HK1114962A - Channel calibration for a time division duplexed communication system - Google Patents

Description

Channel calibration for time division duplex communication systems

Background

I. Field of the invention

The present invention relates generally to communication, and more specifically to techniques for calibrating downlink and uplink channel responses in a Time Division Duplex (TDD) communication system.

II. background

In a wireless communication system, data transmission between an access point and a user terminal is performed over a wireless channel. The downlink and uplink may use the same or different frequency bands depending on the system design. The downlink (or forward link) refers to the communication link from the access points to the user terminals, and the uplink (or reverse link) refers to the communication link from the user terminals to the access points. If two frequency bands are available, separate frequency bands may be allocated to the downlink and uplink using Frequency Division Duplexing (FDD) techniques. If only one frequency band is available, the downlink and uplink may share the same frequency band using Time Division Duplexing (TDD).

To achieve high performance, it is often necessary to know the frequency response of the wireless channel. For example, the access point may require a downlink response to perform spatial processing for downlink data transmission to the user terminal (described below). The downlink channel response may be estimated by the user terminal based on pilots transmitted by the access point. The user terminal may then send the downlink channel response estimate back to the access point for its use. For such channel estimation schemes, the need to send the pilot on the downlink and send the channel estimate back to the access point incurs additional delay and resources.

For a TDD system with a common frequency band, the downlink and uplink channel responses may be assumed to be reciprocal to each other. That is, ifHRepresenting the channel response matrix from array A to array B, the reciprocal channel means that the coupling from array B to array A is byH ^TIs given inH ^TRepresentation matrixHThe transposing of (1). Thus, for a TDD system, the channel response for one link may be based on anotherThe pilot sent on one link. For example, the uplink channel response may be estimated based on pilots received via the uplink, and a transpose of the uplink channel response estimate may be used as the estimate of the downlink channel response.

However, the frequency response of the transmit and receive chains at the access point is typically different from the frequency response of the transmit and receive chains at the user terminal. In particular, the frequency response of the transmit and receive chains for uplink transmissions may be different from the frequency response of the transmit and receive chains for downlink transmissions. The "effective" downlink channel response (including the response of the applicable transmit and receive chains) will thus differ from the reciprocity of the effective uplink channel response due to differences in these transmit and receive chains (i.e., the effective channel response is not reciprocal). If the reciprocity of the obtained channel response for one link is used for spatial processing on the other link, any difference in the frequency responses of the transmit and receive chains will represent an error that, if not determined or accounted for, may degrade performance.

Accordingly, there is a need in the art for techniques to calibrate the downlink and uplink in a TDD communication system.

Summary of the invention

Techniques are provided herein to calibrate downlink and uplink channels to account for differences in frequency responses of transmit and receive chains at an access point and a user terminal. After calibration, the obtained estimate of the channel response of one link may be used to obtain an estimate of the channel response of the other link. This may simplify channel estimation and spatial processing.

In a particular embodiment, a method of calibrating downlink and uplink in a wireless TDD multiple-input multiple-output (MIMO) communication system is provided. According to this method, a pilot is transmitted on the uplink channel and used to derive an estimate of the uplink channel response. A pilot is also transmitted on the downlink channel and used to derive an estimate of the downlink channel response. A correction factor for the access point and a correction factor for the user terminal are then determined based on the downlink and uplink channel response estimates. The access point may apply its correction factor to its transmit side, or to its receive side, or to both transmit and receive sides. The user terminal may also apply its correction factor to its transmitting side, or to its receiving side, or to both the transmitting and receiving sides. In case the access point applies its correction factor and the user terminal also applies its correction factor, the response of the calibrated downlink and uplink channels is approximately reciprocal. These correction factors may be determined using matrix ratio calculations or Minimum Mean Square Error (MMSE) calculations for the downlink and uplink channel response estimates as explained below.

This calibration may be performed in real time based on transmission over the air. Each user terminal in the system may perform calibration in a manner that a correction factor thereof is derived by one or more access points. Similarly, each access point may perform calibration in a manner in which its correction factors are derived by one or more user terminals. For an Orthogonal Frequency Division Multiplexing (OFDM) system, this calibration may be performed on a set of frequency subbands to obtain a correction factor for each frequency subband in the set. Calibration factors for other "uncalibrated" frequency sub-bands may be interpolated based on the obtained calibration factors for the "calibrated" frequency sub-bands.

Various aspects and embodiments of the invention are described in further 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.

Fig. 1 shows the transmit and receive chains at an access point and a user terminal in a MIMO system.

Fig. 2A shows the application of correction factors at both the transmitting and receiving sides at the access point and user terminal.

Fig. 2B illustrates applying correction factors to the transmit side at both the access point and the user terminal.

Fig. 2C illustrates applying correction factors to the receiving side at both the access point and the user terminal.

Fig. 3 illustrates a process of calibrating downlink and uplink channel responses in a TDD MIMO-OFDM system.

Fig. 4 shows a process of deriving an estimate of a correction vector from downlink and uplink channel response estimates.

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

Fig. 6 is a block diagram of a Transmit (TX) spatial processor.

Detailed description of the invention

These calibration techniques described herein may be used for various wireless communication systems. Further, these techniques may be used for single-input single-output (SISO) systems, multiple-input single-output (MISO) systems, single-input multiple-output (SIMO) systems, and multiple-input multiple-output (MIMO) systems.

MIMO systems employing multiple (N)_TMultiple) transmitting antenna and multiple (N)_RMultiple) receive antennas for data transmission. From this N_TA transmitting antenna and N_RThe MIMO channel formed by the receiving antennas can be decomposed into N_SA separate channel of which N_R≤min{N_T，N_R}. This N_SEach of the independent channels is also referred to as a spatial channel 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. This typically requires an accurate estimate of the channel response between the transmitter and the receiver.

Fig. 1 shows a block diagram of transmit and receive chains at an access point 102 and a user terminal 104 in a MIMO system. For this system, the downlink and uplink share the same frequency band in a time division duplex manner.

For the downlink, at access point 102, the symbols (from the "transmit" vector)x _dnRepresented) is processed by transmit chain 114 and is driven from N_apThe antennas 116 transmit over a wireless channel. At the user terminal 104, the downlink signal consists of N_utReceived by antennas 152 and processed by receive chains 154 to obtain received symbols (represented by "receive" vectors)r _dnRepresentation). The processing by transmit chain 114 typically includes digital to analog conversion, amplification, filtering, frequency up conversion, and so forth. The processing by receive chain 154 typically includes frequency down conversion, amplification, filtering, analog to digital conversion, and so forth.

For the uplink, at the user terminal 104, the symbols (from the transmit vector)x _upRepresented) is processed by transmit chain 164 and is driven from N_utThe antennas 152 transmit over a wireless channel. At access point 102, the uplink signal consists of N_apReceived by antenna 116 and processed by a receive chain 124 to obtain received symbols (from received vectors)r _upRepresentation).

For the downlink, the received vector at the user terminal may be expressed as:

r _dn＝R _ut HT _ap x _dnin the formula (1)

Whereinx _dnIs to have a value corresponding to N at the slave access point_apN of symbols transmitted by antennas_apA transmission vector of each entry;

r _dnis to have a value corresponding to N at the user terminal_utN of symbols received on antennas_utA received vector of entries;

T _apis with N at the access point_apTransmit chain for individual antennasN of entries corresponding to associated complex gains_ap×N_apA diagonal matrix;

R _utis provided with N at the subscriber terminal_utN of entries corresponding to complex gains associated with receive chains of individual antennas_ut×N_utA diagonal matrix;

His N corresponding to downlink_ut×N_apA channel response matrix.

The response of the transmit and receive chains, as well as the response of the wireless channel, is typically a function of frequency. For simplicity, these responses are assumed to be flat fading (i.e., flat frequency response).

For the uplink, the receive vector at the access point may be expressed as:

r _up＝R _ap H ^T T _ut x _upin the formula (2)

Whereinx _upIs corresponding to N at the slave user terminal_utA transmission vector of symbols transmitted by the antennas;

r _upis for N at the access point_apA received vector of symbols received on the antennas;

T _utis provided with N at the subscriber terminal_utN of entries corresponding to complex gains associated with transmit chains of individual antennas_ut×N_utA diagonal matrix;

R _apis with N at the access point_apN of entries corresponding to complex gains associated with receive chains of individual antennas_ap×N_apA diagonal matrix;

H ^Tis N corresponding to the uplink_ap×N_utChannel responseAnd (4) matrix.

For TDD systems, there is typically a high correlation between downlink and uplink channel responses, since the downlink and uplink channels share the same frequency band. Thus, the downlink and uplink channel response matrices may be assumed to be reciprocal to each other (or transpose to each other) and are respectively represented asHAndH ^Tas shown in formulas (1) and (2). However, the response of the transmit and receive chains at the access point is typically not equal to the response of the transmit and receive chains at the user terminal. These differences thus lead to the following inequalityR _ap H ^T T _ut≠(R _ut HT _ap)^T。

From equations (1) and (2), the "effective" downlink and uplink channel responses, including the responses of the applicable transmit and receive chainsH _dnAndH _upcan be expressed as:

H _dn＝R _ut HT _apandH _up＝R _ap H ^T T _ut. Formula (3)

Combining these two equations in equation set (3) results in the following relationship:

formula (4)

Rearrangement formula (4) gives the following formula:

or

Formula (5)

WhereinAnd isFormula (5) may also be expressed as:

H _up K _ut＝(H _dn K _ap)^T. Formula (6)

The left hand side of equation (6) represents one form of the calibrated uplink channel response, while the right hand side represents the transpose of one form of the calibrated downlink channel response. The diagonal matrix is divided as shown in equation (6)K _utAndK _apthe application to the effective downlink and uplink responses enables the calibrated channel responses for the downlink and uplink to be expressed as transposes of each other. N corresponding to access point_ap×N_apDiagonal matrixK _apIs the receive chain responseR _apAnd transmit chain responseT _apRatio of (or to) to) Wherein the ratio is taken on a meta-by-meta basis. Similarly, N corresponding to user terminal_ut×N_utDiagonal matrixK _utIs the receive chain responseR _utAnd transmit chain responseT _utThe ratio of.

Fig. 2A illustrates applying correction matrices to both transmit and receive sides at an access point and a user terminal to account for differences in transmit and receive chains at the access point and the user terminal. On the downlink, the transmit vector is first sent by unit 112x _dnAnd matrixK _tapMultiplication. The processing of the downlink by transmit chain 114 and receive chain 154 is the same as that shown in fig. 1. The output of the receive chain 154 is coupled to a matrix by unit 156K _rutMultiplying to provide a received vector for the downlinkr _dn. On the uplink, the transmit vector is first sent by unit 162x _upAnd matrixK _ttutMultiplication. The processing of the uplink by the transmit chain 164 and the receive chain 124 is the same as that shown in fig. 1. The output of the receive chain 124 is coupled to a matrix by unit 126K _rapMultiplying to provide a received vector for the uplinkr _up。

In the case where correction matrices are applied at the access point and user terminal as shown in fig. 2A, the calibrated downlink and uplink channel responses may be expressed as:

H _cdn＝K _rut R _ut HT _ap K _tapandH _cup＝K _rap R _ap H ^T T _ut K _tut. Formula (7)

If it is notThe two equations in equation set (7) can be combined as follows:

formula (8)

The terms in rearrangement formula (8) give the following formula:

formula (9)

In equation (9) a diagonal matrix has been utilizedAAndBis a characteristic ofAB＝BATo reorganize the diagonal matrix.

Equation (9) indicates that the calibrated downlink and uplink channel responses can be obtained by satisfying the following conditions:

and formula (10a)

Formula (10b)

Where a is an arbitrary complex proportionality constant.

In general, correction factors for an access point may be applied to a transmitting side and/or a receiving side at the access point. Similarly, the correction factor for the user terminal may be applied to the transmitting side and/or the receiving side at the user terminal. For a given station, which may be an access point or a user terminal, the correction matrix for that station may be divided into a correction matrix for the transmitting side and a correction matrix for the receiving side. The correction matrix for one side (which may be the transmitting side or the receiving side) may be an identity matrixIOr an optional matrix. The correction matrix on the other side will thus be uniquely specified. These correction matrices need not directly account for transmit and/or receive chain errors that typically cannot be measured.

Table 1 lists 9 possible configurations for applying correction factors at the access point and the user terminal. For configuration 1, the correction factor is applied to both the transmit and receive sides at the access point, and also to both the transmit and receive sides at the user terminal. For configuration 2, the correction factor is applied to the transmit-only side at both the access point and the user terminal, whereK _tap＝K _ap，K _rap＝I，K _tut＝K _utAnd is andK _rut＝I. For configuration 3, the correction factor is applied to the receive-only side at both the access point and the user terminal, whereK _tap＝I，And isK _tut＝I. Other configurations are shown in table 1.

TABLE 1

Configuration of	Access point		User terminal
					Sending	Receiving	Sending	Receiving
	1	K tap	K rap	K tut					K rut
2					K ap	I	K ut	I
	3	I	K ap -1	I					K ut -1
4					K ap	I	I	K ut -1
	5	I	K ap -1	K ut					I
6					K tap	K rap	K ut	I
	7	K tap	K rap	I					K ut -1
8					K ap	I	K tut	K rut
	9	I	K ap -1	K tut					K rut

FIG. 2B illustrates the application of the correction matrix to the transmitting side of configuration 2K _apAndK _utto account for differences in transmit and receive chains at the access point and the user terminal. On the downlink, the transmit vector is first sent by unit 112x _dnAnd a correction matrixK _apMultiplication. The subsequent processing of the downlink by transmit chain 114 and receive chain 154 is the same as that shown in fig. 1. On the uplink, the transmit vector is first sent by unit 162x _upAnd a correction matrixK _utMultiplication. The subsequent processing of the uplink by the transmit chain 164 and the receive chain 124 is the same as that shown in fig. 1. Calibrated downlink and uplink channel responses observed by the user terminal and access point, respectivelyThis can be expressed as:

H _cdn＝H _dn K _apandH _cup＝H _up K _ut. Formula (11)

FIG. 2C shows the application of the correction matrix to the receiving side of configuration 3K _ap ^-1AndK _ut ^-1to account for differences in transmit and receive chains at the access point and the user terminal. On the downlink, the vector is transmittedx _dnIs processed by the transmit chain 114 at the access point. The downlink signal is processed by a receive chain 154 and further combined with a correction matrix by a unit 156 at the user terminalK _ut ^-1Multiplying to obtain a received vectorr _dn. On the uplink, the vector is transmittedx _upProcessed by the transmit chain 164 at the user terminal. The uplink signal is processed by the receive chain 124 and further combined with a correction matrix by a unit 126 at the access pointK _ap ^-1Multiplying to obtain a received vectorr _up. The calibrated downlink and uplink channel responses observed by the user terminal and the access point, respectively, can thus be expressed as:

andformula (12)

As shown in table 1, the correction matrix includes values that can account for differences in the transmit and receive chains at the access point and the user terminal. This will thus allow the channel response of one link to be expressed by the channel response of another link. The calibrated downlink and uplink channel responses may have various forms depending on whether correction factors are applied at the access point and the user terminal. For example, the calibrated downlink and uplink channel responses may be expressed as shown in equations (7), (11), and (12).

Calibration may be performed to determine a matrixK _apAndK _ut. In general, the true channel responseHAnd transmit and receive chain responses are unknown, they cannot be accurately and easily determined. Instead, the effective downlink and uplink channel responsesH _dnAndH _upmay be estimated based on pilots transmitted on the downlink and uplink, respectively, as described below. As a "true" matrixK _apAndK _utis estimated by the correction matrixAndand thus may be based on downlink and uplink channel response estimates as described belowAndto derive. Matrix arrayAndincluding correction factors that account for differences in transmit and receive chains at the access point and the user terminal. Once the transmit and receive chains have been calibrated, the obtained calibrated channel response estimate for one of the chains may be used (e.g.,) To determine an estimate of the calibrated channel response of the other link (e.g.,)。

the calibration techniques described herein may also be used for wireless communication systems employing OFDM. OFDM effectively partitions the overall system bandwidth into several (N)_FMultiple) orthogonal subbands, which are also referred to as tones, subcarriers, frequency bins, or subchannels. With OFDM, each subband is associated with a respective subcarrier that may be modulated with data. For a MIMO system employing OFDM (i.e., a MIMO-OFDM system), each subband of each spatial channel may be considered an independent transmission channel.

This calibration may be performed in various ways. For clarity, a specific calibration scheme is described below for a TDD MIMO-OFDM system.

Fig. 3 shows a flow diagram of one embodiment of a process 300 for calibrating downlink and uplink channel responses in a TDD MIMO-OFDM system. First, the user terminal acquires the timing and frequency of the access point using an acquisition procedure defined for the system (block 310). The user terminal may then send a message to initiate calibration with the access point, or may be initiated by the access point. Calibration may be performed by the access point in parallel with the registration/authentication of the user terminal (e.g., during call setup) and may also be performed at any time of warranty.

Calibration may be performed for all subbands used for data transmission (which may be referred to as "data" subbands). Subbands not used for data transmission (e.g., guard subbands) typically do not need to be calibrated. However, since the frequency response of the transmit and receive chains at the access point and user terminal is flat over most of the subbands of interest, and since adjacent subbands are likely to be correlated, calibration may be performed for only a subset of these data subbands. If not all data subbands are aligned, the subbands to be aligned (which are referred to as "designated" subbands) may be signaled to the access point (e.g., in a message sent to initiate alignment).

To perform calibration, the user terminal sends a MIMO pilot to the access point on these designated subbands (block 312). The generation of the MIMO pilot is described in detail below. The duration of the uplink MIMO pilot transmission may depend on the number of designated subbands. For example, if calibration is performed for 4 subbands, 8 OFDM symbols may be sufficient, while more subbands will require more (e.g., 20) OFDM symbols. The total transmit power is typically fixed. If the MIMO pilot is sent on a small number of subbands, each of these subbands may use a larger amount of transmit power and the SNR for each subband is higher. Conversely, if the MIMO pilot is transmitted on a large number of subbands, a smaller amount of transmit power may be used per subband and the SNR for each subband may be poor. If the SNR for each subband is not high enough, more OFDM symbols may be transmitted for the MIMO pilot and combined at the receiver to obtain a higher overall SNR for that subband.

An access point receives the uplink MIMO pilot and derives an estimate of the uplink channel response for each designated subbandWhere k represents the subband index. Channel estimation based on MIMO pilots is explained below. The uplink response estimate is quantized and sent to the user terminal (block 314). Each matrixThe entry in (1) is for the uplink of subband k at N_utA transmitting antenna and N_apComplex channel gain between the receive antennas. The channel gains of all matrices are scaled by a particular scaling factor that is common across all designated subbands to achieve the desired dynamic range. For example, each matrixThe channel gain in (b) may be inversely scaled by the maximum channel gain of all the designated subband matrices so that the magnitude of the maximum channel gain is 1. The absolute channel gain is not important since the purpose of the calibration is to normalize the gain/phase difference between the downlink and uplink channels. If the channel gain uses a 12-bit complex value (i.e., with 12-bit in-phase (I) and 12-bit quadrature (Q) components), the downlink channel response estimate may be at 3 · N_ut·N_ap·N_sbAre sent to the user terminal in bytes, where "3" corresponds to a total of 24 bits used to represent the I and Q components, and N_sbIs the number of specified subbands.

The user terminal also receives a downlink MIMO pilot transmitted by the access point (block 316) and derives an estimate of the downlink channel response for each designated subband based on the received pilot(block 318). The user terminal then estimates based on the uplink and downlink responsesAndto determine a correction factor for each specified subbandAnd(block 320).

To derive the correction factors, the downlink and uplink channel responses for each sub-band are assumed to be reciprocal with gain/phase corrections to account for differences in transmit and receive chains at the access point and user terminal as follows:

H _up(k)K _ut(k)＝(H _dn(k)K _ap(k))^Tk is equal to K, formula (13)

Where K represents a set of all data subbands. Since only estimates of valid downlink and uplink channel responses are available for a given subband during calibration, equation (13) can be rewritten as:

formula (14)

Where K' represents a set of all designated subbands. A correction vector may be definedIncludedN of (A)_utAnd a diagonal element. In this way,andare equivalent. Similarly, a correction vector may be definedIncludedN of (A)_apAnd a diagonal element.Andare also equivalent.

Calibration factorAndestimation from the channel can be done in various waysAndincluding by matrix ratio calculation and MMSE calculation. Both of these calculation methods are described in further detail below. Other calculation methods may also be used and are within the scope of the invention.

A.Matrix ratio calculation

FIG. 4 illustrates the use of matrix ratio calculations to estimate channel responses from uplink and downlinkAndderiving correction vectorsAndis shown in the flowchart of one embodiment of process 320 a. Process 320a may be used for block 320 in fig. 3.

First, calculate an N for each specified subband_ut×N_apMatrix arrayC(k) (block 412) the following:

for K ∈ K', equation (15)

Wherein this ratio is taken on a bin-by-bin basis.C(k) Can thus be calculated as:

for i 1, … …, N_utAnd j is 1, … …, N_apEquation (16)

WhereinAndare respectivelyAndthe (i, j) (row, column) th element of (a), and c_i，j(k) Is thatC(k) The (i, j) th element of (1).

In one embodiment, a correction matrix for an access pointIs defined as being equal toC(k) Is normalized and is derived by block 420.C(k) Each line of (a) is first scaled by scaling N in the line by the first element in the line_apEach of the individual elements is normalized (block 422). Thus, ifIs line i of C (k), then line normalizedCan be expressed as:

formula (17)

The average of these normalized rows is then determined to be the N_utThe sum of the normalized rows divided by N_ut(block 424). Correction vectorIs set equal to this average (block 426), which may be expressed as:

for K ∈ K'. Formula (18)

Because of this normalization, the result of the normalization,is a unit cell.

In one embodiment, correction vectors for user terminalsIs defined as being equal toC(k) Is determined and is derived by block 430.C(k) First by a vectorIs represented by K_ap，j，j(k) To scale each element in the column for normalization (block 432). Thus, ifIs thatC(k) The jth column element of (1), then the normalized columnCan be expressed as:

formula (19)

The inverse of the normalized columns is then averaged to the N_apThe sum of the inverses of the normalized columns divided by N_ap(block 434). Correction vectorIs set equal to this average (block 436), which may be expressed as:

formula (20)

Wherein the normalized columnThe inversion of (b) is performed on a bin-by-bin basis.

B.MMSE calculation

For MMSE calculation, correction factorAndis estimated from the downlink and uplink channel responsesAndderived to minimize a Mean Square Error (MSE) between the calibrated downlink channel response and the calibrated uplink channel response. This condition can be expressed as:

for K ∈ K, equation (21)

It can also be written as:

for K e K, the number of bits in the bit is,

whereinBecause of the fact thatIs a diagonal matrix.

Formula (21) isIs set equal to a unit cell or K_ap，0，0(k) 1. Without such constraint, it will be in the matrixAnda trivial solution is obtained if all the elements of (a) are set equal to 0. In formula (21)First, a matrix is obtainedY(k) Is composed ofThen obtain the matrixY(k) N of (A)_ap·N_utThe square of the absolute value of each of the entries. Whereby the mean square error (or squared error, since divided by N_ap·N_utOmitted) is equal to all N_ap·N_utThe sum of the individual squared values.

Performing MMSE computation on each specified subband to obtain a correction factor for that subbandAndthe MMSE calculation for one subband is explained below. For simplicity, the subband index k is omitted in the following description. For simplicity, the downlink will also beChannel response estimationIs given as { a_ijEstimating uplink channel responseIs given as { b_ij}, matrixIs given as { u_iAnd matrixHas a diagonal element of { v_jWhere i is 1, … …, N_apAnd j is 1, … …, N_ut。

The rewritable mean square error from equation (21) is as follows:

formula (22)

Is also subjected to u₁Constraint of 1. The minimum mean square error can be obtained by taking the partial derivatives of equation (22) for u and v and setting these two partial derivatives to 0. The result of these operations is the following system of equations:

for i 2, … …, N_apAnd formula (23a)

For j 2, … …, N_utAnd formula (I) and23b)

in the formula (23a), u₁1 thus there is no partial derivation for this case, and the index i goes from 2 to N_ap。

The set (N) of equations (23a) and (23b)_ap+N_ut-1) the equations can be more conveniently expressed in matrix form as follows:

A _y＝zequation (24)

Wherein

And is

Matrix arrayAComprising (N)_ap+N_ut-1) rows, wherein the first N_apLine-1 corresponds to N from equation set (23a)_ap1 equations and finally N_utThe rows correspond to N from equation set (23b)_utAn equation. In particular, a matrixAThe first row of (a) is generated with i-2 from equation set (23a), the second row is generated with i-3, and so on. Matrix arrayAN of (2)_apRows are generated from equation set (23b) with j equal to 1, and so on, and the last row is with j equal to N_utAnd (4) generating. As shown above, the matrixAEntries and vectors ofzCan be based on a matrixAndis obtained from the entry in (1).

The correction factor being included in the vectoryIn, vectoryCan be obtained as follows:

y＝A ^-1 z. Formula (25)

The result of the MMSE computation is a correction matrix that minimizes the mean square error in the calibrated downlink and uplink channel responses as shown in equation (21)Anddue to the matrixAndis based on downlink and uplink channel response estimatesAndobtained, thus correcting the matrixAndis dependent on the channel estimateAndthe quality of (c). The MIMO pilot may be averaged at the receiver to obtain a pairAndmore accurate estimation of.

Correction matrix obtained based on MMSE calculationAndgenerally better than a correction matrix calculated based on matrix ratios, especially when some of the channel gains are small and measurement noise can significantly degrade the channel gains.

C.Post-calculation

Regardless of the particular calculation method chosen for use, after calculation of the correction matrix is complete, the user terminal sends the access point correction vectors for all the specified subbands to the access pointIf it is notUsing a 12-bit complex value for each correction factor, the correction vectors for all the specified sub-bandsCan be in 3 (N)_ap-1)·N_sbIs sent to the access point in bytes, where "3" is the total of 24 bits for the I and Q components, (N)_ap-1) is due to each vectorIs equal to a unit cell and thus does not need to be transmitted, and N_sbIs the number of specified subbands. If the header is set to 2⁹With-1 +511, then a 12dB headroom is available (since the maximum positive 12-bit signed value is 2)¹¹-1 ═ 2047), which would allow a 12-bit value to accommodate up to 12dB of gain mismatch between downlink and uplink. If the downlink and uplink are matched to within 12dB and the first element is normalized to the value 511, the absolute value of the other elements should not be greater than 511 · 4 — 2044 and can be represented by 12 bits.

Obtaining a pair of correction vectors per a given subbandAndthis correction may not be performed for all data subbands. For example, a correction may be performed for every n subbands, where n may be determined based on the expected response of the transmit and receive chains (e.g., n may be 2, 4, 8, 16, etc.). Correction may also be performed for non-uniformly distributed sub-bands. For example, more subbands near the band edges may be calibrated since there may be more filter roll-offs at the edges of the passband, which will produce more mismatch in the transmit and receive chains. In general, any number of subbands and any distribution of subbands may be calibrated and this is within the scope of the present invention.

If calibration is not performed for all data subbands, a correction factor for an "uncalibrated" subband may be obtained by interpolating the correction factor obtained for the specified subband. The access point can be for K ∈ KPerforming interpolation to obtain a correction vector for K ∈ KSimilarly, the user terminal may be for K ∈ KPerforming interpolation to obtain a correction vector for K ∈ K

Thereafter, the access point and the user terminal use their respective correction vectorsAndor a corresponding correction matrixAndwhere K ∈ K. The access point may correct the matrix based thereonAnd derives a correction matrix for its transmission side under the constraint shown in equation (10a)And a correction matrix for its receiving sideSimilarly, the user terminal may correct the matrix based thereonAnd derives a correction matrix for its transmission side under the constraint shown in equation (10b)And a correction matrix for its receiving side

Correction matrixAnd a correction matrixEach may be divided into two matrices to improve dynamic range, reduce quantization error, address transmit and receive chain limitations, and so on. If there is a known imbalance at the transmit side, the transmit side correction matrix may attempt to eliminate this imbalance. For example, if one antenna has a smaller power amplifier, the transmit power of the antenna with the stronger power amplifier may be reduced by applying an appropriate correction matrix to the transmit side. However, operating the transmit side at lower power levels results in performance loss. Adjustments can thus be made on the receive side to compensate for this known transmit imbalance. If both the transmit and receive chains have less gain for a given antenna due to, for example, less antenna gain, then this calibration does not adjust this antenna because the transmit and receive sides are matched.

The above described correction scheme of obtaining a correction factor vector for each of the access point and the user terminal allows a "compatible" correction vector to be derived for the access point when calibration is performed by different user terminals. If the access point has been calibrated (e.g., by one or more other user terminals), the current correction vector may be updated with the newly derived correction vector.

For example, if two user terminals perform this calibration procedure simultaneously, the calibration results from the two user terminals may be averaged to improve performance. However, calibration is typically performed for one user terminal at a time. The second user terminal will thus observe the downlink to which the correction vector for the first user terminal has been applied. In this case, the product of the second correction vector and the original correction vector may be used as the new correction vector, or "weighted average" (described below) may also be used. The access point typically uses a single correction vector for all user terminals rather than different correction vectors for different user terminals (although this is also achievable). Updates from multiple user terminals or sequential updates from one user terminal can be treated in the same way. The updated vector can be applied directly (by multiplication). Alternatively, if some averaging is desired to reduce measurement noise, a weighted average may be used as described below.

If the access point uses the correction vectorTo transmit a new correction vector for the user terminal to determine based thereonThe current and new correction vectors are multiplied to derive an updated correction vectorCorrection vectorAndmay be derived by the same or different user terminals. In one embodiment, the updated correction vector is defined asWhere this multiplication is element-wise. In another embodiment, the updated correction vector is defined asWhere α is a factor used to provide a weighted average (e.g., 0 < α < 1). If the calibration updates are infrequent, alpha values close to 1 may perform best. A smaller alpha value is better if the calibration updates are frequent but noisy. Updated correction vectorAnd then available to the access point until they are updated again.

As shown in equations (10a) and (10b), the correction factor for a given station (which may be an access point or a user terminal) accounts for the responses of the transmit and receive chains at that station. The access point may perform calibration with the first user terminal to derive its calibration factors and then use these calibration factors to communicate with the second user terminal without performing calibration with the second user terminal. Similarly, the user terminal may perform calibration with the first access point to derive its correction factors and then use these correction factors to communicate with the second access point without performing calibration with the second access point. This may reduce overhead for calibration for access points communicating with multiple user terminals and user terminals communicating with multiple access points, since not every pair of communication stations needs calibration.

In the above description, the correction vector of K ∈ K' is derived by the user terminalAndand a vectorIs sent back to the access point. This scheme advantageously distributes the calibration process among the user terminals of the multiple access system. However, the calibration vectorAndit can also be derived by the access point, which then will vectorBack to the user terminal and this is within the scope of the invention.

The above calibration scheme enables each user terminal to calibrate its transmit and receive chains in real time via radio transmission. This allows high performance to be achieved for user terminals with different frequency responses without the need for strict frequency response specifications or performing calibration at the factory. The access point may be calibrated by multiple user terminals to provide improved accuracy.

D.Gain considerations

Calibration may be performed based on "normalized" gains for the downlink and uplink channels, which are gains given with respect to the noise floor at the receiver. The use of normalized gain enables the characteristics of one link (e.g., channel gain and SNR per spatial channel) to be obtained based on the gain measurement of the other link after the downlink and uplink have been calibrated.

The access point and the user terminal may initially balance their receiver input levels so that the noise levels on the receive paths of the access point and the user terminal are approximately the same. This balancing may be achieved by estimating the noise floor, for example by finding the portion of the received TDD frame (which is the unit of downlink/uplink transmission) that has the smallest average power over a particular time duration (e.g., one or two symbol periods). In general, there is no transmission at a time just before the start of each TDD frame because any uplink data must be received by the access point, then a receive/transmit turnaround time is required before the access point transmits on the downlink. Depending on the interference environment, the noise floor may be determined based on multiple TDD frames. The downlink and uplink channel responses are then measured against this noise floor. More specifically, the channel gain for a given subband for a given transmit and receive antenna pair may first be obtained as, for example, the ratio of the received pilot symbols to the transmitted pilot symbols for that subband for that transmit and receive antenna pair. The normalized gain is then equal to the measured gain divided by the noise floor.

A large difference in the normalized gain of the access point and the normalized gain of the user terminal may result in a correction factor for the user terminal that is significantly different than unity. Correction factors for access pointsSub-near unity since the matrixIs set to 1.

If the correction factor for the user terminal is significantly different from a unit cell, the user terminal may not be able to apply the calculated correction factor. This is because the user terminal has a constraint on its maximum transmit power and may not be able to increase its transmit power for large correction factors. Furthermore, reducing the transmit power for small correction factors is generally undesirable as it reduces the achievable data rate.

The user terminal may thus transmit using a scaled version of the calculated correction factor. The scaled calibration factor may be obtained by scaling the calculated correction factor by a particular scaling value that may be set equal to the gain increment (difference or ratio) between the downlink and uplink channel responses. This gain increase may be calculated as the average of the difference (or increase) in normalized gain for the downlink and uplink. The calibration value (or gain increment) used for the correction factor may be sent to the access point along with the calculated correction factor for the access point.

With the correction factor and the scaling value or gain increment, the downlink channel characteristics can be determined from the measured uplink channel response and vice versa. The gain delta may be updated if the noise floor at the access point or user terminal changes, and the updated gain delta may be sent to the other party in a message.

In the above description, the calibration results in two sets (or vectors or matrices) of correction factors corresponding to each sub-band, one of which setsUsed by the access point, and another groupBy the user terminalAnd (4) end use. The access point may apply its correction factor to the transmitting side and/or the receiving side as described aboveThe user terminal may also apply its correction factor to the transmitting side and/or the receiving sideIn general, calibration is performed such that the calibrated downlink and uplink channel responses are reciprocal, regardless of where the correction factor is applied.

2.MIMO pilot

For calibration, a MIMO pilot is transmitted by the user terminal on the uplink to enable the access point to estimate the uplink channel response, and a MIMO pilot is transmitted by the access point on the downlink to enable the user terminal to estimate the downlink channel response. The MIMO pilot frequency is formed by the slave N_TN transmitted by transmitting antenna_TEach pilot transmission constitutes a pilot, where the pilot transmission from each transmit antenna may be identified by the receiving station. The MIMO pilot may be generated and transmitted in various manners. The uplink and downlink may use the same or different MIMO pilots. In any case, the MIMO pilots used by the downlink and uplink are known at both the access point and the user terminal.

In one embodiment, the MIMO pilot is formed from N_TA particular OFDM symbol (denoted "P") transmitted by each of the transmit antennas, where there is N for the downlink_T＝N_apAnd for the uplink there is N_T＝N_ut. For each transmit antenna, the same P OFDM symbol is sent in each symbol period designated for MIMO pilot transmission. However, the P OFDM symbol for each antenna is covered with a different N-chip Walsh sequence assigned to that antenna, where N ≧ N for the downlink_apAnd N ≧ N for the uplink_ut. Walsh covering maintains this N_TOrthogonality between transmitting antennas, andthe receiver can distinguish the transmitting antennas.

P OFDM symbol includes for the N_sbOne modulation symbol for each of the assigned subbands. The POFDM symbol thus includes N, which may be selected to facilitate channel estimation by the receiver_sbA particular "word" of individual modulation symbols. This word may also be defined to minimize peak-to-peak variation in the transmitted MIMO pilot. This in turn can reduce the amount of distortion and non-linearity produced by the transmit and receive chains, which in turn leads to improved accuracy of the channel estimation.

For clarity, a particular MIMO pilot is described below with respect to a particular MIMO OFDM system. For this system, the access point and the user terminal each have 4 transmit/receive antennas. The system bandwidth is divided into 64 orthogonal subbands, or N_FThey are given an index of +31 to-32, 64. Of these 64 subbands, 48 subbands (e.g., with indices of ± {1, …, 6, 8, …,20, 22, …, 26}) are used for data, 4 subbands (e.g., with indices of ± {7, 21}) are used for pilot and possibly also for signaling, the DC subband (with index of 0) is not used, and the remaining subbands are not used to serve as guard subbands. This OFDM subband structure is described in publicly available 1999 month 9 IEEE standard 802.11a entitled "Part 11: wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: a High-speed Physical Layer in the 5GHz Band (section 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specification: High speed Physical Layer in the 5GHz Band) document is described in further detail.

The P OFDM symbol includes a set of 52 QPSK modulation symbols corresponding to the 48 data subbands and 4 pilot subbands. This P OFDM symbol may be given as follows:

p (real) · {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, -1, 1, -1, 1, 1, 0, 1, -1, -1, -1, -1, -1, 1, -1, -1, -1, -1, 1, 1, 1, -1, -1, 1, -1, 1, -1, 1, -0, 0, 0, 0, 0, 0},

p (imaginary) · {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, 0, -1, -1, -1, -1, 1, 1, -1, 1, -1, 1, -1, 1, 1, 1, 1, 1, -1, 1, 0, 0, 0, 0, 0},

where g is the gain of the pilot. The values in parentheses are given for subband indices-32 to-1 (corresponding to the first row) and 0 to +31 (corresponding to the second row). Thus, the first row of P (real) and P (imaginary) indicates that symbol (-1-j) was transmitted in subband-26, symbol (-1+ j) was transmitted in subband-25, and so on. The second line of P (real) and P (imaginary) indicates that symbol (1-j) was transmitted in subband 1, that symbol (-1-j) was transmitted in subband 2, and so on. The MIMO pilot may also use other OFDM symbols.

In one embodiment, 4 transmit antennas are assigned Walsh sequences W for MIMO pilots₁＝1111，W₂＝1010，W₃1100, and W₄1001. For a given Walsh sequence, a value of "1" indicates that a P OFDM symbol is transmitted, and a value of "0" indicates that a-P OFDM symbol is transmitted (i.e., each of the 52 modulation symbols in P is inverted).

Table 2 lists the symbols spanning 4 symbol periods or N_psThe MIMO pilot of 4 transmits OFDM symbols transmitted from each of the 4 transmit antennas.

TABLE 2

OFDM code element	Antenna 1	Antenna 2	Antenna 3	Antenna 4
					1	+P	+P	+P	+P
2	+P	-P	+P	-P
					3	+P	+P	-P	-P
4	+P	-P	-P	+P

For longer MIMO pilot transmissions, the Walsh sequence corresponding to each transmit antenna is simply repeated. For the set of Walsh sequences, MIMO pilot transmission occurs over an integer multiple of 4 symbol periods to ensure orthogonality among the 4 transmit antennas.

The receiver may derive an estimate of the channel response based on the received MIMO pilots by performing complementary processing. Specifically, to recover the pilot transmitted from transmit antenna i and received by receive antenna j, the pilot received by receive antenna j is first processed with the Walsh sequence assigned to transmit antenna i in a manner complementary to the Walsh covering performed at the transmitter. All N of the MIMO pilot are then accumulated_psThe decovered OFDM symbols for one symbol period, where the accumulation is performed separately for each of the 52 subbands used to carry the MIMO pilot. The result of this accumulation isk + -1, …, 26, which is an estimate of the effective channel response from transmit antenna i to receive antenna j, including the response of the transmit and receive chains, corresponding to the 52 data and pilot subbands.

The same processing may be performed to recover the pilot from each transmit antenna at each receive antenna. The pilot processing provides N_ap·N_utValues that are effective channel response estimates for each of the 52 channelsH _up(k) OrH _dn(k) Is used as the element of (1).

In another embodiment, the MIMO pilot uses a Fourier matrixF. The fourier matrix may have arbitrary square dimensions, e.g. 3 × 3,4 × 4, 5 × 5, and so on. The elements of an nxn fourier matrix may be expressed as:

and m is 1, …, N.

Each transmitting antenna is assigned withFOne column of (c). The elements in the assigned column are used to multiply the pilot symbols over different time intervals in a similar manner as the elements of the Walsh sequence. In general, any orthogonal matrix whose elements have unity magnitude may be used to multiply the pilot symbols of the MIMO pilot.

In yet another embodiment applicable to a MIMO-OFDM system, the subbands available for transmission are divided into N_TNon-overlapping or disjoint subsets. For each transmit antenna, pilot symbols are transmitted on a subset of subbands in each time interval. Each transmit antenna may be at N corresponding to the duration of the MIMO pilot_TCyclically traverse the N in time intervals_TA subset of. The MIMO pilot may also be transmitted in other manners.

Regardless of how the MIMO pilot is transmitted, channel estimation may be performed by the access point and the user terminal during calibration to obtain effective uplink channel response estimates, respectivelyAnd effective downlink channel response estimationThese estimates are then used to derive correction factors as described above.

3.Spatial processing

Correlation between downlink and uplink channel responses can be exploited to simplify channel estimation and spatial processing at access points and user terminals for TDD MIMO systems and TDD MIMO-OFDM systems. This simplification would be feasible after performing calibration to account for differences in the transmit and receive chains. As mentioned above, the calibrated channel response is:

for the downlink isAnd formula (26a)

For the uplink isFormula (26b)

The approximation of the last equation in equation (26b) is due to the use of an estimate of the actual correction factor.

Channel response matrix for each sub-bandH(k) May be "diagonalized" to obtain N corresponding to the subband_sAn eigenmode. These eigenmodes may be considered orthogonal spatial channels. This diagonalization may be achieved by performing a channel response matrixH(k) By singular value decomposition, or by performingH(k) Is the correlation matrix ofR(k)＝H ^H(k)H(k) By eigenvalue decomposition.

Calibrated uplink channel response matrixH _cup(k) The singular value decomposition of (a) may be expressed as:

formula (27)

WhereinU _ap(k) Is thatH _cup(k) N of the left eigenvector of_ut×N_utA unitary matrix;

∑(k) is thatH _cup(k) N of singular value of_ut×N_apA diagonal matrix; and

V _ut(k) is thatH _cup(k) N of the right eigenvector of_ap×N_apA unitary matrix.

Unitary matrixMBy characteristics ofM ^H M＝IAnd (5) characterizing. Accordingly, the calibrated downlink channel response matrixH _cdn(k) The singular value decomposition of (a) may be expressed as:

formula (28)

Matrix arrayV _ut ^*(k) AndU _ap ^*(k) thereby also respectivelyH _cdn(k) Wherein "+" denotes the complex conjugate. Matrix arrayV _ut(k)、V _ut ^*(k)、V _ut ^T(k) And, andV _ut ^H(k) is a matrixV _ut(k) In different forms of, andU _ap(k)、U _ap ^*(k)、U _ap ^T(k) and, andU _ap ^H(k) is also a matrixU _ap(k) Different forms of (1). For simplicity, the matrix is described belowU _ap(k) AndV _ut(k) the references may also refer to their various other forms. Matrix arrayU _ap(k) AndV _ut(k) used for spatial processing by the access point and the user terminal, respectively, and denoted by their subscripts.

Singular value decomposition is further detailed by Gilbert Strang in academy 1980 in a second edition of the book entitled "Linear algebra and Its Applications," the contents of which are incorporated herein by reference.

The user terminal may estimate a calibrated downlink channel response based on the MIMO pilot sent by the access point. The user terminal may then perform a calibrated downlink channel response estimate for K ∈ KIs decomposed to obtain a diagonal matrix of K ∈ KAndleft eigenvector matrix ofV _ut ^*(k) In that respect This singular value decomposition can be given asWhere the header ("*") above each matrix indicates that it is an estimate of the actual matrix.

Similarly, the access point may estimate a calibrated uplink channel response based on the MIMO pilot sent by the user terminal. The access point may then perform a calibrated uplink channel response estimate for K ∈ KIs decomposed to obtain a diagonal matrix of K ∈ KAndleft eigenvector matrix ofThis singular value decomposition can be given as

Because of the reciprocal channel and calibration, the matrix can be obtained by only performing this singular value decomposition by the user terminal or access pointAndand both. If performed by the user terminal, the matrixIs used for spatial processing at the user terminal, and the matrixMay be sent back to the access point.

The access point may also be based on a transmission from the user terminalDerived from the transmitted guided referenceAndsimilarly, the user terminal may also derive the matrix based on the steered reference sent by the access pointAndthis guided reference is described in detail in commonly assigned U.S. patent application serial No. 10/693,419 entitled MIMO WLAN System filed on 23/10/2003.

Matrix arrayAndn available in MIMO channel_STransmitting independent data streams on eigenmodes, where N_S≤min{N_ap，N_ut}. The spatial processing for transmitting multiple data streams on the downlink and uplink is described below.

A.Uplink spatial processing

The spatial processing by the user terminal for uplink transmissions may be expressed as:

formula (29)

Whereinx _up(k) Is the uplink transmit vector for subband k; and is

s _up(k) Is of up to N_SN corresponding to sub-band k to be_SA data vector of non-zero entries of modulation symbols transmitted on the eigenmodes.

Other processing may also be performed on the modulation symbols prior to transmission. For example, channel inversion can be applied across the data subbands (e.g., for each eigenmode) to approximately equalize the received SNRs for all data subbands. This spatial processing can thus be expressed as:

formula (30)

WhereinW _up(k) Is a matrix with weights corresponding to the (optional) inversion of the uplink channel.

Channel inversion may also be performed by assigning transmit power to each subband before modulation occurs, in which case the vectors _up(k) Including channel inversion coefficients, and a matrixW _up(k) May be omitted from equation (30). In the following description, a matrix is used in the equationsW _up(k) Indicating that channel inversion coefficients are not included in the vectors _up(k) In (1). There is no matrix in the equationW _up(k) May indicate that (1) no channel inversion was performed or (2) channel inversion was performed and incorporated into the vectors _up(k) In (1).

Channel inversion may be performed as described in the aforementioned U.S. patent application serial No. 10/693,419 and in commonly assigned U.S. patent application serial No. 10/229,209 entitled "Coded MIMO System with Selective channel inversion Applied Per Eigenmode" filed on 27/8/2002.

The received uplink transmission at the access point may be expressed as:

formula (31)

Whereinr _up(k) Is the receive vector for the uplink for subband k;

n(k) is Additive White Gaussian Noise (AWGN) for subband k; and

x _up(k) as shown in equation (29).

Receiver spatial processing (or spatial matched filtering) of received uplink transmissions at an access point may be expressed as:

formula (32)

WhereinIs a data vector transmitted by a user terminal on the uplinks _up(k) Is estimated, andis post-processed noise. Equation (32) transmits the vector assuming that channel inversion is not performed at the user terminalx _up(k) As shown in equation (29), and receiving a vectorr _up(k) As shown in equation (31).

B.Downlink spatial processing

The spatial processing of downlink transmissions by an access point may be expressed as:

formula (33)

Whereinx _dn(k) Is a transmission vector, ands _dn(k) is the data vector for the downlink.

Other processing (e.g., channel inversion) may also be performed on the modulation symbols prior to transmission. This spatial processing can thus be expressed as:

formula (34)

WhereinW _dn(k) Is a matrix with weights corresponding to the inversion of the (optional) downlink channel.

The received downlink transmission at the user terminal may be expressed as:

formula (35)

The receiver spatial processor (or spatial matched filtering) of the received downlink transmission at the user terminal may be expressed as:

formula (36)

Equation (36) the vector is transmitted assuming that no channel inversion is performed at the access pointx _dn(k) As shown in equation (33), and receiving the vectorr _dn(k) As shown in equation (35).

Table 3 summarizes the spatial processing performed at the access point and the user terminal for data transmission and reception. Table 3 assumes that at the transmitter the data is transmitted byW(k) Other processing is performed. However, if no other processing is performed, thenW(k) Simply equal to the identity matrix.

TABLE 3

In the above description, and as shown in table 3, the correction matrices are used at the access point on the transmitting side and the receiving side, respectivelyAndone of the two correction matrices may be set equal to the identity matrix. Using correction matrices at a transmitting side and a receiving side, respectively, at a user terminalAndthese two momentsOne of the arrays may also be set equal to the identity matrix. Correction matrixAndcan be associated with the weight matrixW _dn(k) AndW _up(k) are combined to obtain a gain matrixG _dn(k) AndG _up(k) whereinAnd is

C.Data transmission on a link

Data transmission on a given link may also be achieved by applying a correction matrix at the transmitting station and using an MMSE receiver at the receiving station. For example, data transmission on the downlink may be achieved by applying correction factors at the transmit-only side of the access point and using an MMSE receiver at the user terminal. For simplicity, this description is for a single subband, and the subband index k is omitted in the equations. The calibrated downlink and uplink channel responses may be given as:

H _cup(k)＝R _ap H ^T T _ut＝H _upand formula (37)

Formula (38)

The user terminal transmits a pilot on the uplink, which the access point uses to derive an estimate of the uplink channel response. The access point performs uplink channel response estimation as shown in equation (27)And deriving the matrixThe access point then usesSpatial processing is performed to transmit data on the eigenmodes of the MIMO channel as shown in equation (33).

The downlink transmission received at the user terminal may be expressed as:

r _dn＝H _dn x _dn+n. Formula (39)

Equation (39) indicates that no correction factor is applied at the user terminal. The user terminal derives the MMSE spatial filter matrix as follows:

formula (40)

WhereinAnd is

*_nnIs the autocovariance matrix of the noise.

If the noise is AWGN, thenWherein sigma_n ²Is the variance of the noise. The user terminal may derive based on pilots sent with the data by the access pointH _edn。

The user terminal then performs MMSE spatial processing as follows:

formula (41)

Whereinn _mmseIncluding MMSE filtered noise and residual crosstalk, andis a data vectors _dnIs estimated. From MMSE spatial filter matrixMIs a non-normalized estimate of the data symbols. The user terminal can exchangeMultiplying by a scaling matrixD，D＝[diag[MH _edn]]^-1To obtain a normalized estimate of the data symbols.

If the user terminal applies the correction matrix on its receiving sideThe total downlink channel response will beH _odn＝K _rut H _edn. On the receiving side at the user terminal a correction matrix is appliedK _rutMMSE spatial filter momentMatrix ofCan be expressed as:

formula (42)

The inverse in equation (42) may be rearranged as follows:

formula (43)

Substituting formula (43) for formula (42) yields the following formula:

formula (44)

On the receiving side at the user terminal a correction matrix is appliedK _rutIn this case, the received downlink transmission at the user terminal may be expressed as:

formula (45)

The user terminal then performs MMSE spatial processing as follows:

formula (46)

Equations (45) and (46) indicate that the user terminal can obtain the same performance as the MMSE receiver regardless of whether the correction factor is applied at the user terminal. MMSE processing implicitly accounts for any mismatch between the transmit and receive chains at the user terminal. If no correction factor is applied on the receiving side at the user terminal, thenH _ednDeriving MMSE spatial matched filters, if correction factors are appliedH _odnAnd (6) derivation.

Similarly, data transmission on the uplink can be achieved by applying correction matrices on the transmit side and/or the receive side at the user terminal and using an MMSE receiver at the access point.

4.MIMO-OFDM system

Fig. 5 illustrates a block diagram of one embodiment of an access point 502 and a user terminal 504 within a TDD MIMO-OFDM system. For simplicity, the following description assumes that the access point and the user terminal are each equipped with 4 antennas that can be used for data transmission and reception.

On the downlink, at access point 502, a Transmit (TX) data processor 510 receives traffic data (i.e., information bits) from a data source 508 and signaling and other information from a controller 530. TX data processor 510 formats, codes, interleaves, and modulates (i.e., symbol maps) the received data and generates a stream of modulation symbols for each spatial channel used for data transmission. A TX spatial processor 520 receives the streams of modulation symbols from TX data processor 510 and performs spatial processing to provide 4 streams of transmit symbols, one for each antenna. TX spatial processor 520 also multiplexes pilot symbols as appropriate (e.g., for calibration).

Each Modulator (MOD)522 receives and processes a respective one of the transmit symbol streams to generate a respective one of the OFDM symbol streams. Each OFDM symbol stream is further processed by a transmit chain within a modulator 522 to generate a corresponding downlink modulated signal. The 4 downlink modulated signals from modulators 522a through 522d are then transmitted from 4 antennas 524a through 524d, respectively.

At user terminal 504, antennas 552 receive the transmitted downlink modulated signals and each antenna provides a received signal to a respective one of demodulators (DEMODs) 554. Each demodulator 554, which includes a receive chain, performs complementary processing to that performed at modulator 522 and provides received symbols. A Receive (RX) spatial processor 560 performs spatial processing on the received symbols from all demodulators 554 and provides recovered symbols, which are estimates of the modulation symbols transmitted by the access point. An RX data processor 570 processes (e.g., symbol demaps, deinterleaves, and decodes) the recovered symbols and provides decoded data. The decoded data may include recovered traffic data, signaling, and so forth, which are provided to a data sink 572 for storage and/or to controller 580 for further processing.

Controllers 530 and 580 control the operation of various processing units at the access point and user terminal, respectively. Memory units 532 and 582 store data and program codes used by controllers 530 and 580, respectively.

During calibration, RX spatial processor 560 provides downlink channel response estimates derived based on MIMO pilots transmitted by the access pointRX data processor 570 provides an uplink channel response estimate derived by the access point and transmitted on the downlinkController 580 receives the channel response estimateAndderiving correction matricesAndand will matrixAnd provided to a TX data processor 590 for transmission back to the access point. The controller 580 further bases on the correction matrixDeriving correction matricesAndwhereinAndmay be an identity matrix, a correction matrixIs provided to a TX spatial processor 592 and applies a correction matrixProvided to RX spaceAnd a processor 560.

The processing for the uplink may be the same as or different from the processing for the downlink. Data and signaling are processed (e.g., coded, interleaved, and modulated) by a TX data processor 590 and further spatially processed by a TX spatial processor 592, and pilot symbols are multiplexed into TX spatial processor 592. These pilot and modulation symbols are further processed by a modulator 554 to generate an uplink modulated signal, which is then transmitted via antenna 552 to the access point.

At access point 10, the uplink modulated signals are received by antennas 524, demodulated by a demodulator 522, and processed by a RX spatial processor 540 and a RX data processor 542 in a manner complementary to the processing performed by the user terminals. During calibration, RX spatial processor 560 provides uplink channel estimates derived based on MIMO pilots transmitted by user terminalsMatrix arrayIs received by controller 530 and provided to TX data processor 510 for transmission to a user terminal.

Fig. 6 illustrates a block diagram of a TX spatial processor 520a, which may be used for TX spatial processors 520 and 592 in fig. 5. For simplicity, the following description assumes that all 4 eigenmodes are selected for use.

Within processor 520a, a demultiplexer 632 receives 4 streams of modulation symbols (denoted as s) to be transmitted on 4 eigenmodes₁(n) to s₄(N)), applying each stream to correspond to N_DN of data sub-bands_DOf the sub-streams, and provides 4 modulation symbol sub-streams for each data subband to a respective one of TX subband spatial processors 640. Each processor 640 performs processing, for example, as shown in equations (29), (30), (33), or (34) for one sub-band.

In each TX subband spatial processor 640, 4Modulating a sub-stream of symbols (denoted as s)₁(k) To s₄(k) Are provided to 4 beamformers 650a to 650d corresponding to the 4 eigenmodes of the associated subband. Each beamformer 650 performs beamforming to transmit one symbol stream on one eigenmode of one subband. Each beamformer 650 receives a symbol substream s_m(k) And using eigenvectors of the associated eigenmodesv _m(k) To perform beamforming. Within each beamformer 650, the modulation symbols are provided to 4 multipliers 652a through 652d, which 4 multipliers also receive the eigenvectors of the associated eigenmodesv _m(k) 4 elements v of_m，1(k)、v_m，2(k)、v_m，3(k) And v_m，4(k) In that respect Eigenvectorsv _m(k) Is a matrix corresponding to the downlinkAnd is a matrix corresponding to the uplinkColumn m. Each multiplier 652 multiplies a scaled modulation symbol by its eigenvector value v_m，j(k) And provides "beamformed" symbols. Multipliers 652a through 652d provide the 4 beamformed symbol substreams (to be transmitted from the 4 antennas) to summers 660a through 660d, respectively.

Each summer 660 receives and sums 4 beamformed symbols for 4 eigenmodes over each symbol period and provides a preprocessed symbol for an associated transmit antenna. Adders 660a through 660d provide 4 pre-processed symbol substreams corresponding to the 4 transmit antennas to buffers/multiplexers 670a through 670d, respectively. Each buffer/multiplexer 670 receives pilot symbols and symbols from a corresponding N_DThe pre-processed symbols for TX subband spatial processor 640 for the data subbands. Each buffer/multiplexer 670 then multiplexes pilot symbols, pre-processed symbols, and zero symbols for the pilot subbands, data subbands, and unused subbands, respectively, to form N symbols over the symbol period_FOne code elementThe sequence of (a). During calibration, pilot symbols are transmitted on designated subbands. Multipliers 668a through 668d are respectively assigned Walsh sequences W for these 4 antennas as described above and shown in table 2₁To W₄Covering the pilot symbols corresponding to the 4 antennas. Each buffer/multiplexer 670 provides a stream of symbols to a corresponding one of multipliers 672.

Multipliers 672a to 672d also receive correction factor K, respectively₁(k)、K₂(k)、K₃(k) And K₄(k) In that respect The correction factor for each data subband k corresponding to the downlinkAnd corresponding to the uplinkThe diagonal elements of (1). Each multiplier 672 with its correction factor K_m(k) To scale its input symbols and provide transmit symbols. Multipliers 672a through 672d provide 4 transmit symbol streams for the 4 transmit antennas.

Spatial processing and OFDM modulation are further described in detail in the aforementioned U.S. patent application serial No. 10/693,419.

The calibration techniques described herein may be implemented by various means. For example, these techniques may be implemented in hardware, software, or a combination thereof. For a hardware implementation, the calibration techniques may be implemented within one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions recited herein, or a combination thereof, at the access point and the user terminal.

For a software implementation, these calibration techniques may be performed by modules (e.g., procedures, functions, and so on) that perform the functions described herein. The software codes may be stored in memory units (e.g., memory units 532 and 582 in fig. 5) and executed by processors (e.g., by controllers 530 and 580, as appropriate). The memory unit may be implemented within or external to the processor, in which case it can be communicatively coupled to the processor via various means as is known in the art.

Subtitles are included herein for ease of reference and to aid in locating certain sections. These headings are not intended to limit the scope of the concepts described therein under, and these concepts will have applicability in other sections throughout the entire specification.

The previous description of the disclosed 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 departing from the spirit or scope of the invention. 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 (40)

1. A method of calibrating a communication link in a wireless Time Division Duplex (TDD) communication system, comprising:

obtaining a channel response estimate for a downlink channel from an access point to a user terminal;

obtaining a channel response estimate for an uplink channel from the user terminal to the access point;

determining a correction factor for the access point and a correction factor for the user terminal based on the channel response estimates for the downlink and uplink channels, the correction factor for the access point and the correction factor for the user terminal being used to obtain a calibrated downlink channel response and a calibrated uplink channel response.

2. The method of claim 1, further comprising:

applying the correction factor for the access point on either a transmit side or a receive side of the access point or both the transmit and receive sides.

3. The method of claim 1, further comprising:

applying the correction factor for the user terminal at a transmitting side, or at a receiving side, or at both the transmitting and receiving sides of the user terminal.

4. The method of claim 1, wherein determining the correction factor for the access point and the correction factor for the user terminal comprises

Determining the correction factor for the access point and the correction factor for the user terminal based on:

whereinIs a matrix of the channel response estimates for the downlink channel,

is a matrix of the channel response estimates for the uplink channel,

is a matrix of the correction factors for the access point,

is a matrix of said correction factors for the user terminal, and

"T" denotes transpose.

5. The method of claim 4, further comprising:

deriving a correction factor for a transmit side of the access point and a correction factor for a receive side of the access point based on:

whereinIs the matrix of the correction factors for the transmitting side of the access point, an

Is a matrix of the correction factors for the receive side of the access point.

6. The method of claim 5, further comprising:

will matrixOr matrixSet as the identity matrix.

7. The method of claim 5, further comprising: will matrixOr matrixSet to an arbitrary matrix.

8. The method of claim 4, further comprising:

deriving a correction factor for a transmitting side of the user terminal and a correction factor for a receiving side of the user terminal based on:

whereinIs a matrix of said correction factors for the transmitting side of the user terminal, an

Is a matrix of the correction factors for the receiving side of the user terminal.

9. The method of claim 4, wherein the determining the correction factor for the access point and the correction factor for the user terminal comprises

Computing the matrix C as a matrixAnd matrixA meta ratio of

Deriving a matrix based on the matrix CAnd

10. the method of claim 9, wherein the derivation matrixIncluding normalizing each of a plurality of rows of the matrix C, determining an average of the plurality of normalized rows of the matrix C, and constructing the matrix C based on the average of the plurality of normalized rows

11. The method of claim 9, wherein the derivation matrixIncludes normalizing each of a plurality of columns of the matrix C, determining an average of inverses of the plurality of normalized columns of the matrix C, and constructing the matrix C based on the average of inverses of the plurality of normalized columns

12. The method of claim 4, further comprising: deriving matrices based on Minimum Mean Square Error (MMSE) computationAnd

13. the method of claim 4, further comprising: derivation based on Minimum Mean Square Error (MMSE) calculationMatrix arrayAndto minimize the Mean Square Error (MSE) given by

|H _up K _ut-(H _dn K _ap)^T|²。

14. The method of claim 1, wherein the determining a correction factor for an access point and a correction factor for a user terminal comprises

Deriving a first set of matrices of correction factors for the access point corresponding to the first set of frequency subbands, and

interpolating the first set of matrices to obtain a second set of matrices of correction factors for the access point corresponding to a second set of frequency subbands.

15. The method of claim 1, wherein the determining the correction factor for the access point and the correction factor for the user terminal comprises

Deriving a first set of matrices of correction factors for the user terminal corresponding to the first set of frequency subbands, and

interpolating the first set of matrices to obtain a second set of matrices of correction factors for the user terminal corresponding to a second set of frequency subbands.

16. The method of claim 1, further comprising:

transmitting a pilot on the uplink channel, wherein the uplink channel response estimate is derived based on the pilot transmitted on the uplink channel; and

receiving a pilot on the downlink channel, wherein the downlink channel response estimate is derived based on the pilot received on the downlink channel.

17. An apparatus in a wireless Time Division Duplex (TDD) communication system, comprising:

means for obtaining a channel response estimate for a downlink channel from an access point to a user terminal;

means for obtaining a channel response estimate for an uplink channel from the user terminal to the access point; and

means for determining a correction factor for the access point and a correction factor for the user terminal based on the channel response estimates for the downlink and uplink channels, the correction factor for the access point and the correction factor for the user terminal used to obtain a calibrated downlink channel response and a calibrated uplink channel response.

18. The apparatus of claim 17, further comprising:

means for applying the correction factor for the access point on a transmit side, or a receive side, or both the transmit and receive sides of the access point.

19. The apparatus of claim 17, further comprising:

means for deriving a correction factor for a transmit side of the access point and a correction factor for a receive side of the access point based on the correction factor for the access point.

20. The apparatus of claim 17, further comprising:

means for applying the correction factor for the user terminal on a transmit side, or a receive side, or both the transmit and receive sides of the user terminal.

21. The apparatus of claim 17, further comprising:

means for deriving a correction factor for a transmit side of the user terminal and a correction factor for a receive side of the user terminal based on the correction factor for the user terminal.

22. The apparatus of claim 17, wherein the means for determining the correction factor for the access point and the correction factor for the user terminal comprises

Means for performing Minimum Mean Square Error (MMSE) calculations on the channel response estimates for the downlink and uplink channels to determine the correction factor for the access point and the correction factor for the user terminal.

23. The apparatus of claim 17, wherein the means for determining the correction factor for the access point and the correction factor for the user terminal comprises

Means for performing a matrix ratio calculation on channel response estimates for the downlink and uplink channels to determine the correction factor for the access point and the correction factor for the user terminal.

24. A method of calibrating a communication link in a wireless Time Division Duplex (TDD) multiple-input multiple-output (MIMO) communication system, comprising:

transmitting a pilot on a first communication link from a first station to a second station;

obtaining a channel response estimate for the first communication link derived based on the pilot transmitted on the first communication link;

receiving a pilot from the second station on a second communication link;

deriving a channel response estimate for the second communication link based on the pilot received on the second communication link; and

determining a correction factor for the first station and a correction factor for the second station based on channel response estimates for the first and second communication links, the correction factor for the first station and the correction factor for the second station used to obtain a calibrated channel response for the first communication link and a calibrated channel response for the second communication link.

25. The method of claim 24, further comprising:

applying the correction factor for the first station at either the transmit side or the receive side or both the transmit and receive sides of the first station.

26. The method of claim 24, further comprising:

transmitting the correction factor for the second station to the second station.

27. The method of claim 24, further comprising:

updating a correction factor for the first station based on calibration with a plurality of second stations.

28. An apparatus in a wireless Time Division Duplex (TDD) communication system, comprising:

a transmit spatial processor configured to transmit a first pilot on a first communication link from a first station to a second station;

a receive spatial processor configured to receive a second pilot on a second communication link from the second station, to derive a channel response estimate for the second communication link based on the received second pilot, and to receive a channel response estimate for the first communication link derived based on the transmitted first pilot; and

a controller to determine a correction factor for the first station and a correction factor for the second station based on channel response estimates for the first and second communication links, the correction factor for the first station and the correction factor for the second station used to obtain a calibrated channel response for the first communication link and a calibrated channel response for the second communication link.

29. The apparatus of claim 28, wherein the controller performs a Minimum Mean Square Error (MMSE) calculation on channel response estimates for the first and second communication links to determine the correction factor for the first station and the correction factor for the second station.

30. The apparatus of claim 28, wherein the controller performs a matrix ratio calculation on channel response estimates for the first and second communication links to determine the correction factor for the first station and the correction factor for the second station.

31. The apparatus of claim 28 wherein the controller derives correction factors for the transmit spatial processor and correction factors for the receive spatial processor based on the correction factor for the first station.

32. The apparatus of claim 28, wherein the controller updates the correction factor for a first station based on calibration with a plurality of second stations.

33. A method of transmitting data in a wireless Time Division Duplex (TDD) multiple-input multiple-output (MIMO) communication system, comprising:

applying a correction factor for a first station at a transmit side, or a receive side, or both the transmit and receive sides of the first station;

transmitting a pilot on a first communication link from the first station to a second station; and

receiving a data transmission sent over a second communication link from the second station to the first station, wherein the data transmission is spatially processed based on a channel response estimate for the first communication link derived from the pilot sent over the first communication link.

34. The method of claim 33, further comprising:

the received data transmission is spatially processed with a matched filter.

35. The method of claim 33, wherein the receiving the data transmission sent on the second communication link comprises receiving the data transmission sent on the second communication link

Receiving the data transmission sent on a plurality of eigenmodes of the second communication link.

36. The method of claim 33, wherein the second station applies a correction factor to a transmit side, or a receive side, or both the transmit and receive sides at the second station.

37. An apparatus in a wireless Time Division Duplex (TDD) multiple-input multiple-output (MIMO) communication system, comprising:

means for applying a correction factor to a transmit side, or a receive side, or both the transmit and receive sides at the first station;

means for transmitting a pilot on a first communication link from the first station to a second station; and

means for receiving a data transmission sent over a second communication link from the second station to the first station, wherein the data transmission is spatially processed based on a channel response estimate for the first communication link derived from the pilot sent over the first communication link.

38. An apparatus in a wireless Time Division Duplex (TDD) multiple-input multiple-output (MIMO) communication system, comprising:

a transmit processor for transmitting a pilot on a first communication link from a first station to a second station; and

a receive processor to receive a data transmission sent on a second communication link from the second station to the first station, wherein the data transmission is spatially processed based on a channel response estimate for the first communication link derived from the pilot sent on the first communication link, and the transmit processor applies a correction factor to the pilot sent, or the receive processor applies a correction factor to the data transmission received, or the transmit processor applies a correction factor to the pilot sent and the receive processor applies a correction factor to the data transmission received.

39. A method of transmitting data in a wireless Time Division Duplex (TDD) multiple-input multiple-output (MIMO) communication system, comprising:

transmitting a pilot on a first communication link from a first station to a second station;

receiving a data transmission sent over a second communication link from the second station to the first station, wherein the second station applies a correction factor to a transmit side, or a receive side, or both the transmit and receive sides at the second station, and the data transmission is spatially processed based on a channel response of the first communication link derived from the pilot sent over the first communication link; and

processing the received data transmission with a Minimum Mean Square Error (MMSE) receiver at the first station.

40. The method of claim 39, wherein the receiving the data transmission sent on the second communication link comprises

Receiving the data transmission sent on a plurality of eigenmodes of the second communication link.