CN103441963B - The optimum decision delay time search method of least mean-square error decision feedback equalization system - Google Patents
- ️Wed Aug 10 2016
Info
-
Publication number
- CN103441963B CN103441963B CN201310349561.8A CN201310349561A CN103441963B CN 103441963 B CN103441963 B CN 103441963B CN 201310349561 A CN201310349561 A CN 201310349561A CN 103441963 B CN103441963 B CN 103441963B Authority
- CN
- China Prior art keywords
- mse
- mid
- delay
- decision
- search method Prior art date
- 2013-08-12 Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims abstract description 7
- 230000001934 delay Effects 0.000 claims description 6
- 238000004891 communication Methods 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000000342 Monte Carlo simulation Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Landscapes
- Noise Elimination (AREA)
Abstract
本发明涉及数字通信技术领域，尤其涉及一种最小均方误差判决反馈均衡系统的最优判决时延搜索方法，在训练阶段，接收端首先根据导频信号获得信道冲击响应和噪声方差信息，然后接收端按一定算法搜索得到最优判决时延，从而设置系统的滤波器系数，之后系统进入发送数据阶段。本发明当判决反馈滤波器长度小于信道阶数时，为MMSE‑DFE系统提供了一种最优判决时延的搜索方法，该搜索方法可以得到最优或者次优的判决时延，与遍历搜索方法取得的系统均方误差性能相近，但是显著的降低了运算复杂度。
The present invention relates to the technical field of digital communication, in particular to a minimum mean square error decision feedback equalization system optimal decision delay search method, in the training phase, the receiving end first obtains the channel impulse response and noise variance information according to the pilot signal, and then The receiving end searches for the optimal decision delay according to a certain algorithm, so as to set the filter coefficient of the system, and then the system enters the stage of sending data. When the length of the decision feedback filter is smaller than the channel order, the present invention provides a search method for the optimal decision delay for the MMSE-DFE system. The performance of the system mean square error obtained by the method is similar, but the computational complexity is significantly reduced.
Description
技术领域technical field
本发明涉及数字通信技术领域，尤其涉及一种最小均方误差判决反馈均衡系统的最优判决时延搜索方法。The invention relates to the technical field of digital communication, in particular to an optimal decision delay search method for a minimum mean square error decision feedback equalization system.
背景技术Background technique
在数字通信系统中，由于信道的多径效应，传输的符号间会产生码间干扰。为了克服码间干扰和提高通信系统的性能，在接收端需要采用均衡技术。最小均方误差判决反馈均衡器（MMSE-DFE，Minimum Mean Square Error-Decision Feedback Equalizer）是其中一种重要的均衡技术。MMSE-DFE系统接收端由两个滤波器组成，即前馈滤波器和判决反馈滤波器。判决时延是此系统设计的一个重要参数，它决定接收端两个滤波器的系数配置，进而对系统的数据检测性能有重要影响。判决时延的最优取值受到前馈、判决反馈滤波器的长度，信噪比等因素影响。In a digital communication system, due to the multipath effect of the channel, there will be intersymbol interference between the transmitted symbols. In order to overcome the intersymbol interference and improve the performance of the communication system, an equalization technique needs to be adopted at the receiving end. Minimum Mean Square Error-Decision Feedback Equalizer (MMSE-DFE, Minimum Mean Square Error-Decision Feedback Equalizer) is one of the important equalization techniques. The receiving end of the MMSE-DFE system consists of two filters, namely the feedforward filter and the decision feedback filter. The decision delay is an important parameter in the design of this system, which determines the coefficient configuration of the two filters at the receiving end, and has an important impact on the data detection performance of the system. The optimal value of decision delay is affected by factors such as feedforward, length of decision feedback filter, and signal-to-noise ratio.
美国《国际电气与电子工程师协会信息论学报》（“MMSE decision-feedbackequalizers：finite-length results”，IEEE Transactions on Information Theory，1995，41(4)：961-975）指出当前馈滤波器长度Nf等于信道阶数v时，最优判决时延是Nf-1。美国《国际电气与电子工程师协会信号处理快报》（“Optimum decision delay of thefinite-length DFE”，IEEE Signal Processing Letters，2004，54(2)：701-711）证明了如果反馈滤波器长度Nb足够大并且Nb≥v时，最优判决时延是Nf-1。因此，在Nb≥v时，最优判决时延问题已经基本解决。但是在Nb<v时，最优判决时延与每次信道的具体实现有关，无法在理论上求得具体表达式。美国《国际电气与电子工程师协会信息论学报》（“The effect ofdecision delay in finite-length decision feedback equalization”，IEEETransactions on Information Theory，1996，42(2)：618-621）提出了用遍历搜索的方法找到在Nb<v时的最优判决时延。但是，每在一个选定的判决时延计算一次系统均方误差（MSE，Mear Square Error），需要一次矩阵求逆运算，因此遍历搜索的方法计算量较大，其实际应用可能受限。The American Institute of Electrical and Electronics Engineers Information Theory Journal ("MMSE decision-feedback equalizers: finite-length results", IEEE Transactions on Information Theory, 1995, 41(4): 961-975) pointed out that the feedforward filter length Nf is equal to the channel When the order is v, the optimal decision delay is Nf-1. The American Institute of Electrical and Electronics Engineers Signal Processing Letters ("Optimum decision delay of the finite-length DFE", IEEE Signal Processing Letters, 2004, 54(2): 701-711) proves that if the feedback filter length N b is sufficient When is large and N b ≥ v, the optimal decision delay is N f -1. Therefore, when Nb≥v, the optimal decision delay problem has basically been solved. However, when N b < v, the optimal decision delay is related to the specific realization of each channel, and the specific expression cannot be obtained theoretically. "The effect of decision delay in finite-length decision feedback equalization", IEEE Transactions on Information Theory, 1996, 42(2): 618-621) proposed a method of traversal search to find Optimal decision delay when Nb<v. However, every calculation of the system mean square error (MSE, Mear Square Error) at a selected decision delay requires a matrix inversion operation, so the method of traversal search requires a large amount of calculation, and its practical application may be limited.
发明内容Contents of the invention
本发明的目的在于提供一种最小均方误差判决反馈均衡系统的最优判决时延搜索方法，从而对接收端的滤波器系数进行最优或者次优的设置，提高整个系统的数据检测性能。The purpose of the present invention is to provide an optimal decision delay search method for a minimum mean square error decision feedback equalization system, thereby setting optimal or suboptimal filter coefficients at the receiving end and improving the data detection performance of the entire system.
为了实现上述的目的，采用如下的技术方案：一种最小均方误差判决反馈均衡系统的最优判决时延搜索方法，所述方法包括以下步骤：In order to achieve the above object, adopt the following technical scheme: a minimum mean square error decision feedback equalization system optimal decision delay search method, the method includes the following steps:
S1在训练阶段，发送端发送导频信号，接收端估计出信道冲击响应和噪声方差；S1 In the training phase, the sending end sends pilot signals, and the receiving end estimates the channel impulse response and noise variance;
S2初始化三个时延变量ta＝0、tb＝Nf+v-1和tmid＝int[(ta+tb)/2]，这里int[·]表示取实数的整数部分，Nf是系统接收机中的前馈滤波器长度，v是信道阶数，然后计算三个判决时延下的系统均方误差MSE，即MSE(ta)、MSE(tb)和MSE(tmid)；S2 initializes three delay variables t a =0, t b =N f +v-1 and t mid =int[(t a +t b )/2], where int[ ] represents the integer part of the real number, Nf is the feed-forward filter length in the system receiver, v is the channel order, and then calculate the system mean square error MSE under the three decision delays, namely MSE(t a ), MSE(t b ) and MSE(t mid );
S3判断三个MSE是否满足终止条件，若满足终止条件，则最优判决时延取三个MSE中的最小值对应的判决时延并终止循环，否则继续；S3 judges whether the three MSEs meet the termination condition. If the termination condition is satisfied, the optimal decision delay takes the decision delay corresponding to the minimum value among the three MSEs and terminates the loop, otherwise continue;
S4若MSE(ta)≤MSE(tmid)且MSE(tmid)≥MSE(tb)，则最优判决时延取MSE(ta)、MSE(tb)中的最小值对应的判决时延并终止循环，否则继续；S4 If MSE(t a )≤MSE(t mid ) and MSE(t mid )≥MSE(t b ), the optimal decision delay is the one corresponding to the minimum value of MSE(t a ) and MSE(t b ). Judgment time delay and terminate the loop, otherwise continue;
S5根据时延更新算法更新三个时延变量ta、tb和tmid；S5 updates the three delay variables t a , t b and t mid according to the delay update algorithm;
S6计算新的MSE(ta)、MSE(tb)和MSE(tmid)，返回S3。S6 calculates new MSE(t a ), MSE(t b ) and MSE(t mid ), and returns to S3.
本发明适用于MMSE-DFE系统，MMSE-DFE系统的接收信号可以表示为：The present invention is applicable to MMSE-DFE system, and the received signal of MMSE-DFE system can be expressed as:
ythe y kk == ΣΣ ll == 00 vv hh ll xx kk -- ll ++ nno kk
这里hl是第l径信道冲击响应，v是信道阶数，而信道的多径数为v+1。xk是发送信号，nk是功率为的高斯加性白噪声。接收端包括前馈滤波器和判决反馈滤波器两部分，它们的长度分别是Nf和Nb，其系数可以分别表示为： b = [ 1 b 1 · · · b N b ] T , w = w 0 w 1 · · · w N f - 1 T . 如果系统预先设置的判决时延是Δ，则经过两个滤波器后的检测信号是发送信号xi的Δ时延版本。系统的均方误差定义为判决之前的信号与判决之后的信号的均方误差，经推导可得：Here h l is the channel impulse response of the lth path, v is the order of the channel, and the multipath number of the channel is v+1. x k is the transmitted signal, n k is the power of Gaussian additive white noise. The receiving end includes two parts, the feedforward filter and the decision feedback filter. Their lengths are N f and N b respectively, and their coefficients can be expressed as: b = [ 1 b 1 &Center Dot; &Center Dot; &Center Dot; b N b ] T , w = w 0 w 1 &Center Dot; &Center Dot; &Center Dot; w N f - 1 T . If the system preset decision delay is Δ, the detection signal after two filters is the Δ-delayed version of the transmitted signal xi . The mean square error of the system is defined as the signal before the decision and the signal after the decision The mean square error of can be obtained by derivation:
MSEMSE (( ΔΔ )) == 11 ee 00 TT RR ΔΔ -- 11 ee 00
其中e0是一个仅首位元素为1、其余元素为零的列向量，符号(.)T表述转置操作，矩阵RΔ表示为：Among them, e 0 is a column vector with only the first element being 1 and the remaining elements being zero. The symbol (.) T expresses the transposition operation, and the matrix R Δ is expressed as:
RR ΔΔ == 00 ΔΔ ×× (( NN bb ++ 11 )) II NN bb ++ 11 00 tt ×× (( NN bb ++ 11 )) TT RR xx // ythe y -- 11 00 ΔΔ ×× (( NN bb ++ 11 )) II NN bb ++ 11 00 tt ×× (( NN bb ++ 11 ))
RR xx // ythe y == Hh Hh Hh // σσ ww 22 ++ II
这里I是单位阵，hi是第i径的信道冲击响应，v是信道阶数，Nb是反馈滤波器长度。可以发现，系统MSE以判决时延Δ为参数，为了使系统检测性能最优，需要找到使系统MSE最小的Δ。当最优Δ找到，就可以设置最优反馈滤波器系数和前馈滤波器系数为：Here I is the identity matrix, h i is the channel impulse response of the i-th path, v is the channel order, and N b is the length of the feedback filter. It can be found that the system MSE takes the decision delay Δ as a parameter. In order to optimize the system detection performance, it is necessary to find the minimum system MSE Δ. When the optimal Δ is found, the optimal feedback filter coefficients and feedforward filter coefficients can be set as:
bb OPTOPT == RR ΔΔ -- 11 ee 00 ee 00 TT RR ΔΔ -- 11 ee 00 ,, ww OPTOPT == (( HHHH Hh ++ II σσ ww 22 )) -- 11 Hh bb OPTOPT
系统的操作分成两个阶段：训练阶段和发送数据阶段。在训练阶段，先发送导频信号，接收端估计出信道冲击响应h和噪声方差采用一定算法寻找最优判决时延，得出反馈滤波器系数和前馈滤波器系数并进行设置，然后系统进入发送数据阶段。The operation of the system is divided into two phases: training phase and sending data phase. In the training phase, the pilot signal is sent first, and the receiver estimates the channel impulse response h and the noise variance Use a certain algorithm to find the optimal decision delay, get the feedback filter coefficients and feedforward filter coefficients and set them, and then the system enters the data sending stage.
上述方案中，所述终止条件为小于门限ε，所述门限ε为预先设定值。In the above scheme, the termination condition is is smaller than the threshold ε, and the threshold ε is a preset value.
上述方案中，所述时延更新算法为：In the above scheme, the delay update algorithm is:
若MSE(ta)≤MSE(tmid)≤MSE(tb)，则更新tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；If MSE(t a )≤MSE(t mid )≤MSE(t b ), then update t b =t mid , t mid =int[(t a +t b )/2], and t a remains unchanged;
若MSE(ta)≥MSE(tmid)≥MSE(tb)，则更新ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；If MSE(t a )≥MSE(t mid )≥MSE(t b ), then update t a =t mid , t mid =int[(t a +t b )/2], and t b remains unchanged;
若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≤MSE(tb)，则更新If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≤MSE(t b ), update
tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；t b =t mid , t mid =int[(t a +t b )/2], t a remains unchanged;
若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≥MSE(tb)，则更新If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≥MSE(t b ), update
ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变。t a =t mid , t mid =int[(t a +t b )/2], t b remains unchanged.
与现有技术相比，本发明当判决反馈滤波器长度小于信道阶数时，为MMSE-DFE系统提供了一种最优判决时延的搜索方法，该搜索方法可以得到最优或者次优的判决时延，与遍历搜索方法取得的系统均方误差性能相近，但是显著的降低了运算复杂度。Compared with the prior art, when the length of the decision feedback filter is smaller than the channel order, the present invention provides a search method for the optimal decision delay for the MMSE-DFE system, and the search method can obtain optimal or suboptimal The decision delay is similar to the system mean square error performance achieved by the ergodic search method, but the computational complexity is significantly reduced.
附图说明Description of drawings
图1为本发明的流程图；Fig. 1 is a flow chart of the present invention;
图2为系统信噪比为20dB时本发明与遍历搜索方法的MSE性能对比；Fig. 2 is the MSE performance comparison of the present invention and the traverse search method when the system signal-to-noise ratio is 20dB;
图3为不同的系统信噪比时本发明与遍历搜索方法的MSE性能对比。Fig. 3 is a comparison of MSE performance between the present invention and the ergodic search method at different system signal-to-noise ratios.
具体实施方式detailed description
下面结合附图和实施例对本发明作进一步的描述。The present invention will be further described below in conjunction with the accompanying drawings and embodiments.
本发明的流程如图1所示，包括以下步骤：Flow process of the present invention is shown in Figure 1, comprises the following steps:
S1在训练阶段，发送端发送导频信号，接收端估计出信道冲击响应和噪声方差；S1 In the training phase, the sending end sends pilot signals, and the receiving end estimates the channel impulse response and noise variance;
S2初始化三个时延变量ta＝0、tb＝Nf+v-1和tmid＝int[((ta+tb)/2]，这里int[·]表示取实数的整数部分，Nf是系统接收机中的前馈滤波器长度，v是信道阶数，然后计算三个判决时延下的系统均方误差MSE，即MSE(ta)、MSE(tb)和MSE(tmid)；S2 initializes three delay variables t a =0, t b =N f +v-1 and t mid =int[((t a +t b )/2], where int[ ] represents the integer part of the real number , N f is the length of the feedforward filter in the system receiver, v is the channel order, and then calculate the system mean square error MSE under the three decision delays, namely MSE(t a ), MSE(t b ) and MSE (t mid );
S3判断三个MSE是否满足终止条件，若满足终止条件，则最优判决时延取三个MSE中的最小值对应的判决时延并终止循环，否则继续，终止条件为小于门限ε，门限ε为预先设定值；S3 judges whether the three MSEs meet the termination conditions. If the termination conditions are met, the optimal decision delay takes the decision delay corresponding to the minimum value among the three MSEs and terminates the loop. Otherwise, the termination condition is is less than the threshold ε, the threshold ε is a preset value;
S4若MSE(ta)≤MSE(tmid)且MSE(tmid)≥MSE(tb)，则最优判决时延取MSE(ta)、MSE(tb)中的最小值对应的判决时延并终止循环，否则继续；S4 If MSE(t a )≤MSE(t mid ) and MSE(t mid )≥MSE(t b ), the optimal decision delay is the one corresponding to the minimum value of MSE(t a ) and MSE(t b ). Judgment time delay and terminate the loop, otherwise continue;
S5根据时延更新算法更新三个时延变量ta、tb和tmid，时延更新算法为：S5 updates the three delay variables t a , t b and t mid according to the delay update algorithm, and the delay update algorithm is:
若MSE(ta)≤MSE(tmid)≤MSE(tb)，则更新tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；If MSE(t a )≤MSE(t mid )≤MSE(t b ), then update t b =t mid , t mid =int[(t a +t b )/2], and t a remains unchanged;
若MSE(ta)≥MSE(tmid)≥MSE(tb)，则更新ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；If MSE(t a )≥MSE(t mid )≥MSE(t b ), then update t a =t mid , t mid =int[(t a +t b )/2], and t b remains unchanged;
若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≤MSE(tb)，则更新If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≤MSE(t b ), update
tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；t b =t mid , t mid =int[(t a +t b )/2], t a remains unchanged;
若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≥MSE(tb)，则更新If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≥MSE(t b ), update
ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；t a =t mid , t mid =int[(t a +t b )/2], t b remains unchanged;
S6计算新的MSE(ta)、MSE(tb)和MSE(tmid)，返回S3。S6 calculates new MSE(t a ), MSE(t b ) and MSE(t mid ), and returns to S3.
实施例一在信噪比为20dB时，本发明与遍历搜索方法的性能对比。Embodiment 1 When the signal-to-noise ratio is 20 dB, the performance of the present invention is compared with that of the traverse search method.
在Matlab仿真环境下，使用蒙特卡洛方法对本发明所提方法的系统均方误差进行仿真计算。系统的信噪比定义为：Under the Matlab simulation environment, the Monte Carlo method is used to simulate and calculate the system mean square error of the method proposed in the present invention. The signal-to-noise ratio of the system is defined as:
SNRSNR == EE. (( || xx ii || 22 )) ΣΣ ll == 00 vv || hh ll || 22 // σσ ww 22
其中hl是第l径信道冲击响应，v是信道阶数，是噪声方差。不失一般性，假设发送信号功率为1。仿真中采用多径Rayleigh信道，信道径数为10而其阶数为9，每径功率为0.1。系统的前馈滤波器长度Nf设置为12，判决反馈滤波器长度Nb为8，从而Nb<v=9。where h l is the channel impulse response of the lth path, v is the channel order, is the noise variance. Without loss of generality, it is assumed that the transmitted signal power is 1. The multipath Rayleigh channel is adopted in the simulation, the number of channel paths is 10 and its order is 9, and the power of each path is 0.1. The feed-forward filter length N f of the system is set to 12, and the decision feedback filter length N b is 8, so N b <v=9.
进行100次信道实现，对于每次循环均采用如下步骤处理：Perform 100 channel implementations, and use the following steps for each cycle:
S1在训练阶段，系统先发送导频信号，由接收端估计出信道冲击响应h和噪声方差假设估计是足够精确的；S1 In the training phase, the system first sends a pilot signal, and the receiver estimates the channel impulse response h and noise variance Assuming estimates are sufficiently precise;
S2初始化三个时延变量ta＝0，tb＝12，tmid＝6，计算三个判决时延下的系统均方误差MSE，即MSE(ta)、MSE(tb)和MSE(tmid)；S2 initializes three delay variables t a =0, t b =12, t mid =6, and calculates the system mean square error MSE under the three decision delays, namely MSE(t a ), MSE(t b ) and MSE (t mid );
S3判断是否小于门限0.05，如果满足此条件，则最优判决时延取三个MSE中的最小值对应的判决时延，否则继续下一步骤；S3 judgment Whether it is less than the threshold 0.05, if this condition is met, the optimal decision delay is the decision delay corresponding to the minimum value among the three MSEs, otherwise continue to the next step;
S4若MSE(ta)≤MSE(tmid)且MSE(tmid)≥MSE(tb)，则最优判决时延取MSE(ta)、MSE(tb)中的最小值对应的判决时延，否则继续下一步骤；S4 If MSE(t a )≤MSE(t mid ) and MSE(t mid )≥MSE(t b ), the optimal decision delay is the one corresponding to the minimum value of MSE(t a ) and MSE(t b ). Judgment delay, otherwise continue to the next step;
S5根据MSE(ta)，MSE(tb)和MSE(tmid)的值对三个时延变量进行更新：S5 updates the three delay variables according to the values of MSE(t a ), MSE(t b ) and MSE(t mid ):
若MSE(ta)≤MSE(tmid)≤MSE(tb)，则更新tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；If MSE(t a )≤MSE(t mid )≤MSE(t b ), then update t b =t mid , t mid =int[(t a +t b )/2], and t a remains unchanged;
若MSE(ta)≥MSE(tmid)≥MSE(tb)，则更新ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；If MSE(t a )≥MSE(t mid )≥MSE(t b ), then update t a =t mid , t mid =int[(t a +t b )/2], and t b remains unchanged;
若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≤MSE(tb)，则更新If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≤MSE(t b ), update
tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；t b =t mid , t mid =int[(t a +t b )/2], t a remains unchanged;
若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≥MSE(tb)，则更新If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≥MSE(t b ), update
ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；t a =t mid , t mid =int[(t a +t b )/2], t b remains unchanged;
S6计算新的MSE(ta)、MSE(tb)和MSE(tmid)，返回S3。S6 calculates new MSE(t a ), MSE(t b ) and MSE(t mid ), and returns to S3.
上述循环结束后，得到一个最优或者次优的判决时延和相应的MSE。After the above cycle ends, an optimal or suboptimal decision delay and a corresponding MSE are obtained.
如图2所示，在每一次信道实现，本发明得到的MSE性能均接近遍历搜索方法。对于遍历搜索方法，每次需要搜索13次判决时延并且计算13次MSE才可以找到最优的MSE和相应的判决时延。本发明方法仅需要约5.39次MSE计算，因此本发明可以显著节省运算量。As shown in Fig. 2, in each channel implementation, the MSE performance obtained by the present invention is close to that of the ergodic search method. For the traversal search method, it is necessary to search 13 times of decision delay and calculate 13 times of MSE each time to find the optimal MSE and corresponding decision delay. The method of the present invention only needs about 5.39 MSE calculations, so the present invention can significantly save the amount of computation.
实施例二在不同的系统信噪比时，本发明与遍历搜索方法的性能对比。Embodiment 2 The performance comparison between the present invention and the ergodic search method under different system signal-to-noise ratios.
在Matlab仿真环境下，使用蒙特卡洛方法对本发明所提方法的系统均方误差进行仿真计算。系统的判决反馈滤波器长度Nb设置为7，其他部分的设置与实施例一相同。对于每个信噪比，均仿真产生500次信道实现，在每一次循环的操作步骤与实施例一相同，得到500个MSE值以后，求平均值，如下表所示：Under the Matlab simulation environment, the Monte Carlo method is used to simulate and calculate the system mean square error of the method proposed in the present invention. The length Nb of the decision feedback filter of the system is set to 7, and the settings of other parts are the same as those in the first embodiment. For each signal-to-noise ratio, 500 channel realizations are simulated, and the operation steps in each cycle are the same as in Embodiment 1. After obtaining 500 MSE values, calculate the average value, as shown in the following table:
SNR(dB)SNR(dB) 55 1010 1515 2020 2525 本发明this invention 5.065.06 5.265.26 5.445.44 5.285.28 5.315.31 遍历搜索traverse search 1414 1414 1414 1414 1414
表1本发明方法和遍历搜索方法平均计算MSE的次数。Table 1 The average number of times MSE is calculated by the method of the present invention and the traversal search method.
如图3所示，在信噪比为5dB到25dB的区域，遍历搜索和提出的方法几乎具有相同的MSE性能。从表1可以看出，遍历搜索方法在每个信噪比需要搜索14次判决时延，计算14个MSE值，从而得到最优的判决时延和相应的MSE。但是，本发明方法在各个信噪比处仅需要平均搜索不到6次判决时延和进行不到6次的MSE计算，因此，本发明方法相比遍历搜索方法显著的节省了运算量。As shown in Fig. 3, the ergodic search and the proposed method have almost the same MSE performance in the region of SNR from 5dB to 25dB. It can be seen from Table 1 that the traversal search method needs to search 14 decision delays for each SNR and calculate 14 MSE values, so as to obtain the optimal decision delay and corresponding MSE. However, the method of the present invention requires less than 6 average search delays and less than 6 MSE calculations at each SNR. Therefore, the method of the present invention significantly saves the amount of computation compared with the traversal search method.
Claims (3)
1.一种最小均方误差判决反馈均衡系统的最优判决时延搜索方法，其特征在于，所述方法包括以下步骤：1. a minimum mean square error decision feedback equalization system optimal decision delay search method, is characterized in that, said method comprises the following steps: S1在训练阶段，发送端发送导频信号，接收端估计出信道冲击响应和噪声方差；S1 In the training phase, the sending end sends pilot signals, and the receiving end estimates the channel impulse response and noise variance; S2初始化三个时延变量ta＝0、tb＝Nf+v-1和tmid＝int[(ta+tb)/2]，其中，Nf是前馈滤波器长度，v是信道阶数，计算三个判决时延下的系统均方误差MSE，即MSE(ta)、MSE(tb)和MSE(tmid)；S2 initializes three delay variables t a =0, t b =N f +v-1 and t mid =int[(t a +t b )/2], where N f is the length of the feedforward filter, v is the channel order, calculate the system mean square error MSE under the three decision delays, namely MSE(t a ), MSE(t b ) and MSE(t mid ); S3判断三个MSE是否满足终止条件，若满足终止条件，则最优判决时延取三个MSE中的最小值对应的判决时延并终止循环，否则继续；S3 judges whether the three MSEs meet the termination condition. If the termination condition is satisfied, the optimal decision delay takes the decision delay corresponding to the minimum value among the three MSEs and terminates the loop, otherwise continue; S4若MSE(ta)≤MSE(tmid)且MSE(tmid)≥MSE(tb)，则最优判决时延取MSE(ta)、MSE(tb)中的最小值对应的判决时延并终止循环，否则继续；S4 If MSE(t a )≤MSE(t mid ) and MSE(t mid )≥MSE(t b ), the optimal decision delay is the one corresponding to the minimum value of MSE(t a ) and MSE(t b ). Judgment time delay and terminate the loop, otherwise continue; S5根据时延更新算法更新三个时延变量ta、tb和tmid，所述时延更新算法为：S5 updates the three delay variables t a , t b and t mid according to the delay update algorithm, and the delay update algorithm is: 若MSE(ta)≤MSE(tmid)≤MSE(tb)，则更新tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；If MSE(t a )≤MSE(t mid )≤MSE(t b ), then update t b =t mid , t mid =int[(t a +t b )/2], and t a remains unchanged; 若MSE(ta)≥MSE(tmid)≥MSE(tb)，则更新ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；If MSE(t a )≥MSE(t mid )≥MSE(t b ), then update t a =t mid , t mid =int[(t a +t b )/2], and t b remains unchanged; 若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≤MSE(tb)，则更新tb＝tmid、tmid＝int[(ta+tb)/2]，ta不变；If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≤MSE(t b ), update t b =t mid , t mid =int[ (t a +t b )/2], t a remains unchanged; 若MSE(ta)≥MSE(tmid)、MSE(tmid)≤MSE(tb)且MSE(ta)≥MSE(tb)，则更新ta＝tmid、tmid＝int[(ta+tb)/2]，tb不变；If MSE(t a )≥MSE(t mid ), MSE(t mid )≤MSE(t b ) and MSE(t a )≥MSE(t b ), update t a =t mid , t mid =int[ (t a +t b )/2], t b remains unchanged; S6计算新的MSE(ta)、MSE(tb)和MSE(tmid)，返回S3。S6 calculates new MSE(t a ), MSE(t b ) and MSE(t mid ), and returns to S3. 2.根据权利要求1所述的搜索方法，其特征在于，所述终止条件为小于门限ε。2. The search method according to claim 1, wherein the termination condition is less than the threshold ε. 3.根据权利要求2所述的搜索方法，其特征在于，所述门限ε为预先设定值。3. The search method according to claim 2, wherein the threshold ε is a preset value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310349561.8A CN103441963B (en) | 2013-08-12 | 2013-08-12 | The optimum decision delay time search method of least mean-square error decision feedback equalization system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310349561.8A CN103441963B (en) | 2013-08-12 | 2013-08-12 | The optimum decision delay time search method of least mean-square error decision feedback equalization system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103441963A CN103441963A (en) | 2013-12-11 |
| CN103441963B true CN103441963B (en) | 2016-08-10 |
Family
ID=49695629
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310349561.8A Expired - Fee Related CN103441963B (en) | 2013-08-12 | 2013-08-12 | The optimum decision delay time search method of least mean-square error decision feedback equalization system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103441963B (en) |
Citations (2)
* Cited by examiner, † Cited by third party| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1705300A (en) * | 2004-06-02 | 2005-12-07 | 美国博通公司 | System and method for adjusting multiple control loops using common criteria |
| US7382827B2 (en) * | 2001-09-18 | 2008-06-03 | Broadcom Corporation | Computation of decision feedback equalizer coefficients with constrained feedback tap energy |
-
2013
- 2013-08-12 CN CN201310349561.8A patent/CN103441963B/en not_active Expired - Fee Related
Patent Citations (2)
* Cited by examiner, † Cited by third party| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7382827B2 (en) * | 2001-09-18 | 2008-06-03 | Broadcom Corporation | Computation of decision feedback equalizer coefficients with constrained feedback tap energy |
| CN1705300A (en) * | 2004-06-02 | 2005-12-07 | 美国博通公司 | System and method for adjusting multiple control loops using common criteria |
Non-Patent Citations (5)
* Cited by examiner, † Cited by third party| Title |
|---|
| Equalisation with adaptive time lag;Y. Gong等;《IEE Proceedings Communications》;20051031;第152卷(第5期);第661-667页 * |
| Optimized Decision Delays in Finite-Length MIMO DFE;Greg Day等;《IEEE SIGNAL PROCESSING LETTERS》;20050531;第12卷(第5期);第391-394页 * |
| The Effect of Decision Delay in Finite-Length Decision Feedback Equalization;Paul A. Voois等;《IEEE Transactions on Information Theory》;20020831;第42卷(第2期);第618-621页 * |
| 反馈滤波器长度对判决反馈均衡器性能的影响;郭继昌 等;《天津大学学报》;20090228;第42卷(第2期);第145-148页 * |
| 时变信道追踪结构自适应DFE-RLS均衡器;黄本雄 等;《华中科技大学学报（自然科学版）》;20080430;第36卷(第4期);第1-4页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103441963A (en) | 2013-12-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10659260B2 (en) | 2020-05-19 | Decision feedback equalization processing device and method |
| CN108712353B (en) | 2021-05-11 | Soft iteration channel estimation method |
| CN105553898A (en) | 2016-05-04 | Equalizer and feedback equalization method |
| CN104506465A (en) | 2015-04-08 | Power line communication channel estimation method |
| CN103368885B (en) | 2016-07-06 | The fusion method that a kind of frequency domain bidirectional iteration is balanced |
| CN105634545A (en) | 2016-06-01 | Interference elimination method based on matrix decomposition in faster-than-Nyquist communication system |
| CN103763228B (en) | 2017-01-25 | Combination optimization self-adaptive frequency domain blind equalization method and system |
| CN108881080A (en) | 2018-11-23 | A kind of anti-ICI detection method of OFDM based on sliding window and deep learning |
| CN102790746B (en) | 2014-10-15 | Channel estimation method for OFDM (orthogonal frequency division multiplexing) system |
| CN106452652A (en) | 2017-02-22 | A multipath interference suppression method based on a chaotic wireless communication system |
| Ali et al. | 2013 | Legendre based equalization for nonlinear wireless communication channels |
| CN103441963B (en) | 2016-08-10 | The optimum decision delay time search method of least mean-square error decision feedback equalization system |
| CN110213184B (en) | 2020-06-26 | Adaptive channel blind equalization method based on modified cost function |
| CN103384229A (en) | 2013-11-06 | Error feedback network design method applicable to blind equalization of complex domain signals |
| Kumar et al. | 2014 | Linear block equalizers in rayleigh fading channel with normalized channel impulse response |
| CN103117969B (en) | 2016-05-11 | A kind of fractional lower-order statistics mould value transform wavelet frequency domain multimode blind balance method |
| CN106487400B (en) | 2019-11-26 | Uncoupling single antenna Interference Suppression System and method based on fractionally spaced equalizer |
| CN102299875B (en) | 2014-01-22 | A Wavelet Multi-mode Blind Equalization Method Introducing Immune Optimization Support Vector Machine |
| CN103428130B (en) | 2016-08-10 | A kind of minimum Mean Square Error Linear equalization methods eliminating impulsive noise |
| CN102137052A (en) | 2011-07-27 | Variable step length least mean square channel equilibrium method based on gradient vector |
| CN101651500B (en) | 2013-02-27 | Improved Blind Equalization Method for T/2 Fractional Interval Coordinate Transformation |
| JP2008510323A (en) | 2008-04-03 | A method for calculating the filter coefficient of the equalizer of a communication receiver using the Hermite optimization method. |
| CN108418771A (en) | 2018-08-17 | A Channel Equalization Algorithm Combining Time Domain and Frequency Domain |
| CN114374587B (en) | 2022-11-29 | Channel time domain equalization method based on frame |
| CN103685102B (en) | 2016-09-28 | A kind of time domain equalization implementation method of broadband signal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 2013-12-11 | C06 | Publication | |
| 2013-12-11 | PB01 | Publication | |
| 2014-01-08 | C10 | Entry into substantive examination | |
| 2014-01-08 | SE01 | Entry into force of request for substantive examination | |
| 2016-08-10 | C14 | Grant of patent or utility model | |
| 2016-08-10 | GR01 | Patent grant | |
| 2020-07-31 | CF01 | Termination of patent right due to non-payment of annual fee | |
| 2020-07-31 | CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160810 Termination date: 20190812 |