at the access point for downlink transmission can be represented as

Khap(k)-"Kar(Kyu)"O"ar(K)vap(K), for

KeEK (28) equation where hap(K) is the transmission vector, and zats(K) is the data vector for the downlink.

Again, additional processing (such as channel inversion) may be performed on the modulation symbols prior to transmission. At the same time, spatial processing can be expressed as

Khap(k)-"Kar(yu"O"ar(Yu)MMap(K)vap(K), for KeK (29) equation where MMap(k) is the weight matrix for (optional) downlink inversion.

The received downlink transmission at the user terminal can be expressed as

Tats(K)--Nap(K)Khap(K)--p(K), for KeK equation (30)

Matsyuk vap(K) kpc) where hap(K) is the transmission vector represented by equation (28).

The spatial processing (or matched filtering) at the user terminal for the received downlink transmission can be expressed as

K sn groin t ;

Mane V (M ek (UK) sr YeoovAinE I e B NO in (VOYU), for keK (31) ev (K) NVK) equation

Equation (31) assumes that the channel inversion is not performed in the transmitter and that the received vector

Tats(K) has the form represented by equation (30).

Table 2 presents the spatial processing at the access point and the user terminal for data transmission and reception. Table 2 assumes that additional M/(K) processing is performed at the transmitter. However, if such additional processing is not performed, M/(K) can be treated as a unit matrix.

Table 2

SHI Ascending line. ! C Descending line !

AI terminal Transmission: ! Reception: ! which set g Y au, h cha k o: as

Access point. Reception: Transfer: ku vy lna vod V AN for Kob pz'syuvsv

In the above description and as shown in Table 2, correction matrices "Kav(yu)" and "Ka(Kyu)" are used for spatial processing during transmission at the access point and the user terminal, respectively. This can simplify the overall spatial processing, since in any case (for example, for channel inversion) scaling of the modulation symbols may be required, and the correction matrices "Kar(K) and "Ky(K) can be combined with the weighting matrices UMapc(Kk) and UMur(K) to obtain matrices Sap(K) and (Zor(K)) amplification, where

Sap(k)-MMap(K) Kar(K) and Syr(k)-UMor(K)"Ky(k). Processing can also be performed in such a way that correction matrices are used for spatial processing at reception (instead of spatial processing at 4. MIMO-OBOM system

Fig. 5 is a block diagram of an embodiment of the access point 502 and the user terminal 504 in the TOU MIMO-OREOM system. For simplicity, the following description assumes that both the access point and the user terminal are equipped with four receive/transmit antennas.

In the case of downlink at access point 502, transmission (TX) data processor 510 receives traffic data (i.e., information bits) from data source 508 and signaling and other information from controller 530. TX data processor 510 formats, encodes, interleaves, and modulates (i.e. symbol mapping) of data to provide a stream of modulation symbols for each proprietary mode used for data transmission. TX spatial processor 520 receives streams of modulation symbols from TX data processor 510 and performs spatial processing to provide four streams of transmission symbols, one stream for each antenna. The TX spatial processor 520 also performs additional multiplexing of the pilot symbols as needed (eg, for calibration).

Each modulator (MOB) 522 receives and processes a corresponding stream of transmission symbols to provide a corresponding OROM symbol stream. Each OROM symbol stream is further processed in the transmission circuit in modulator 522 to provide a corresponding modulated downlink signal. Then, the four modulated signals from the modulator 522a-5224 are transmitted through the four antennas 524a-524a, respectively.

At the user terminal 504, antennas 522 receive the transmitted modulated downlink signals and each antenna provides the received signal to a corresponding demodulator (OEMOB) 554. Each demodulator 554 (which includes a receive circuit) performs processing complementary to that performed in the modulator 522, and provides accepted characters. The receiving (EX) spatial processor 560 then performs spatial processing from all demodulators 554 to provide recovered symbols that are estimates of the modulation symbols transmitted by the access point. During EX calibration, the spatial processor 560 provides an estimate of the calibrated downlink, Neac((K), based on the MIMO pilot signal transmitted by the access point.

EX data processor 570 processes (eg, performs reverse symbol mapping, reverse interleaving, and decoding) the recovered symbols to provide decoded data. The decoded data may include recovered traffic data, signaling, etc. and is provided to the data consumer 572 for storage and/or to the controller 580 for further processing. During EX calibration, the data processor 570 provides an estimate of the calibrated uplink, "Neur(K), which is output at the access point and transmitted on the downlink.

Controllers 530 and 580 control the operation of various processing units in the access point and the user terminal, respectively. During calibration, the controller 580 can take estimates of "Nesap(K) and "Neor(K) of channel responses, output correlation matrices "Kar(K) and "Kia(Kk), provide matrices "Ky(K) to TX spatial processor 592 for transmission on the uplink and provide the "Kar(k)" matrix to the TX processor 590 data for transmission to the access point.

Memory devices 532 and 582 store data and program codes used by controllers 530 and 580, respectively.

Uplink processing may be the same as or different from downlink processing. Data and signaling are processed (eg, encoded, interleaved, and modulated) in TX data processor 590 with further spatial processing in TX spatial processor 592, which also performs additional multiplexing of pilot symbols during calibration. Pilot symbols and modulation symbols are further processed in modulators 554 to generate modulated uplink signals, which are then transmitted via antennas 552 to the access point.

At access point 110, modulated uplink signals are received by antennas 524, demodulated in demodulators 522, and processed in EX spatial processor 540 and EX data processor 542 in a manner complementary to that performed in the user terminal. During EX calibration, the spatial processor 560 also provides an estimate of the calibrated uplink channel based on the MIMO pilot signal transmitted by the user terminal. in the user terminal.

Fig.b6 is a block diagram of TX spatial processor 520a, which can be used as TX spatial processors 520 and 592 of Fig.5. For simplicity, the following description assumes that all four custom mods are selected for use.

In processor 520a, demultiplexer 632 accepts four streams of modulation symbols (labeled 51(n)-54(n)) for transmission in four eigenmodes, demultiplexes each stream into Mo substreams for Mo subbands of data, and provides four substreams of modulation symbols for each subband into the appropriate TX spatial processor 640 subband. Each processor 640 performs the processing described by equation (24), (25), (28) or (29) for one subband.

In each subband TX spatial processor 640, four substreams (denoted 51(K)-54(K)) of modulation symbols are provided to four multipliers 642a-6424, which also accept gain d(k), o2(K), 9z(K) and 94(K) for the four eigenmodes of the corresponding subband. In the case of the downlink, the four gains for each data subrange are diagonal elements of the corresponding matrix Sap(k), where Sap(K) - Kar(Y) or Sap(k)-UMar(K) Kar(K). In the case of the ascending line, the gain represents the diagonal elements of the matrix Ср(к), where Сур(к) - Ки(КУ) or Сур(к) -UMor(К) Ки(кК). Each multiplier 642 scales its modulation symbols by a corresponding gain dt(K) to provide scaled modulation symbols.

Multipliers 642a-642a provide four streams of scaled modulation symbols to four beamformers 650a-650a, respectively.

Each beamformer 650 performs beamforming to transmit one sub-stream of symbols in one eigenmode of one sub-band. Each beamformer 650 accepts one substream of 5t(K) scaled symbols and performs beamforming using its own vector mt(K) for the corresponding eigenmode. In each beamformer 650, the scaled modulation symbols are fed into four multipliers 652a-652a, which also accept the four elements uUt(K), Mt.2(K), Uta(K) and mt«(K), the eigenvector mt(K) for appropriate own fashion. The eigenvector mi(K) represents the th column of the matrix "O"ar(K) for the downlink, and represents the th column of the matrix "Ud(K) for the uplink. Each multiplier 652 then performs the multiplication of the scaled modulation symbols to the appropriate mt/(K) value of the eigenvector to provide the “beamforming” symbols.Multipliers 652a-6524 provide four substreams of beamforming symbols (which are intended to be transmitted over the four antennas) to adders 6bOa-6604, respectively.

Each adder 660 receives and sums the four beamforming symbols for the four eigenmodes for each symbol period to provide a preprocessed symbol for the corresponding transmit antenna. Adders 6bb0a-660a provide four substreams of pre-processed symbols for four transmission antennas to buffers/multiplexers 670a-6704, respectively.

Each buffer/multiplexer 670 receives pilot symbols and preprocessed symbols from TX spatial processors 640 subbands for Mo subbands of data. Each buffer/multiplexer 670 then multiplexes pilot symbols, preprocessed symbols, and zeros for pilot subbands, data subbands, and unused subbands, respectively, to form a sequence of Me transmission symbols for a given symbol period. During calibration, pilot symbols are transmitted on the assigned subbands. Multipliers bbva-66v8va cover the pilot symbols for the four antennas with Walsh sequences UMi-U/4, respectively, assigned to the four antennas, as described above and shown in Table 1. Each buffer/multiplexer 670 provides a stream of x(n) transmission symbols for one transmission antennas, and the stream of transmission symbols contains serially connected sequences of Me transmission symbols.

Spatial processing and OEBOM modulation are described in more detail in the aforementioned provisional US patent application Mob0/421309.

In various embodiments of the present invention set forth in this description, peer-to-peer communication between different user terminals (SHT or 5TA) within the same basic service area (855) or different B55 can be implemented, as described below. A ST or 5TA that has performed calibration with a single access point (AP) is a member of the basic service area (855). One access point is a common node for all OTs in B55. The calibration methods described above facilitate the following types of communication: () the OT in the B55 can use directional transmission (TX) to communicate directly with the AP on the uplink (U), and the AP can use directional transmission (TX) to communicate with a downlink OT (01. (its OT in B55 can directly exchange data with another ST in the same B55 using forward communication. In this case, such peer-to-peer communication must be initiated in advance, since none of the OTs has information about the channel between them. In various implementations, the pre-initiation procedure works as follows: - the initiator of the peer line is the AR source (VAR) and the other OT is the OT receiver (0ST) - the VAR sends a MIMO pilot signal for the ST together with a line establishment request that contains B55 IO and pAR I. The request must be sent in general mode (ie, with transmission diversity).

IO, his B55 I and some baud rate indicator for use in OAR. - Then SAR can use forward transmission on 0 and OT can use forward transmission on CI. Speed control and tracking can be carried out by dividing gears into segments 0. and ЦИ. with sufficient time intervals between them to perform the processing. (ii) OTs belonging to one B55 (eg B551) can use forward transmission to OTs belonging to another B55 (eg B552), even if each has calibrated with a different AR. However, in this case there is an uncertainty in the phase angle (for each subband). This is due to the fact that the calibration procedure described above establishes a ratio that is unique to the AP with which the calibration was performed. The specified ratio is a complex constant, a(k, i) - ZARTX (0) 9АРАХ(О) where K is a subrange index and | represents the AP index, and 0 represents the index of the reference antenna (for example, antenna 0) used in the AP. In one embodiment, this constant is common to all STs in a given B55, but is independent for different B55s.

As a result, when the ST with B551 exchanges data with the OT in B552, forward communication without correction or compensation for this constant can result in a phase shift and scaling of the amplitude of the entire own system. The phase shift can be determined by using a pilot signal (directed or non-directional) and removed in the receivers of each respective ST. In one embodiment, the amplitude correction or compensation may be a conventional scaling of the SSS and may be removed by estimating the noise level in each receiver, which may affect the selection of the transmission rate.

In various embodiments, peer-to-peer exchange between OTs belonging to different B55s can be performed as follows: - the initiator of the peer line (for example, OT B551) is the AP source (AP), and another OT (for example, OT

B552) is an OT-receiver (0SHT). - The VAR sends the MIMO pilot signal for the ST together with the line establishment request, which contains B55 IO and pAR I. The request must be sent in general mode (ie, with transmission diversity). - The IOT responds by sending a pilot signal sent by the MIMO and a confirmation containing the OT

IO, his B55 I and some baud rate indicator for use in OAR. - The VAR receiver (Kx) can estimate the phase shift for the uplink (UP) and apply a correction constant for each subband. The OAR may then use forward downlink transmission (0) but must include the forward reference preamble in at least the first forwarded packet to enable the receiver FT (Ex) to perform phase shift correction or compensation in the OI for each subband. A forward reference preamble may not be required for subsequent transmissions on 0. Rate control and tracking may be accomplished by splitting transmissions into 01 and CI segments with sufficient time intervals between them for processing.

The calibration methods described in this description can be implemented using various means.

For example, these methods can be implemented in the form of hardware, software, or a combination thereof. In the case of implementation in the form of hardware, the methods can be implemented at the access point and the user terminal in one or more application-oriented integrated circuits (ABIS), digital signal processors (DSPs), digital signal processing devices (DSPs), programmable logic devices (PLDs 0), internally programmable gate arrays (ERBA), processors, controllers, microcontrollers, other electronic units made with the possibility of performing the functions outlined in this description or their combination.

In the case of implementation in the form of software tools, calibration methods can be implemented using modules (for example, procedures, functions, etc.) that perform the functions outlined in this description.

Program codes can be stored in a memory device (eg, memory devices, 532 and 582 of Fig.5) and executed by a processor (eg, controllers 530 and 580, respectively).

The storage device can be made in the processor or external to the processor, and in this case it can be connected with the possibility of data exchange with the processor using various means known in the art.

Headings are included in this description for reference and to aid in locating specific sections. These headings should not be construed as limiting the scope of the concepts in the sections they are titled, and these concepts may be used in other sections throughout the specification.

The above description of the disclosed embodiments is provided to enable any person skilled in the art to utilize the present invention. Various modifications to the specified embodiments should be apparent to those skilled in the art, and the general principles set forth herein are applicable to other embodiments without departing from the spirit and scope of the present invention. Thus, the present invention is not to be limited to the embodiments disclosed herein, but rather is intended to be given the broadest scope consistent with the principles and novel features disclosed herein.

That's Kim Jin's roofs

Ne pe "Be ta. pepknefttnnkya mm t ote psrsomyutkyaya V re

The same Ei-pen gi tv the same and 'N. Yannia Sov and -- it's no Irvovamkoi el lanhnht -- llemnAyn

EE pom | heh

FAIRY lomyuyum Homy meme forever. de te te me te den k7 ti y Sh te 7 Eh ee . and puyuyeovnnkih and DEN shi and

MN A sh with Zh

Mb TO What are you swarming

Ouch! "h what

Timing and frequency 8 of system 1 are received

The MIMO pilot signal is transmitted on the assigned subbands to the access point.

Do they accept the otsika odklaku shchi and the ascending channel from ? access folders

They accept piloz-sisnal on 6 steady lines from the access point

They perform an evaluation of the call. ! downlink, based 96 on the downlink body signal

Determine the equalization factors: for the point of access and the terminal from the user, based on wai. evaluations of uplink and downlink responses

Fig. 3

The beginning of I zi and z

They calculate the matrix E as 2 decodes of feedback estimates of the outgoing and outgoing channels rn tk tyuki t pn yuchtyu nt menu ny ny ty ky ky ty nt y i 2 i Feed each line 422

H matrix C and ! and eat

Н Determine the average for p E ! normalized lines and N

R; ; with R n ; Determine the correction vector '. yeah

Н Because the melon point will be reached as an equal! on average, for normalized strings, the phoneme zh nyan fem mk mae chek ty voy yayu ny vosav ovo X

Dt same her netto tyh tet tet sia tt

And - about I

І Normalize each column 12 " " of the matrix С К-

Eh

And 5 x ry, and mo, Determine the average for and N' inverse values of the formed

H columns of H with 4 AND Y

Н Determine the vector vector Н

GI Mk' Ky for the terminal as equal to . on average for inverse , ; values of normalized columns and z i ' i ozh ey pila kana ana nn y

The end of fig. 4 ke i a ze "i i-

B 5 uki otoi de a ischa stu a -k pesto, ryh roku M stuk re wash / Eh ohzhaki B tu lhn- Winn ve KN YN ek pe DMK Di V penya blyu How - ve same - how in mloteitko Inn pan y, mob THIS Monday and how" you and rya V, ud y KV B with U vor rya EN SYA AN still so. yr

I 1X erzhsttogog cut, li DplYh ŽI klyu 1 Х 2 і Х t antiptetti tyti v cheki --ik

N xh pyt TEMI ZhK r, zh i Y ! Fairy ze

N c) which 1. SCHI zom N ska chi N be K

Ша яни я зів уко "be ZE іі коди гі YO у смюхких 1 else where. Е і і - Nisha: mouth other І in the attraction of tli | o | sleep ' svi Я АЮ Хв І --l I Mu dezkntih ha іe: wai not

Not in it

I vfuyuy pdhiszh hk, I , mol yukh t caught ko.