CN1798115B

CN1798115B - Method for compensating channel distortion in communication system and iterative decision feedback equalizer thereof

Info

Publication number: CN1798115B
Application number: CN 200410065850
Authority: CN
Inventors: 徐景; 程时昕; 周志刚; 陈明
Original assignee: Individual
Current assignee: Individual
Priority date: 2004-12-22
Filing date: 2004-12-22
Publication date: 2010-04-21
Anticipated expiration: 2024-12-22
Also published as: CN1798115A

Abstract

The invention discloses a method for compensating channel distortion in a communication system and an iterative decision feedback equalizer thereof. The compensation method mainly comprises the steps of defining the length of the feedforward filter, giving an approximate solution of the feedforward filter in a discrete Fourier transform domain and the like. The iterative decision feedback equalizer mainly comprises the following components and signal flows: the received signal is output to the feedforward filter after passing through the pulse shaping filter and the sampler, the output signal of the feedforward filter is superposed with the output of the feedback filter, and then is input to the decision device after passing through the descrambler and the despreader, the hard decision of the decision device is input to the spread spectrum and the scrambler to obtain the feedback signal corresponding to each transmitting antenna and each code channel, the feedback signals of all the transmitting antennas are input to each feedback filter, the iteration is stopped after the iteration reaches the preset number, the decision device outputs the log-likelihood ratio information to the channel decoder, and the channel decoder obtains the information bit after decoding. The design method and the equalizer thereof have no error propagation phenomenon, and the realization complexity is approximate to a linear function of the channel memory length, and the method can be used for a communication system.

Description

Method for Compensating Channel Distortion in Communication System and Iterative Decision Feedback Equalizer

一技术领域a technical field

本发明涉及一种补偿通信系统中的信道失真和消除共信道干扰的方法和设备，特别是一种低复杂度的补偿通信系统信道失真的方法和迭代判决反馈均衡器。The invention relates to a method and equipment for compensating channel distortion in a communication system and eliminating co-channel interference, in particular to a low-complexity method for compensating channel distortion in a communication system and an iterative decision feedback equalizer.

二背景技术Two background technology

在通信系统中，信道的失真将导致符号间干扰，接收机必须采取一定的技术克服多径衰落。在第二代无线通信系统GSM中，采用维特比均衡器克服多径衰落，该技术的主要缺点是实现复杂度为信道记忆长度的指数函数，因此不适合宽带通信系统。在第三代无线通信系统WCDMA中，用时域线性均衡器[见GLOBECOM 1999年，Vol.1a，第467-471页，K.Hooli，M.Latva-aho，M.Juntti的“Multiple access interference suppressionwith linear chip equalizers in WCDMA downlink receivers”]补偿信道失真，从而恢复码道间的正交性，该技术的主要缺点是其实现复杂度为信道记忆长度的三次函数，且其性能受残余符号间干扰的影响。现有技术的缺陷和不足：常用的补偿信道失真的技术有维特比均衡器、时域线性均衡器，判决反馈均衡器。维特比均衡器、时域线性均衡器、判决反馈均衡器的主要缺陷为实现复杂度高，不适合未来的宽带通信，且判决反馈均衡存在错误传播现象。A.Burg[见VTC秋季，2003年10月，第468-472页，A.Burg，M.Rupp，S.Haene，D.Perels，N.Felber，W.Fichtner的“Low complexityfrequency-domain equalization of MIMO channels with applicatiohs to MIMO-CDMAsystems”]，[见Signals，Systems & Computers，The Thrity-Seventh AsilomarConference on，2003年11月，Vol.2，第1266-1272页，A.Burg，M.Rupp，；N.Felber，W.Fichtner的“Practical low complexity linear equalization for MIMO-CDMA systems”]提出的均衡器存在性能平台，在实际系统中无法应用。In a communication system, channel distortion will lead to intersymbol interference, and the receiver must adopt certain techniques to overcome multipath fading. In the second-generation wireless communication system GSM, the Viterbi equalizer is used to overcome multipath fading. The main disadvantage of this technology is that the implementation complexity is an exponential function of the channel memory length, so it is not suitable for broadband communication systems. In the third-generation wireless communication system WCDMA, a time-domain linear equalizer is used [see GLOBECOM 1999, Vol.1a, pages 467-471, K.Hooli, M.Latva-aho, M.Juntti "Multiple access interference suppression with linear chip equalizers in WCDMA downlink receivers”] to compensate for channel distortion, thereby restoring the orthogonality between code channels. The main disadvantage of this technique is that its implementation complexity is a cubic function of the channel memory length, and its performance is affected by residual intersymbol interference. Influence. Defects and deficiencies of the prior art: Commonly used techniques for compensating channel distortion include Viterbi equalizers, time-domain linear equalizers, and decision feedback equalizers. The main disadvantages of Viterbi equalizer, time-domain linear equalizer and decision feedback equalizer are high implementation complexity, not suitable for future broadband communication, and error propagation phenomenon exists in decision feedback equalization. A. Burg [See VTC Fall, October 2003, pp. 468-472, "Low complexityfrequency-domain equalization of MIMO channels with applicatiohs to MIMO-CDMAsystems”], [see Signals, Systems & Computers, The Thrity-Seventh Asilomar Conference on, November 2003, Vol.2, pp. 1266-1272, A. Burg, M. Rupp,; The equalizer proposed by N.Felber, W.Fichtner's "Practical low complexity linear equalization for MIMO-CDMA systems"] has a performance platform and cannot be applied in actual systems.

三发明内容Three invention content

本发明所要解决的技术问题是通信系统中信道失真和共信道干扰。它克服了现有技术的主要缺陷：复杂度至少为信道记忆长度的三次函数；存在错误传播现象；或存在性能平台。The technical problem to be solved by the invention is channel distortion and co-channel interference in the communication system. It overcomes the main drawbacks of the prior art: the complexity is at least a cubic function of the channel memory length; there is an error propagation phenomenon; or there is a performance platform.

本发明的技术方案：一种通信系统信道失真补偿方法，是对迭代判决反馈均衡器的第l次迭代滤波器的设计方法，它包括以下步骤：Technical scheme of the present invention: a kind of communication system channel distortion compensation method is to the design method of the 1st iterative filter of iterative decision feedback equalizer, and it comprises the following steps:

(1)定义前馈滤波器长度L_e；(1) Define the feedforward filter length L _e ;

(2)给出无限长迭代判决反馈均衡器的设置，即无限长迭代判决反馈均衡器的第i根发射天线前馈滤波器频域表达式为 (2) The setting of the infinite length iterative decision feedback equalizer is given, that is, the frequency domain expression of the i-th transmitting antenna feedforward filter of the infinite length iterative decision feedback equalizer is

(3)给出无限长迭代判决反馈均衡器前馈滤波器在离散傅立叶变换域的近似解，即无限长前馈滤波器的近似解在离散傅立叶变换域的表达式为其中

δ[n]为冲击函数；(3) The approximate solution of the infinite-length iterative decision feedback equalizer feed-forward filter in the discrete Fourier transform domain is given, that is, the approximate solution of the infinite-length feed-forward filter in the discrete Fourier transform domain is expressed as in

δ[n] is the impact function;

(4)对应步骤3设计的前馈滤波器，基于输出信干噪比最大的准则，给出最优反馈滤波器设置，即对应于第u个码道、第i根发射天线符号的最优反馈滤波器为(4) Corresponding to the feedforward filter designed in step 3, based on the criterion of the maximum output signal-to-interference-noise ratio, the optimal feedback filter setting is given, that is, the optimal feedback filter setting corresponding to the u-th code channel and the i-th transmit antenna symbol The feedback filter is

${b b}_{Sub Sub,, i i,, t t,, u u}^{l l} = = {ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t}$

$= = {ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} {Σ Σ}_{n no = = 11}^{{N N}_{r r}} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i,, n no}^{l l}))}^{* *} {((H h))}_{((n no - - 11)) ((22 J J + + L L)) + + 11 : : n no ((22 J J + + L L)),, t t}$

$= = {ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} {Σ Σ}_{n no = = 11}^{{N N}_{r r}} {Σ Σ}_{p p = = 11}^{22 J J + + L L} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i,, n no}^{l l}))}_{p p}^{* *} {h h}_{n no,, floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11} [[L L - - 11 - - t t + + p p]]$

t≠(i-1)(2D+1)+D+1t≠(i-1)(2D+1)+D+1

其中in

${\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l} = = [\begin{matrix} {\overset{&OverBar; &OverBar;}{w w}}_{Sub Sub,, 11,, i i}^{l l} \\ . . \\ . . \\ . . \\ {\overset{&OverBar; &OverBar;}{w w}}_{Sub Sub {,, N N}_{r r},, i i}^{l l} \end{matrix}],,$

是

的离散傅立叶逆变换；

yes

Inverse discrete Fourier transform of ;

(5)根据迭代判决反馈均衡器的第l次迭代输出信干噪比，计算归一化相关系数，即第l次、第i根发射天线迭代输出信干噪比为(5) Calculate the normalized correlation coefficient according to the output SINR of the iterative decision feedback equalizer in the l-th iteration, that is, the output SINR of the l-th and i-th transmit antenna iterations is

${SINR SINR}_{Sub Sub,, i i}^{l l} = = {| | {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{((i i - - 11)) ((22 D D. + + 11)) + + D D. + + 11} | |}^{22} / / (({Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) ((22 D D. + + 11)) + + D D. + + 11}^{22 D D. + + 11} ((\frac{U u}{N N} - -$

$\frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {(({ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11}))}^{22})) {| | {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t} | |}^{22} + + {σ σ}_{v v}^{22} {| | | | {\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l} | | | |}^{22}))$

其次利用第l次、第i根发射天线迭代输出信干噪比计算归一化相关系数，对应任意调制方式的归一化相关系数可根据如下公式进行计算Secondly, the normalized correlation coefficient is calculated by using the iterative output signal-to-interference-noise ratio of the lth and i-th transmitting antennas, and the normalized correlation coefficient corresponding to any modulation method can be calculated according to the following formula

${ρ ρ}_{{m m}_{t t},, u u}^{l l} = = E E. < < {d d}_{{m m}_{t t},, u u} [[n no]] {(({\overset{^^}{d d}}_{{m m}_{t t},, u u}^{l l} [[n no]]))}^{* *} > >$

$= = \underset{c c,, e e &Element; &Element; Π Π}{Σ Σ} P P {{{d d}_{{m m}_{t t},, u u} [[n no]] = = c c,, {\overset{^^}{d d}}_{{m m}_{t t},, u u}^{l l} [[n no]] = = e e}} {ce ce}^{* *}$

$= = \underset{c c,, e e &Element; &Element; Π Π}{Σ Σ} P P {{{\overset{^^}{d d}}_{{m m}_{t t},, u u}^{l l} [[n no]] = = e e | | {d d}_{{m m}_{t t},, u u} [[n no]] = = c c}} P P {{{d d}_{{m m}_{t t},, u u} [[n no]] = = c c}} {ce ce}^{* *};;$

其中in

(.)^T，(.)^*分别表示转置和共轭转置；(.) ^T , (.) ^* represent transpose and conjugate transpose respectively;

()指W_Sub，i；() refers to W _{Sub, i} ;

(H)_t表示矩阵H的第t列；(H) _t represents the tth column of matrix H;

${R R}^{l l - - 11} [[ω ω]] = = {Σ Σ}_{{m m}_{t t} = = 11}^{{M m}_{t t}} ((\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {| | {ρ ρ}_{{m m}_{t t},, u u}^{l l - - 11} | |}^{22})) {H h}_{{m m}_{t t}} [[ω ω]] {H h}_{{m m}_{t t}}^{* *} [[ω ω]] + + {σ σ}_{v v}^{22} {I I}_{{N N}_{r r}};;$

H为信道状态信息，

H is the channel state information,

${H h}_{{n no}_{r r},, {m m}_{t t}} = = [\begin{matrix} {h h}_{{n no}_{r r},, {m m}_{t t}} [[L L - - 11]] & . . . . . . & {h h}_{{n no}_{r r},, {m m}_{t t}} [[00]] \\ {h h}_{{n no}_{r r},, {m m}_{t t}} [[L L - - 11]] & . . . . . . & {h h}_{{n no}_{r r},, {m m}_{t t}} [[00]] \\ {h h}_{{n no}_{r r},, {m m}_{t t}} [[L L - - 11]] & . . . . . . & {h h}_{{n no}_{r r},, {m m}_{t t}} [[00]] \end{matrix}]$

${\overset{^^}{Q Q}}^{l l} [[f f]] = = {[\begin{matrix} {\overset{^^}{H h}}^{l l} [[22 f f]] & α α {\overset{^^}{H h}}^{l l} [[22 f f + + 11]] \end{matrix}]}^{T T};;$

σ_v ²为方差；σ _v ² is the variance;

是N_r维的单位矩阵；

is the identity matrix of N _r dimensions;

floor(x)表示取其整数部分；floor(x) means to take its integer part;

N_r为接收天线个数；N _r is the number of receiving antennas;

(H)_{(n-1)(2J+L)+1：n(2J+L)，t}表示矩阵H的第t列、从第(n-1)(2J+L)+1行到n(2J+L)行的元素组成的列；(H) _{(n-1)(2J+L)+1: n(2J+L), t} represents the tth column of matrix H, from row (n-1)(2J+L)+1 to n( 2J+L) column consisting of elements of row;

L为信道记忆长度；L is the channel memory length;

J表示观察窗长度；J represents the length of the observation window;

D＝L+J-1为判决延时；D=L+J-1 is the decision delay;

N为扩频因子；N is the spreading factor;

为第m_t根发射天线与第n_r根接收天线的信道冲激响应；

is the channel impulse response of the m _t transmitting antenna and the n _r receiving antenna;

U为激活的码道数；U is the number of activated code channels;

为第m_t根发射天线、第u个码道c_u[n]承载的符号；

is the symbol carried by the m _t transmit antenna and the u code channel c _u [n];

m_t为第m_t根发射天线；m _t is the m _tth transmitting antenna;

u为码道个数；u is the number of code channels;

II表示调制符号集，c和e表示星座图上的点；II represents the modulation symbol set, and c and e represent points on the constellation diagram;

为第m_t根发射天线、第u个码道、第(l-1)级、第k个时刻的硬判决； is the hard decision of the m _t root transmitting antenna, the u code channel, the (l-1) level, and the k th moment;

M_t为发射天线根数；M _t is the number of transmitting antennas;

${H h}_{{m m}_{t t}} [[ω ω]] = = {[[{H h}_{11,, {m m}_{t t}} [[ω ω]],, . . . . . .,, {H h}_{{N N}_{r r},, {m m}_{t t}} [[ω ω]]]]}^{T T};;$

${H h}_{{n no}_{r r},, {m m}_{t t}} [[ω ω]] = = {Σ Σ}_{l l = = 00}^{L L - - 11} {h h}_{{n no}_{r r},, {m m}_{t t}} [[l l]] {e e}^{- - jωl jωl};;$

${\overset{^^}{H h}}^{l l} [[f f]] = = H h [[f f]] {Σ Σ}^{l l - - 11};;$

(H[f])_n，m＝H_n，m[f]；(H[f]) _{n, m} = H _{n, m} [f];

${Σ Σ}^{l l - - 11} = = diag diag (([[\sqrt{\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {| | {ρ ρ}_{11,, u u}^{l l - - 11} | |}^{22}},, . . . . . .,, \sqrt{\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {| | {ρ ρ}_{{M m}_{t t},, u u}^{l l - - 11} | |}^{22}}]]));;$

为的2L_e长度离散傅立叶变换； for The 2L _e- length discrete Fourier transform;

α＝U[1]/U[0]；α=U[1]/U[0];

$U u [[f f]] = = {Σ Σ}_{n no = = 00}^{22 {L L}_{e e} - - 11} u u [[t t]] {e e}^{\frac{- - j j 22 πnf πnf}{22 {L L}_{e e}}};;$

$u u [[t t]] = = \{\begin{matrix} 11,, & 00 \leq \leq t t < < {L L}_{e e} \\ 00,, & {L L}_{e e} \leq \leq t t < < 22 {L L}_{e e} \end{matrix} . .$

本发明的通信系统信道补偿方法的迭代判决反馈均衡器，它包括：脉冲成形滤波器和采样器1，前馈滤波器2、解扰器3、解扩器4、判决器5、扩频和加扰器6、反馈滤波器7、信道译码器8，其信号流程：接收信号经脉冲成形滤波器和采样器1后，输出给前馈滤波器2-i，前馈滤波器2-i的输出信号与反馈滤波器7-i的输出相迭加，去除干扰成份，去除干扰成份的信号经解扰器3-i、解扩器4-i-u后，输入判决器5-i-u，判决器5-i-u的硬判决经扩频和加扰器6-i-u后，得到对应第i根发射天线、第u个码道的反馈信号，所有发射天线的反馈信号都输入每一个反馈滤波器7，迭代次数达到预置次数后，停止迭代，判决器5输出对数似然比信息给信道译码器8，信道译码器8经译码得到信息比特。The iterative decision feedback equalizer of communication system channel compensation method of the present invention, it comprises: pulse shaping filter and sampler 1, feedforward filter 2, descrambler 3, despreader 4, decider 5, spread spectrum and Scrambler 6, feedback filter 7, channel decoder 8, its signal process: after the received signal passes through the pulse shaping filter and sampler 1, it is output to the feedforward filter 2-i, and the feedforward filter 2-i The output signal of the output signal and the output of the feedback filter 7-i are superimposed, and the interference component is removed, and after the signal of the interference component is passed through the descrambler 3-i and the despreader 4-i-u, it is input into the decision unit 5-i-u, and the decision unit After the hard decision of 5-i-u is spread spectrum and scrambler 6-i-u, the feedback signal corresponding to the i-th transmitting antenna and the u-th code channel is obtained, and the feedback signals of all transmitting antennas are input into each feedback filter 7, After the number of iterations reaches the preset number of times, the iteration is stopped, and the decision unit 5 outputs log-likelihood ratio information to the channel decoder 8, and the channel decoder 8 decodes to obtain information bits.

本发明的有益效果：不存在错误传播现象；由于利用快速离散傅立叶变换算法更新滤波器设置，其实现复杂度近似为信道记忆长度的线性函数。表1给出了均衡器的实现复杂度比较。其中L为信道记忆长度，J表示观察窗长度，D＝L+J-1为判决延时，M_t和N_r分别为发射天线个数和接收天线个数。从表1可以看出，时域线性均衡器的实现复杂度是信道记忆长度的三次函数，当信道记忆长度比较大时，时域线性均衡器是不可实现的，而本发明提出的迭代判决反馈均衡器的实现复杂度近似为信道记忆长度的线性函数。而现有的时域判决反馈均衡器的实现复杂度比时域线性均衡器还要高，且在低信噪比范围存在错误传播现象。The beneficial effect of the present invention is that there is no error propagation phenomenon; since the fast discrete Fourier transform algorithm is used to update the filter settings, the implementation complexity is approximately a linear function of the channel memory length. Table 1 shows the implementation complexity comparison of the equalizer. Where L is the channel memory length, J is the observation window length, D=L+J-1 is the decision delay, M _t and N _r are the number of transmitting antennas and receiving antennas respectively. As can be seen from Table 1, the implementation complexity of the time-domain linear equalizer is a cubic function of the channel memory length. The implementation complexity of the equalizer is approximately a linear function of the channel memory length. However, the implementation complexity of the existing time-domain decision feedback equalizer is higher than that of the time-domain linear equalizer, and there is an error propagation phenomenon in the range of low signal-to-noise ratio.

表1均衡器实现复杂度Table 1 Equalizer Implementation Complexity

复数乘法次数Number of complex multiplications 时域线性均衡器Temporal Linear Equalizer Nr2(2J+L)2Mt(2D+1)+Nr3(2J+L)3/3Nr2(2J+L)2Mt(2D+1)+Nr 3(2J+L)3/3 迭代判决反馈均衡器Iterative Decision Feedback Equalizer MtNr(2J+L)(log2(4J+2L)-1)+Nr2Mt(2J+L)(Ls+1)+Nr3(Ls+1)(2J+L)/3+2Mt2Nr(2J+L)(Ls+1)+(2Ls+1)NrMt(2J+2L)/2(log2(2J+2L)-1)MtNr(2J+L)(log2(4J+2L)-1)+Nr2Mt(2J+L)(Ls+1)+Nr3(Ls+1)(2J+L)/3+2Mt2Nr(2J+L)(L s+1)+(2Ls+1)NrMt(2J+2L)/2 (log2(2J+2L)-1)

本发明应用范围为克服通信系统中的多径衰落和消除共信道干扰。The scope of application of the invention is to overcome multipath fading and eliminate co-channel interference in a communication system.

本发明提出的迭代判决反馈均衡器的实现复杂度近似为信道记忆长度的线性函数，且不存在错误传播现象和平台效应，能使系统频谱利用率显著提高。The implementation complexity of the iterative decision feedback equalizer proposed by the invention is approximately a linear function of the channel memory length, and there is no error propagation phenomenon and platform effect, which can significantly improve the system spectrum utilization rate.

四附图说明Four drawings

下面结合附图和具体实施方式对本发明作进一步详细说明。The present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments.

图1是本发明的迭代判决反馈均衡器的框图。FIG. 1 is a block diagram of the iterative decision feedback equalizer of the present invention.

图2是反馈滤波器的框图。Figure 2 is a block diagram of a feedback filter.

图3是80MHz传输带宽、载频2.5GHz、最大时延扩展2.4us信道时延功率分布即三角分布。Figure 3 shows the channel delay power distribution of 80MHz transmission bandwidth, carrier frequency of 2.5GHz, and maximum delay extension of 2.4us, that is, triangular distribution.

图4表示在单发单收10MHz传输带宽的CDMA系统中，线性均衡器和迭代判决反馈均衡器的性能比较。实线表示迭代判决反馈均衡器性能曲线，点线表示线性均衡器性能曲线。Figure 4 shows the performance comparison between the linear equalizer and the iterative decision feedback equalizer in a CDMA system with a single-send and single-receive 10MHz transmission bandwidth. The solid line represents the performance curve of the iterative decision feedback equalizer, and the dotted line represents the performance curve of the linear equalizer.

图5表示在单发单收10MHz传输带宽的CDMA系统中迭代判决反馈均衡器的性能。实线表示迭代判决反馈均衡器前馈滤波器长度为66时的性能曲线，点线表示迭代判决反馈均衡器前馈滤波器长度为82时的性能曲线，点划线表示迭代判决反馈均衡器前馈滤波器长度可变时的性能曲线。Fig. 5 shows the performance of the iterative decision feedback equalizer in a CDMA system with a single-send and a single-receive 10MHz transmission bandwidth. The solid line represents the performance curve of the iterative decision feedback equalizer when the length of the feedforward filter is 66, the dotted line represents the performance curve of the iterative decision feedback equalizer when the length of the feedforward filter is 82, and the dotted line represents the performance curve of the iterative decision feedback equalizer. Performance curves for variable feed filter lengths.

图6表示在(2Tx，2Rx)10MHz传输带宽的CDMA系统中，线性均衡器和迭代判决反馈均衡器的性能比较，实线表示迭代判决反馈均衡器的性能曲线，点线表示时域线性均衡器性能曲线。Figure 6 shows the performance comparison between the linear equalizer and the iterative decision feedback equalizer in a CDMA system with (2Tx, 2Rx) 10MHz transmission bandwidth, the solid line represents the performance curve of the iterative decision feedback equalizer, and the dotted line represents the time-domain linear equalizer performance curve.

图7表示单发单收80MHz传输带宽的CDMA系统中迭代判决反馈均衡器的性能，前馈滤波器长度为576。Figure 7 shows the performance of the iterative decision feedback equalizer in a CDMA system with a single-send and single-receive 80MHz transmission bandwidth, and the length of the feedforward filter is 576.

图8表示(2Tx，2Rx)80MHz传输带宽的CDMA系统中迭代判决反馈均衡器的性能。前馈滤波器长度为576。Figure 8 shows the performance of an iterative decision feedback equalizer in a CDMA system with (2Tx, 2Rx) 80MHz transmission bandwidth. The feedforward filter length is 576.

五具体实施方式Five specific implementation methods

一种补偿通信系统信道失真方法，是对迭代判决反馈均衡器的第l次迭代滤波器的设计方法，它包括以下步骤：A kind of compensation communication system channel distortion method is to the design method of the lth iteration filter of iterative decision feedback equalizer, and it comprises the following steps:

(2)给出无限长迭代判决反馈均衡器的设置；(2) Provide the settings of the infinitely long iterative decision feedback equalizer;

(3)给出无限长迭代判决反馈均衡器前馈滤波器在离散傅立叶变换域的近似解；(3) Give the approximate solution of the feedforward filter of the infinite-length iterative decision feedback equalizer in the discrete Fourier transform domain;

(4)对应步骤3设计的前馈滤波器，基于输出信干噪比最大的准则，给出最优反馈滤波器设置；(4) Corresponding to the feed-forward filter designed in step 3, based on the maximum criterion of the output signal-to-interference-noise ratio, the optimal feedback filter setting is given;

(5)根据迭代判决反馈均衡器的第l次迭代输出信干噪比，计算归一化相关系数；补偿通信系统信道失真方法是：(5) Calculate the normalized correlation coefficient according to the 1st iteration output SINR of the iterative decision feedback equalizer; the compensation communication system channel distortion method is:

无限长迭代判决反馈均衡器的第i根发射天线前馈滤波器频域表达式为The frequency domain expression of the feedforward filter of the i-th transmit antenna of the infinite-length iterative decision feedback equalizer is

${W W}_{i i}^{l l} [[ω ω]] = = {(({(({R R}^{l l - - 11} [[ω ω]]))}^{T T}))}^{- - 11} {(({H h}_{i i}^{* *} [[ω ω]]))}^{T T} - - - - - - ((11))$

其中(.)^T，(.)^*分别表示转置和共轭转置。无限长前馈滤波器的近似解在离散傅立叶变换域的表达式为where (.) ^T , (.) ^* denote transpose and conjugate transpose, respectively. The approximate solution of the infinite feed-forward filter is expressed in the discrete Fourier transform domain as

${W W}_{Sub Sub}^{l l} [[f f]] = = {(({(({\overset{^^}{Q Q}}^{l l} [[f f]]))}^{* *} {\overset{^^}{Q Q}}^{l l} [[f f]] + + {σ σ}_{v v}^{22} {I I}_{{N N}_{r r}}))}^{- - 11} (({(({\overset{^^}{Q Q}}^{l l} [[f f]]))}^{* *} T T [[f f]])) - - - - - - ((22))$

对应于第u个码道、第i根发射天线符号的最优反馈滤波器为The optimal feedback filter corresponding to the u-th code channel and the i-th transmit antenna symbol is

${b b}_{Sub Sub,, i i,, t t,, u u}^{l l} = = {ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t} - - - - - - ((33))$

t≠(i-1)(2D+1)+D+1t≠(i-1)(2D+1)+D+1

第l次、第i根发射天线迭代输出信干噪比为The output SINR of the lth and i-th transmitting antenna iterations is

${SINR SINR}_{Sub Sub,, i i}^{l l} = = {| | {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{((i i - - 11)) ((22 D D. + + 11)) + + D D. + + 11} | |}^{22} / / (({Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) ((22 D D. + + 11)) + + D D. + + 11}^{22 D D. + + 11} ((\frac{U u}{N N} - - - - - - - - ((44))$

${ρ ρ}_{i i,, u u}^{l l} = = E E. < < {d d}_{i i,, u u} [[n no]] {(({\overset{^^}{d d}}_{i i,, u u}^{l l} [[n no]]))}^{* *} > > - - - - - - ((55))$

$= = \underset{c c,, e e &Element; &Element; Π Π}{Σ Σ} P P {{{d d}_{i i,, u u} [[n no]] = = c c,, {\overset{^^}{d d}}_{i i,, u u}^{l l} [[n no]] = = e e}} {ce ce}^{* *}$

$= = \underset{c c,, e e &Element; &Element; Π Π}{Σ Σ} P P {{{\overset{^^}{d d}}_{i i,, u u}^{l l} [[n no]] = = e e | | {d d}_{i i,, u u} [[n no]] = = c c}} P P {{{d d}_{i i,, u u} [[n no]] = = c c}} {ce ce}^{* *}$

考虑多发多收的码分多址通信系统，接收机如图1所示，经脉冲成形和切片级采样1后，第n_r根接收天线的接收信号为Considering the CDMA communication system with multiple transmission and multiple reception, the receiver is shown in Figure 1. After pulse shaping and slice-level sampling 1, the received signal of the n _rth receiving antenna is

${r r}_{{n no}_{r r}} [[n no]] = = {Σ Σ}_{{m m}_{t t} = = 11}^{{M m}_{t t}} {Σ Σ}_{l l = = 00}^{L L - - 11} {h h}_{{m m}_{t t},, {n no}_{r r}} [[l l]] {s the s}_{{m m}_{t t}} [[n no - - l l]] + + {v v}_{{n no}_{r r}} [[n no]] - - - - - - ((66))$

其中发射信号定义为where the transmitted signal is defined as

${s the s}_{{m m}_{t t}} [[n no]] = = q q [[n no]] {Σ Σ}_{u u = = 11}^{U u} {Σ Σ}_{k k = = 00}^{+ + \infty \infty} {A A}_{{m m}_{t t},, u u} {d d}_{{m m}_{t t},, u u} [[k k]] {c c}_{u u} [[n no - - Nk Nk]] - - - - - - ((77))$

U为激活的码道数，

第m_t根发射天线、第u个码道c_u[n]承载的符号，它具有归一化能量，N为扩频因子，表示功率控制因子，q[n]为扰码序列。

为第m_t根发射天线与第n_r根接收天线的信道冲激响应，

为复加性白高斯噪声，方差为σ_v ²，L为信道记忆长度。接收信号的矩阵表示为U is the number of active code channels,

The symbol carried by the m _t transmit antenna and the u code channel c _u [n] has normalized energy, N is the spreading factor, Indicates the power control factor, and q[n] is the scrambling code sequence.

is the channel impulse response of the m _t transmitting antenna and the n _r receiving antenna,

It is the compound additive white Gaussian noise, the variance is σ _v ² , and L is the channel memory length. The matrix of the received signal is expressed as

r[n]＝Hs[n]+v[n] (8)r[n]=Hs[n]+v[n] (8)

其中in

$H h = = [\begin{matrix} {H h}_{1,1 1,1} & {H h}_{1,2 1,2} & . . . . . . & {H h}_{11,, {M m}_{t t}} \\ {H h}_{2,1 2,1} & {H h}_{2,2 2,2} & . . . . . . & {H h}_{22,, {M m}_{t t}} \\ . . & . . & . . & . . \\ . . & . . & . . & . . \\ . . & . . & . . & . . \\ {H h}_{{N N}_{r r},, 11} & {H h}_{{N N}_{r r},, 22} & . . . . . . & {H h}_{{N N}_{r r},, {M m}_{t t}} \end{matrix}] - - - - - - ((99))$

$s the s [[n no]] = = {[[{s the s}_{11} [[n no]],, . . . . . .,, {s the s}_{{M m}_{t t}} [[n no]]]]}^{T T} - - - - - - ((1010))$

$v v [[n no]] = = {[[{v v}_{11} [[n no]],, . . . . . .,, {v v}_{{N N}_{r r}} [[n no]]]]}^{T T} - - - - - - ((1111))$

$r r [[n no]] = = {[[{r r}_{11} [[n no]],, . . . . . .,, {r r}_{{N N}_{r r}} [[n no]]]]}^{T T} - - - - - - ((1212))$

${r r}_{{n no}_{r r}} [[n no]] = = {[[{r r}_{{n no}_{r r}} [[n no - - J J]],, . . . . . .,, {r r}_{{n no}_{r r}} [[n no + + L L - - 11 + + J J]]]]}^{T T} - - - - - - ((1313))$

${s the s}_{{m m}_{t t}} [[n no]] = = {[[{s the s}_{{m m}_{t t}} [[n no - - D D.]],, . . . . . .,, {s the s}_{{m m}_{t t}} [[n no + + D D.]]]]}^{T T} - - - - - - ((1414))$

${H h}_{{n no}_{r r},, {m m}_{t t}} = = [\begin{matrix} {h h}_{{n no}_{r r},, {m m}_{t t}} [[L L - - 11]] & . . . . . . & {h h}_{{n no}_{r r},, {m m}_{t t}} [[00]] \\ {h h}_{{n no}_{r r},, {m m}_{t t}} [[L L - - 11]] & . . . . . . & {h h}_{{n no}_{r r},, {m m}_{t t}} [[00]] \\ {h h}_{{n no}_{r r},, {m m}_{t t}} [[L L - - 11]] & . . . . . . & {h h}_{{n no}_{r r},, {m m}_{t t}} [[00]] \end{matrix}] - - - - - - ((1515))$

${v v}_{{n no}_{r r}} [[n no]] = = {[[{v v}_{{n no}_{T T}} [[n no - - J J]],, . . . . . .,, {v v}_{{n no}_{r r}} [[n no + + L L - - 11 + + J J]]]]}^{T T} - - - - - - ((1616))$

J表示观察窗长度，D＝L+J-1为判决延时。传统的时域均衡器的主要缺陷是实现复杂度非常高和存在错误传播现象，最近A.Burg提出了一种低复杂度的线性均衡器，但其性能存在平台效应，不适合应用于实际通信系统。为了克服平台效应和避免错误传播现象。根据信道状态信息H，第l次迭代滤波器设计如下：J represents the length of the observation window, and D=L+J-1 is the decision delay. The main defect of the traditional time-domain equalizer is that the implementation complexity is very high and there is an error propagation phenomenon. Recently, A.Burg proposed a low-complexity linear equalizer, but its performance has a platform effect, which is not suitable for practical communication system. In order to overcome the platform effect and avoid the phenomenon of error propagation. According to the channel state information H, the l-th iterative filter design is as follows:

(1)定义前馈滤波器长度L_e，一般L_e为信道记忆长度的3至5倍。(1) Define the length L _e of the feedforward filter, and generally L _e is 3 to 5 times the length of the channel memory.

(2)给出无限长迭代判决反馈均衡器的设置，第i根发射天线前馈滤波器频域表达式为(2) Given the settings of the infinitely long iterative decision feedback equalizer, the frequency domain expression of the i-th transmit antenna feedforward filter is

${W W}_{i i}^{l l} [[ω ω]] = = {(({(({R R}^{l l - - 11} [[ω ω]]))}^{T T}))}^{- - 11} {(({H h}_{i i}^{* *} [[ω ω]]))}^{T T} - - - - - - ((1717))$

其中in

${R R}^{l l - - 11} [[ω ω]] = = {Σ Σ}_{{m m}_{t t} = = 11}^{{M m}_{t t}} ((\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {| | {ρ ρ}_{{m m}_{t t},, u u}^{l l - - 11} | |}^{22})) {H h}_{{m m}_{t t}} [[ω ω]] {H h}_{{m m}_{t t}}^{* *} [[ω ω]] + + {σ σ}_{v v}^{22} {I I}_{{N N}_{r r}} - - - - - - ((1818))$

${H h}_{{m m}_{t t}} [[ω ω]] = = {[[{H h}_{11,, {m m}_{t t}} [[ω ω]],, . . . . . .,, {H h}_{{N N}_{r r},, {m m}_{t t}} [[ω ω]]]]}^{T T} - - - - - - ((1919))$

${H h}_{{n no}_{r r},, {m m}_{t t}} [[ω ω]] = = {Σ Σ}_{l l = = 00}^{L L - - 11} {h h}_{{n no}_{r r},, {m m}_{t t}} [[l l]] {e e}^{- - jωl jωl} - - - - - - ((2020))$

为第(l-1)次、第m_t根发射天线、第u个码道的重新映射符号与真实符号的归一化相关系数，

是N_r维的单位矩阵。 is the normalized correlation coefficient between the remapped symbol and the real symbol of the (l-1)th time, the _mtth transmit antenna, and the uth code channel,

is the identity matrix of N _r dimensions.

(3)无限长前馈滤波器的近似解在离散傅立叶变换域的表达式为(3) The approximate solution of the infinite feed-forward filter is expressed in the discrete Fourier transform domain as

${W W}_{Sub Sub}^{l l} [[f f]] = = {(({(({\overset{^^}{Q Q}}^{l l} [[f f]]))}^{* *} {\overset{^^}{Q Q}}^{l l} [[f f]] + + {σ σ}_{v v}^{22} {I I}_{{N N}_{r r}}))}^{- - 11} (({(({\overset{^^}{Q Q}}^{l l} [[f f]]))}^{* *} T T [[f f]])) - - - - - - ((21 twenty one))$

其中in

${\overset{^^}{Q Q}}^{l l} [[f f]] = = {[\begin{matrix} {\overset{^^}{H h}}^{l l} [[22 f f]] & α α {\overset{^^}{H h}}^{l l} [[22 f f + + 11]] \end{matrix}]}^{T T} - - - - - - ((22 twenty two))$

$T T [[f f]] = = {[\begin{matrix} T T [[22 f f]] {I I}_{{M m}_{t t}} & T T [[22 f f + + 11]] {I I}_{{M m}_{t t}} \end{matrix}]}^{T T} - - - - - - ((23 twenty three))$

$T T [[f f]] = = {Σ Σ}_{n no = = 00}^{22 {L L}_{e e} - - 11} δ δ [[n no - - D D.]] {e e}^{\frac{- - j j 22 πnf πnf}{22 {L L}_{e e}}} - - - - - - ((24 twenty four))$

$u u [[t t]] = = \{\begin{matrix} 11,, & 00 \leq \leq t t < < {L L}_{e e} \\ 00,, & {L L}_{e e} \leq \leq t t < < 22 {L L}_{e e} \end{matrix} - - - - - - ((2525))$

$U u [[f f]] = = {Σ Σ}_{n no = = 00}^{22 {L L}_{e e} - - 11} u u [[t t]] {e e}^{\frac{- - j j 22 πnf πnf}{22 {L L}_{e e}}} - - - - - - ((2626))$

α＝U[1]/U[0] (27)α＝U[1]/U[0] (27)

${\overset{^^}{H h}}^{l l} [[f f]] = = H h [[f f]] {Σ Σ}^{l l - - 11} - - - - - - ((2828))$

(H[f])_n，m＝H_n，m[f] (29)(H[f]) _{n, m} = H _{n, m} [f] (29)

${Σ Σ}^{l l - - 11} = = diag diag (([[\sqrt{\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {| | {ρ ρ}_{11,, u u}^{l l - - 11} | |}^{22}},, . . . . . .,, \sqrt{\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {| | {ρ ρ}_{{M m}_{t t},, u u}^{l l - - 11} | |}^{22}}]])) - - - - - - ((3030))$

其中为

的2L_e长度离散傅立叶变换。对应第i根发射天线符号的前馈滤波器为in for

The 2L _e- length discrete Fourier transform. The feedforward filter corresponding to the i-th transmit antenna symbol is

${\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l} = = [\begin{matrix} {\overset{&OverBar; &OverBar;}{w w}}_{Sub Sub,, 11,, i i}^{l l} \\ . . \\ . . \\ . . \\ {\overset{&OverBar; &OverBar;}{w w}}_{Sub Sub {,, N N}_{r r},, i i}^{l l} \end{matrix}] - - - - - - ((3131))$

其中

是

的离散傅立叶逆变换。in

yes

Inverse discrete Fourier transform of .

(4)对应步骤3设计的前馈滤波器，基于输出信干噪比最大的准则，对应于第u个码道、第i根发射天线符号的最优反馈滤波器为(4) Corresponding to the feed-forward filter designed in step 3, based on the criterion of the maximum output signal-to-interference-noise ratio, the optimal feedback filter corresponding to the u-th code channel and the i-th transmit antenna symbol is

$= = {ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} {Σ Σ}_{n no = = 11}^{{N N}_{r r}} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, n no,, i i}^{l l}))}^{* *} {((H h))}_{((n no - - 11)) ((22 J J + + L L)) + + 11 : : n no ((22 J J + + L L)),, t t}$

$= = {ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} {Σ Σ}_{n no = = 11}^{{N N}_{r r}} {Σ Σ}_{p p = = 11}^{22 J J + + L L} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, n no,, i i}^{l l}))}_{p p}^{* *} {h h}_{n no,, floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11} [[L L - - 11 - - t t + + p p]] - - - - - - ((3232))$

t≠(i-1)(2D+1)+D+1t≠(i-1)(2D+1)+D+1

其中floor(x)表示取其整数部分，(H)_t表示矩阵H的第t列，(H)_{(n-1)(2J+L)+1：n(2J+L)，t}表示矩阵H的第t列、从第(n-1)(2J+L)+1行到n(2J+L)行的元素组成的列。Where floor(x) means to take its integer part, (H) _t means the tth column of matrix H, (H) _{(n-1)(2J+L)+1:n(2J+L), t} means matrix H The t-th column of , a column composed of elements from the (n-1)(2J+L)+1th row to the n(2J+L)th row.

(5)利用第l次、第i根发射天线迭代输出信干噪比计算归一化相关系数。首先计算第l次、第i根发射天线迭代输出信干噪比(5) Calculate the normalized correlation coefficient by using the iterative output signal-to-interference-noise ratio of the lth and i-th transmitting antennas. Firstly, calculate the SINR output of the lth and ith transmitting antenna iterations

$\frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {(({ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11}))}^{22})) {| | {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t} | |}^{22})) + + {σ σ}_{v v}^{22} {| | | | {\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l} | | | |}^{22} - - - - - - ((3333))$

其次利用第l次、第i根发射天线迭代输出信干噪比计算归一化相关系数，对应任意调制方式的归一化相关系数可根据如下公式进行计算：Secondly, the normalized correlation coefficient is calculated by using the iterative output signal-to-interference-noise ratio of the lth and i-th transmitting antennas, and the normalized correlation coefficient corresponding to any modulation mode can be calculated according to the following formula:

${ρ ρ}_{i i,, u u}^{l l} = = E E. < < {d d}_{i i,, u u} [[n no]] {(({\overset{^^}{d d}}_{i i,, u u}^{l l} [[n no]]))}^{* *} > >$

$= = \underset{c c,, e e &Element; &Element; Π Π}{Σ Σ} P P {{{d d}_{i i,, u u} [[n no]] = = c c,, {\overset{^^}{d d}}_{i i,, u u}^{l l} [[n no]] = = e e}} {ce ce}^{* *} - - - - - - ((3434))$

其中II表示调制符号集，c和e表示星座图上的点。本发明给出如下QPSK和16QAM调制方式的归一化相关系数计算公式，Where II represents the set of modulation symbols, and c and e represent points on the constellation diagram. The present invention provides the normalized correlation coefficient computing formula of following QPSK and 16QAM modulation mode,

其中in

$Q Q ((x x)) = = \frac{11}{\sqrt{22 π π}} {&Integral; &Integral;}_{x x}^{+ + \infty \infty} {e e}^{- - {t t}^{22} / / 22} dt dt - - - - - - ((3636))$

${Q Q}_{1616 QAM QAM,, i i,, 11} = = Q Q ((\sqrt{\frac{11}{55} {SINR SINR}_{Sub Sub,, i i}^{l l}})) - - - - - - ((3737))$

${Q Q}_{1616 QAM QAM,, i i,, 22} = = Q Q ((33 \sqrt{\frac{11}{55} {SINR SINR}_{Sub Sub,, i i}^{l l}}))$

${Q Q}_{1616 QAM QAM,, i i,, 33} = = Q Q ((55 \sqrt{\frac{11}{55} {SINR SINR}_{Sub Sub,, i i}^{l l}}))$

本领域的研究人员可以不需付出创新劳动，由公式(34)得到其他调制方式的归一化相关系数计算公式，从而设计其他调制方式的前馈和反馈滤波器。从滤波器的设计来看，前馈和反馈滤波器都可以用快速傅立叶算法实现，其滤波器设计复杂度为O(L_e/2(log2(L_e)-1))量级。另一个非常重要的特点是，没有对判决反馈作完全正确的假设，用归一化相干系数来度量硬判决的可靠性，而且前一级硬判决的可靠性，又用于设计当前级的前馈滤波器，逐级提高输出的可靠性，从而克服了平台效应和避免了错误传播现象。下面介绍第i根发射天线、第u个码道的信号检测过程。如图1，脉冲成形和切片级采样后的信号经前馈滤波2-i后，前馈滤波器2-i输出为Researchers in this field can obtain the normalized correlation coefficient calculation formulas of other modulation methods from formula (34) without any innovative work, so as to design feedforward and feedback filters of other modulation methods. From the point of view of filter design, both feedforward and feedback filters can be realized by fast Fourier algorithm, and the complexity of filter design is O(L _e /2(log2(L _e )-1)) order of magnitude. Another very important feature is that there is no completely correct assumption on the decision feedback, the reliability of the hard decision is measured by the normalized coherence coefficient, and the reliability of the hard decision of the previous stage is used to design the previous stage of the current stage. Feed filter, improve the reliability of the output step by step, thereby overcoming the platform effect and avoiding the phenomenon of error propagation. The following describes the signal detection process of the i-th transmitting antenna and the u-th code channel. As shown in Figure 1, after the signal after pulse shaping and slice-level sampling is fed-forward filtered 2-i, the output of feed-forward filter 2-i is

${\overset{~ ~}{y the y}}_{i i}^{l l} [[n no]] = = {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} r r [[n no]] - - - - - - ((3838))$

$= = {γ γ}_{i i}^{l l} {s the s}_{i i} [[n no]] + + {\overset{~ ~}{v v}}_{i i} [[n no]]$

其中in

${\overset{~ ~}{v v}}_{i i} [[n no]] = = = = {Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) * * ((22 D D. + + 11)) + + D D. + + 11}^{{M m}_{t t} ((22 D D. + + 11))} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t} {s the s}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11} [[n no - - D D. + + mod mod ((((t t - - 11)),, 22 D D. + + 11))]] - - - - - - ((3939))$

$+ + {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} v v [[n no]]$

${γ γ}_{i i}^{l l} = = {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{((i i - - 11)) ((22 D D. + + 11)) + + D D. + + 11} - - - - - - ((4040))$

s_i[n]为第i根天线发射的信号。对应于第m_t根发射天线、第u个码道的判决反馈信号可由判决器5-m_t-u重构为s _i [n] is the signal transmitted by the i-th antenna. The decision feedback signal corresponding to the m _t transmitting antenna and the u th code channel can be reconstructed by the decision unit 5-m _t -u as

${\overset{^^}{s the s}}_{{m m}_{t t},, u u}^{l l - - 11} [[n no]] = = q q [[n no]] {Σ Σ}_{k k = = 00}^{+ + \infty \infty} A A {\overset{^^}{d d}}_{{m m}_{t t},, u u}^{l l - - 11} [[k k]] {c c}_{u u} [[n no - - Nk Nk]] - - - - - - ((4141))$

其中为第m_t根发射天线、第u个码道、第(l-1)级、第k个时刻的硬判决。如图2，对应第m_t根发射天线的反馈信号输入反馈滤波器7-i-m_t1或反馈滤波器7-i-m_t2，并根据调制方式选择反馈滤波器。反馈滤波器7-i的输出与前馈滤波器2-i输出相迭加可得，in is the hard decision of the m _t transmit antenna, the u code channel, the (l-1) level, and the k time. As shown in Figure 2, the feedback signal corresponding to the m _t transmitting antenna is input to the feedback filter 7-im _t 1 or the feedback filter 7-im _t 2, and the feedback filter is selected according to the modulation mode. The output of the feedback filter 7-i is superimposed with the output of the feedforward filter 2-i,

${\overset{&OverBar; &OverBar;}{y the y}}_{i i}^{l l} [[n no]] = = {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} r r [[n no]] - - - - - - ((4242))$

$- - {Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) * * ((22 D D. + + 11)) + + D D. + + 11}^{{M m}_{t t} ((22 D D. + + 11))} {Σ Σ}_{u u = = 11}^{U u} {b b}_{i i,, t t,, u u}^{l l} {\overset{^^}{s the s}}_{floor floor ((((k k - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} [[n no - - D D. + + mod mod ((((t t - - 11)),, 22 D D. + + 11))]]$

$= = {γ γ}_{i i}^{l l} {s the s}_{i i} [[n no]] + + {\overset{&OverBar; &OverBar;}{v v}}_{i i} [[n no]]$

其中in

${\overset{&OverBar; &OverBar;}{v v}}_{i i}^{l l} [[n no]] = = {Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) * * ((22 D D. + + 11)) + + D D. + + 11}^{{M m}_{t t} ((22 D D. + + 11))} {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t} {s the s}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11} [[n no - - D D. + + mod mod ((((t t - - 11)),, 22 D D. + + 11))]] - - - - - - ((4343))$

$- - {Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) * * ((22 D D. + + 11)) + + D D. + + 11}^{{M m}_{t t} ((22 D D. + + 11))} {Σ Σ}_{u u = = 11}^{U u} {b b}_{i i,, t t,, u u}^{l l} {\overset{^^}{s the s}}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11} [[n no - - D D. + + mod mod ((((t t - - 11)),, 22 D D. + + 11))]]$

残余干扰加噪声v_i ^l[n]的功率为The power of residual interference plus noise v _i ^l [n] is

${P P}_{{\overset{&OverBar; &OverBar;}{v v}}_{i i}^{l l}} = = {Σ Σ}_{t t = = 11,, t t &NotEqual; &NotEqual; ((i i - - 11)) ((22 D D. + + 11)) + + D D. + + 11}^{22 D D. + + 11} ((\frac{U u}{N N} - - \frac{11}{N N} {Σ Σ}_{u u = = 11}^{U u} {(({ρ ρ}_{floor floor ((((t t - - 11)) / / ((22 D D. + + 11)))) + + 11,, u u}^{l l - - 11}))}^{22})) {| | {(({\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l}))}^{* *} {((H h))}_{t t} | |}^{22} + + {σ σ}_{v v}^{22} {| | | | {\overset{&OverBar; &OverBar;}{W W}}_{Sub Sub,, i i}^{l l} | | | |}^{22})) - - - - - - ((4444))$

在上述推导过程中，用到以下假设In the above derivation process, the following assumptions are used

$E E. < < {d d}_{{i i}_{11},, {u u}_{11}} [[{n no}_{11}]] {(({\overset{^^}{d d}}_{{i i}_{22},, {u u}_{22}}^{l l - - 11} [[{n no}_{22}]]))}^{* *} > > = = {ρ ρ}_{{i i}_{11},, {u u}_{11}}^{l l - - 11} δ δ [[{n no}_{11} - - {n no}_{22}]] δ δ [[{u u}_{11} - - {u u}_{22}]] δ δ [[{i i}_{11} - - {i i}_{22}]] - - - - - - ((4545))$

$E E. < < {\overset{^^}{d d}}_{{i i}_{11},, {u u}_{11}} [[{n no}_{11}]] {(({\overset{^^}{d d}}_{{i i}_{22},, {u u}_{22}}^{l l - - 11} [[{n no}_{22}]]))}^{* *} > > = = δ δ [[{n no}_{11} - - {n no}_{22}]] δ δ [[{u u}_{11} - - {u u}_{22}]] δ δ [[{i i}_{11} - - {i i}_{22}]] - - - - - - ((4646))$

$E E. < < {\overset{^^}{d d}}_{{i i}_{11},, {u u}_{11}}^{l l - - 11} [[{n no}_{11}]] {(({v v}_{{n no}_{r r}} [[{n no}_{22}]]))}^{* *} > > = = 00 - - - - - - ((4747))$

经过解扰3和解扩4后，对应于第i根发射天线、第u个码道的符号判决变量为After descrambling 3 and despreading 4, the symbol decision variable corresponding to the i-th transmit antenna and the u-th code channel is

${\overset{&OverBar; &OverBar;}{d d}}^{l l} [[k k]] = = \frac{11}{\sqrt{N N}} {Σ Σ}_{n no = = Nk Nk}^{N N ((k k + + 11)) - - 11} {c c}_{u u} [[n no - - Nk Nk]] {((q q [[n no]]))}^{* *} {\overset{&OverBar; &OverBar;}{y the y}}_{i i}^{l l} [[n no]] - - - - - - ((4848))$

$= = {γ γ}_{i i}^{l l} {d d}_{i i,, u u} [[k k]] + + \frac{11}{\sqrt{N N}} {Σ Σ}_{n no = = Nk Nk}^{N N ((k k + + 11)) - - 11} {c c}_{u u} [[n no - - Nk Nk]] {((q q [[kn k n]]))}^{* *} {\overset{&OverBar; &OverBar;}{v v}}_{i i}^{l l} [[n no]]$

根据公式(48)可以得到第i根发射天线、第u个码道的硬判决和对数似然比信息。特别需要注意上述滤波过程都可以用快速傅立叶变换算法实现，因此滤波过程的实现复杂度为O(L_e/2(log2(L_e)-1))。初始级可采用A.Burg提出的均衡器的输出。本发明只给出基于并行的迭代判决反馈均衡器方法和设置，本领域的研究人员可以不需付出创新劳动，给出基于串行和分组的迭代判决反馈均衡器方法和设置。表2给出了CDMA系统参数，表3和图3分别给出了10MHz和80MHz传输带宽信道时延功率分布。According to formula (48), the hard decision and log likelihood ratio information of the i-th transmitting antenna and the u-th code channel can be obtained. In particular, it should be noted that the above filtering process can be realized by fast Fourier transform algorithm, so the implementation complexity of the filtering process is O(L _e /2(log2(L _e )-1)). The initial stage can use the output of the equalizer proposed by A. Burg. The present invention only provides the method and setting of the iterative decision feedback equalizer based on parallelism, and researchers in the field can provide the method and setting of the iterative decision feedback equalizer based on serial and grouping without innovative labor. Table 2 shows the CDMA system parameters, and Table 3 and Figure 3 show the power distribution of 10MHz and 80MHz transmission bandwidth channel time delay respectively.

表2系统参数Table 2 System parameters

表3 10MHz传输带宽、载频2.5GHz信道时延功率分布Table 3 10MHz transmission bandwidth, carrier frequency 2.5GHz channel delay power distribution

本发明的补偿通信系统信道失真方法的迭代判决反馈均衡器，它包括：脉冲成形滤波器和采样器1，前馈滤波器2、解扰器3、解扩器4、判决器5、扩频和加扰器6、反馈滤波器7、信道译码器8，其信号流程：接收信号经脉冲成形滤波器和采样器1后，输出给前馈滤波器2-i，前馈滤波器2-i的输出信号与反馈滤波器7-i的输出相迭加，去除干扰成份，去除干扰成份的信号经解扰器3-i、解扩器4-i-u后，输入判决器5-i-u，判决器5-i-u的硬判决经扩频和加扰器6-i-u后，得到对应第i根发射天线、第u个码道的反馈信号，所有发射天线的反馈信号都输入每-个反馈滤波器7，迭代次数达到预置次数后，停止迭代，判决器5输出对数似然比信息给信道译码器8，信道译码器8经译码得到信息比特。The iterative decision feedback equalizer of the compensation communication system channel distortion method of the present invention, it comprises: pulse shaping filter and sampler 1, feed-forward filter 2, descrambler 3, despreader 4, decision device 5, spread spectrum And scrambler 6, feedback filter 7, channel decoder 8, its signal flow: receive signal after pulse shaping filter and sampler 1, output to feedforward filter 2-i, feedforward filter 2-i The output signal of i is superimposed with the output of the feedback filter 7-i to remove the interference component. After the signal of the removal of the interference component is passed through the descrambler 3-i and the despreader 4-i-u, it is input to the decision device 5-i-u, and the judgment After the hard decision of the device 5-i-u passes through the spread spectrum and scrambler 6-i-u, the feedback signal corresponding to the i-th transmitting antenna and the u-th code channel is obtained, and the feedback signals of all transmitting antennas are input into each feedback filter 7. After the number of iterations reaches the preset number of times, the iteration is stopped, and the decision unit 5 outputs the log-likelihood ratio information to the channel decoder 8, and the channel decoder 8 decodes to obtain information bits.

反馈滤波器7-i的信号流程如下：第m_t根发射天线的反馈信号输入反馈滤波器7-i-m_t1或反馈滤波器7-i-m_t2，且根据调制方式选择反馈滤波器7-i-m_t1或反馈滤波器7-i-m_t2，所有反馈滤波器的输出相迭加，得到反馈滤波器7-i的输出，该输出与前馈滤波器2-i的输出相迭加。The signal flow of the feedback filter 7-i is as follows: the feedback signal of the m _t transmitting antenna is input to the feedback filter 7-im _t 1 or the feedback filter 7-im _t 2, and the feedback filter 7-im t is selected according to the modulation mode _t 1 or feedback filter 7-im _t 2, all feedback filters The outputs of the feedforward filter 2-i are superimposed to obtain the output of the feedback filter 7-i.

迭代判决反馈均衡器的诸器件均为已知，本领域技术人员根据设计步骤和信号流程即可实现本发明。All components of the iterative decision feedback equalizer are known, and those skilled in the art can realize the present invention according to the design steps and signal flow.

结合图4、5、6、7、8，在以下的性能比较中，CDMA通信系统总处于满负荷状态，即所有的码道都被激活。图4给出时域线性均衡器和迭代判决反馈均衡器的性能比较。时域线性均衡器的滤波器长度和迭代判决反馈均衡器的前馈滤波器长度都为50，从方程(18)可以算出反馈滤波器长度为64。从图4可以看出在初始级(A.Burg提出的均衡器)不能正常工作，本发明提出的迭代判决反馈均衡器经过二次迭代处理能取得令人满意的性能，若目标误比特率为0.03，与时域线性均衡器相比较，迭代判决反馈均衡器能取得4.5dB性能增益。由于初始级性能较差，迭代判决反馈均衡器在误比特率0.01存在性能平台，值得庆幸的是在未加编码的通信系统中，最感兴趣的误比特率范围为0.1至0.01之间。在单发单收10MHz传输带宽的CDMA系统中，线性均衡器和迭代判决反馈均衡器的性能比较。实线表示迭代判决反馈均衡器性能曲线，点线表示线性均衡器性能曲线。With reference to Figures 4, 5, 6, 7, and 8, in the following performance comparisons, the CDMA communication system is always at full load, that is, all code channels are activated. Fig. 4 shows the performance comparison of time-domain linear equalizer and iterative decision feedback equalizer. The filter length of the time-domain linear equalizer and the feedforward filter length of the iterative decision feedback equalizer are both 50, and the feedback filter length can be calculated as 64 from equation (18). As can be seen from Figure 4, the initial stage (the equalizer proposed by A.Burg) cannot work normally, and the iterative decision feedback equalizer proposed by the present invention can obtain satisfactory performance through secondary iteration processing. If the target bit error rate is 0.03, compared with the time-domain linear equalizer, the iterative decision feedback equalizer can achieve 4.5dB performance gain. Due to the poor performance of the initial stage, the iterative decision feedback equalizer has a performance plateau at a bit error rate of 0.01. Fortunately, in uncoded communication systems, the most interesting bit error rate range is between 0.1 and 0.01. In the CDMA system with single transmit and single receive 10MHz transmission bandwidth, the performance comparison between linear equalizer and iterative decision feedback equalizer. The solid line represents the performance curve of the iterative decision feedback equalizer, and the dotted line represents the performance curve of the linear equalizer.

图5给出单发单收10MHz传输带宽的CDMA系统中，不同长度迭代判决反馈均衡器的性能。实线表示迭代判决反馈均衡器前馈滤波器长度为66时的性能曲线，点线表示迭代判决反馈均衡器前馈滤波器长度为82时的性能曲线，点划线表示迭代判决反馈均衡器前馈滤波器长度可变时的性能曲线。从图5可以看出增加初始级滤波器长度能消除性能平台和改善性能。为了降低实现复杂度，可以采用前馈滤波器长度66的迭代判决反馈均衡器，与前馈滤波器长度82的迭代判决反馈均衡器相比较，其导致的性能损失可以忽略不计。Figure 5 shows the performance of the iterative decision feedback equalizer with different lengths in a CDMA system with a single-send and single-receive 10MHz transmission bandwidth. The solid line represents the performance curve of the iterative decision feedback equalizer when the length of the feedforward filter is 66, the dotted line represents the performance curve of the iterative decision feedback equalizer when the length of the feedforward filter is 82, and the dotted line represents the performance curve of the iterative decision feedback equalizer. Performance curves for variable feed filter lengths. It can be seen from Figure 5 that increasing the initial stage filter length can eliminate the performance plateau and improve performance. In order to reduce the implementation complexity, an iterative decision feedback equalizer with a feedforward filter length of 66 can be used. Compared with an iterative decision feedback equalizer with a feedforward filter length of 82, the performance loss caused by it can be ignored.

图6给出(2Tx，2Rx)10MHz传输带宽的CDMA系统中，线性均衡器和迭代判决反馈均衡器的性能比较，实线表示迭代判决反馈均衡器的性能曲线，点线表示时域线性均衡器性能曲线。迭代判决反馈均衡器的设置为：在初始级前馈滤波器长度为164，以后各级前馈滤波器长度为100。时域线性均衡器滤波器长度为100。从性能比较可以看出，迭代判决反馈均衡器可以取得令人满意的性能增益。Figure 6 shows the performance comparison between the linear equalizer and the iterative decision feedback equalizer in a CDMA system with (2Tx, 2Rx) 10MHz transmission bandwidth, the solid line represents the performance curve of the iterative decision feedback equalizer, and the dotted line represents the time-domain linear equalizer performance curve. The settings of the iterative decision feedback equalizer are: the length of the feedforward filter in the initial stage is 164, and the length of the feedforward filter in the subsequent stages is 100. The time-domain linear equalizer has a filter length of 100. It can be seen from the performance comparison that the iterative decision feedback equalizer can achieve satisfactory performance gain.

图7和图8给出了传输带宽为80MHz时迭代判决反馈均衡器的性能。信道记忆长度为192，由于时域线性均衡器的实现复杂度为信道记忆长度的三次方，在工程上很难实现，因此末给出时域线性均衡器的性能。从图7和图8可以看出迭代判决反馈均衡器总能取得令人满意的性能。图7给出在单发单收80MHz传输带宽的CDMA系统中迭代判决反馈均衡器的性能。前馈滤波器长度为576。图8在(2Tx，2Rx)80MHz传输带宽的CDMA中迭代判决反馈均衡器的性能从图7和图8可以看出迭代判决反馈均衡器总能取得令人满意的性能。Figure 7 and Figure 8 show the performance of the iterative decision feedback equalizer when the transmission bandwidth is 80MHz. The channel memory length is 192. Since the implementation complexity of the time-domain linear equalizer is the third power of the channel memory length, it is difficult to realize in engineering, so the performance of the time-domain linear equalizer is not given. It can be seen from Figure 7 and Figure 8 that the iterative decision feedback equalizer can always achieve satisfactory performance. Figure 7 shows the performance of the iterative decision feedback equalizer in a CDMA system with a single-send and a single-receive 80MHz transmission bandwidth. The feedforward filter length is 576. Figure 8 Performance of iterative decision feedback equalizer in (2Tx, 2Rx) 80MHz transmission bandwidth CDMA It can be seen from Figure 7 and Figure 8 that the iterative decision feedback equalizer can always achieve satisfactory performance.

本发明的技术关键点和欲保护点：迭代判决反馈均衡器为无限长切片迭代判决反馈均衡器的近似解。在每一级，当有限长前馈滤波器给定后，反馈滤波器使残余干扰功率最小，由均衡器的输出信噪比来计算归一化相干系数，当前的归一化相干系数用于更新有限长前馈滤波器和反馈滤波器。滤波器的实现都可以利用快速离散傅立叶变换算法实现。The technical key point and protection point of the present invention: the iterative decision feedback equalizer is an approximate solution of the infinite length slice iterative decision feedback equalizer. At each stage, when the finite-length feedforward filter is given, the feedback filter minimizes the residual interference power, and the normalized coherence coefficient is calculated from the output signal-to-noise ratio of the equalizer, and the current normalized coherence coefficient is used for Update finite-length feedforward and feedback filters. The realization of the filter can be realized by fast discrete Fourier transform algorithm.

Claims

1. communication system channel distortion compensating method is that it may further comprise the steps to the method for designing of the l time iterative filter of feedback equalizer through iteration decision:

(1) definition feedforward filter length L _e

(2) provide the setting of endless feedback equalizer through iteration decision, promptly the i transmit antennas feedforward filter frequency-domain expression of endless feedback equalizer through iteration decision is

W_{i}^{l} [ω] = {({(R^{l - 1} [ω])}^{T})}^{- 1} {(H_{i}^{*} [ω])}^{T};

(3) provide the approximate solution of endless feedback equalizer through iteration decision feedforward filter in discrete Fourier transform domain, promptly the approximate solution of endless feedforward filter in the expression formula of discrete Fourier transform domain is

W_{Sub}^{l} [f] = {({({\hat{Q}}^{l} [f])}^{*} {\hat{Q}}^{l} [f] + σ_{v}^{2} I_{N_{r}})}^{- 1} ({({\hat{Q}}^{l})}^{*} T [f]),

Wherein

T [f] = {[\begin{matrix} T [2 f] I_{M_{t}} & T [2 f + 1] I_{M_{t}} \end{matrix}]}^{T}

T [f] = Σ_{n = 0}^{{2 L}_{e} - 1} δ [n - D] e^{- \frac{j 2 πnf}{{2 L}_{e}}},

δ [n] is an impulse function;

(4) feedforward filter of corresponding step 3 design, the criterion based on output Signal to Interference plus Noise Ratio maximum provides optimum feedback filter setting, and promptly the optimum feedback filter corresponding to u code channel, i transmit antennas symbol is

b_{Sub, i, t, u}^{l} = ρ_{floor ((t - 1) / (2 D + 1)) + 1, u}^{l - 1} {({\overset{&OverBar;}{W}}_{Sub, i}^{l})}^{*} {(H)}_{t}

= ρ_{floor ((t - 1) / (2 D + 1)) + 1, u}^{l - 1} Σ_{n = 1}^{N_{r}} {({\overset{&OverBar;}{W}}_{Sub, n, i}^{l})}^{*} {(H)}_{(n - 1) (2 J + L) + 1 : n (2 J + L), t}

= ρ_{floor ((t - 1) / (2 D + 1)) + 1, u}^{l - 1} Σ_{n = 1}^{N_{r}} Σ_{p = 1}^{2 J + L} {({\overset{&OverBar;}{W}}_{Sub, n, i}^{l})}_{p}^{*} h_{n, floor ((t - 1) / (2 D + 1)) + 1} [L - 1 - t + p]

t≠(i-1)(2D+1)+D+1

Wherein

{\overset{&OverBar;}{W}}_{Sub, i}^{l} = [\begin{matrix} {\overset{&OverBar;}{W}}_{Sub, 1, i}^{l} \\ . \\ . \\ . \\ {\overset{&OverBar;}{W}}_{Sub, N_{r}, i}^{l} \end{matrix}],

{\overset{&OverBar;}{W}}_{Sub, n_{r}, i}^{l} = {[w_{Sub, n_{r}, i}^{l} [L_{e} - 1], w_{Sub, n_{r}, i}^{l} [L_{e} - 2], . . ., w_{Sub, n_{r}, i}^{l} [0]]}^{*}, w_{Sub, n_{r}, i}^{l} [t]

Be

Inverse discrete Fourier transformer inverse-discrete;

(5) according to the l time iteration output Signal to Interference plus Noise Ratio of feedback equalizer through iteration decision, calculate normalizated correlation coefficient, promptly the l time, i transmit antennas iteration are exported Signal to Interference plus Noise Ratio and are

{SINR}_{Sub, i}^{l} = {| {({\overset{&OverBar;}{W}}_{Sub, i}^{l})}^{*} {(H)}_{(i - 1) (2 D + 1) + D + 1} |}^{2} / (Σ_{t = 1, t &NotEqual; (i - 1) (2 D + 1) + D + 1}^{2 D + 1} (\frac{U}{N} -

\frac{1}{N} Σ_{u = 1}^{U} {(ρ_{floor ((t - 1) / (2 D + 1)) + 1, u}^{l - 1})}^{2}) {| {({\overset{&OverBar;}{W}}_{Sub, i}^{l})}^{*} {(H)}_{t} |}^{2} + σ_{v}^{2} {| | {\overset{&OverBar;}{W}}_{Sub, i}^{l} | |}^{2})

Next utilizes the l time, i transmit antennas iteration output Signal to Interference plus Noise Ratio calculating normalizated correlation coefficient, and the corresponding normalizated correlation coefficient of modulation system arbitrarily can calculate according to following formula

ρ_{m_{t}, u}^{l} = E < d_{m_{t}, u} [n] {({\hat{d}}_{m_{t}, u}^{l} [n])}^{*} >

= \underset{c, e &Element; Π}{Σ} P {d_{m_{t}, u} [n] = c, {\hat{d}}_{m_{t}, u}^{l} [n] = e} {ce}^{*}

= \underset{c, e &Element; Π}{Σ} P {{\hat{d}}_{m_{t}, u}^{l} [n] = e | d_{m_{t}, u} [n] = c} P {d_{m_{t}, u} [n] = c} {ce}^{*};

Wherein

(.) ^T, (.) ^*Represent transposition and conjugate transpose respectively;

() refers to W _{Sub, i}

(H) _tThe t row of representing matrix H;

R^{l - 1} [ω] = Σ_{m_{t} = 1}^{M_{t}} (\frac{U}{N} - \frac{1}{N} Σ_{u = 1}^{U} {| ρ_{m_{t}, u}^{l - 1} |}^{2}) H_{m_{t}} [ω] H_{m_{t}}^{*} [ω] + σ_{v}^{2} I_{N_{r}};

H is a channel condition information,

H = [\begin{matrix} H_{1,1} & H_{1,2} & . . . & H_{1, M_{t}} \\ H_{2,1} & H_{2,2} & . . . & H_{2, M_{t}} \\ . & . & . & . \\ . & . & . & . \\ . & . & . & . \\ H_{N_{r}, 1} & H_{N_{r}, 2} & . . . & H_{N_{r}, M_{t}} \end{matrix}];

H_{n_{r}, m_{t}} = [\begin{matrix} h_{n_{r}, m_{t}} [L - 1] & . . . & h_{n_{r}, m_{t}} [0] \\ h_{n_{r}, m_{t}} [L - 1] & . . . & h_{n_{r}, m_{t}} [0] \\ h_{n_{r}, m_{t}} [L - 1] & . . . & h_{n_{r}, m_{t}} [0] \end{matrix}]

{\hat{Q}}^{l} [f] = {[\begin{matrix} {\hat{H}}^{l} [2 f] & α {\hat{H}}^{l} [2 f + 1] \end{matrix}]}^{T};

σ _v ²Be variance;

Be N _rThe unit matrix of dimension;

Its integer part is got in floor (x) expression;

N _rBe the reception antenna number;

(H) _{(n-1) (2J+L)+and 1:n (2J+L), t}The t of representing matrix H row, from (n-1) (2J+L)+1 row formed to element of n (2J+L) row of row;

L is the channel memory span;

J represents observation window length;

D=L+J-1 is the judgement time-delay;

N is a spreading factor;

Be m _tTransmit antennas and n _rThe channel impulse response of root reception antenna;

The code channel number of U for activating;

Be m _tTransmit antennas, a u code channel c _uThe symbol of [n] carrying;

m _tBe m _tTransmit antennas;

U is the code channel number;

II represents the modulation symbol collection, and c and e represent the point on the planisphere;

Be m _tTransmit antennas, a u code channel, (l-1) level, the hard decision in a k moment;

M _tBe the transmitting antenna radical;

H_{m_{t}} [ω] = {[H_{{1, m}_{t}} [ω], . . ., H_{N_{r}, m_{t}} [ω]]}^{T};

H_{n_{r}, m_{t}} [ω] = Σ_{l = 0}^{L - 1} h_{n_{r}, m_{t}} [l] e^{- jωl};

{\hat{H}}^{l} [f] = H [f] Σ^{l - 1};

(H[f]) _n，m＝H _n，m[f]；

Σ^{l - 1} = diag ([\sqrt{\frac{U}{N} - \frac{1}{N} Σ_{u = 1}^{U} {| ρ_{1, u}^{l - 1} |}^{2}}, . . ., \sqrt{\frac{U}{N} - \frac{1}{N} Σ_{u = 1}^{U} {| ρ_{M_{t}, u}^{l - 1} |}^{2}}]);

For

2L _eThe discrete length Fourier transform;

α＝U[1]/U[0]；

U [f] = Σ_{n = 0}^{2 L_{e} - 1} u [t] e^{\frac{- j 2 πnf}{{2 L}_{e}}};

u [t] = \{\begin{matrix} 1, & 0 \leq t < L_{e} \\ 0, & L_{e} \leq t < 2 L_{e} \end{matrix} .

2. realize the feedback equalizer through iteration decision of claim 1 communication system channel distortion compensating method, it comprises: pulse shaping filter and sampler [1], also comprise feedforward filter [2], descrambler [3], despreader [4], decision device [5], spread spectrum and scrambler [6], feedback filter [7], channel decoder [8], it is characterized in that signal flow: after the received signal passages through which vital energy circulates is washed into mode filter and sampler [1], export to feedforward filter 2-i, the output of the output signal of feedforward filter 2-i and feedback filter 7-i is superposition mutually, remove interfering components, the signal of removing interfering components is through descrambler 3-i, behind the despreader 4-i-u, input decision device 5-i-u, the hard decision of decision device 5-i-u is behind spread spectrum and scrambler 6-i-u, obtain corresponding i transmit antennas, the feedback signal of u code channel, the feedback signal of all transmitting antennas is all imported whenever-individual feedback filter [7], after iterations reaches and presets number of times, stop iteration, decision device [5] output log-likelihood ratio information is given channel decoder [8], and channel decoder [8] obtains information bit through decoding.

3. according to claim The feedback equalizer through iteration decision of channel distortion compensation method is characterized in that the signal flow of feedback filter 7-i is as follows: m in described realization claim 1 communication system _tThe feedback signal input feedback filter 7-i-m of transmit antennas _t1 or feedback filter 7-i-m _t2, and according to modulation system selection feedback filter 7-i-m _t1 or feedback filter 7-i-m _t2, all feedback filters Output phase superposition, obtain the output of feedback filter 7-i, this output and the output of feedforward filter 2-i is superposition mutually.