CN1674482A

CN1674482A - Method and apparatus for detecting normalized iterative soft interference cancelling signal

Info

Publication number: CN1674482A
Application number: CN 200510038645
Authority: CN
Inventors: 高西奇; 尤肖虎; 王闻今
Original assignee: Southeast University
Current assignee: Huawei Technologies Co Ltd
Priority date: 2005-04-01
Filing date: 2005-04-01
Publication date: 2005-09-28
Anticipated expiration: 2025-04-01
Also published as: CN100373840C

Abstract

The normalized iterative soft interference cancellation signal detection method relates to a signal detection method for an iterative receiver of a multi-antenna wireless communication system. In the transmission iterative receiver, at the receiving end, the received multi-antenna signals are firstly subjected to normalized matching and combining in the space-time domain, and then iterative interference cancellation detection is performed to obtain the estimated value of the signal and the estimated interference noise variance, and then the solution Harmonic decoding, the decoder performs decoding to obtain the soft information of the bit, reconstructs the mean value and variance of the signal, and feeds it back to the detector for interference cancellation, demodulation and decoding. The method and the iterative detection and decoding method of the device have high performance; the normalization method not only makes the dynamic range of data in the calculation process smaller, which is beneficial to fixed-point realization, but also reduces the number of variables when understanding the mapping, which is beneficial to query table implementation.

Description

Normalized Iterative Soft Interference Cancellation Signal Detection Method and Device

技术领域technical field

本发明涉及一种通过使用多个发送/接收天线来传输高速数据的宽带移动通信系统，尤其涉及一种用于多天线无线通信系统迭代接收机的信号检测方法。所涉及的无线通信系统具有一个或一个以上的接收天线和一个或一个以上的发送天线。The present invention relates to a broadband mobile communication system for transmitting high-speed data by using multiple transmitting/receiving antennas, and in particular to a signal detection method for an iterative receiver of a multi-antenna wireless communication system. The wireless communication system involved has one or more receiving antennas and one or more transmitting antennas.

背景技术Background technique

在一定的容错率的条件下，多天线发送多天线接收技术现在被用来提高无线通信系统的传输效率，在给定传输速率的条件下降低传输功率，或者给定传输功率的条件下提高信号的传输速率。在载波频率高，而天线之间距离较远的情况下，各个发送天线至各个接收天线之间的路径损耗可以近似认为独立，在这种情况下，出现了多种多天线的发送/接收方法，主要可以空分复用和空时编码。空分复用系统在各个发送天线发送独立的数据流，比如V-BLAST(垂直的贝尔实验室分层空时结构)，故高空时编码利用正交性设计，同一信号在各个发送天线以不同形式发送，以获得发送分集，简化接收机的复杂度。Under the condition of a certain error tolerance rate, multi-antenna transmission and multi-antenna reception technology is now used to improve the transmission efficiency of wireless communication systems, reduce the transmission power under the condition of a given transmission rate, or improve the signal under the condition of a given transmission power. transmission rate. When the carrier frequency is high and the distance between the antennas is long, the path loss between each transmitting antenna and each receiving antenna can be considered to be approximately independent. In this case, a variety of multi-antenna transmission/reception methods have emerged , mainly for space division multiplexing and space-time coding. The space division multiplexing system sends independent data streams at each transmitting antenna, such as V-BLAST (Vertical Bell Labs Layered Space-Time Structure), so the high-altitude space-time coding uses orthogonality design, and the same signal is transmitted at different transmitting antennas. In order to obtain transmit diversity and simplify the complexity of the receiver.

在宽带系统中，符号时间间隔小于多径信道的多径时延扩展，因此接收信号在时间上会存在符号间干扰。多载波系统比如正交频分复用等等，在一定程度上能够解决这个问题。由于存在多个发送天线，在每个接收天线上会存在多天线的信号之间的干扰。我们设发送天线数为N，接收天线数为M，信道的最大多径时延扩展为L，用s_n(k)表示第n根发送天线在k时刻的发送信号，用r_m(k)和z_m(k)分别表示第m根接收天线第k时刻接收到的信号和加性高斯白噪声，h_m，n(l)为时延为l时第n根发送天线到第m根接收天线的信道冲击响应系数。则有：In a broadband system, the symbol time interval is smaller than the multipath delay spread of the multipath channel, so there will be intersymbol interference in the received signal in time. Multi-carrier systems such as Orthogonal Frequency Division Multiplexing can solve this problem to a certain extent. Due to the presence of multiple transmit antennas, there will be interference between the signals of the multiple antennas on each receive antenna. We assume that the number of transmitting antennas is N, the number of receiving antennas is M, and the maximum multipath delay of the channel is extended to L. Use s _n (k) to represent the transmission signal of the nth transmitting antenna at time k, and use r _m (k) and z _m (k) respectively represent the signal received by the mth receiving antenna at the kth moment and the additive white Gaussian noise, h _m,n (l) is the time delay from the nth transmitting antenna to the mth receiving antenna when the time delay is l. The channel impulse response coefficient of the antenna. Then there are:

${r r}_{m m} ((k k)) = = {Σ Σ}_{l l = = 00}^{L L - - 11} {Σ Σ}_{n no = = 11}^{N N} {h h}_{m m,, n no} ((l l)) \cdot &Center Dot; {s the s}_{n no} ((k k - - l l)) + + {z z}_{m m} ((k k)) - - - - - - [[11]]$

迭代检测译码方法是多天线系统中以低复杂度获得接近最优联合检测译码的行之有效的方法。在这种方法中，检测器与译码器之间进行多次迭代后才对发送信息比特进行判决。而在非末次迭代的过程中，译码器计算出每个比特的软信息(通常用对数似然比来表示)并且反馈给检测器帮助检测器进行检测，在译码器提供反馈软信息的条件下，检测器能够获得更为精确的输出结果，从而使得译码器获得更优的译码性能，这样逐次迭代以获得更优的性能。目前基于迭代检测译码方法有很多检测器被作了研究，主要有最大后验概率(MAP)算法，基于最小均方误差准则(MMSE)算法，近似最大似然方法例如球形译码等等，这些方法性能都相对比较好，但是复杂度依然很高。本发明给出基于匹配合并的迭代软干扰抵消检测方法是降低复杂度的有效检测方法。The iterative detection and decoding method is an effective method to obtain near-optimal joint detection and decoding with low complexity in multi-antenna systems. In this method, a decision is made on the transmitted information bits after multiple iterations between the detector and the decoder. In the non-last iteration process, the decoder calculates the soft information of each bit (usually represented by the logarithmic likelihood ratio) and feeds it back to the detector to help the detector to detect, and the decoder provides feedback soft information Under the condition of , the detector can obtain more accurate output results, so that the decoder can obtain better decoding performance, so that iteratively obtains better performance. At present, many detectors based on iterative detection and decoding methods have been studied, mainly including the maximum a posteriori probability (MAP) algorithm, based on the minimum mean square error criterion (MMSE) algorithm, approximate maximum likelihood methods such as spherical decoding, etc. The performance of these methods is relatively good, but the complexity is still high. The invention provides that the iterative soft interference cancellation detection method based on matching and merging is an effective detection method for reducing complexity.

发明内容Contents of the invention

技术问题：本发明的目的是提供一种用于无线通信接收机的归一化迭代软干扰抵消信号检测方法和装置，该方法和装置具有空时匹配合并和迭代软干扰抵消相比其他多天线接收方法具有较低复杂度；迭代检测译码的方法具有较高的性能；归一化方法不仅使得计算过程中的数据动态范围变小，利于定点实现，而且减少了解映射时变元的个数，有利于查表实现。Technical problem: The object of the present invention is to provide a normalized iterative soft interference cancellation signal detection method and device for a wireless communication receiver, which has space-time matching and combination and iterative soft interference cancellation compared to other multi-antenna The receiving method has lower complexity; the iterative detection and decoding method has higher performance; the normalization method not only makes the dynamic range of data in the calculation process smaller, which is beneficial to fixed-point implementation, but also reduces the number of variables when understanding the mapping , which is conducive to the realization of table lookup.

技术方案：在多天线无线传输迭代接收机中，检测器可使用本发明提出的归一化空时匹配合并的迭代软干扰抵消的检测方法，在接收端首先对接收到的多天线信号进行空时域上的归一化匹配合并，然后进行迭代干扰抵消检测得到信号的估计值和估计的干扰噪声方差，再进行解调和译码，译码器进行译码得到比特的软信息，对信号进行均值和方差重建，反馈给检测器重新进行干扰抵消，解调和译码。Technical solution: In the multi-antenna wireless transmission iterative receiver, the detector can use the iterative soft interference cancellation detection method of normalized space-time matching and combining proposed by the present invention, and at the receiving end, the received multi-antenna signal is first space-time Normalized matching and merging in the time domain, and then iterative interference cancellation detection to obtain the estimated value of the signal and the estimated variance of interference noise, and then demodulation and decoding, the decoder decodes to obtain the soft information of the bit, and the signal Perform mean and variance reconstruction, feed back to the detector for interference cancellation, demodulation and decoding.

基于迭代软干扰抵消检测的Turbo接收方法的步骤如下：The steps of the Turbo receiving method based on iterative soft interference cancellation detection are as follows:

步骤1：归一化空时匹配合并；Step 1: Normalized space-time matching and merging;

步骤2：如果是第一次检测译码迭代，执行以下分步骤，后续迭代执行步骤3；Step 2: If it is the first detection and decoding iteration, perform the following sub-steps, and perform step 3 in subsequent iterations;

21)将信号均值方差初始化均值为0，方差为1，21) Initialize the mean value and variance of the signal to be 0 and the variance to be 1,

22)进行多级串行干扰抵消，22) performing multi-stage serial interference cancellation,

23)计算比特似然比，23) Calculate the bit likelihood ratio,

24)进入步骤4；24) Go to step 4;

步骤3：Step 3:

31)根据译码器反馈的比特似然比重建信号均值和方差，31) Reconstruct the signal mean and variance according to the bit likelihood ratio fed back by the decoder,

32)进行一级并行干扰抵消，32) performing one-level parallel interference cancellation,

33)计算比特似然比，33) Calculate the bit likelihood ratio,

34)进入步骤4；34) Go to step 4;

步骤4：Step 4:

41)反交织比特似然比，进行译码，41) Anti-interleaving bit likelihood ratio, decoding,

42)如果是最后一次检测译码迭代，转入步骤5。如果否，转入分步骤3，42) If it is the last detection and decoding iteration, go to step 5. If not, go to sub-step 3,

43)进行一定次数的迭代译码，保留中间软信息给下次检测译码迭代的译码做初始化；输出译码后比特似然比并进行反交织，43) Perform a certain number of iterative decoding, retain the intermediate soft information to initialize the decoding of the next detection and decoding iteration; output the bit likelihood ratio after decoding and perform deinterleaving,

44)进入步骤3；44) Go to step 3;

步骤5：Step 5:

对比特似然比进行判决，得到判决后的输出比特序列。The bit likelihood ratio is judged to obtain the judged output bit sequence.

所述的检测装置包括：软输入软输出检测器，交织和反交织器，软输入软输出译码器，判决器；软输入软输出检测器包括空时合并单元、干扰抵消单元、软解调单元、均值方差单元，空时合并单元的输入端接输入信号“r_m(K)”，空时合并单元的输出端接干扰抵消单元，干扰抵消单元的输出端接软解调单元，软解调单元的输出端分两路，其中一路接均值方差单元，均值方差单元的输出端接干扰抵消单元，软解调单元输出端的另一路接交织和反交织器中的反交织器，反交织器的输出端接软输入软输出译码器，软输入软输出译码器的输出端分两路，其中一路接交织和反交织器中的交织器，另一路输出到判决器。The detection device comprises: a soft input and soft output detector, an interleaving and deinterleaving device, a soft input and soft output decoder, and a decision device; a soft input and soft output detector includes a space-time merging unit, an interference cancellation unit, a soft demodulation unit, mean variance unit, the input terminal of the space-time merging unit is connected to the input signal "r _m (K)", the output terminal of the space-time merging unit is connected to the interference canceling unit, the output terminal of the interference canceling unit is connected to the soft demodulation unit, and the soft demodulation unit The output end of the modulation unit is divided into two paths, one of which is connected to the mean variance unit, the output end of the mean variance unit is connected to the interference cancellation unit, and the other path of the output end of the soft demodulation unit is connected to the deinterleaver in the interleaving and deinterleaver, and the deinterleaver The output terminal of the soft-input and soft-output decoder is connected to the soft-input and soft-output decoder, and the output of the soft-input and soft-output decoder is divided into two paths, one of which is connected to the interleaver in the interleaving and deinterleaver, and the other is output to the decision device.

空时合并单元：对多天线信号进行匹配合并，在多径信道下对多径信号进行匹配合并。其中空时合并单元使用归一化的空时合并方法。Space-time combining unit: perform matching and combining on multi-antenna signals, and perform matching and combining on multipath signals under multipath channel. The space-time merging unit uses a normalized space-time merging method.

检测器单元：去除干扰，得到去干扰后的信号和估计的干扰噪声的方差，并且计算每个比特的软信息(通常用对数似然比来表示)。其中检测器单元使用软输入软输出的检测方法，即检测器单元可以使用译码器反馈的软信息。检测器单元使用迭代软干扰抵消检测，首次检测时，信号的均值方差使用查表的方法快速实现。Detector unit: removes the interference, obtains the variance of the de-interferenced signal and the estimated interference noise, and calculates the soft information of each bit (usually represented by a log-likelihood ratio). The detection method in which the detector unit uses soft input and soft output means that the detector unit can use the soft information fed back by the decoder. The detector unit uses iterative soft interference cancellation detection. When the first detection is performed, the mean value and variance of the signal are quickly realized by a look-up table method.

交织器和反交织器单元：反交织器把检测得到的比特似然比排成译码器的顺序。而交织器把译码器输出的似然比重新按照检测器的要求排列。Interleaver and deinterleaver unit: The deinterleaver arranges the detected bit likelihood ratios into the order of the decoder. The interleaver rearranges the likelihood ratios output by the decoder according to the requirements of the detector.

软输入软输出译码器单元：把反交织后的比特似然比按照编码块根据编码器的约束进行译码，得到新的译码后的似然比，在非末次迭代中将其反馈给检测器，而在末次迭代中将用于信息比特的判决。Soft-input and soft-output decoder unit: Decode the deinterleaved bit likelihood ratio according to the encoding block according to the constraints of the encoder, obtain a new decoded likelihood ratio, and feed it back to detector, which will be used for information bit decisions in the final iteration.

基于迭代软干扰抵消检测的迭代接收机Iterative Receiver Based on Iterative Soft Interference Cancellation Detection

本发明所述的多天线系统中的迭代检测译码接收装置主要可以分为软输入软输出检测器，交织和反交织器，软输入软输出译码器等几部分。The iterative detection and decoding receiving device in the multi-antenna system of the present invention can be mainly divided into several parts such as a soft input and soft output detector, an interleaving and deinterleaving device, and a soft input and soft output decoder.

1、软输入软输出检测器可以分为归一化空时匹配合并，干扰抵消，比特似然比计算和信号均值方差重建四个部分。其中空时匹配合并部分把接收信号按照空间和时间维度上把信号能量收集，得到合并后的信号和相应的信号分量和干扰信号的系数。干扰抵消部分根据空时合并得到的信号和相应的系数，还有重建的均值方差，去除干扰，并且计算剩余噪声和干扰的方差。比特似然比计算部分根据去除干扰后的信号和剩余噪声和干扰的方差计算比特的似然比，并且把它送到译码器。信号的均值方差重建模块用译码器反馈的或者检测得到的似然比重建信号的均值方差。1. The soft-input and soft-output detector can be divided into four parts: normalized space-time matching and merging, interference cancellation, bit likelihood ratio calculation and signal mean variance reconstruction. The space-time matching and merging part collects the signal energy of the received signal according to the space and time dimensions, and obtains the combined signal and the corresponding signal components and coefficients of the interference signal. The interference cancellation part removes the interference based on the signal obtained by the space-time combination and the corresponding coefficients, as well as the mean variance of the reconstruction, and calculates the residual noise and the variance of the interference. The bit likelihood ratio calculation part calculates the bit likelihood ratio according to the variance of the interference-removed signal and residual noise and interference, and sends it to the decoder. The mean variance reconstruction module of the signal uses the likelihood ratio fed back by the decoder or detected to reconstruct the mean variance of the signal.

2、反交织器把检测得到的比特似然比排成译码器的顺序。而交织器把译码器输出的似然比重新按照检测器的要求排列。2. The deinterleaver arranges the detected bit likelihood ratios into the order of the decoder. The interleaver rearranges the likelihood ratios output by the decoder according to the requirements of the detector.

3、译码器把反交织后的比特似然比按照编码块根据编码器的约束进行译码，得到新的译码后的似然比，在非末次迭代中将其反馈给检测器，而在末次迭代中将用于信息比特的判决。3. The decoder decodes the deinterleaved bit-likelihood ratio according to the encoding block according to the constraints of the encoder, and obtains a new decoded likelihood ratio, which is fed back to the detector in the non-last iteration, while In the last iteration it will be used for the decision of the information bits.

有益效果：本发明提出的迭代接收方法具有如下优点：Beneficial effects: the iterative receiving method proposed by the present invention has the following advantages:

1.同传统的检测译码级联的接收机相比，性能大大的改善，在频谱效率不变的情况下大大提高了功率效率。使用高阶调制能在功率效率不变的情况下大大提高频谱效率。1. Compared with the traditional detection and decoding cascade receiver, the performance is greatly improved, and the power efficiency is greatly improved under the condition that the spectrum efficiency remains unchanged. Using higher order modulation can greatly improve spectral efficiency while maintaining power efficiency.

2.同其他的迭代检测译码方法相比，基于归一化的空时匹配合并的迭代软干扰抵消检测方法有效地降低了检测器的复杂度，降低了数据的动态范围，不仅使复杂度随发送天线数和径数成线性增长，而且避免了MMSE必须的求逆运算，使得算法更为稳定。2. Compared with other iterative detection and decoding methods, the iterative soft interference cancellation detection method based on normalized space-time matching and merging effectively reduces the complexity of the detector and the dynamic range of the data, which not only makes the complexity It grows linearly with the number of transmitting antennas and the number of paths, and avoids the necessary inverse operation of MMSE, making the algorithm more stable.

3.归一化方法不仅使得计算过程中的数据动态范围变小，利于定点实现，而且减少了解映射时变元的个数，有利于查表实现。3. The normalization method not only reduces the dynamic range of data in the calculation process, which is beneficial to fixed-point implementation, but also reduces the number of variables when understanding the mapping, which is beneficial to the implementation of table lookup.

本发明提出的接收方法适合用于各类无线传输系统，主要包括The receiving method proposed by the present invention is suitable for various wireless transmission systems, mainly including

1.窄带多天线传输系统，即信道对于信号为频率平坦衰落。1. Narrowband multi-antenna transmission system, that is, the channel is frequency-flat fading for the signal.

2.单载波多天线传输系统，即信道对于信号为频率选择性衰落，而传输采用单载波方式。2. Single-carrier multi-antenna transmission system, that is, the channel is frequency-selective fading for the signal, and the transmission adopts a single-carrier mode.

3.多载波多天线传输系统，比如OFDM，广义多载波系统。3. Multi-carrier multi-antenna transmission system, such as OFDM, generalized multi-carrier system.

4.不仅可以用于扩频系统，也可以用于非扩频系统。4. It can be used not only in spread spectrum systems, but also in non-spread spectrum systems.

5.多址方式可以采用CDMA，TDMA，FDMA等各种。5. Multiple access methods can adopt CDMA, TDMA, FDMA, etc.

6.可以用于任意两个多天线无线设备之间的通信。6. Can be used for communication between any two multi-antenna wireless devices.

附图说明Description of drawings

图1为迭代软干扰抵消检测的迭代接收机框图。其中有：软输入软输出检测器1、空时合并单元11、干扰抵消单元12、软解调单元13、均值方差单元14，交织和反交织器2、反交织器21、交织器22，软输入软输出译码器3，判决器4。Figure 1 is a block diagram of an iterative receiver for iterative soft interference cancellation detection. Among them: soft input and soft output detector 1, space-time combining unit 11, interference cancellation unit 12, soft demodulation unit 13, mean value variance unit 14, interleaving and deinterleaver 2, deinterleaver 21, interleaver 22, soft Input soft output decoder 3, decider 4.

具体实施方式 Detailed ways

归一化迭代软干扰抵消信号检测方法是在多天线无线传输迭代接收机中，在接收端首先对接收到的多天线信号进行空时域上的归一化匹配合并，然后进行迭代干扰抵消检测得到信号的估计值和估计的干扰噪声方差，再进行解调和译码，译码器进行译码得到比特的软信息，对信号进行均值和方差重建，反馈给检测器重新进行干扰抵消，解调和译码。The normalized iterative soft interference cancellation signal detection method is in the iterative receiver of multi-antenna wireless transmission. At the receiving end, the received multi-antenna signals are firstly subjected to normalized matching and combining in the space-time domain, and then iterative interference cancellation detection is performed. Obtain the estimated value of the signal and the estimated variance of the interference noise, and then demodulate and decode. The decoder decodes to obtain the soft information of the bit, reconstructs the mean value and variance of the signal, and feeds back to the detector for interference cancellation again. Harmonic decoding.

基于迭代软干扰抵消检测的迭代接收方法的步骤如下：The steps of the iterative receiving method based on iterative soft interference cancellation detection are as follows:

步骤2：如果是第一次检测译码迭代，执行以下分步骤，后续迭代执行步骤3，Step 2: If it is the first detection and decoding iteration, perform the following sub-steps, and perform step 3 in subsequent iterations,

23)计算比特似然比，23) Calculate the bit likelihood ratio,

24)进入步骤4；24) Go to step 4;

步骤3：Step 3:

33)计算比特似然比，33) Calculate the bit likelihood ratio,

34)进入步骤4；34) Go to step 4;

步骤4：Step 4:

42)如果是最后一次检测译码迭代，转入步骤5，如果否，转入分步骤3，42) If it is the last detection and decoding iteration, go to step 5, if not, go to sub-step 3,

44)进入步骤3；44) Go to step 3;

步骤5：Step 5:

假设系统有N根发送天线，M根接收天线。信息比特经过纠错编码(包括使用迭代译码的级联码或基于图的码)，然后再经过比特交织和调制，分发到各个发送天线。在各个发送天线上，可以把符号序列直接发送，也可以插入循环前缀并且经过IFFT(反快速傅立叶变换)，成为OFDM(正交频分复用)符号进行发送。Suppose the system has N transmitting antennas and M receiving antennas. The information bits are encoded for error correction (including concatenated codes using iterative decoding or graph-based codes), then bit-interleaved and modulated, and distributed to the individual transmit antennas. On each transmit antenna, the symbol sequence can be directly transmitted, or a cyclic prefix can be inserted and subjected to IFFT (Inverse Fast Fourier Transform) to become an OFDM (Orthogonal Frequency Division Multiplexing) symbol for transmission.

令：make:

$\underset{&OverBar; &OverBar;}{r r} ((k k)) = = [\begin{matrix} r r ((k k)) \\ r r ((k k + + 11)) \\ . . \\ . . \\ . . \\ r r ((k k + + L L - - 11)) \end{matrix}],, u u ((k k)) = = [\begin{matrix} s the s ((k k - - L L + + 11)) \\ . . \\ . . \\ . . \\ s the s ((k k)) \\ . . \\ . . \\ . . \\ s the s ((k k + + L L - - 11)) \end{matrix}],, \underset{&OverBar; &OverBar;}{z z} ((k k)) = = [\begin{matrix} z z ((k k)) \\ z z ((k k + + 11)) \\ . . \\ . . \\ . . \\ z z ((k k + + L L - - 11)) \end{matrix}]$

则公式[1]可以表示为Then formula [1] can be expressed as

r(k)＝H·u(k)+ z(k) [2] r (k) = H u (k) + z (k) [2]

其中， r(k)为ML×1的信号s(k)的完全观测信号矢量，H为ML×(2L-1)N的信道卷积矩阵，u(k)为(2L-1)N×1的发送信号向量， z(k)为ML×1的噪声向量。Among them, r (k) is the complete observation signal vector of ML×1 signal s(k), H is the channel convolution matrix of ML×(2L-1)N, u(k) is (2L-1)N× 1 is the transmitted signal vector, and z (k) is the noise vector of ML×1.

在接收端，需要用导频或已知序列先对信道冲击响应进行估计，在得到信道冲击响应后，开始迭代检测译码。首先对发送的信号进行归一化的空时匹配合并，即At the receiving end, it is necessary to first estimate the channel impulse response with a pilot frequency or a known sequence, and start iterative detection and decoding after obtaining the channel impulse response. First, the normalized space-time matching and merging is performed on the transmitted signals, that is,

${x x}_{n no} ((k k)) = = {ρ ρ}_{n no}^{- - 11} \cdot \cdot {h h}_{N N ((L L - - 11)) + + n no}^{H h} \cdot \cdot \underset{&OverBar; &OverBar;}{r r} ((k k)) - - - - - - [[33]]$

公式[3]中h_N(L-1)+n为矩阵H的第N(L-1)+n列， $ρ_{n} = h_{N (L - 1) + n}^{H} \cdot h_{N (L - 1) + n} .$ 且令h _N(L-1)+n in formula [3] is the N(L-1)+nth column of matrix H, $ρ_{no} = h_{N (L - 1) + no}^{h} \cdot h_{N (L - 1) + no} .$ and order

为矩阵H除去h_N(L-1)+n后剩下的部分，为u(k)除去s_n(k)后剩下的部分，即：

is the part left after removing h _N(L-1)+n for the matrix H, The remaining part after removing s _n (k) for u(k), namely:

${\overset{~ ~}{H h}}_{N N ((L L - - 11)) + + n no} = = [[{h h}_{11},, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, {h h}_{N N ((L L - - 11)) + + n no - - 11},, {h h}_{N N ((L L - - 11)) + + n no + + 11},, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, {h h}_{N N ((22 L L - - 11))}]]$

${u u}_{&OverBar; &OverBar;}^{~ ~}_{n no} ((k k)) = = {[[{s the s}_{11} ((k k - - L L + + 11)),, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, {s the s}_{n no - - 11} ((k k)),, {s the s}_{n no + + 11} ((k k)),, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, {s the s}_{N N} ((k k + + L L - - 11))]]}^{T T}$

1、首次检测1. First detection

将初始化信号的均值为0，方差为1。利用重建的均值方差对合并后的信号干扰抵消，去除干扰，且计算剩余干扰噪声的方差。will initialize the signal to have a mean of 0 and a variance of 1. The combined signal interference is canceled by using the reconstructed mean variance, the interference is removed, and the variance of the remaining interference noise is calculated.

${\overset{~ ~}{x x}}_{n no} ((k k)) = = {x x}_{n no} ((k k)) - - {F f}_{n no}^{H h} \cdot \cdot E E. [[{u u}_{&OverBar; &OverBar;}^{~ ~}_{n no} ((k k))]] - - - - - - [[44]]$

${σ σ}_{I I,, n no}^{22} ((k k)) = = {F f}_{n no}^{H h} \cdot \cdot cov cov (({u u}_{&OverBar; &OverBar;}^{~ ~}_{n no} ((k k)))) \cdot \cdot {F f}_{n no} + + {ρ ρ}_{n no}^{- - 11} \cdot \cdot {σ σ}_{n no}^{22} - - - - - - [[55]]$

其中 $F_{n}^{H} = ρ_{n}^{- 1} \cdot h_{N (L - 1) + n}^{H} {\tilde{H}}_{N (L - 1) + n} .$ 公式[4]去除了其他信号的干扰，公式[5]计算了剩余干扰噪声的方差。我们假设剩余干扰服从高斯分布，则将信号传输等效成高斯信道，如下：in $f_{no}^{h} = ρ_{no}^{- 1} \cdot h_{N (L - 1) + no}^{h} {\tilde{h}}_{N (L - 1) + no} .$ Equation [4] removes the interference of other signals, and Equation [5] calculates the variance of the remaining interference noise. We assume that the residual interference obeys a Gaussian distribution, then the signal transmission is equivalent to a Gaussian channel, as follows:

${\overset{~ ~}{x x}}_{n no} ((k k)) = = {s the s}_{n no} ((k k)) + + {n no}^{' '} - - - - - - [[66]]$

然后根据信号调制方式对其进行软解调，得到每个编码比特的似然比。用比特似然比来重建信号的均值方差如下：Then it is soft-demodulated according to the signal modulation method to obtain the likelihood ratio of each coded bit. Using the bit likelihood ratio to reconstruct the mean variance of the signal is as follows:

$E E. [[{s the s}_{n no} ((k k))]] = = \underset{α α}{Σ Σ} α α \cdot \cdot P P (({s the s}_{n no} ((k k)) = = α α)) - - - - - - [[77]]$

$cov cov [[{s the s}_{n no} ((k k))]] = = \underset{α α}{Σ Σ} {| | α α | |}^{22} \cdot &Center Dot; P P (({s the s}_{n no} ((k k)) = = α α)) - - E E. {[[{s the s}_{n no} ((k k))]]}^{22} - - - - - - [[88]]$

设符号α由比特d₀，d₁，…d_Mc-1映射而成，则Suppose the symbol α is mapped from bits d ₀ , d ₁ ,...d _Mc-1 , then

$P P (({s the s}_{n no} ((k k)) = = α α)) = = {Π Π}_{k k = = 00}^{{M m}_{C C} - - 11} \frac{11}{22} [[11 + + {\overset{~ ~}{d d}}_{k k} \cdot &Center Dot; tanh tanh ((L L (({d d}_{k k})) / / 22))]] - - - - - - [[99]]$

利用公式[7]，公式[8]和公式[9]重建信号的均值和方差后，重新用公式[4]，公式[5]进行干扰抵消，再进行软解调。如此循环迭代2～4之后进入译码。在这里需要注意的是，在首次检测时使用的是串行干扰抵消，即在当前符号被解调后就更新其均值和方差，在后续符号的干扰抵消中即使用更新后的均值方差。After using formula [7], formula [8] and formula [9] to reconstruct the mean and variance of the signal, use formula [4] and formula [5] again to cancel the interference, and then perform soft demodulation. Enter decoding after loop iterations 2 to 4 in this way. It should be noted here that serial interference cancellation is used in the first detection, that is, the mean and variance of the current symbol are updated after demodulation, and the updated mean and variance are used in interference cancellation of subsequent symbols.

利用公式[7]，公式[8]，公式[9]进行均值方差重建需要先对信号进行软解调，而且计算复杂度高，且随着调制阶数呈指数增长。我们可以根据归一化的特性进一步对均值方差重建过程进行简化，均值和方差仅仅为公式[6]中

和σ_I，n ²(k)函数，即Using formula [7], formula [8], and formula [9] to reconstruct the mean variance requires soft demodulation of the signal first, and the computational complexity is high, and it grows exponentially with the modulation order. We can further simplify the mean-variance reconstruction process according to the normalized characteristics, and the mean and variance are only in the formula [6]

and σ _{I, n} ² (k) functions, namely

$E E. [[{s the s}_{n no} ((k k))]] = = {f f}_{11} (({\overset{~ ~}{x x}}_{n no} ((k k)),, {σ σ}_{I I,, n no}^{22} ((k k)))) - - - - - - [[1010]]$

$cov cov [[{s the s}_{n no} ((k k))]] = = {f f}_{22} (({\overset{~ ~}{x x}}_{n no} ((k k)),, {σ σ}_{I I,, n no}^{22} ((k k)))) - - - - - - [[1111]]$

其中f₁(□)和f₂(□)为二维函数，在实际系统的实现中可以用查表或者查表结合插值的方法来实现。这种方法不仅具有较低的复杂度，而且系统采用不同调制方式时只需改变映射表，具有复杂度低并且复杂度恒定的优点。Among them, f ₁ (□) and f ₂ (□) are two-dimensional functions, which can be realized by table lookup or table lookup combined with interpolation in the actual system implementation. This method not only has lower complexity, but also only needs to change the mapping table when the system adopts different modulation modes, which has the advantages of low complexity and constant complexity.

2、后续检测2. Follow-up detection

将译码器反馈回来的比特似然比根据公式[7]，公式[8]和公式[9]重建信号的均值和方差。再根据公式[4]和公式[5]进行对所有信号进行干扰抵消。得到信号估计值和相对应的噪声干扰的方差后再进行软解调，获得比特似然比。在后续检测中，我们使用了1级并行干扰抵消，即干扰抵消前不根据新的检测结果更新信号的均值和方差。The bit likelihood ratio fed back by the decoder is used to reconstruct the mean and variance of the signal according to formula [7], formula [8] and formula [9]. Then perform interference cancellation for all signals according to formula [4] and formula [5]. After obtaining the variance of the signal estimation value and the corresponding noise interference, soft demodulation is performed to obtain the bit likelihood ratio. In the follow-up detection, we use 1-level parallel interference cancellation, that is, the mean and variance of the signal are not updated according to the new detection results before the interference cancellation.

本发明的归一化迭代软干扰抵消信号检测方法的检测装置包括：软输入软输出检测器1，交织和反交织器2，软输入软输出译码器3，判决器4；软输入软输出检测器1包括空时合并单元11、干扰抵消单元12、软解调单元13、均值方差单元14，空时合并单元11的输入端接输入信号“r_m(K)”，空时合并单元11的输出端接干扰抵消单元12，干扰抵消单元12的输出端接软解调单元13，软解调单元13的输出端分两路，其中一路接均值方差单元14，均值方差单元14的输出端接干扰抵消单元12，软解调单元13输出端的另一路接交织和反交织器2中的反交织器21，反交织器21的输出端接软输入软输出译码器，软输入软输出译码器的输出端分两路，其中一路接交织和反交织器2中的交织器22，另一路输出到判决器4。The detection device of the normalized iterative soft interference cancellation signal detection method of the present invention comprises: a soft input and soft output detector 1, an interleaving and deinterleaving device 2, a soft input and soft output decoder 3, and a decision device 4; soft input and soft output Detector 1 comprises space-time merging unit 11, interference cancellation unit 12, soft demodulation unit 13, mean value variance unit 14, the input terminal of space-time merging unit 11 is connected with input signal "r _m (K) ", and space-time merging unit 11 The output terminal of the interference canceling unit 12 is connected, and the output terminal of the interference canceling unit 12 is connected to the soft demodulation unit 13, and the output terminal of the soft demodulation unit 13 is divided into two paths, wherein one path is connected to the mean value variance unit 14, and the output end of the mean value variance unit 14 Connect interference canceling unit 12, the other road of soft demodulation unit 13 output terminals connects interleaving and deinterleaver 21 in deinterleaver 2, the output terminal of deinterleaver 21 connects soft input soft output decoder, soft input soft output decoder The output end of the encoder is divided into two paths, one of which is connected to the interleaver 22 in the interleaving and deinterleaver 2, and the other is output to the decision unit 4.

基于迭代软干抵消检测的迭代接收处理是在多天线无线传输迭代接收机中，在接收端首先对接收到的多天线信号进行空时域上的归一化匹配合并，然后进行迭代干扰抵消检测得到信号的估计值和估计的干扰噪声方差，再进行解调和译码，译码器进行译码得到比特的软信息，对信号进行均值和方差重建，反馈给检测器重新进行干扰抵消，解调和译码。The iterative reception processing based on iterative soft interference cancellation detection is in the multi-antenna wireless transmission iterative receiver. At the receiving end, the received multi-antenna signals are firstly normalized and combined in the space-time domain, and then the iterative interference cancellation detection is performed. Obtain the estimated value of the signal and the estimated variance of the interference noise, and then demodulate and decode. The decoder decodes to obtain the soft information of the bit, reconstructs the mean value and variance of the signal, and feeds back to the detector for interference cancellation again. Harmonic decoding.

基于迭代软干扰抵消检测的迭代接收处理步骤如下：The iterative reception processing steps based on iterative soft interference cancellation detection are as follows:

即根据公式[3]，先计算系数 $ρ_{n} = Σ_{l = 0}^{L - 1} Σ_{m = 1}^{M} \cdot | h_{m, n} (l) |$ 和系数 $f_{n, n^{'}} (l) = ρ_{n}^{- 1} \cdot Σ_{l^{'} = \max {0, l}}^{\min {L - 1, l + L - 1}} Σ_{m = 1}^{M} h_{m, n}^{*} (l^{'}) \cdot h_{m, n^{'}} (l^{'} - l),$ 其中l＝-(L-1)，…，L-1，然后计算 $x_{n} (k) = ρ_{n}^{- 1} \cdot Σ_{l = 0}^{L - 1} Σ_{m = 1}^{M} h_{m, n}^{*} (l) \cdot r_{m} (k + l) .$ That is, according to the formula [3], first calculate the coefficient $ρ_{no} = Σ_{l = 0}^{L - 1} Σ_{m = 1}^{m} \cdot | h_{m, no} (l) |$ and coefficient $f_{no, {no}^{'}} (l) = ρ_{no}^{- 1} \cdot Σ_{l^{'} = \max {0, l}}^{\min {L - 1, l + L - 1}} Σ_{m = 1}^{m} h_{m, no}^{*} (l^{'}) \cdot h_{m, {no}^{'}} (l^{'} - l),$ where l=-(L-1),...,L-1, then calculate $x_{no} (k) = ρ_{no}^{- 1} \cdot Σ_{l = 0}^{L - 1} Σ_{m = 1}^{m} h_{m, no}^{*} (l) &Center Dot; r_{m} (k + l) .$

2.1)将信号均值方差初始化均值为0，方差为1，即 s_n(k)＝0， $σ_{s, n}^{2} (k) = 1 .$ 2.1) Initialize the mean value and variance of the signal to be 0 and the variance to be 1, that is, s _n (k)=0, $σ_{the s, no}^{2} (k) = 1 .$

2.2)进行多级串行干扰抵消，设B为信号块长，T为干扰抵消级数，令n＝0，k＝0，t＝0。2.2) Perform multi-stage serial interference cancellation, let B be the signal block length, T be the number of interference cancellation stages, set n=0, k=0, t=0.

2.2.1)计算 ${\tilde{x}}_{n} (k) = x_{n} (k) - Σ_{l = 1}^{L - 1} f_{n, n^{'}} (- l) \cdot s_{n} (k - l) - Σ_{l = 1}^{L - 1} f_{n, n^{'}} (l) \cdot s_{n} (k + l)$ 和2.2.1) Calculation ${\tilde{x}}_{no} (k) = x_{no} (k) - Σ_{l = 1}^{L - 1} f_{no, {no}^{'}} (- l) &Center Dot; {the s}_{no} (k - l) - Σ_{l = 1}^{L - 1} f_{no, {no}^{'}} (l) \cdot {the s}_{no} (k + l)$ and

${σ σ}_{I I,, n no}^{22} ((k k)) = = {ρ ρ}_{n no}^{- - 11} \cdot \cdot {σ σ}_{z z}^{22} + + {Σ Σ}_{l l = = 11}^{L L - - 11} [[| | {f f}_{n no,, {n no}^{' '}} ((l l)) {| |}^{22} {\cdot \cdot σ σ}_{s the s,, {n no}^{' '}}^{22} ((k k + + l l)) + + {| | {f f}_{n no,, {n no}^{' '}} ((- - l l)) | |}^{22} \cdot \cdot {σ σ}_{s the s,, {n no}^{' '}}^{22} ((k k - - l l))]]$

2.2.2)根据 $P (x_{n} (k) | s_{n} (k) = α) = \frac{1}{\sqrt{2 π} σ_{I, n} (k)} \cdot \exp (- \frac{{({\tilde{x}}_{n} (k) - α)}^{2}}{σ_{I, n}^{2} (k)})$ 计算信号s_n(k)为每个发送符号的概率。根据 ${\overset{&OverBar;}{s}}_{n} (k) = \underset{α}{Σ} α \cdot P (x_{n} (k) | s_{n} (k) = α)$ 和2.2.2) According to $P (x_{no} (k) | {the s}_{no} (k) = α) = \frac{1}{\sqrt{2 π} σ_{I, no} (k)} \cdot \exp (- \frac{{({\tilde{x}}_{no} (k) - α)}^{2}}{σ_{I, no}^{2} (k)})$ Compute the probability of signal s _n (k) for each transmitted symbol. according to ${\overset{&OverBar;}{the s}}_{no} (k) = \underset{α}{Σ} α &Center Dot; P (x_{no} (k) | {the s}_{no} (k) = α)$ and

${σ σ}_{s the s,, n no}^{22} ((k k)) = = \underset{α α}{Σ Σ} {| | α α | |}^{22} \cdot \cdot P P (({x x}_{n no} ((k k)) | | {s the s}_{n no} ((k k)) = = α α {)) - - \overset{&OverBar; &OverBar;}{s the s}}_{n no}^{22} ((k k))$

k＝k+1，若k＝B，则n＝n+1，k＝0；若n＞N，则n＝1，t＝t+1，若t＝T，则进入步骤2.3，否则进入步骤2.2.1。k=k+1, if k=B, then n=n+1, k=0; if n>N, then n=1, t=t+1, if t=T, then enter step 2.3, otherwise enter Step 2.2.1.

2.3)计算比特似然比2.3) Calculate bit likelihood ratio

$L L ((b b)) = = log log \frac{\underset{α α : : b b = = + + 11}{Σ Σ} P P (({x x}_{n no} ((k k)) | | {s the s}_{n no} ((k k)) = = α α))}{\underset{α α : : b b = = - - 11}{Σ Σ} P P (({x x}_{n no} ((k k)) | | {s the s}_{n no} ((k k)) = = α α))}$

2.4)进入步骤4；2.4) Go to step 4;

步骤3：Step 3:

3.1)根据译码器反馈的比特似然比重建信号均值和方差，首先根据3.1) Reconstruct the signal mean and variance according to the bit likelihood ratio fed back by the decoder, first according to

$P (s_{n} (k) = α) = Π_{k = 0}^{M_{C} - 1} \frac{1}{2} [1 + {\tilde{d}}_{k} \cdot \tanh (L (d_{k}) / 2)]$ 由反馈的比特似然比计算符号概率，然后根据 ${\overset{&OverBar;}{s}}_{n} (k) = \underset{α}{Σ} α \cdot P (s_{n} (k) = α)$ 和 $σ_{s, n}^{2} (k) = \underset{α}{Σ} {| α |}^{2} \cdot P (s_{n} (k) = α) - {\overset{&OverBar;}{s}}_{n}^{2} (k)$ 计算信号均值和方差。 $P ({the s}_{no} (k) = α) = Π_{k = 0}^{m_{C} - 1} \frac{1}{2} [1 + {\tilde{d}}_{k} \cdot \tanh (L (d_{k}) / 2)]$ The symbol probability is calculated from the feedback bit likelihood ratio, and then according to ${\overset{&OverBar;}{the s}}_{no} (k) = \underset{α}{Σ} α \cdot P ({the s}_{no} (k) = α)$ and $σ_{the s, no}^{2} (k) = \underset{α}{Σ} {| α |}^{2} \cdot P ({the s}_{no} (k) = α) - {\overset{&OverBar;}{the s}}_{no}^{2} (k)$ Computes signal mean and variance.

3.2)进行一级并行干扰抵消3.2) Perform one-level parallel interference cancellation

计算 ${\tilde{x}}_{n} (k) = x_{n} (k) - Σ_{l = 1}^{L - 1} f_{n, n^{'}} (- l) \cdot s_{n} (k - l) - Σ_{l = 1}^{L - 1} f_{n, n^{'}} (l) \cdot s_{n} (k + l)$ 和calculate ${\tilde{x}}_{no} (k) = x_{no} (k) - Σ_{l = 1}^{L - 1} f_{no, {no}^{'}} (- l) &Center Dot; {the s}_{no} (k - l) - Σ_{l = 1}^{L - 1} f_{no, {no}^{'}} (l) &Center Dot; {the s}_{no} (k + l)$ and

${σ σ}_{I I,, n no}^{22} ((k k)) = = {ρ ρ}_{n no}^{- - 11} \cdot &Center Dot; {σ σ}_{z z}^{22} + + {Σ Σ}_{l l = = 11}^{L L - - 11} [[{| | {f f}_{n no,, {n no}^{' '}} ((l l)) | |}^{22} \cdot &Center Dot; {σ σ}_{s the s,, {n no}^{' '}}^{22} ((k k + + l l)) + + {| | {f f}_{n no,, {n no}^{' '}} ((- - l l)) | |}^{22} \cdot &Center Dot; {σ σ}_{s the s,, {n no}^{' '}}^{22} ((k k - - l l))]]$

3.3)计算比特似然比，(同步骤2.3)3.3) Calculate bit likelihood ratio, (same as step 2.3)

3.4)进入步骤4；3.4) Go to step 4;

步骤4：Step 4:

4.1)反交织比特似然比，进行译码，4.1) Inverse interleaving bit likelihood ratio for decoding,

4.2)如果是最后一次检测译码迭代，转入步骤5，如果否，转入分步骤3，4.2) If it is the last detection and decoding iteration, go to step 5, if not, go to sub-step 3,

4.3)进行一定次数的迭代译码，保留中间软信息给下次检测译码迭代的译码做初始化；输出译码后比特似然比并进行反交织，4.3) Perform a certain number of iterative decoding, retain the intermediate soft information to initialize the decoding of the next detection and decoding iteration; output the bit likelihood ratio after decoding and perform deinterleaving,

4.4)进入步骤3；4.4) Go to step 3;

步骤5：Step 5:

其中步骤2.2.2计算均值方差可用查表来实现，即Wherein the step 2.2.2 calculates the mean variance and can be realized by looking up the table, that is

${\overset{&OverBar;}{s}}_{n} (k) = f_{1} ({\tilde{x}}_{n} (k), σ_{I, n}^{2} (k))$ 和 $σ_{s, n}^{2} (k) = f_{2} ({\tilde{x}}_{n} (k), σ_{I, n}^{2} (k)),$ f₁(·)和f₂(·)均事先做成表储存起来。 ${\overset{&OverBar;}{the s}}_{no} (k) = f_{1} ({\tilde{x}}_{no} (k), σ_{I, no}^{2} (k))$ and $σ_{the s, no}^{2} (k) = f_{2} ({\tilde{x}}_{no} (k), σ_{I, no}^{2} (k)),$ Both f ₁ (·) and f ₂ (·) are stored as tables in advance.

Claims

1. A normalized iterative soft interference cancellation signal detection method, characterized in that the detection method is based on the iterative detection and decoding receiving process of iterative soft interference cancellation detection. In the multi-antenna wireless transmission iterative receiver, at the receiving end first Perform normalized matching and combining in the space-time domain on the received multi-antenna signals, and then perform iterative interference cancellation detection to obtain the estimated value of the signal and the estimated interference noise variance, and then perform demodulation and decoding, and the decoder performs decoding The code obtains the soft information of the bit, reconstructs the mean value and variance of the signal, and feeds it back to the detector for interference cancellation, demodulation and decoding.

2. The normalized iterative soft interference cancellation signal detection method according to claim 1, characterized in that the iterative detection decoding reception processing steps based on iterative soft interference cancellation detection are as follows:

Step 1: Normalized space-time matching and merging;

Step 2: If it is the first detection and decoding iteration, perform the following sub-steps, and perform step 3 in subsequent iterations,

21) Initialize the mean value and variance of the signal to be 0 and the variance to be 1,

22) performing multi-stage serial interference cancellation,

23) Calculate the bit likelihood ratio,

24) Go to step 4;

Step 3:

31) Reconstruct the signal mean and variance according to the bit likelihood ratio fed back by the decoder,

32) performing one-level parallel interference cancellation,

33) Calculate the bit likelihood ratio,

34) Go to step 4;

Step 4:

41) Anti-interleaving bit likelihood ratio, decoding,

42) If it is the last detection and decoding iteration, go to step 5, if not, go to sub-step 3,

43) Perform a certain number of iterative decoding, retain the intermediate soft information to initialize the decoding of the next detection and decoding iteration; output the bit likelihood ratio after decoding and perform deinterleaving,

44) Go to step 3;

Step 5:

The bit likelihood ratio is judged to obtain the judged output bit sequence.

3. A detection device suitable for the normalized iterative soft interference cancellation signal detection method according to claim 1, characterized in that the detection device comprises: a soft input soft output detector (1), interleaving and deinterleaving Device (2), soft input and soft output decoder (3), decision device (4); soft input and soft output detector (1) includes space-time merging unit (11), interference cancellation unit (12), soft demodulation Unit (13), mean value variance unit (14), the input terminal of the space-time merging unit (11) is connected to the input signal "rm (K)", and the output terminal of the space-time merging unit (11) is connected to the interference canceling unit (12), The output terminal of the interference cancellation unit (12) is connected to the soft demodulation unit (13), and the output terminal of the soft demodulation unit (13) is divided into two paths, wherein one path is connected to the mean value variance unit (14), and the output of the mean value variance unit (14) Interference canceling unit (12) is connected to the terminal, another road of the soft demodulation unit (13) output terminal is connected to the deinterleaver (21) in the deinterleaver (2), and the output terminal of the deinterleaver (21) is connected to the soft input Soft output decoder, the output end of the soft input and soft output decoder is divided into two paths, one of which is connected to the interleaver (22) in the interleaving and deinterleaver (2), and the other path is output to the decision device (4).