CN100362755C

CN100362755C - A Sign Estimation Method

Info

Publication number: CN100362755C
Application number: CNB2004100594684A
Authority: CN
Inventors: 魏立梅
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2004-06-28
Filing date: 2004-06-28
Publication date: 2008-01-16
Anticipated expiration: 2024-06-28
Also published as: CN1716794A

Abstract

The present invention discloses a method for estimating symbols. The method is to use the posterior probabilities of various symbols obtained from the previous symbol estimation as the prior information for this symbol estimation in the application of multiple symbol estimations. The prior information is synthesized during the current symbol estimation to obtain the estimated value of the current symbol. After the method of the present invention is applied, since the prior information of the previous estimate is considered in the symbol estimation, the accuracy of the symbol estimation can be improved under the condition that the number of estimations is constant or reduced, and can be achieved in a shorter time. Higher accuracy, thereby increasing the correctness of data transmission in the network, and can reduce the time used for data transmission.

Description

A Sign Estimation Method

技术领域 technical field

本发明涉及无线通信技术，特别是涉及一种符号的估计方法。The present invention relates to wireless communication technology, in particular to a symbol estimation method.

背景技术 Background technique

多输入多输出(Multiple Input Multiple Output，MIMO)技术是无线通信领域的重大技术突破，它能够在不增加带宽的情况下成倍地提高通信系统的容量和频谱利用率。MIMO技术在发送端和接收端采用多天线(天线阵列)同时发送和接收信号。由于各发射天线同时发送的信号占用同一个频带，因而通信带宽并没有增加。每个发送天线和每个接收天线之间存在一个空间信道。如果每个空间信道的信道冲击响应独立，则MIMO系统通过多个发送天线和多个接收天线可以在发送端和接收端之间创建多个并行的独立的空间信道。通过这些并行的空间信道独立地传输信息，MIMO系统的传输数据率必然成倍增加。1998年G.J.Foschini和M.J.Gans在文献中充分论证了上述结论，并定量指出：假设MIMO系统有M根发送天线和N根接收天线，在窄带慢衰落信道下，就可以建立N×M阶信道矩阵。该矩阵的元素为独立同分布的复高斯随机变量。MIMO系统可以获得的信道容量将是单输入单输出(Single Input Single Output，SISO)系统的min(M，N)倍，且总的发射功率保持不变，此处min(M，N)表示取M和N中的最小值。Multiple Input Multiple Output (MIMO) technology is a major technological breakthrough in the field of wireless communication, which can double the capacity and spectrum utilization of the communication system without increasing the bandwidth. MIMO technology uses multiple antennas (antenna arrays) at the transmitter and receiver to simultaneously transmit and receive signals. Since the signals transmitted simultaneously by each transmit antenna occupy the same frequency band, the communication bandwidth does not increase. A spatial channel exists between each transmit antenna and each receive antenna. If the channel impulse response of each spatial channel is independent, the MIMO system can create multiple parallel independent spatial channels between the transmitting end and the receiving end through multiple transmitting antennas and multiple receiving antennas. By independently transmitting information through these parallel spatial channels, the transmission data rate of MIMO systems must be multiplied. In 1998, G.J.Foschini and M.J.Gans fully demonstrated the above conclusions in the literature, and quantitatively pointed out: Assuming that the MIMO system has M transmitting antennas and N receiving antennas, under the narrowband slow fading channel, an N×M order channel matrix can be established . The elements of this matrix are independent and identically distributed complex Gaussian random variables. The channel capacity that a MIMO system can obtain will be min(M, N) times that of a Single Input Single Output (SISO) system, and the total transmit power remains unchanged, where min(M, N) means taking The minimum of M and N.

鉴于MIMO系统的高频谱效率，MIMO系统的解调方法就成为研究的重点。自1996年以来，先后出现了许多MIMO系统的解调方法。其中，有一类并行垂直分层空时检测方法，这类方法具有如下特点：In view of the high spectral efficiency of the MIMO system, the demodulation method of the MIMO system becomes the focus of research. Since 1996, there have been many demodulation methods for MIMO systems. Among them, there is a class of parallel vertical layered space-time detection methods, which have the following characteristics:

采用多级并行干扰对消结构进行每个数据流的符号检测。在每一级并行干扰对消结构中，首先得到每个数据流中符号的决策统计量，然后由决策统计量得到符号的估计值。由符号的估计值可以得到该符号对其他数据流的干扰。对某一个数据流来讲，从接收信号中减掉其他数据流的符号的干扰，就得到该数据流的更新的接收信号。按照这种方式，在每级并行干扰对消结构中通过并行处理得到每个数据流的更新的接收信号。所有数据流的更新的接收信号就是下一级并行干扰对消结构的输入信号，用于下一级并行干扰对消结构中相应数据流的符号检测。由于更新的接收信号中其他数据流的干扰已经被消除，在下一级并行干扰对消结构中再一次进行该数据流符号的检测时，该数据流的符号的估计的性能就更好。Symbol detection for each data stream is performed using a multi-stage parallel interference cancellation structure. In each level of parallel interference cancellation structure, the decision statistic of symbols in each data stream is firstly obtained, and then the estimated value of the symbol is obtained from the decision statistic. From the estimated value of a symbol, the interference of the symbol to other data streams can be obtained. For a certain data stream, the received signal of the data stream is updated by subtracting the interference of symbols of other data streams from the received signal. In this way, an updated received signal for each data stream is obtained by parallel processing in each stage of the parallel interference cancellation structure. The updated received signals of all data streams are input signals of the next-level parallel interference cancellation structure, and are used for symbol detection of corresponding data streams in the next-level parallel interference cancellation structure. Since the interference of other data streams in the updated received signal has been eliminated, when the detection of the symbols of the data stream is performed again in the next-level parallel interference cancellation structure, the performance of estimating the symbols of the data stream will be better.

多用户检测是多个用户信号的联合检测，并行干扰对消多用户检测方法是一类重要的多用户检测方法。并行干扰对消多用户检测方法采用类似于上述MIMO系统的多级并行干扰对消结构进行多个用户信号的联合检测。在每级并行干扰对消结构中，首先得到每个用户发送符号的多径合并结果，然后由该符号的多径合并结果得到该符号的估计值。由符号的估计值可以得到该符号对其他用户的干扰。对某一个用户来讲，从接收信号减掉其他用户的符号的干扰，就得到该用户的更新的接收信号。由于更新的接收信号中其他用户的符号的干扰已经被消除，在下一级干扰对消结构中再一次进行该用户的符号检测时，该用户的符号的估计的性能就更好。Multi-user detection is the joint detection of multiple user signals, and the parallel interference cancellation multi-user detection method is an important class of multi-user detection methods. The parallel interference cancellation multi-user detection method adopts the multi-level parallel interference cancellation structure similar to the above-mentioned MIMO system to perform joint detection of multiple user signals. In each level of parallel interference cancellation structure, the multipath combination result of the symbol sent by each user is obtained first, and then the estimated value of the symbol is obtained from the multipath combination result of the symbol. The interference of the symbol to other users can be obtained from the estimated value of the symbol. For a certain user, the updated received signal of the user is obtained by subtracting the interference of symbols of other users from the received signal. Since the interference of other user's symbols in the updated received signal has been eliminated, when the user's symbol is detected again in the next-level interference cancellation structure, the performance of the user's symbol estimation will be better.

综上所述，在MIMO系统和多用户检测中，需要对MIMO系统的一个数据流或多用户检测系统的一个用户的符号进行不止一次的估计，每次符号估计都是在上一次干扰对消以后进行。To sum up, in the MIMO system and multi-user detection, it is necessary to estimate the symbols of a data stream of the MIMO system or a user of the multi-user detection system more than once, and each symbol estimation is based on the previous interference cancellation to be carried out later.

在现有技术中，通过如下方法对符号进行估计。In the prior art, symbols are estimated by the following method.

设被发送的符号为a，且a∈A＝{A₁，A₂，…，A_K}。A＝{A₁，A₂，…，A_K}是所有可能的发送符号构成的集合。Let the transmitted symbol be a, and a∈A={A ₁ , A ₂ , . . . , A _K }. A={A ₁ , A ₂ , . . . , A _K } is a set of all possible symbols to be sent.

在对符号a进行第i(i＝1，2，…，S，S是符号a被估计的次数)次估计时，符号a的观测信号为：When the ith (i=1, 2, ..., S, S is the number of times symbol a is estimated) estimation is performed on symbol a, the observed signal of symbol a is:

y_i＝a+v_i y _i =a+v _i

其中，v_i是高斯白噪声。Among them, v _i is Gaussian white noise.

在MIMO系统中观测信号y_i是符号a的决策统计量，在多用户检测系统中观测信号y_i是符号a的多径合并结果。In the MIMO system, the observed signal y _i is the decision statistic of the symbol a, and in the multi-user detection system, the observed signal y _i is the multipath combination result of the symbol a.

可以通过硬判决和软判决两种方式获得符号a的第i次估计值 The i-th estimated value of symbol a can be obtained in two ways: hard decision and soft decision

在硬判决方式下，满足下式：Under hard judgment, Satisfies the following formula:

${\overset{^^}{a a}}_{i i} = = arg arg \underset{{A A}_{k k} &Element; &Element; A A}{max max} ((P P (({A A}_{k k} | | {y the y}_{i i}))))$

$= = arg arg \underset{{A A}_{k k} &Element; &Element; A A}{max max} ((P P (({y the y}_{i i} | | {A A}_{k k})) P P (({A A}_{k k})))) - - - - - - ((11))$

其中，表达式 $\arg \max_{A_{k} &Element; A} (P (A_{k} | y_{i}))$ 的值是使P(A_k|y_i)最大的A_k，并将该A_k作为a_i的估计值

表达式

\arg \max_{A_{k} &Element; A} (P (y_{i} | A_{k}) P (A_{k}))

的值是使P(y_i|A_k)P(A_k)最大的A_k；在公式(1)中，表达式P(A|B)表示发生B时A发生的条件概率，所以P(A_k|y_i)表示接收端得到y_i时A_k的条件概率，即A_k的后验概率，P(y_i|A_k)表示发送符号为A_k时y_i的条件概率，P(A_k)表示发送符号为A_k的先验概率。where the expression

\arg \max_{A_{k} &Element; A} (P (A_{k} | {the y}_{i}))

The value of is A _k that maximizes P(A _k |y _i ), and uses this A _k as the estimated value of a _i

expression

\arg \max_{A_{k} &Element; A} (P ({the y}_{i} | A_{k}) P (A_{k}))

The value of is A _k that maximizes P(y _i |A _k )P(A _k ); in formula (1), the expression P(A|B) represents the conditional probability of A happening when B occurs, so P( A _k |y _i ) indicates the conditional probability of A _k when the receiver gets y _i , that is, the posterior probability of A _k , P(y _i |A _k ) indicates the conditional probability of y _i when the transmitted symbol is A _k , P( A _k ) represents the prior probability that the transmitted symbol is A _k .

假设 $p (A_{k}) = \frac{1}{K}, k = 1,2, \cdot \cdot \cdot, K,$ 并经过公式推导，可以得到如下结果：suppose $p (A_{k}) = \frac{1}{K}, k = 1,2, \cdot \cdot \cdot, K,$ And after formula derivation, the following results can be obtained:

${\overset{^^}{a a}}_{i i} = = arg arg \underset{{A A}_{k k} &Element; &Element; A A}{min min} {| | | | {y the y}_{i i} - - {A A}_{k k} | | | |}^{22} - - - - - - ((11 a a))$

的值是使‖y_i-A_k‖²最小的A_k。例如，如果‖y_i-A₂‖²最小，那么的值为A₂。

The value of is A _k that minimizes ‖y _i -A _k ^‖2 . For example, if ‖y _i -A ₂ ‖ ² is the smallest, then The value of is A ₂ .

硬判决方式实现比较简单，计算量小，但估计的准确率低。The implementation of the hard decision method is relatively simple, and the amount of calculation is small, but the accuracy of the estimation is low.

在软判决方式下，

满足下式：Under soft judgment,

Satisfies the following formula:

${\overset{^^}{a a}}_{i i} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} P P (({A A}_{k k} | | {y the y}_{i i}))}{{Σ Σ}_{k k = = 11}^{K K} P P (({A A}_{k k} | | {y the y}_{i i}))} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} P P (({y the y}_{i i} | | {A A}_{k k})) P P (({A A}_{k k}))}{{Σ Σ}_{k k = = 11}^{K K} P P (({y the y}_{i i} | | {A A}_{k k} | |)) P P (({A A}_{k k}))} - - - - - - ((22))$

假设 $p (A_{k}) = \frac{1}{K}, k = 1,2, \cdot \cdot \cdot, K,$ 并经过公式推导，可得到如下结果：suppose $p (A_{k}) = \frac{1}{K}, k = 1,2, &Center Dot; &Center Dot; &Center Dot;, K,$ And after formula derivation, the following results can be obtained:

${\overset{^^}{a a}}_{i i} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} f f (({y the y}_{i i} | | {A A}_{k k}))}{{Σ Σ}_{k k = = 11}^{K K} f f (({y the y}_{i i} | | {A A}_{k k}))} - - - - - - ((22 a a))$

其中，β_i称为校正因子，用以校正信道估计不理想造成的符号估计的偏差和干扰对消的偏差；f(y_i|A_k)表示发送符号为A_k时接收信号y_i的概率密度函数。Among them, β _i is called the correction factor, which is used to correct the deviation of symbol estimation and interference cancellation caused by the unsatisfactory channel estimation; f(y _i |A _k ) represents the probability of receiving signal y _i when the transmitted symbol is A _k density function.

若发送的符号是复数，则：If the symbol sent is complex, then:

$f f (({y the y}_{i i} | | {A A}_{k k})) = = \frac{11}{{22 πσ πσ}_{i i}^{22}} {e e}^{\frac{{(({YR YR}_{i i} - - {AR AR}_{k k}))}^{22} + + {(({YI YI}_{i i} - - {AI AI}_{k k}))}^{22}}{{22 σ σ}_{i i}^{22}}} - - - - - - ((22 b b))$

其中，AR_k、AI_k分别为A_k的实部和虚部；YR_i、YI_i分别为y_i的实部和虚部。Among them, AR _k and AI _k are the real part and imaginary part of A _k respectively; YR _i and YI _i are the real part and imaginary part of y _i respectively.

由于高斯白噪声v_i的实部VR_i和虚部VI_i分别服从正态分布，即N(0，σ_i ²)分布，所以可用如下方法确定σ_i：分别求出VR_i和VI_i的方差，再将两个方差值平均得到σ_i，或者求复噪声的方差，再除以二得到σ_i。Since the real part VR _i and the imaginary part VI _i of Gaussian white noise v _i respectively obey the normal distribution, that is, the N(0, σ _i ² ) distribution, the following method can be used to determine σ _i : Calculate the VR _i and VI _i respectively variance, and then average the two variance values to get σ _i , or find the variance of the complex noise and divide by two to get σ _i .

当发送的符号是复数时，公式(2)中的β_i也是复数。When the transmitted symbol is a complex number, β _i in formula (2) is also a complex number.

若发送的符号是实数，则：If the sent symbol is a real number, then:

$f f (({y the y}_{i i} | | {A A}_{k k})) = = \frac{11}{\sqrt{22 π π} {σ σ}_{i i}} {e e}^{\frac{{(({YR YR}_{i i} - - {AR AR}_{k k}))}^{22}}{{22 σ σ}_{i i}^{22}}} - - - - - - ((22 c c))$

其中，AR_k为A_k的数值，是实数；YR_i为y_i的实部，通常，如果A_k是实数，则y_i也是实数。Wherein, AR _k is the numerical value of A _k , which is a real number; YR _i is the real part of y _i , usually, if A _k is a real number, then y _i is also a real number.

如果A_k为实数，那么在确定σ_i时将高斯白噪声v_i看作实数，即只考虑v_i的实部VR_i。VR_i服从正态分布分布，即N(0，σ_i ²)分布，所以求出VR_i的方差即可得到σ_i。If A _k is a real number, then the Gaussian white noise v _i is regarded as a real number when determining σ _i , that is, only the real part VR _i of v _i is considered. VR _i obeys normal distribution, that is, N(0, σ _i ² ) distribution, so σ _i can be obtained by calculating the variance of VR _i .

当发送的符号是实数时，公式(2)中的β_i也是实数。When the transmitted symbol is a real number, β _i in formula (2) is also a real number.

下面说明β_i的取值，β_i的取值与符号的信噪比(Signal-Noise Rate，SNR)密切相关，是符号的SNR的函数。The value of β _i will be described below. The value of β _i is closely related to the signal-noise ratio (Signal-Noise Rate, SNR) of the symbol, and is a function of the SNR of the symbol.

通常，在接收端，符号的SNR要达到一定的数值，才能使解映射和译码后的比特误块率(Block Error Rate，BLER)的性能满足业务质量的要求。对于给定的BLER数值，可以通过仿真确定：为使译码以后BLER性能达到要求，符号所需要达到的最低SNR数值S_MIN。并根据具体情况设置两个SNR区间分界点，分别为T₁＝S_MIN-δ₁，T₂＝S_MIN-δ₂。其中，δ₁＞0，δ₂＞δ₁。Usually, at the receiving end, the SNR of the symbol must reach a certain value, so that the performance of the Block Error Rate (Block Error Rate, BLER) after demapping and decoding can meet the requirement of service quality. For a given BLER value, it can be determined through simulation: in order to make the BLER performance meet the requirements after decoding, the minimum SNR value S _MIN that the symbol needs to achieve. And set two SNR interval boundary points according to specific conditions, which are respectively T ₁ =S _MIN -δ ₁ and T ₂ =S _MIN -δ ₂ . Wherein, δ ₁ >0, δ ₂ >δ ₁ .

如果符号的SNR大于等于阈值T₁＝S_MIN-δ₁，则认为该符号的信道估计比较准确，可以近似认为β_i＝1，也就是说，在干扰对消中不考虑信道估计的偏差对干扰对消造成的影响。If the SNR of the symbol is greater than or equal to the threshold T ₁ =S _MIN -δ ₁ , the channel estimation of the symbol is considered to be relatively accurate, and it can be approximately considered that β _i =1, that is, the deviation of the channel estimation is not considered in interference cancellation. The impact of interference cancellation.

如果符号的SNR小于等于阈值T₂＝S_MIN-δ₂，就认为该符号的SNR太低，使信道估计非常差，可以近似认为β_i＝0，也就是说，由于信道质量太差，所以本次估计值为0，由于估计值为0属于无效的估计值，所以该估计结果不参加干扰对消处理。If the SNR of the symbol is less than or equal to the threshold T ₂ =S _MIN -δ ₂ , it is considered that the SNR of the symbol is too low, making the channel estimation very poor, and it can be approximately considered that β _i =0, that is, because the channel quality is too poor, so This time the estimated value is 0, since the estimated value of 0 is an invalid estimated value, the estimated result does not participate in the interference cancellation process.

如果符号的SNR大于阈值T₂＝S_MIN-δ₂且小于阈值T₁＝S_MIN-δ₁，可以通过COSSAP仿真或MATLAB仿真优化确定在SNR区间(T₂，T₁)下β_i的具体取值α_i。需要注意的是，δ₁、δ₂取值直接影响β_i对符号估计偏差的校正的精度和SNR优化区间的大小。可以根据对校正精度和优化计算量的需要确定。If the SNR of the symbol is greater than the threshold T ₂ =S _MIN -δ ₂ and less than the threshold T ₁ =S _MIN -δ ₁ , the specific value of β _i in the SNR interval (T ₂ , T ₁ ) can be determined through COSSAP simulation or MATLAB simulation optimization. Take the value α _i . It should be noted that the values of δ ₁ and δ ₂ directly affect the accuracy of correction of β _i to the symbol estimation deviation and the size of the SNR optimization interval. It can be determined according to the need for correction accuracy and optimization calculation amount.

综上所述，β_i取值如下：To sum up, the value of β _i is as follows:

${β β}_{i i} = = \{\begin{matrix} 00,, & SNR SNR \leq \leq {T T}_{22} \\ {α α}_{i i},, & {T T}_{22} < < SNR SNR < < {T T}_{11} \\ 11,, & SNR SNR &GreaterEqual; &Greater Equal; {T T}_{11} \end{matrix} - - - - - - ((22 d d))$

SNR在大于T2且小于T1的区间，也可以将该区间分为若干子区间，用[SNR_i-1，SNR_i]表示，其中 ${SNR}_{i} = T_{2} + \frac{i (T_{1} - T_{2})}{I}, i = 1,2, \cdot \cdot \cdot, I,$ I是根据需要确定的SNR优化区间(T₂，T₁)的个数，在划分子区间的情况下，在每个子区间[SNR_i-1，SNR_i]内分别通过COSSAP仿真或MATLAB仿真优化得到各自区间的α_i值。If the SNR is in the interval greater than T2 and less than T1, the interval can also be divided into several sub-intervals, represented by [SNR _i-1 , SNR _i ], where ${SNR}_{i} = T_{2} + \frac{i (T_{1} - T_{2})}{I}, i = 1,2, \cdot \cdot \cdot, I,$ I is the number of SNR optimization intervals (T ₂ , T ₁ ) determined according to the needs. In the case of dividing sub-intervals, each sub-interval [SNR _i-1 , SNR _i ] is optimized by COSSAP simulation or MATLAB simulation Get the α _i values of the respective intervals.

从以上的分析可以看出，以硬判决方式估计符号实现比较简单，计算量分别通过COSSAP仿真或MATLAB仿真优化得到各自区间的α_i值。From the above analysis, it can be seen that it is relatively simple to estimate the symbol by hard decision, and the calculation amount is optimized by COSSAP simulation or MATLAB simulation to obtain the α _i value of each interval.

从以上的分析可以看出，以硬判决方式估计符号实现比较简单，计算量小；以软判决方式估计符号实现比较复杂，计算量大。但是在实际应用中，软判决方式估计符号的准确率比硬判决方式估计符号的准确率高。From the above analysis, it can be seen that estimating symbols by hard decision is relatively simple and requires less calculation; while estimating symbols by soft decision is more complicated and requires a lot of calculation. But in practical application, the accuracy rate of estimating symbols by soft decision method is higher than that by hard decision method.

上述符号的估计的硬判决公式和软判决公式是在假设所有可能的符号被发送的先验概率相等的条件下求得的，因为在公式(1a)和公式(2a)中，没有体现出所有可能的符号被发送的先验概率，也就是说，所有可能的符号被发送的概率相同。当没有任何先验信息时，假设符号a取所有可能值的概率相等，在这种假设下进行符号估计是符号估计的最不利的情况。如果可以获得符号a取值的一些先验信息，就可以利用先验信息对符号a进行更准确地估计，不仅估计的准确率较高，也可以减少符号估计的次数或级数。The hard-decision formula and soft-decision formula for the estimation of the above symbols are obtained under the assumption that all possible symbols are transmitted with equal prior probabilities, because in formula (1a) and formula (2a), not all The prior probability of possible symbols being sent, that is, all possible symbols are sent with equal probability. Symbol estimation under this assumption is the most unfavorable case of symbol estimation when there is no prior information, assuming that the symbol a takes all possible values with equal probability. If some prior information about the value of the symbol a can be obtained, the symbol a can be estimated more accurately by using the prior information. Not only the estimation accuracy is higher, but also the number or series of symbol estimation can be reduced.

发明内容 Contents of the invention

本发明的主要目的在于提供一种符号的估计方法，在进行多次符号估计时提高符号估计的准确度。The main purpose of the present invention is to provide a method for estimating symbols, which improves the accuracy of symbol estimation when multiple symbol estimations are performed.

本发明的目的是通过如下技术方案实现的：The purpose of the present invention is achieved through the following technical solutions:

一种符号的估计方法，包括如下步骤：A symbol estimation method, comprising the steps of:

A、根据符号的观测信号和符号可能性集合中的所有符号的先验概率对符号进行估计，得到本次估计值，同时确定本次估计中所有符号的本次后验概率；A. Estimate the symbols according to the observed signals of the symbols and the prior probabilities of all the symbols in the symbol possibility set to obtain the current estimated value, and at the same time determine the current posterior probabilities of all the symbols in the current estimation;

B、判断是否达到预定的估计次数，如果是，输出本次符号估计的估计值，然后结束；否则，将所有符号的本次后验概率作为下一次符号估计的先验概率，然后返回步骤A进行下一次符号估计。B. Judging whether the predetermined number of estimates is reached, if yes, output the estimated value of this symbol estimation, and then end; otherwise, use the current posterior probability of all symbols as the prior probability of the next symbol estimation, and then return to step A Do the next symbol estimate.

所述符号的观测信号为符号的决策统计量或多径合并结果。The observed signal of the symbol is a decision statistic or a multipath combination result of the symbol.

在第一次估计时，设置符号可能性集合中的所有符号的先验概率相同。At the first estimate, all symbols in the set of symbol likelihoods are set to have the same prior probability.

所述预定的估计次数为2到4次。The predetermined number of estimations is 2 to 4 times.

步骤A所述根据符号的观测信号和符号可能性集合中的所有符号的先验概率对符号进行估计的方法是：The method for estimating symbols according to the observed signals of symbols and the prior probabilities of all symbols in the symbol possibility set described in step A is:

采用如下公式获得第i次估计的估计值：The estimated value of the i-th estimate is obtained using the following formula:

${\overset{^^}{a a}}_{i i} = = arg arg \underset{{A A}_{k k} &Element; &Element; {A A}^{' '}}{min min} (({Σ Σ}_{j j = = 11}^{i i} \frac{{| | | | {y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{σ σ}_{j j}^{22}}))$

其中，i为估计的次数，j的值为1到i，为第i次估计的估计值，y_j为第j次估计的决策统计量，其值为符号值与高斯白噪声之和，A′为符号可能性集合，A_k为符号可能性集合中的符号，σ_j ²是高斯白噪声的方差。Among them, i is the number of estimates, and the value of j is 1 to i, is the estimated value of the i-th estimate, y _j is the decision statistic of the j-th estimate, its value is the sum of the symbol value and Gaussian white noise, A' is the symbol possibility set, A _k is the symbol possibility set Symbol, σ _j ² is the variance of Gaussian white noise.

${\overset{^^}{a a}}_{i i} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} \cdot \cdot exp exp ((- - {Σ Σ}_{j j = = 11}^{i i} \frac{| | | | {{y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{22 σ σ}_{j j}^{22}}))}{{Σ Σ}_{k k = = 11}^{K K} exp exp ((- - {Σ Σ}_{j j = = 11}^{i i} \frac{{| | | | {y the y}_{i i} - - {A A}_{k k} | | | |}^{22}}{{22 σ σ}_{j j}^{22}}))}$

其中，i为估计的次数，j的值为1到i，

为第i次估计的估计值，y_j为第j次估计的决策统计量，其值为符号值与高斯白噪声之和，A_k为符号可能性集合中的符号，σ_j ²是高斯白噪声的方差，β_i为由传输信道的信号噪声比SNR确定的信道校正因子。Among them, i is the number of estimates, and the value of j is 1 to i,

is the estimated value of the i-th estimate, y _j is the decision statistic of the j-th estimate, its value is the sum of the symbol value and Gaussian white noise, A _k is the symbol in the symbol possibility set, σ _j ² is Gaussian white The variance of the noise, β _i is the channel correction factor determined by the signal-to-noise ratio SNR of the transmission channel.

所述由传输信道的SNR确定信道校正因子β_i是在SNR区间中通过COSSAP仿真或MATLAB仿真确定的。The channel correction factor β _i determined by the SNR of the transmission channel is determined through COSSAP simulation or MATLAB simulation in the SNR interval.

所述高斯白噪声的方差σ_j ²为：当高斯白噪声是复数时，σ_j ²为高斯白噪声的实部的方差和虚部的方差的平均值；当高斯白噪声是实数时，σ_j ²为高斯白噪声的实部的方差。The variance σ _j of the Gaussian white noise is: when the Gaussian white noise is a complex number, _σj ² is the mean value of ^the variance of the real part and the variance of the imaginary part of the Gaussian white noise; when the Gaussian white noise is a real number, σj _j ² is the variance of the real part of Gaussian white noise.

该方法用于多输入多输出系统或多用户检测的并行干扰对消技术中。The method is used in the parallel interference canceling technology of multiple-input multiple-output system or multi-user detection.

本发明提供了一种符号的估计方法，该方法是根据符号的决策统计量或符号的多径合并结果和符号可能性集合中的所有符号的先验概率对符号进行估计，得到本次估计值，并确定本次估计中所有符号的本次后验概率，然后将所有符号的本次后验概率作为下一次估计的先验概率，代入下一次估计中。现有技术的方法在每次估计时都假设所有符号的先验概率均相同，仅根据符号的决策统计量或多径合并结果来估计符号。从本发明和现有技术的对比来看，本发明的方法由于将前次估计所得的所有符号的后验概率作为本次估计的所有符号的先验概率，显著提高了符号估计的准确性，同时，在达到同样的估计准确度的前提下，可以显著减少多次符号估计所用的时间，从而增加了网络中数据传输的正确性，并可减少传输数据所需的时间。The present invention provides a symbol estimation method, which is to estimate the symbol according to the decision statistics of the symbol or the multipath combination result of the symbol and the prior probability of all symbols in the symbol possibility set, and obtain the current estimated value , and determine the current posterior probabilities of all symbols in this estimation, and then use the current posterior probabilities of all symbols as the prior probability of the next estimation, and substitute them into the next estimation. The method in the prior art assumes that the prior probability of all symbols is the same at each estimation, and only estimates the symbols according to the decision statistics of the symbols or the result of multipath combination. From the comparison between the present invention and the prior art, the method of the present invention significantly improves the accuracy of symbol estimation because the posterior probability of all symbols obtained in the previous estimation is used as the prior probability of all symbols estimated this time. At the same time, under the premise of achieving the same estimation accuracy, the time used for multiple symbol estimation can be significantly reduced, thereby increasing the correctness of data transmission in the network and reducing the time required for data transmission.

附图说明 Description of drawings

图1是根据本发明的符号的估计方法流程图。Fig. 1 is a flow chart of a symbol estimation method according to the present invention.

具体实施方式 Detailed ways

为了使本发明的目的、技术方案和优点更清楚，下面结合附图和具体实施方式对本发明作进一步描述。In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

本发明的符号的估计方法是在硬判决和软判决中，考虑前次估计的结果作为先验信息。The symbol estimation method of the present invention considers the result of previous estimation as prior information in hard decision and soft decision.

图1是根据本发明的符号的估计方法流程图，从图1中可以看出，该方法包括如下步骤：Fig. 1 is the estimation method flow chart of symbol according to the present invention, as can be seen from Fig. 1, this method comprises the following steps:

步骤101：设置估计次数值S，设置估计次数变量i＝1。Step 101: Set the value S of the number of estimates, and set the variable i=1 for the number of estimates.

步骤102：设置符号可能性集合中的所有符号的第i次先验概率相同。Step 102: Set the i-th prior probability of all symbols in the symbol possibility set to be the same.

步骤103：根据符号的第i次观测信号和第i次先验概率对符号进行第i次估计，得到第i次估计值，同时确定第i次估计中所有符号的第i次后验概率。Step 103: Estimate the symbol for the i-th time according to the i-th observed signal and the i-th prior probability of the symbol to obtain the i-th estimated value, and determine the i-th posterior probability of all symbols in the i-th estimate.

步骤104：设置i＝i+1。Step 104: Set i=i+1.

步骤105：判断i是否大于S，如果是，转步骤107；如果不是，转到步Step 105: judge whether i is greater than S, if yes, go to step 107; if not, go to step

步骤106：将所有符号的第(i-1)次后验概率作为第i次估计的先验概率，然后返回步骤103。Step 106: Use the (i-1)th posterior probability of all symbols as the ith estimated prior probability, and then return to step 103 .

步骤107：将第(i-1)次符号估计值作为该符号的检测结果输出，结束符号估计过程。Step 107: output the (i-1)th symbol estimation value as the detection result of the symbol, and end the symbol estimation process.

优选地，符号估计次数S为2到4时就已经可以获得很好的符号估计性能，再增加S的数值，符号估计的性能也不会显著提高。Preferably, good symbol estimation performance can be obtained when the number of symbol estimation times S is 2 to 4, and if the value of S is increased, the performance of symbol estimation will not be significantly improved.

下面说明如何具体应用本发明的方法。和现有技术一样，本发明的方法也可采取硬判决和软判决两种判决方式。How to specifically apply the method of the present invention is described below. Like the prior art, the method of the present invention can also adopt two judgment modes of hard judgment and soft judgment.

y_i＝a+v_i y _i =a+v _i

其中，v_i是高斯白噪声。Among them, v _i is Gaussian white noise.

在MIMO系统中，观测信号y_i是符号a的决策统计量，在多用户检测系统中，观测信号y_i是符号a的多径合并结果。In the MIMO system, the observed signal y _i is the decision statistic of the symbol a, and in the multi-user detection system, the observed signal y _i is the multipath combination result of the symbol a.

在硬判决方式下，符号a的第i(i＝1，2，…，S，S是符号估计的次数)次估计值

满足下式：In the hard decision mode, the ith (i=1, 2, ..., S, S is the number of times of symbol estimation) estimated value of symbol a

Satisfies the following formula:

$= = arg arg \underset{{A A}_{k k} &Element; &Element; A A}{max max} ((P P (({y the y}_{i i} | | {A A}_{k k})) P P (({A A}_{k k} | | {y the y}_{i i - - 11})))) - - - - - - ((33))$

从公式(3)可以看出，用P(A_k|y_i-1)代替了公式(1)中的P(A_k))，即，公式(3)考虑了上一次符号估计时所得到的发送符号为A_k的后验概率。第一次估计符号时，仍然假设发送符号为A_k的先验概率相等，即为，仍然按照公式(1a)进行符号估计；在第二次及以后的符号估计时，把前一次符号估计中计算得到的发送符号为A_k的概率，即后验概率，作为本次估计的先验信息。It can be seen from formula (3) that P(A _k |y _i-1 ) is used instead of P(A _k )) in formula (1), that is, formula (3) takes into account the obtained The sent symbol of is the posterior probability of A _k . When estimating the symbol for the first time, it is still assumed that the prior probability of the transmitted symbol being A _k is equal, that is, , still perform symbol estimation according to formula (1a); in the second and subsequent symbol estimations, the probability that the transmitted symbol calculated in the previous symbol estimation is A _k , that is, the posterior probability, is used as the prior estimation of this estimation test information.

对公式(3)进一步推导，可得第i次符号估计时的估计值为：Further deriving the formula (3), the estimated value of the i-th symbol estimation can be obtained as:

${\overset{^^}{a a}}_{i i} = = arg arg \underset{{A A}_{k k} &Element; &Element; A A}{min min} ((Sum Sum ((i i,, {A A}_{k k})))) - - - - - - ((33 a a))$

其中，in,

$Sum Sum ((i i,, {A A}_{k k})) = = Sum Sum ((i i - - 11,, {A A}_{k k})) + + \frac{{| | | | {y the y}_{i i} - - {A A}_{k k} | | | |}^{22}}{{σ σ}_{i i}^{22}} - - - - - - ((33 b b))$

$Sum Sum ((i i - - 11,, {A A}_{k k})) = = {Σ Σ}_{j j = = 11}^{i i - - 11} \frac{{| | | | {y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{σ σ}_{j j}^{22}} - - - - - - ((33 c c))$

将公式(3b)和公式(3c)代入公式(3a)，可得到：Substituting formula (3b) and formula (3c) into formula (3a), we can get:

${\overset{^^}{a a}}_{i i} = = arg arg \underset{{A A}_{k k} &Element; &Element; A A}{min min} (({Σ Σ}_{j j = = 11}^{i i} \frac{{| | | | {y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{σ σ}_{j j}^{22}})) - - - - - - ((33 d d))$

在第(i-1)次符号估计时，计算得到Sum(i-1，A_k)，k＝1，2，…，K。在第i次符号估计时，直接用前一次符号估计时计算得到的Sum(i-1，A_k)，k＝1，2，…，K，按照公式(3b)计算得到Sum(i，A_k)，k＝1，2，…，K，然后按照公式(3a)求出符号a的估计

。其中，Sum(i，A_k)，k＝1，2，…，K在第(i+1)次符号估计时还要使用。其中，公式(3b)中Sum(i-1，A_k)，k＝1，2，…，K是第(i-1)次估计中后验概率信息在第i次符号估计中的具体体现。At the (i-1)th symbol estimation, Sum(i-1, A _k ), k=1, 2, . . . , K is calculated. In the i-time symbol estimation, directly use the Sum(i-1, A _k ) calculated in the previous symbol estimation, k=1, 2, ..., K, and calculate Sum(i, A k ) according to the formula (3b) _k ), k=1, 2, ..., K, then find the estimate of symbol a according to formula (3a)

. Wherein, Sum(i, A _k ), k=1, 2, . . . , K is also used in the (i+1)th symbol estimation. Among them, Sum(i-1, A _k ) in formula (3b), k=1, 2, ..., K is the specific embodiment of the posterior probability information in the i-th symbol estimation in the (i-1)th estimation .

在软判决方式下，

满足下式：Under soft judgment,

Satisfies the following formula:

${\overset{^^}{a a}}_{i i} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} P P (({A A}_{k k} | | {y the y}_{i i}))}{{Σ Σ}_{k k = = 11}^{K K} P P (({A A}_{k k} | | {y the y}_{i i}))} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} P P (({y the y}_{i i} | | {A A}_{k k})) P P (({A A}_{k k} | | {y the y}_{i i - - 11}))}{{Σ Σ}_{k k = = 11}^{K K} P P (({y the y}_{i i} | | {A A}_{k k} | |)) P P (({A A}_{k k} | | {y the y}_{i i - - 11}))} - - - - - - ((44))$

从公式(4)可以看出，用P(A_k|y_i-1)代替了公式(2)中的P(A_k)，即，公式(4)考虑了上一次符号估计时所得到的发送符号为A_k的后验概率。第一次估计符号时，仍然假设发送符号为A_k的概率相等，即为

，并按照公式(2a)计算符号估计值；在第二次及以后的符号估计时，把前一次估计所得到的发送符号为A_k的概率，即后验概率，作为本次估计的先验信息。It can be seen from formula (4) that P(A _k ) in formula (2) is replaced by P(A _k |y _i-1 ), that is, formula (4) takes into account the obtained The posterior probability that the transmitted symbol is A _k . When estimating the symbol for the first time, it is still assumed that the probability of the transmitted symbol being A _k is equal, that is,

, and calculate the symbol estimation value according to the formula (2a); in the second and subsequent symbol estimation, the probability that the transmitted symbol obtained in the previous estimation is A _k , that is, the posterior probability, is used as the prior estimation of this estimation information.

对公式(4)进一步推导，可得第i次符号估计时的估计值为：Further deriving the formula (4), the estimated value of the i-th symbol estimation can be obtained as:

${\overset{^^}{a a}}_{i i} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} Sum Sum ((i i {,, A A}_{k k}))}{{Σ Σ}_{k k = = 11}^{K K} Sum Sum ((i i {,, A A}_{k k}))} - - - - - - ((44 a a))$

其中，in,

$Sum Sum ((i i,, {A A}_{k k})) = = Sum Sum ((i i - - 11,, {A A}_{k k})) exp exp ((\frac{{| | | | {y the y}_{i i} - - {A A}_{k k} | | | |}^{22}}{{22 σ σ}_{i i}^{22}})) - - - - - - ((44 b b))$

$Sum Sum ((i i - - 11,, {A A}_{k k})) = = exp exp ((- - {Σ Σ}_{j j = = 11}^{i i - - 11} \frac{{| | | | {y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{22 σ σ}_{j j}^{22}})) - - - - - - ((44 c c))$

将公式(4b)和公式(4c)代入公式(4a)，可得到：Substituting formula (4b) and formula (4c) into formula (4a), we can get:

${\overset{^^}{a a}}_{i i} = = {β β}_{i i} \frac{{Σ Σ}_{k k = = 11}^{K K} {A A}_{k k} \cdot &Center Dot; exp exp ((- - {Σ Σ}_{j j = = 11}^{i i} \frac{| | | | {{y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{22 σ σ}_{j j}^{22}}))}{{Σ Σ}_{k k = = 11}^{K K} exp exp ((- - {Σ Σ}_{j j = = 11}^{i i} \frac{{| | | | {y the y}_{j j} - - {A A}_{k k} | | | |}^{22}}{{22 σ σ}_{j j}^{22}}))} - - - - - - ((44 d d))$

在第(i-1)次符号估计时，计算得到Sum(i-1，A_k)，k＝1，2，…，K。在第i次符号估计时，直接用前一次符号估计时计算得到的Sum(i-1，A_k)，k＝1，2，…，K，按照公式(4b)计算得到Sum(i，A_k)，k＝1，2，…，K，然后按照公式(4a)求出符号a的估计

。其中，Sum(i，A_k)，k＝1，2，…，K在第(i+1)次符号估计时还要使用。At the (i-1)th symbol estimation, Sum(i-1, A _k ), k=1, 2, . . . , K is calculated. In the i-time symbol estimation, directly use the Sum(i-1, A _k ) calculated during the previous symbol estimation, k=1, 2,..., K, and calculate Sum(i, A k ) according to the formula (4b) _k ), k=1, 2, ..., K, then find the estimate of symbol a according to formula (4a)

. Wherein, Sum(i, A _k ), k=1, 2, . . . , K is also used in the (i+1)th symbol estimation.

从公式(3)和公式(4)可以看出，每次符号估计都加入上一次符号估计的信息作为本次符号估计的先验信息。其中，公式(4b)中Sum(i-1，A_k)，k＝1，2，…，K是第(i-1)次估计中后验概率信息在第i次符号估计中的具体体现。It can be seen from formula (3) and formula (4) that the information of the previous symbol estimation is added to each symbol estimation as the prior information of the symbol estimation this time. Among them, Sum(i-1, A _k ) in formula (4b), k=1, 2, ..., K is the specific embodiment of the posterior probability information in the i-th symbol estimation in the (i-1)th estimation .

在公式(4a)中，β_i称为校正因子，用以校正信道估计不理想造成的符号估计的偏差和干扰对消的偏差，在本发明的方法中，β_i值的确定和现有技术的方法一样，在此不予以说明。In formula (4a), β _i is called a correction factor, which is used to correct the deviation of symbol estimation and the deviation of interference cancellation caused by unsatisfactory channel estimation. In the method of the present invention, the determination of β _i value is the same as that of the prior art The method is the same and will not be described here.

在具体的实施过程中可对根据本发明的方法进行适当的改进，以适应具体情况的具体需要。因此可以理解，根据本发明的具体实施方式只是起示范作用，并不用以限制本发明的保护范围，例如，本发明不局限于在MIMO系统或多用户检测系统中应用，还可以适用于一切需要进行多级或多次符号估计的系统或方法。Appropriate improvements can be made to the method according to the present invention in the specific implementation process to meet the specific needs of specific situations. Therefore, it can be understood that the specific implementation manners according to the present invention are only exemplary, and are not intended to limit the protection scope of the present invention. For example, the present invention is not limited to applications in MIMO systems or multi-user detection systems, and can also be applied to all needs A system or method for performing multi-level or multiple symbol estimation.

Claims

1. A method for estimating symbols, the method comprising the steps of:

A. estimating the symbol according to the observation signal of the symbol and the prior probability of all the symbols in the symbol possibility set to obtain the estimation value of this time, and simultaneously determining the posterior probability of this time of all the symbols in the estimation;

B. judging whether the preset estimation times are reached, if so, outputting the estimation value of the symbol estimation, and then ending; otherwise, the posterior probability of the time of all the symbols is used as the prior probability of the next symbol estimation, and then the step A is returned to carry out the next symbol estimation.

2. The method of estimating symbols of claim 1, wherein the observed signal of the symbol is a decision statistic of the symbol or a multipath combining result.

3. The method of estimating symbols according to claim 1, wherein the prior probabilities of all symbols in the symbol likelihood set are set to be the same at the time of the first estimation.

4. The symbol estimation method according to claim 1, wherein the predetermined number of estimations is 2 to 4.

5. The method for estimating symbols according to claim 1, wherein the method for estimating symbols according to the observed signal of symbols and the prior probabilities of all symbols in the symbol possibility set in step a is:

obtaining an estimated value of the ith estimation by adopting the following formula:

where i is the estimated degree and j has a value of 1 to i _i Is an estimated value of the i-th estimation, y _j The decision statistic for the j-th estimation is the sum of the symbol value and white Gaussian noise, A' is the symbol probability set, A _k For symbols in the set of symbol possibilities, σ _j ² Is the variance of gaussian white noise.

6. The method of estimating symbols according to claim 1, wherein the method of estimating symbols according to their observed signals and prior probabilities of all symbols in the symbol likelihood set in step a is:

where i is the estimated degree and j has a value of 1 to i _i Is an estimate of the i-th estimate, y _j The decision statistic for the j-th estimate, whose value is the sum of the symbol value and Gaussian white noise, A _k For symbols in a set of symbol possibilities, σ _j ² Is the variance of Gaussian white noise, beta _i Is a channel correction factor determined by the signal-to-noise ratio SNR of the transmission channel.

7. Method for estimating symbols according to claim 6, characterized in that said channel correction factor β is determined from the SNR of the transmission channel _i Is determined by cossa simulation or MATLAB simulation in the SNR interval.

8. Method for estimating symbols according to claim 5 or 6, characterized in that the variance σ of said Gaussian white noise _j ² Comprises the following steps: when Gaussian white noise is complex, σ _j ² Is the average of the variance of the real part and the variance of the imaginary part of the Gaussian white noise; when Gaussian white noise is a real number, σ _j ² Is the variance of the real part of gaussian white noise.

9. The method of estimating symbols of claim 1, wherein the method is used in a parallel interference cancellation technique for multi-input multi-output systems or multi-user detection.