TWI336193B - Method and apparatus for space-frequency equalization for oversampled received signals - Google Patents

Description

1336193 IX. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present disclosure relates generally to communications, and more particularly to techniques for performing equalization at a receiver in a communication system. N [Prior Art] In a communication system, a transmitter typically processes (e.g., encodes, interleaves, symbol maps, spreads, and blends) traffic data to produce a sequence of wafers. The relay then processes the wafer sequence to produce a radio frequency (RF) signal and transmits the RF signal via a communication channel. The communication channel distort the transmitted RF signal with a channel response and additionally degrades the signal with noise and interference from other transmitters. The previous receiver receives the R F signal transmitted by the user and processes the received r F (4) to obtain a sample. The receiver can perform equalization of the samples to obtain an estimate of the wafers transmitted by the pirates. The receiver then processes (e.g., de-mixes, despreads, demodulates, deinterleaves, and decodes) the wafer estimates to obtain decoded data. The equalization performed by the receiver typically determines the quality of the wafers and The overall performance has a big impact. Therefore, there is a need in the art for the technology to be implemented in a way that achieves good performance. SUMMARY OF THE INVENTION Techniques for performing on-frequency are described herein. The inter-day frequency equalization of the parent frequency specialization and space equalization ^ pa ^, combining the signal components in the two and frequency dimensions and the equalization of the signals spatially combines the signal components. Also described in accordance with one embodiment of the present invention includes at least one processor and _ 110773.doc 1336193: the appearance of the memory. The processor is derived from multiple pure antennas and oversampled = equalizer coefficients for multiple signal replicas (or spectral replicas). The (etc.) operator uses the equalizer coefficients to pass the input symbols for the plurality of signal replicas to obtain an output symbol. According to a further embodiment, a method is provided in which equalizer coefficients for a plurality of signal replicas are obtained from a plurality of receiving antennas and oversampling. The method uses the equalizer coefficient (4) to feed the input symbols of the plurality of signal replicas. A device is described in accordance with yet another embodiment. The device includes means for equalizer coefficients from a plurality of receive antennas and oversampling for multiple signal replicas. The apparatus further includes means for using the equalizer coefficients to pass the input symbols for the copies of the signals.

According to yet another embodiment, an apparatus is described that includes at least one processor and a memory. The (etc.) processor obtains input symbols for a plurality of (M) signal replicas from a plurality (8) of receiving antennas and a plurality of (c) oversamplings, where m is equal to R times C. The (equal) processor obtains equalizer coefficients for the M signal replicas, and the input symbols for the M signal replicas are passed over the four coefficients and will be used for the M signals The filtered symbols are combined to obtain an output symbol. According to yet another embodiment, a method is provided in which input symbols for μ signal replicas are obtained from R receive antennas and C oversampling. The equalizer coefficients for the copies of the signals are obtained. The equalizer coefficients are used to transition the input symbols for the one of the t-number replicas. The filtered symbols for the replicas of the μ signals are combined to obtain an output symbol. 110773.doc β) According to yet another embodiment, a device is described that includes means for obtaining an input symbol for a plurality of signal replicas from R receiving 'antennas and C times oversampling, for obtaining for Μ a component of the equalizer coefficients of the plurality of signal replicas, means for using the equalizer coefficients for the rounded symbols for the copies of the plurality of signals, and for combining the copies of the signals for the signals The components that have passed the symbol. According to yet another embodiment, an apparatus is described that includes at least one MEMS and a memory. The (etc.) processor constructs at least one spatial frequency equalizer and at least one spatial equalizer. Each spatial frequency equalizer combines the signal components in spatial and frequency dimensions. Each spatial equalizer combines signal components in a spatial dimension. According to yet another embodiment, a method is provided in which a signal component is combined in a spatial and frequency dimension for a first set of at least one frequency bin; Two sets of at least one frequency slot combine signal components in a spatial dimension. According to yet another embodiment, an apparatus is described, the apparatus comprising means for combining signal components in a spatial and frequency dimension for a set of at least - frequency slots; and for spatially dimensioning a second set of at least one frequency slot A component that combines signal components. Various aspects and embodiments of the invention are described in further detail below. [Embodiment] The term "exemplary" means " as an example, an example or a description ". Any embodiment described herein as "exemplary," is not necessarily to be construed as preferred or advantageous over other embodiments. For the sake of clarity, the following nomenclature is used for most of the following descriptions. In the lower case of the index with the 110773.doc for the sample period, the time domain scalar is shown, for example, (n) in the uppercase with the exponent k for the frequency bin, the frequency domain scalar 'for example 'H(k) The vector is represented in bold lowercase, for example, the package, and the matrix is represented in bold uppercase, for example, 迀. Figure 1 shows a communication with two transmitters 110? (and 11〇y and one receiver 15〇) System 100. Transmitter 11 is equipped with a single-antenna core, transmitter H〇y is equipped with a plurality of (Τ) antennas 112 & to 1121, and receiver 15 is equipped with a plurality of (R) antennas 152a to 152r. The single antenna of the transmitter 及10χ and the R antennas of the receiver 15〇 form a single input multiple output (SIM〇) channel. The T antennas of the transmitter u〇y and the R antennas of the receiver 150 form a multiple input multiple output. (ΜΙΜΟ) channel. For the transmitter 11〇x&u〇y, there is a single-input single-output (SIS〇) channel between each pair of transmitter/receiver antennas. The SIS〇 channel can be characterized by a time domain channel. Impulse response h(n) or frequency domain channel frequency response H(k). A Fourier Transform (FFT) or a κ-point Discrete Fourier Transform (DFT) converts a time domain representation into a frequency domain representation, which can be expressed as: where π-1" in the index is due to the exponents η and k starting from 1 instead of 〇 Initially, a frequency domain representation can be converted to a time domain representation by K-point inverse FFT (IFFT) or a K-point inverse DFT (IDFT), which can be expressed as:

咖)=丄.|//(moxibustion).e_y2jr(A:1)(n-1)/K

Eq(2) 110773.doc 1336193 Figure 2 shows a signal flow 200 for data transmission from a single antenna transmitter n to multiple antenna receivers 150. Receiver 150 uses the reception difference' which is to receive a single data stream with multiple receive antennas. For simplicity, Figure 2 shows a case where the receiver 150 has two antennas and oversampling the received signal from each receiving antenna twice (2x). Figure 2 shows the use of a spatial segment frequency domain equalizer (FDE) that performs equalization in the frequency domain. The term

''Spatial Section'' refers to sampling at a higher rate than the rate required by the Nyquist sampling theorem.] Transmitter 110x processes the traffic data and generates the transmission wafer x(n) at a crystal monthly rate where n' is for the wafer. The index of the cycle. The transmitter can append a cyclic prefix to each Κ/2 transport tile. The loop prefix is a repetition of the data block. The I5 knife uses its anti-shock frequency to select the inter-symmetry interference caused by fading. The 忒 frequency selective fading is a frequency response that is not flat on the system bandwidth. . . In an actual system, the transmitter transmits a sequence of transmission wafers to the receiver. For signal flow 2, a one liter sampler 21 inserts zero after each transfer wafer and produces a transfer sample X(8) at a sample rate that is twice the wafer rate for 2 oversamplings, where η is One of the sample periods is an index. The transmission samples are sent from the single-transmitting antenna and are passed through the sim to the two receiving 1嫂 lines##-the channel pulse in the block^ and the additional noise η(4) via the adder The SIS channel modeling for the two = line. For the second receiving antenna: Li Guangjing is modeled by a channel impulse response in the block and by additional noise n2 (4). For each connection) receiving day 110773.doc 1336193 line K where the channel impulse response (8) includes any pulse shaping chopper at the transmitter, the propagation channel, any front-end chopper at the receiver The effect of etc. Receiver 150 digitizes the received signal from each of the receive antennas at twice the rate of the wafer and obtains the input samples at the sample rate (not shown in Figure 2). If the transmitter attaches any cyclic prefix to each data block, the receiver can remove the cyclic prefix. The time domain input samples from the first receiving antenna are input by a unit 23A using a κ point #attachment/attachment. (4) Converting to the frequency domain to obtain frequency domain input symbolization), where k == i, ..., κ. As shown in Figure 4A, the 2 oversampling results in two signal spectral replicas per receive antenna. A copy of the two redundant signals in the oversampled spectrum for each receive antenna is represented as a lower copy (L) and a higher copy (9). A signal replica can also be referred to as a spectral replica or as some other term. Enter K/2 input 4 No. 1^(1^(where k = l, "., K/2) as the symbol R, L(k) for the lower replica, where k=1,... , κ/2, and provide it to the equalizer | 240a. Enter the last κ/2 input symbol 丨 〇 (] 〇 (where k = K / 2 + i, ..., q is expressed for the k-question copy The symbol Ri u(k), which is ten k=l,...,κ/2, and provides it to a specialization ^ 242 a. Similarly, a unit 230b uses a point fft/df The time domain input sample r 2 (n) from the second receiving antenna is converted into a frequency domain to obtain a frequency domain input nickname R2(k), where k is called, ..., κ. The input symbol R2(k) (where k=l, . . . , K/2) is represented as a symbol for the lower replica, where k—], . . . , K/2, and is supplied to the equalizer 240b. Enter K/2 at the end of the symbol R2(k) (where k=K/2+1, .··, κ) is * for the higher copy I10773.doc 1336193 ruler 2,1; 1〇, where k=l,...,Κ/2, and extract it to the equalizer 242b. The equalizer 240a filters its input symbol with its coefficients (1〇 and provides the 遽Symbol of which "*,, table A conjugate complex number is used. The equalizer 242a filters its input symbols h, u(k) with its coefficients and provides a filtered symbol Yi.iKk). The equalizer 240b filters its coefficient ash 2*丄(8) The symbols R2, L(k) are input and the filtered symbols Y2, L(k) are provided. The equalizer 242b filters its input symbol R2 u(k) with its coefficient w^k) and provides a _ filtered symbol Y2, u(k). An adder 244a will be separately from the equalizer to add the filtered symbols Yl, L(k) and Y2, L(k) with 240b and provide output for the lower replica. Symbol YL(k). An adder 244b adds the filtered symbols γ丨, u(k) and Y2, u(k) from equalizers 242a and 242b, respectively, and provides output symbols for the more replicas. Yu(k). A unit 250 performs a K-point IFFT/IDFT on the output symbols YL(k) and Yu(k) and provides an output sample y(n) at a sample rate. A downsampler 252 discards the output samples every other time. The output samples yrhV are provided at a wafer rate. The output samples are further processed to obtain Β decoded data. Figure 3 shows a frequency domain signal flow 300 for receiving differences for a spatial segment FDE. Signal Flow 300 The signal flow in Figure 2 is equivalent and is also used in the case of having two receive antennas and oversampling twice. The transmitter 110x processes the traffic data and generates a transmission crystal. In an actual system, the transmitter The transmit wafer sequence is sent to the receiver and no FFT/DFT is performed. However, for signal flow 3 〇〇 , a unit 3 〇 is difficult for the transfer wafers, a Κ / 2 point FFT / DFT is performed and a frequency domain transmission symbol (8) is provided, where hU., K/2. The transmission symbol (8) is transmitted from the single transmit antenna 110773.doc 1336193 and via the SIMO channel to the two receive; antenna. By (1) the frequency response H], L(k) in block 320a and the additional noise N through the adder 324a for the lower replica, L(k) and (2) the frequency in block 322a The siso channel for the first receive antenna is modeled in response to Hi u(k) and via additional noise Ni, u(k) for one of the higher replica adders 326a. A unit 328a converts - time domain noise ηι(η) and provides frequency domain noises lVL(k) and Nl u(k). Similarly, the frequency in (1) the frequency response H2L(k) in block 320b and the additional noise N2, L(k) and (2) in block 322b via one of the lower/replica adders 324b The SIS〇 channel for the second receive antenna is modeled in response to H2, u(k) and via additional noise N2, u(k) complemented by one of the higher replicas 32. A unit 328b converts the time domain noise ~(11) and provides the frequency domain MN2L(k) and N2, u(k). As shown in Figure 3, the transmission symbol (9) is transmitted via all four blocks 320a, 320b, 322a, 322b. At the receiver 150, the first equalizer 340& self-adder 32 receives the frequency domain input symbols R1, L(k), filters the input symbols with the coefficients (4) thereof, and provides the symbol Y1 for the consideration. , L(k). The equalizer 342a receives the input symbol R from the adder, u(k) 'filters the input symbols with their coefficients, and provides a filtered symbol Y, u(k)e, equalizer 34〇 b The adder 32 receives the input symbol Uk), filters the input symbols with its coefficient %, and provides the filtered symbol Y2L(k). The equalizer 342b receives the input symbols R2, u(k) from the adder 32, using its coefficients ash 2^ (magnesizes the input symbols' and provides the filtered symbols Y2iU(k). An adder 344a The filtered symbols Y^iXk) from the equalizers 34〇3 and 342a, respectively, are added to YllJ(k) and provide a 110773.doc filter symbol Y](k) for the first receive antenna. An adder 344b adds the filtered symbols Y2, L(k) and Y2, u(k) from equalizers 340b and 342b, respectively, and provides filtered symbols Y2 for the second receive antenna (k). ). An adder 346 adds the filtered symbols Yi(k) to Y2(k). A gain element 348 scales the output of adder 346 with a 1/2 gain and provides an output symbol y(8). A unit 350 performs a K/2 point IFFT/IDFT on the output symbol Γ(4) and provides a time domain output sample at the wafer rate ("'). In comparing signal flows 200 and 300, by upsampler 210 in Figure 2; c (2 times upsampling followed by a K-point FFT/DFT is equivalent to unit 310 in Figure 3 _') Perform a K/2 point FFT/DFT and copy Z(4) for the lower and higher copies of the oversampled spectrum. In FIG. 2, Y!, L(k) is added to Y2, L(k) by the adder 244a, and Yi^k) is added to Y2, u(k) by the adder 244b. A series of operations performed by unit 250 for a K-point IFFT/IDFT and by a factor 2 of decimator 252 is equivalent to adding Y^dk) and YiMk) by adder 344a in FIG. Y2, L(k) are added to Y2, u(k) by adder 344b, and Y, (k) and Y.2(k) are added by adder 346, and scaled by unit 348. 1/2 and a series of operations of K/2 point IFFT/IDFT performed by unit 350. In FIG. 3, adders 344a and 344b perform spectral summation and adder 346 performs spatial summation. Spectral and spatial summation can also be performed in other ways. For example, in FIG. 3, Y^Jk) can be added to Y2, L(k) to obtain YL(k) (which corresponds to the output of adder 244a in FIG. 2), and Yi, u(k) and Y2 can be obtained. , u(k) is added to obtain Yu(k) (which corresponds to the output of adder 244b in FIG. 2), and the addition can be further scaled by 1/2 to obtain Y(k). 110773.doc • 13· 1336193 Figure 4A shows an exemplary spectrogram of two receive antennas with 2 oversamplings. The data chip is based on the wafer rate fc. The corresponding spectrum has a one-sided bandwidth fc/2, or equivalently has a double-sided bandwidth, and the roll-off of the transmitter's pulse shaping filter determines. The sample rate heart is used to digitize the received signal from each of the receiving antennas at a rate that is twice the wafer rate, or 匕 = 2 匕. For each receive antenna, the lower replica covers the frequency range from 〇〇 to 匕/2, which corresponds to the slot index ' of k = 1 to K/2 and the higher replica covers the frequency range 匕/2 to fs' The groove index at k = K/2 + iK. For the sake of simplicity, Figure 4A shows a similar spectrum for two receive antennas. In general, the spectrum of each receiving antenna r has a shape determined by the frequency response of the antenna. If H1(k) is not equal to H2(k), the spectral patterns of the two receiving antennas may be different, which is usually the case and is used due to reception differences. Four signal replicas are obtained from the redundancy factor 2 of the two receive antennas and the other redundancy factor 2 of the two oversamplings as shown in Figure 4A. Figure 4Α also shows how the four redundant signal components of the four signal replicas should be combined. The two redundant signal components of each receive antenna are separated by [/2 distance "/2 frequency slots. As shown in Figure 4A, a spatial frequency equalizer can be used for each frequency bin k, where 々 = !, ._., K/2. The spatial frequency equalizer for frequency slot k can be two: the receive antenna combination of redundant signal components for slot kAk + κ/2. The spatial frequency equalizer can be used for κ/2 frequency slots. For the sake of clarity, the processing of a frequency bin k is described below. The same processing can be performed for each of Κ/2 frequency bins or where k = 1, milk. For SIMO transmission from transmitter 110x to receiver 150, the frequency domain input symbol at the receiver of 110773.doc 14 1336193 can be expressed as ···

Eq(3) 4 X 1 input symbol to T(k) = h(k)X(k) + n(k) quantity, 蚴)= between 'lW A, lW岣, uW //印(1)]7· For a 4χ1 channel gain vector, (10))=[heart 1^) heart, 1^) heart 1^)~ (mail is a 4><1 noise vector, and "Γ" indicates transposition. The labels U and L represent the higher and lower replicas of each receiving antenna, respectively, and are separated by a κ/2 frequency slot as shown in Fig. 4A. The frequency domain output symbols of the FDE can be expressed as: Y(k ) = wH(k)-r(k),

Eq(4) =(k) · h(k) - X(k) + wH (k) · «(k), = B(k)X(k) + V(k), where <lW/2 <uW/2 π. (9)/2] is a 4x1 line vector for the equalizer coefficient of frequency φ slot!^, knives (4)=冱"(4)dust (4) for scaling, F(nine)= and "(9)·post) for meaning (magic The filtered noise, and "β " denotes a conjugate transpose. In equation (4), the equalizer coefficient includes the scaling factor 1 /2 ^ of the gain element 348 in Figure 3 can be based on a minimum The mean square error (MMSE) technique, a forced zeroing technique maximum ratio combining (MRC) technique, etc. obtain the equalizer coefficients. For the MMSE technique, the equalizer coefficients satisfy the following conditions: II0773.doc (?) 15

Eq(5) where £{} is a double period: j is heavy and 7 is π. Equation (5) minimizes the mean square error between the FDE output y(4) and the transmitted symbol. The MMSE solution of equation (5) can be expressed as: = . "Song gamma + just ' Eq (6) where 邶) = £ {丨 Youzhou 2} is the energy spectrum of the transmission chip X扪, and the group illusion = 玛姒).〆 (called 4X4 noise covariance matrix. You can apply matrix inversion lemma to equation (6). The equalizer coefficient can represent (4): l + S(k)-hH(k) .^j^k) Equation (7) has a 4><4 inverse matrix p(4) for each frequency bin. Equation (7) can be simplified as described below. In a first simplification scheme that can be used if the lower and higher replicas of the oversampled spectrum have uncorrelated noise or have negligible noise correlation, the noise covariance matrix has the following block diagonal versions: RlW 〇 ^

Eq(7) ) where 0 RuW RCW =

Eq(8)Ά)Ά).λ:(8) 〇lc(k) :(Α〇==£{|Ά)丨2} is the noise variance for the replica c from the antenna r; 〇jk)^E {Ni^k)-N24k)}, ' is the noise correlation coefficient between the two receiving antennas,

Tlc\ ee{L,U} is an index for lower and higher replicas, and 110773.doc •16-1336193 re{l,2) is the number of fingers used for the two receiving antennas. (8) is a -2x2 hetero-ifl covariance matrix for two receive antennas of one of the frequency bins k of the signal replica c. The simplification in equation (8) can also be performed if the correlation between the two receiving antennas can be ignored but the spectral correlation of the noise components in the lower and south replicas can be ignored. In this case, the 4 χ 1 vector of equation (3) can be rearranged to obtain the block diagonal matrix shown in equation (8). When R(9) is defined as shown in equation (8), the equalizer coefficient in the five illusion can be

Table is · 2 D(k) where r = l, 2 and c = 4 ¢/

Eq(9) where hcW = [//i, cW Ά)] Γ is a 2x1 channel gain vector for the slot k of the replica c, [h?W-Hi〇^^(k)clc(k).a2c (k)

Glc{k)-〇lc{k)-{\-\pc{k)\2) [h?w^w]2=·do).y ~2,#) ffUk)-alc(k)-( \-\pc(h)\2) , and O(k) = 1 + S(k) [h« (k) · R-] (k). hL (k) + h« (k). ri ( k). hu (k)] The composition of D(k) can be expanded as follows: 丨~(8)I2 +1'c(10)2仃心々)-2 ton (Ming 2 八八(10).'c(8)%(々) °Uk)alc (k)-(\-\pc(k)\2) ·

Eq(10) In a second simplified scheme in which the noise space and the spectrum are uncorrelated and have a spatial and spectrally equal noise variance, the noise covariance matrix E(4) has the following form: m)-c2 {k)l 110773.doc

Where 4(*) = 0^(moxibustion)= /\ (&) = Pu (A) = 0, and 2(4) is the noise variance L is a unit matrix. When defined as shown in _° equation (1〇), the equalizer coefficient in 6) can be expressed as · __, J rc y^j S(k).|^(k)| + CJ2(k ), where r = l, 2 and c = 々, where ||iiW||= ~(*) + //2l(a) 2+ is good 2 2 ' 丨,υ · IliWl is for K(k) S(k).H^j〇— ~~ one... Eq(ll) 2 channel k response vector secret) Su Shi·丨丨-, π” ''..., the consumption. For example, equation (11) As shown, even if the noise space and spectrum are not related to s, F, but four equalizer coefficients ^l, LW /2 H ^ 2, LW /2 } PFjt, (k) 12 Ji W* (k\ η ' 2'υ()/2 is jointly determined by four spatial and spectrally isolated channel gains. The noise components used for the two receiving antennas on the right are irrelevant such that the sub = sub = 0 in equation (8) In a third simplified scheme that can be used, the noise covariance matrix S(8) has the following form: 占(k) = L(k) 0 0 0 0 0L(k) 0 0 0 0 σί#) 〇0 0 0 '00

Eq(12) For the noise covariance matrix shown in #式(12), different noise variables can be obtained for different signal replicas. The equalizer coefficient 竺"(9) can be obtained based on E(9) as defined in equation (12). Other simplifications can be made for other conditions. For example, the noise correlation between two receive antennas can be a frequency invariant such that w = eight and w = /?u. These different simplifications reduce the calculation of the equalizer coefficients shown in equation (7). The signal-to-noise ratio (SNR) for the wafer rate output sample y(n,) can be expressed as: 110773.doc •18, 1336193 Κ/2

SNR ZS(k) ‘wafer—κη 2nFB(k)−l|2S(k) + |F|2.〇2v(k);

Eq(13)

F K/2 K/2 Ϊ > (*) is a scaling factor, and φ ^ is the wafer SNR of the output sample. Equation (4) provides a bias of MMSE for /(4). The scaling factor F can be applied to y (magic or to obtain an unbiased estimate of the phantom or separately. If a data symbol is spread over a single chip with a spread code (eg 'a Walsh code or a VSF code') The symbol SNR of the data symbol can be obtained by multiplying the SNR of the wafer by the spread code length μ. The space segment FDE of the SIM〇 transmission to the two receiving antennas can be extended to SIMO transmission to any number of receiving antennas. The FDE is transmitted from multiple (T) transmit antennas to multiple (R) receive antennas. For clarity, the following describes a 2Χ2 transmission with two transmit antennas, two receive antennas, and two oversamplings. For a transmission from the transmitter 110y to the receiver (9), the frequency domain input symbols of the receiving benefit can be expressed as:

[(4)=石](4).义1W+赵2 W.义2 W+nW ^ , . Eq(14) _中Σ() is a 4χ1 input symbol vector, hook (8) and especially 2 (the illusion is self-transmitting antenna 1 And 2 transmitted symbols, _1 () is a 4x1 channel gain vector for transmitting antenna 1, 110773.doc -19- 1336193 h2 (A) is: τ is used for transmitting antenna 2 4 χ 1 channel gain vector, and 2 ( The magic is a 4x1 noise vector. The second quantity), _, _, and _ have the form shown in equation (3). Two equalizer coefficient vectors 岌]//(*) and meaning can be obtained. The two transmission symbols 4 (9) and 2 (illusion can be recovered for each frequency slot k, respectively. The equalizer coefficient vector can be obtained based on MMSE, forced return to zero, MRC or some other technique.

„The MMSE equalizer coefficients for each antenna; can be expressed as: (k)==---i(k)__uHn\ ^ (8), where t = 1, 2 Eq(15) where 〖 The index of the two transmitting antennas, Η (k) is the transmitting antenna ί 1x4 equalizer coefficient vector, ^(4)-£{丨4(8)丨} is the energy spectrum transmitted from the antenna ί, and 骂(9) is the transmitting antenna / 4 Χ 4 noise and interference covariance matrix. The noise and interference covariance matrix of two transmitting antennas can be expressed as: 里1 (moxibustion) = 52 (magic. 112 (magic.) 1" (magic + far (female), and

.·· ^.i>(k) = S}(k)· h, (k) hf (k) + R(k) Eq(16) Equation (16) represents the noise and interference protocol of the transmitting antenna t The variance matrix 骂(4) includes (1) the interference of the noise covariance matrix that can be used for the two transmit antennas, and (2) the interference of the data stream transmitted from the other transmit antenna Z, which is ^ bW. The inter-stream interference is determined by the channel response vector H of another transmitting antenna r; (4) and the energy spectrum (4). The above simplification for SIMO transmission is generally not applicable to MIM〇 transmission. This is due to the soil, (9) including inter-stream interference from other transmitting antennas. Therefore, even if K(4) is a pair of angular matrices due to spatial and spectral uncorrelated noise, the inter-flow interference H0773.doc -20* 1336193 is usually not a diagonal matrix. Therefore, the executable can be used to invert the matrix to obtain the equation (15), and the equalizer coefficient vector and coffee can be applied to the wheel vector (4) to obtain the output symbols ¥) and Y2 (8), respectively. Estimates for the transmission symbols ^(8) and X2(k), respectively. The frequency domain rounding symbol from FDE can be expressed as . Y,(k) = (k)·r(k) = Bt(k)-X,(k) + Vt(k), where t = i, 2 Eq(17) where K(6) is the estimate of A(8) transmitted from the transmitting antenna t, the interference

WWW·[hang¥) + η(10) is the noise and dryness of the item. The SNR of each transmit antenna can be expressed as: Κ/2

Es.(k)

Eq(18) Continued wafer,,=77^-^--

XnFt-Bt(k)-l|2.St(k) + |Ft|2-a^t(k); k = l where <#) = five {RW|2b five f(4) is the variance • 1 Κη Ί ·] F, ss is a scaling factor of the transmitting antenna t and is the wafer SNR of the transmitting antenna t. Equation (17) provides a winter (4) bias MMSE estimate. The scaling factor can be applied to Q(9) or M"') to obtain an unbiased estimate of the hook (4) or 々 ("') respectively. If a data symbol is spread over a single wafer with a spread code, the chip is used. The SNR is multiplied by the length of the spread spectrum code to obtain the symbol SNR of the data symbol. A spatial frequency equalizer structure can be used for a SIMO transmission or a transmission. As described above, a spatial frequency equalizer will be used for all antennas. The combination of redundant signal components of k and k + N/2. For the case of R=2 and 2 times overtaking 110773.doc 1336193, the vector of each equalizer coefficient 冱〃(*) or 〇t can be separately Executing RW or R, (magic 4x4 matrix inversion. See Figure 4A 'M々) = Fi, lW 丄 (4)] 7· is the passband of the lower replica and the non-zero transition band. Similarly, Hu(4)=[i/卬(8) is a non-zero passband and transition band in the higher copy. ilL(9) or ilu(8) are small or zero for the passband and the frequency band outside the transition band. Therefore, due to k or The signal component of k + K/2 is actually zero 'so for some frequency bins, there are only two φ (instead of four) The redundant signal components to be combined. In one aspect, the combination of the spatial frequency equalizer and the spatial equalizer is used for the K/2 frequency bin to reduce the complexity. The slot k in the lower replica can be used. And each frequency bin k with a non-negligible signal component on the slot k+K/2 in the higher replica uses a spatial frequency equalizer, which can be used only in slots k or higher replicas in the lower replica. A spatial equalizer is used for each frequency bin k in which there is a signal component on slot k+K/2. Figure 4B shows a spectral diagram of two signal replicas for a receive antenna. • As shown in Figure 4B, For each of the frequency bins bk; SKA2, the signal component on the slot k+K/2 in the higher replica is small or zero, and a space equalizer can be used for each of the slots. Each of the slots KA<k$KB, the lower k of the slot k and the signal component of the slot k+K/2 in the higher replica are not negligible and can be used for each of the slots a spatial frequency equalizer. For each of the frequency bins KB<k$K/2i, the signal component on slot k in the lower replica is small or zero and can be Each uses a spatial equalization. If the above simplification does not apply, a 4X4 inverse matrix can be used for each spatial frequency equalizer. If the simplification is not applicable, then each space can be used: 110773. The doc -22- 1336193 uses a 2 x 2 inverse matrix. The use of spatial frequency equalizers and spatial equalizers typically reduces complexity without degrading performance. For a typical case with one transmit antenna and R receive antennas The channel response vector for each transmit antenna t can be defined as follows: , 丄 丄 (幻 = idle·a(*) /f,.2LW...H|ja (ranging as an Rxl vector, fc'.u W e Ugly, .2, U(*)...^, Can(8)] 7' is an Rxl vector, and = (8) ώ(8)]7 is a 2Rxl vector. The noise vectors for these R receive antennas can be defined as follows. K*) —... is an Rxl vector, Πυ(Λ) —ΓΛ^,υ(々)... is an Rx 1 vector, and = is a 2RxI vector. The noise covariance matrix can be defined as follows: 3^(*) = £"{take(*).<(the magic) is an RXR matrix, 屯W=^{SuW.aiia)) is an RxR matrix, (9) Is the _ 2Rx2R matrix. The noise and interference covariance matrix can be defined as follows: ^r,L (k) = Σ5,· (k) h,· L (k) hfL (k) + R ()t) ^ _ ,, L LW is one ;RxR matrix,

^, uW = + R W '=1'M ' uw is an RxR matrix, and =fef) is a) 2Rx2R matrix.

^Γρ, /W •...Eq(19) Each of the frequency slots KA+ 1 to the ruler* The space frequency equalizer of the mother of the earth represents L. The space equalizer of the frequency slots 1 to KC is expressed as : for: n0773.doc

Cs) • 23- 1336193 Ugly S, t(8)-

St(k) l + stWh^(k).^;\k).h((k) -* v'y Eq(20) The spatial frequency of each of the frequency bins KB + 1 to K/2 The equalizer is expressed as ···, (々) = —----^1--- hH f]r. , τ,-l ,,.

Eq(2i) can be obtained by inverting a /Rx2R matrix to obtain the spatial frequency equalizer coefficients in equation (2〇). The space equalizer coefficient <(*) in equation (19) or (21) can be obtained by inverting an RXR matrix. For R = 2, the spatial equalizer coefficients can be obtained based on an alternative one - 2 χ 2 matrix inversion one closed solution.

Eq(22) Eq(23) Eq(24) To further reduce complexity, a common noise and interference covariance matrix can be used for all of the transmit antennas by applying the matrix inversion lemma. Then, the equalizer coefficients in equations (19) '(20) and (21) can be expressed as: 〇)=5, (/〇.ι^α).2Γ», where *Λ * , where KA< hKB, and 2^(*)=^,(*).1^(幻_3^(*), where KB<Jfc:S; K/2, where soil LW = think + Ψυ = ^Si(k) h,· u (k) h^j (k) + R y (jt) τ.] 'And Ψ(Α:)= Z^/Wh^^hfc^ + R^) , '=1 Ο In one implementation In the example, the frequency slot ka&kb can be defined as: ΚΑ=(1-α + ε)_Κ/4 and

Eq(25) K g =(1 + ex — ε) * Κ/4 , Eq(26) where α is the roll-off factor of the pulse shaping filter of the transmitter, and ε is the threshold of the equalizer selection. 110773.doc -24. 丄(10)193 The roll-off factor can be specified by the system, for example for w_cdma, α = 〇 22. The limit ε is determined by using spatial frequency equalization or spatial equalization and can be defined as 〇 meaning^. When £=〇, the space frequency equalizer can be used for the frequency slot α·Κ/2' for the remaining frequency slots (1, .2/2 space equalizer, and the complexity can be greatly reduced without performance Downgrade. If the threshold is increased, the space equalizer can be used for more frequency slots. The complexity is further reduced, but the performance may begin to degrade. You can choose the threshold based on the compromise between complexity and performance. A program for performing spatial frequency equalization is shown. Frequency domain input symbols for multiple (Μ) signal replicas or multiple per-receiving antennas are obtained from multiple receive antennas and multiple (C) oversampling Obtaining a frequency domain input symbol for c signal replicas (block 512). The input symbols of the two signal replicas are obtained by (1) receiving a time domain wheel sample at a C-chip rate for each-receiving antenna and (7) Converting the input samples into the frequency domain for each receive antenna to obtain the input symbols for the c signal replicas of the receive antenna. For example, based on channel and noise estimation and obtained for Μ signals according to mmse law Equalized (four) number of copies (block 514) The input symbols for the copies of the plurality of signals are filtered by the quantizer coefficients (block 516), and the symbols for the copies of the signals are combined to obtain the output symbols (squares). It is difficult to form a knife in the frequency slot k of the replica of the signal, where k is the frequency-index for each signal replica. 曰 right is - smo transmission recovery - data stream, then each signal Replica and to a set of equalizer coefficients. For example, if c=2 and R=2, then four sets of equalizer coefficients <L(KU(4), can be obtained for the four-n0773.doc -25- 1336193 nickname copy. %1 (phantom sum. For the above embodiment, each group includes κ/2 equalizer coefficients for κ/2 frequencies in a signal replica. The frequency slot equalizer coefficients can be used for each The frequency bin k forms a vector of one equalizer coefficient, (9). It can be based on (1) spectral uncorrelated hysteresis of the c signal replicas from each receiving antenna '(2) the antennas of the same size Spatially uncorrelated noise, or (7) assumptions about the space and spectral uncorrelated noise of the 彳5 复 copy Equalizer coefficients. As described above, the calculation of the equalizer coefficients can be simplified with any noise assumption. If multiple (τ) data streams are recovered for a single transmission, then this can be the same for each data stream. The (4) replica obtains the 等 group equalizer coefficient. For each frequency bin k' a noise and interference covariance matrix meaning (8) can be determined for each data stream and used to obtain the equalizer coefficient for the data stream < (heart or, for

For each frequency bin k, a common noise and interference covariance matrix can be determined and equalizer coefficients for all of the data streams are obtained based on the common noise and interference covariance matrix. Μ Use each group of equalizer coefficients for each data stream to transition the input symbols of the signal copies to obtain the 遽 symbols of the copies of the data signals of the data stream. The symbols of the (4) of the difficult signal replicas of each data stream may be combined to obtain the output symbols of the f stream. Conversely, even when C is greater than 2, since only 2R of the components of the majority (4) have an unavoidable signal energy, the reception β does not usually have to combine all of the M=C.R signal components. Usually the cutoff bands of the transmit fishers and the front end filters of the receivers suppress all other: redundant components. Therefore, even at 02, the actual dimension of the spatial frequency equalizer or the coherent moment 110773.doc -26· 丄336193 remains 2R. Figure 6 shows a process 600 for performing equalization without the combination of a spatial frequency equalizer and a spatial equalizer. Spatial frequency equalization is performed for a first set of frequency bins (e.g., frequency bins +1 to KB in Figure 4B) (block 6丨2). This spatial frequency equalization combines signal components in spatial and frequency dimensions. Space equalization is performed for a second set of frequency bins (eg, frequency bins 1 through 1^ and frequency bins "seven to ft/" in Figure 4B) (block 614). Space equalization combines signal components in spatial dimensions The first and second sets of frequency bins are selected from the frequency response, complexity and performance of the transmit pulse shaping filter (block 616). Figure 7 shows the transmitter 11 in the system 100 of Figure i. y and the receiver 15 〇 one block diagram. For the downlink/forward link transmission, the transmitter ii 〇 y is part of a base station, and the receiver 15 is part of a wireless device. Link/reverse link transmission, the transmitter n〇y is part of a wireless device/knife' and the receiver 150 is part of a base station. A base station is usually a fixed station that communicates with the wireless devices. It can also be called a node b, an access

Interleaving and symbol mapping) The traffic data JL provides data symbol modulators 730a through 730t modulation symbols, one of the preamble groups (for example, for M-PSK or 730t. As used herein, a data The symbol is a composite preamble symbol for a preamble modulation symbol, a modulation symbol H0773.doc -27- 1336193 m-qAM) and the preamble is the transmitter and receiver data. Each modulator 730 processes its number and preamble symbols in a manner specified by the system and provides a transport chip (8) to an associated transmitter unit (TMTR) 736. Each transmitter unit 736 processes (e.g., converts to analog, amplifies, filters, and upconverts) its transmitted wafer and produces a modulated signal. The modulated signals from the T transmitter units 736a through 73 are transmitted from the individual antennas 112a through 112t, respectively. • At the & receiver 150, the R antennas 152a through 152r receive the transmission signals via different signal paths and provide the R received signals to R reception scalars (RCVR) 754a through 7541, respectively. Each receiver unit 754 conditions (eg, filters, amplifies, and downconverts) its received signal, digitizes the adjusted signal at a rate that is multiple (eg, twice) at the rate of the wafer and provides a time domain input sample To a related FFT/DFT unit 756. Each unit 756 converts the input samples into a frequency domain and provides a frequency domain input symbol drought (8). A channel and noise estimator 758 can be based on frequency domain input symbols from FFT/DFT unit 756 (as in FIG. 7 and/or from time domain input samples of receiver unit 754 (not shown in Figure 7)) The channel response vector and noise. Channel and noise estimation can be performed in different ways known to the technology towel. A frequency domain equalizer (FDE) 76G obtains equalizer coefficients based on channel response vector and noise estimation. The equalizer coefficients are over the input symbols, combining the filtered symbols spatially or in frequency or only spatially, and providing output symbols to τ demodulators (Demod) 77〇d 77〇t If the transmitter [m〇 changes the symbol in the time domain by 5 weeks, for example, for CDMA, TDMA, and SC-FDMA, then each of the 77G can perform IFFT/IDFT on the output symbols from the FDE 76G. Per - 110773 .doc • 28- Demodulator 770 then processes its (frequency domain or time domain) output symbols in a manner that is complemented by the processing of the modulator 730 and provides data symbol estimates. A receive (RX) data processor 780 processes (eg, Symbol de-mapping, de-interlacing, and decoding) the data symbols are estimated and provided Decoded data. Typically, at transmitter 11 〇y, the processing of demodulator 770 and RX data processor 780 is complemented by modulator 730 and TX data processor 720, respectively. Controller/Processor 740 and 790 The different processing units at the transmitter 〇 〇 y and the receiver 150 are directly operated. The memory 742 and 792 respectively store data and program code for the transmitter 〇 〇 y and the receiver 150. Different communication systems such as CDMA) systems, time-sharing multi-directional proximity (TDMA) systems, frequency-multidirectional proximity (FDMA) systems, orthogonal frequency-multi-directional proximity (OFDMA) systems, single-carrier FDMA (SC-FDMA) systems, etc. Using the equalization techniques described herein, a CDMA system can construct one or more radio technologies such as Wideband CDMA (W-CDMA), cdma2000, etc. cdma2000 covers IS-2000, IS-856, and IS-95 standards. A TDMA system One such as the Global System for Mobile Communications (GSM) radio technology can be constructed. These different radio technologies and standards are known in the art. In one of the associations of the "3rd Generation Partnership Project" (3GPP) W-CDMA and GSM are described in the document. The cdma2000 〇3GPP and 3GPP2 documents are publicly available in a document for the association of "3rd Generation Partnership Project 2" (3GPP2). An OFDMA system is using orthogonal frequency division multiplexing (OFDM). The modulation symbols are transmitted in the frequency domain on the frequency subband. An SC-FDMA system transmits the modulation symbols in the time domain on the orthogonal frequency subband. 1 10773.doc -29- 1336193 The modulator 730 at the transmitter 110O and the demodulator 770 at the receiver 150 perform the processing as specified by the system. For example, modulator 73 can perform processing for CDMA, OFDM, SC-FDMA, etc., or a combination thereof. Those skilled in the art will understand that any of a variety of other techniques and processes may be used to represent information and signals. For example, voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or light particles, or any combination thereof, may be used to represent the materials, instructions, commands, information, signals, bits, symbols φ, and wafers referred to above. Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and operational steps described in connection with the embodiments disclosed herein may be constructed as electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of the hardware and the software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether the hardware or software is used to perform this kind of functionality, depending on the specific application and design and constraints of the entire system, the technician can construct the functionality for each specific application using different methods. It is not to be construed as causing a departure from the scope of the present invention.口 乂 拘 7 δ δ δ δ δ : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : Its modular logic device, discrete closed or transistor logic, discrete hardware set, and the combination of the material functions described herein to implement the processing process can be microprocessors, but instead, processed. . A processor, controller, microcontroller or state machine H0773.doc 1336193 may also be constructed as a combination of computing devices, for example, a combination of a DSp and a microprocessor, a plurality of microprocessors One or more microprocessors and any other such configuration in combination with a DSP core. The method or algorithm steps described in connection with the embodiments disclosed herein may be implemented directly in a software module or a combination of both executed by a stone or processor. A software module, group can be resident in RAM memory, flash memory, r〇m memory, EPROM memory, EEPR〇M memory, scratchpad, hard disk, • removable disk, a CD-R 〇M or any other form of storage medium known in the art. An exemplary storage medium is consuming to the processor such that the device can read information from or write information to the storage medium. Alternatively, the storage medium and the processor may be integral. The processor and the health storage medium can reside in an ASIC. The ASK: can be resident in the -user terminal. Alternatively, the processor and the storage medium may reside as a discrete component: the invention provides a prior description of the disclosed embodiments to enable any person skilled in the art to carry out or to practice the invention. The various general principles defined herein may be applied to other embodiments without departing from the invention. The invention is not intended to be limited to the embodiments shown herein but is not disclosed herein. The broadest range of principles and novel features is the same. [Simple diagram of the diagram] Figure 1 shows a communication vehicle + ', , · the first two transmitters and one receiver. Figure 2 shows the transmission from a single antenna transmitter to the receiver. Figure 3 shows a -signal flow for receiving a difference frequency domain equalizer. 110773.doc * 31 . Figure 4A shows the spectrum of two receiving antennas without 2 oversampling. Figure 4B unfolds • Tian Cang, a spectrum of two signal replicas for a receiving antenna. Figure 5 shows one of the procedures for performing spatial frequency equalization. It is shown that the execution is combined with the equalization of the spatial frequency equalizer and the space equalizer. Figure 7 shows the transmission of multiple antennas [Main component symbol description] 100 System 11 Ox Transmitter 110y Transmitter 112x Antenna 112a·.· U2t Antenna 150 Receiver 152a··· i52r Antenna 200 Signal Flow 210 Up Sampler 224a Adder 224b Adder 230a Unit 230b Unit 240a Equalizer 240b Equalizer 242a Equalizer 242b Equalizer 110773.doc

•32· 1336193

244a adder 244b adder 250 αΟ early 兀252 downsampler/decimator 300 signal flow 310 FFT/DFT unit 324a adder 324b adder 326a adder 326b adder 328a FFT/DFT unit 328b FFT/DFT unit 340a equalization 340b equalizer 342a equalizer 342b equalizer 344a adder 344b adder 346 adder 348 gain element/unit 350 IFFT/IDFT unit 720 transmission (ΤΧ) data processor 730a...730t modulator 736a... 736t Transmitter Unit (TMTR) • 33- lJ0773.doc

A 7401336193

742 754a."754r 756a." 756r 758 760 770a··· 770t 780 790 792 Controller/Processor Memory Receiver Unit (RCVR) FFT/DFT Unit Channel and Noise Estimator Frequency Domain Equalizer ( FDE) Demodulator Receiver (RX) Data Processor Controller / Processor Memory 110773.doc • 34·

Claims (1)

1336193 Patent Application No. 095115384 (Chinese Patent Application No. 095115384) (June 1999) X. Patent Application Range: A device for equalizing spatial frequency of oversampled received signals, comprising: at least one processor configured Obtaining equalizer coefficients for a plurality of signal replicas obtained via a plurality of receive antennas and oversampling, and filtering the input symbols for the plurality of signal replicas using the equalizer coefficients, and in spatial and frequency dimensions Combining the filtered symbols for a first set of at least one frequency bin and combining the filtered symbols in a spatial dimension for a second set of at least one frequency bin; and a s-resonance 'It is consuming to the at least one processor, wherein the number of the first set of at least one frequency bin and the number of the second set of at least one frequency bin are determined based on an equalizer selection threshold. 2. The apparatus of claim 1, wherein the at least one processor obtains the equalizer coefficients based on a minimum mean square error (MMSE) rule. 3. The device of claim 1, wherein the at least one processor filters the input symbols in the frequency domain with the equalizer coefficients. 4. A method for equalizing spatial frequency of an oversampled received signal, comprising: obtaining an equalizer coefficient for a plurality of signal replicas obtained via a plurality of receive antennas and oversampling; using the equalizer Coefficients for filtering the input symbols for the plurality of signal replicas; combining the filtered symbols on the spatial and frequency dimensions for at least 110773-990628.doc a first set of frequency bins; combining and filtering on spatial dimensions The symbols are for a second set of at least one frequency slot, wherein the number of the first set of at least one frequency slot and the number of the second set of at least one frequency slot are based on a first equalizer selection threshold To judge. 5' The method of claim 4, wherein the filtering the input symbols comprises filtering the input symbols in the frequency domain with the equalizer coefficients. 6. Apparatus for equalizing spatial frequency of an oversampled received signal, comprising: means for obtaining equalizer coefficients for a plurality of signal replicas obtained via a plurality of receive antennas and oversampling; Using the equalizer coefficients to filter the components of the input symbols for the plurality of signal replicas; and combining the filtered symbols on the two and frequency dimensions for at least a first set of frequencies and in a spatial dimension Combining the filtered symbols for a component of the second set of at least one frequency bin, the number of the first set of eight to one frequency slots and the number of the first set of at least one frequency 7. The device of claim 6, wherein the means for transitioning the input symbols comprises means for inputting a symbol with the material of the material. 8. A device for equalizing the spatial frequency of an oversampled received signal, comprising: 110773-990628.doc 1336193 </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> R) receiving Line and multiple (C) oversampling to obtain input symbols for multiple (M) signal replicas, where Μ is equal to R multiplied by C ' to obtain equalizer coefficients for the μ signal replicas, etc. The quantizer coefficients filter the input symbols for the two signal replicas and combine them in spatial and frequency dimensions. The filtered symbols are used for the first set of one of the frequency bins and are combined in the spatial dimension to be filtered. The symbols are used for at least one second set of frequency slots; and | a memory coupled to the at least one processor, wherein the number of the first set of frequency bins and the at least one frequency slot The number of the second set is determined based on the equalizer selection threshold. 9. The apparatus of claim 8, wherein the at least one processor receives input samples at a C-fold wafer rate for each receive antenna and for each A receiving antenna converts the input samples into a frequency domain to obtain input symbols for the C signal replicas from the receiving antenna. φ 1 〇 · The device of claim 6 wherein the at least one processor is at least - Each of the data streams The μ group equalizer coefficients of the copies of the signals. U. The device of claim 8, wherein the ruler is equal to two and c is equal to two, and wherein the processor is obtained for four groups of four signal replicas. Equalizer coefficients, each set of equalizer coefficients for a signal replica from a receive antenna. 12. The apparatus of claim 8, wherein the at least one processor combination is for a frequency bin of the 七7 replica a signal component on k, where k is one of the frequency bins η·, such as the device of claim 8, wherein the at least one processor is based on a minimum mean square error H0773-990628.doc The difference (MMSE) rule obtains the equalizer coefficient. The apparatus of claim 8, wherein the at least one processor obtains the assumption based on an uncorrelated noise of a plurality of signal replicas from each of the receiving antennas. Equalizer coefficient. 15. The apparatus of claim 8, wherein the at least one processor obtains the equalizer coefficients based on a hypothesis of uncorrelated noise of the r receiving antennas. 16. The apparatus of claim 8, wherein the at least one processor obtains the equalizer coefficients based on a hypothesis of spatial and spectral uncorrelated noise of the plurality of L-number replicas. 17. The apparatus of claim 8, wherein for each of the τ data streams to be recovered, the D Hai to ν processor obtains a benefit coefficient for the data stream for the data stream, The equalizer coefficients filter the input symbols for the duplicate copies of the signals to obtain the symbols for the copies of the plurality of signals, and combine the filtered elements for the copies of the binary signals The symbol gets the output symbol for the data stream. 18. The apparatus of claim 17, wherein the at least one processor obtains a noise and interference covariance matrix and obtains a matrix based on the noise and interference covariance matrix for each of the data streams The equalizer coefficient of the data stream. In the device of claim 17, wherein for each of the plurality of frequency bins, the device obtains a common noise and interference covariance matrix and is obtained based on the same noise and interference covariance. Wait for the mother of a data stream - the equalizer coefficient. 20. A method for equalizing the frequency of an oversampled received signal, 110773-990628.doc 1336193 - 贤· It contains: From a plurality of (R) receiving antennas and multiple (C) oversampling to obtain input symbols for a plurality of (M) signal replicas, where Μ is equal to R multiplied by c; is obtained for the Μ Equalizer coefficients for a plurality of signal replicas; filtering the input symbols for the one of the plurality of signal replicas with the equalizer coefficients; combining the filtered symbols for at least one frequency bin in spatial and frequency dimensions a first set; and combining the filtered symbols in a spatial dimension for a second set of at least one frequency bin, wherein the number of the first set of at least one frequency bin and the at least one frequency slot The number of the second set is determined based on a first equalizer selection threshold. The method of claim 20, wherein the combined filtered symbols comprise a frequency bin k for combining the copies of the signals. The signal component above, where k is one of the frequency slots. 22. The method of claim 20, wherein the obtaining the equalizer coefficients comprises the equalizer coefficients based on a hypothesis of uncorrelated noise for the M signal replicas from each of the receive antennas. 23. The method of claim 20, wherein the obtaining the equalizer coefficients comprises a space based on the copies of the one of the plurality of signals and an illusion of the optical #uncorrelated noise to obtain the equalizer coefficients. 24. The method of claim 20, wherein the obtaining the equalizer coefficient comprises obtaining the 110773-990628.doc equalizer for each of the signals to/from the feeder. a coefficient, wherein the filtering the input symbols comprises filtering, for each data stream, the input symbols for the one of the plurality of signal replicas with the equalizer coefficients to obtain a copy for the data streams for the data streams A symbol of the transition, and wherein the combination of the filtered symbols comprises combining the filtered symbols for the one of the plurality of signal replicas for each data stream to obtain an output symbol for the data stream. 25. Apparatus for equalizing spatial frequency of an oversampled received signal, comprising: for obtaining a plurality of (Μ) signal replicas from a plurality of (R) receive antennas and a plurality of (c) oversampling a member of the input symbol, where Μ is equal to R times C; means for obtaining equalizer coefficients for the copies of the respective signals; for filtering with the s-equalizer coefficients for the μ signals a component of the input symbols of the replica; and for combining the filtered symbols in a spatial and frequency dimension for a first set of at least one frequency bin and combining the filtered symbols in a spatial dimension Means for a second set of at least one frequency bin, wherein the number of the first set of at least one frequency bin and the number of the second set of at least one frequency bin are determined based on an equalizer selection threshold. 26. The apparatus of claim 25, wherein the means for combining the filtered symbols comprises means for combining signal components for frequency bins 1^ of the plurality of signal replicas, wherein k is a frequency One of the slots index. 27. The apparatus of claim 25, wherein the 110773-990628.doc • 6-28 component package 3 for obtaining the equalizer coefficient is used based on a copy of the M signals from each of the receiving antennas. Off the noise - assuming the component of the equalizer coefficient. Such as &quot;monthly item 25, wherein the means for obtaining the equalizer coefficient is 29.30. 匕3 is used for the &quot;hypothesis based on spatial and spectral uncorrelated noise of the M signal replicas The component that obtains the equalizer coefficients. The device of claim 25, wherein the means for obtaining the equalizer coefficient comprises means for obtaining, for at least - each of the data streams, a factor of 4 for the equalizer replicas , wherein the means for passing the input symbols includes the input symbols for each of the data stream (four) equalizer coefficients for the one of the plurality of signal replicas to obtain the resource stream Means for filtering the symbols of the copies of the signals, and wherein the means for combining the filtered symbols comprises for combining the data stream for the copies of the coffee signals The filtered payout is used to obtain the components for the output symbols of the data stream. An apparatus for equalizing the spatial frequency of an oversampled received signal, comprising: . a processor configured to construct at least one spatial frequency equalization 2 per-space frequency equalizer to combine signals into spatial and frequency dimensions, and configured to construct at least one spatial equalizer, per space, etc. $ combines signal components in spatial dimensions; and. a memory that is coupled to the at least one processor, wherein the at least one processor constructs the at least one spatial frequency equalizes the -first set of crying buccal slots and constructs the at least one space, etc. for frequency The second set of slots, . 110773-99 〇 628.d〇c is poor, at least 1 of which is the first _ set of the groove 咭筮-phase ▲ paper a Xing Wang V neck rate. ,. The number is determined based on the -equalizer selection threshold. 31. If the request item 30 is set to a V V state, a set of spatial frequency equalizers for the frequency bins is constructed and constructed. Two sets of space equalizers for frequency slots. Brother 32. If the request item 3〇1', the middle 5 Xuandi-group frequency slot and the second group frequency:; the frequency is determined by the frequency pulse response m. 33. The apparatus of claim 3, wherein the at least one processor obtains coefficients for the at least one spatial frequency equalizer and the at least one spatial equalizer based on a Least Mean Square Rank (SE) rule. 34. A method for equalizing spatial frequency of an oversampled received signal, comprising: combining a signal component for a first set of at least one frequency slot space and frequency dimension; and a second set of at least one frequency The slot combines the signal components in a spatial dimension, wherein to &gt;, the number of the first set of frequency bins and the number of the first set of at least one frequency bin are determined based on the equalizer selection threshold. 35. The method of claim 34, wherein combining the 5 Xuan S S components in the spatial and frequency dimensions comprises: obtaining an equalizer coefficient based on a minimum mean square error (MMSE) rule; and using the Transformer coefficient filtering input symbols for multiple signal replicas 0 110773-990628.doc 1336193 · (more) 瞥 瞥 36. A device for equalizing the spatial frequency of oversampled received signals, comprising: a member of the first set of at least one frequency slot space and a frequency dimension; and means for combining the signal components in a spatial dimension for a second set of at least one frequency slot, wherein to a frequency The number of the first set of slots and at least one frequency The number of sets of the grooves is determined based on the threshold of the first equalizer selection. 3 7. The apparatus of claim 36 wherein the means for combining the signal components in the spatial and frequency dimensions comprises: TT*1 V_* ||; the law of minimum mean square error (MMSE) Means for obtaining an equalizer coefficient; and means for filtering the input symbols for the plurality of signal replicas with the equalizer coefficients. H0773-990628.doc