阅读说明：本技术 一种面向毫米波应用的改进常数模盲均衡方法 (Improved constant modulus blind equalization method for millimeter wave application ) 是由 王瑜 吴道龙 茅迪 徐媛媛 李晓冬 于 2019-09-01 设计创作，主要内容包括：本发明提供了一种面向毫米波应用的改进常数模盲均衡方法,接收到的基带信号经采样后送入均衡器,根据均衡器输出,采用CMA算法计算误差函数,根据误差函数,对均衡器系数进行更新,经过迭代,直至恢复正确数据。本发明无需依赖于训练序列,在高速连续通信中能够有效提升信道利用效率；采用实、虚部并行处理,在修正幅度失真的同时能够消除相位旋转；结合泄露算法与高阶累积量思想进行系数更新,在步进系数相同的条件下能够改善现有算法迭代后的残留误差。(the invention provides an improved constant modulus blind equalization method for millimeter wave application. The invention can effectively improve the utilization efficiency of the channel in high-speed continuous communication without depending on a training sequence; real and imaginary parts are processed in parallel, so that phase rotation can be eliminated while amplitude distortion is corrected; and the leakage algorithm and the high-order cumulant idea are combined to update the coefficient, so that the residual error after the iteration of the existing algorithm can be improved under the condition that the stepping coefficient is the same.)

1. an improved constant modulus blind equalization method for millimeter wave application is characterized by comprising the following steps:

Step 1, sending a broadband signal: s (n) ═ a (n) + jb (n), where a (n) and b (n) represent the real and imaginary parts of the wideband complex signal s (n), respectively;

step 2, the length of the equalizer is N, the received baseband signal is sampled and then sent to the equalizer, and the complex vector of the signal is X (N) ═ X_n,x_n-1,...,x_n-N+1]^TThe equalizer coefficient is W (n) ═ W₀(n),w₁(n),...,w_N-1(n)]^TAnd setting the initial coefficient of the equalizer according to the following principle:

When the number N is an even number,

when the number N is an odd number,

That is, W (0) ([ 0,0,. ], 0,1,0, ], 0]^T；

Step 3, calculating the output of the equalizer as: y (n) ═ W^T(n)X(n)；

And 4, calculating an error function by adopting a CMA algorithm according to the output of the equalizer:

Wherein R is_cRelating only to the mean of the transmitted symbols, R for PSK, QAM-like modulation_cThe method is a fixed value and comprises the following steps:

R_c＝E[|s^2p(n)|]/E[|s^p(n)|] (4)

taking p as 2, namely:

R_c＝E[|s⁴(n)|]/E[|s²(n)|] (6)

and performing real and imaginary part parallel processing on the error function calculation process, namely:

e(n)＝e_r(n)+je_i(n) (11)

Wherein R is_{c_r}And R_{c_i}each represents a constant R_cReal and imaginary parts of, y_r(n) and y_i(n) represents the real and imaginary parts of the equalizer output signal y (n), respectively, e_r(n) and e_i(n) represents the real and imaginary parts of the error function e (n), respectively;

And step 5, updating the equalizer coefficient according to the error function, wherein the updating process is as follows:

W(n+1)＝αW(n)-μe(n)X^*(n)-β[W(n)-W(n-1)] (12)

Mu is a stepping coefficient, alpha is a leakage coefficient, beta is a high-order correlation coefficient, W (n) is an equalizer coefficient at the current moment, W (n +1) is an equalizer coefficient to be updated at the next moment, and W (n-1) is an equalizer coefficient at the previous moment; and (3) after the coefficient updating is finished, turning to the step (3), calculating the output of the equalizer at the next moment, and entering an iteration process until correct data is recovered.

Technical Field

The invention relates to the field of broadband wireless communication digital signal processing, mainly aims at the problem of transmission quality deterioration caused by phenomena of intersymbol crosstalk, phase rotation and the like in the broadband continuous data transmission process of a wireless communication system, and particularly relates to a method for directly carrying out blind channel estimation according to a received signal and compensating by combining a leakage algorithm and a high-order cumulant idea without depending on a training sequence, which is used for millimeter wave wireless communication broadband digital signal processing.

Background

In an actual system, due to the finite length truncation of a filter and the deviation of timing sampling time, the superposition of front and back tails among codes can be caused, and thus, the crosstalk among the codes is caused. The in-band amplitude and phase of the wideband rf device and the digital-to-analog converter are not consistent, and the amplitude, phase nonlinear fading, multipath channel, etc. of the wideband channel also cause inter-symbol crosstalk, which requires equalization techniques to eliminate or reduce these inter-symbol crosstalks and recover the original signal.

since the nonlinear effects of the actual channel and device are often unknown and the channel is also time-varying, adaptive equalization algorithms are often needed to update the equalizer coefficients in real-time to eliminate the time-varying effects. The traditional adaptive equalization technology can overcome intersymbol interference to a certain extent, but the traditional adaptive equalization technology needs to continuously send known training sequences to periodically train an equalizer, thereby increasing transmission overhead and reducing channel utilization rate. By adopting the blind equalization channel estimation algorithm, the equalizer parameters do not need to be trained periodically, and the channel utilization rate can be greatly improved. In addition, because the blind equalization algorithm is only related to the amplitude of the signal, the requirement on carrier synchronization is reduced, and the realization of an engineering system is facilitated.

In the practical engineering, aiming at the millimeter wave zero intermediate frequency architecture, aiming at the problems of time-varying property, environmental sensitivity, zero intermediate frequency carrier synchronization performance and the like of a millimeter wave system channel, a real-imaginary part parallel processing mode is adopted to eliminate phase deflection errors, and aiming at the problem that the existing CMA algorithm has large residual errors after convergence, the leakage algorithm and the high-order cumulant idea are combined to correct the problem, so that the intersymbol interference of the system is reduced under the condition that the convergence speed of the traditional CMA algorithm is equivalent.

disclosure of Invention

In order to overcome the defects of the prior art, the invention provides an improved constant modulus blind equalization method oriented to millimeter wave application. Aiming at the problem of transmission quality deterioration caused by phenomena such as intersymbol interference, phase rotation and the like in the broadband continuous data transmission process of a wireless communication system, a real-imaginary part parallel mode is adopted, a leakage algorithm and a high-order cumulant idea are introduced, and the problems that the phase deflection cannot be corrected and the residual error after iteration is large in the existing CMA algorithm are solved.

the technical scheme adopted by the invention for solving the technical problem comprises the following steps:

Step 1, sending a broadband signal: s (n) ═ a (n) + jb (n), where a (n) and b (n) represent the real and imaginary parts of the wideband complex signal s (n), respectively;

Step 2, the length of the equalizer is N, the received baseband signal is sampled and then sent to the equalizer, and the complex vector of the signal is X (N) ═ X_n,x_n-1,...,x_n-N+1]^TThe equalizer coefficient is W (n) ═ W₀(n),w₁(n),...,w_N-1(n)]^Tand setting the initial coefficient of the equalizer according to the following principle:

when the number N is an even number,

When the number N is an odd number,

That is, W (0) ([ 0,0,. ], 0,1,0, ], 0]^T；

Step 3, calculating the output of the equalizer as: y (n) ═ W^T(n)X(n)；

And 4, calculating an error function by adopting a CMA algorithm according to the output of the equalizer:

Wherein R is_cRelating only to the mean of the transmitted symbols, R for PSK, QAM-like modulation_cthe method is a fixed value and comprises the following steps:

R_c＝E[|s^2p(n)|]/E[|s^p(n)|] (4)

taking p as 2, namely:

R_c＝E[|s⁴(n)|]/E[|s²(n)|] (6)

and performing real and imaginary part parallel processing on the error function calculation process, namely:

e(n)＝e_r(n)+je_i(n) (11)

Wherein R is_{c_r}And R_{c_i}Each represents a constant R_cReal and imaginary parts of, y_r(n) and y_i(n) represents the real and imaginary parts of the equalizer output signal y (n), respectively, e_r(n) and e_i(n) represents the real and imaginary parts of the error function e (n), respectively;

And step 5, updating the equalizer coefficient according to the error function, wherein the updating process is as follows:

W(n+1)＝αW(n)-μe(n)X^*(n)-β[W(n)-W(n-1)] (12)

Mu is a stepping coefficient, alpha is a leakage coefficient, beta is a high-order correlation coefficient, W (n) is an equalizer coefficient at the current moment, W (n +1) is an equalizer coefficient to be updated at the next moment, and W (n-1) is an equalizer coefficient at the previous moment; and (3) after the coefficient updating is finished, turning to the step (3), calculating the output of the equalizer at the next moment, and entering an iteration process until correct data is recovered.

the invention has the beneficial effects that:

1) The invention can effectively improve the utilization efficiency of the channel in high-speed continuous communication without depending on a training sequence.

2) and real and imaginary parts are processed in parallel, so that the phase rotation can be eliminated while the amplitude distortion is corrected.

3) and the leakage algorithm and the high-order cumulant idea are combined to update the coefficient, so that the residual error after the iteration of the existing algorithm can be improved under the condition that the stepping coefficient is the same.

drawings

Fig. 1 is a schematic diagram of the equalizer operation of the present invention.

fig. 2 is a flow chart of a blind equalization algorithm in the present invention.

Fig. 3 is a schematic diagram illustrating the principle of equalizer coefficient updating in the present invention.

Fig. 4 is a normalized transmit data constellation diagram in embodiment 1 of the present invention.

Fig. 5 is a received signal constellation after complex channel in embodiment 1 of the present invention.

Fig. 6 is a constellation diagram of the signal compensated by the algorithm of the present invention in embodiment 1 of the present invention.

FIG. 7 is a graph comparing the residual error after convergence in embodiment 1 of the present invention with the conventional CMA algorithm.

Detailed Description

the invention is further explained below with reference to the drawings and examples, and a flow chart of the invention is shown in fig. 2.

Step 1, sending a broadband signal: s (n) ═ a (n) + jb (n), where a (n) and b (n) represent the real and imaginary parts of the wideband complex signal s (n), respectively;

Step 2, fig. 1 is a schematic diagram of the operating principle of the equalizer of the present invention, where the length of the equalizer is N, and the complex vector of the received baseband signal, which is sampled and sent to the equalizer, is X (N) ═ X_n,x_n-1,...,x_n-N+1]^TThe equalizer coefficient is W (n) ═ W₀(n),w₁(n),...,w_N-1(n)]^TAnd setting the initial coefficient of the equalizer according to the following principle:

When the number N is an even number,

when the number N is an odd number,

That is, W (0) ([ 0,0,. ], 0,1,0, ], 0]^T；

Step 3, calculating the output of the equalizer as: y (n) ═ W^T(n)X(n)；

and 4, calculating an error function by adopting a CMA algorithm according to the output of the equalizer:

Wherein R is_crelating only to the mean of the transmitted symbols, R for PSK, QAM-like modulation_cthe method is a fixed value and comprises the following steps:

R_c＝E[|s^2p(n)|]/E[|s^p(n)|] (4)

In general, let p be 2, i.e.:

R_c＝E[|s⁴(n)|]/E[|s²(n)|] (6)

in the processing process, in order to correct the channel phase error, the real part and the imaginary part of the error function calculation process are processed in parallel, namely:

e(n)＝e_r(n)+je_i(n) (11)

wherein R is_{c_r}and R_{c_i}each represents a constant R_creal and imaginary parts of, y_r(n) and y_i(n) represents the real and imaginary parts of the equalizer output signal y (n), respectively, e_r(n) and e_i(n) represents the real and imaginary parts of the error function e (n), respectively;

step 5, fig. 3 is a schematic diagram of the principle of updating the equalizer coefficient in the present invention, the equalizer coefficient is updated according to the error function, the leakage algorithm and the high-order cumulant idea are adopted, and the updating process is as follows:

W(n+1)＝αW(n)-μe(n)X^*(n)-β[W(n)-W(n-1)] (12)

Mu is a stepping coefficient, alpha is a leakage coefficient, beta is a high-order correlation coefficient, W (n) is an equalizer coefficient at the current moment, W (n +1) is an equalizer coefficient to be updated at the next moment, and W (n-1) is an equalizer coefficient at the previous moment; and (3) after the coefficient updating is finished, turning to the step (3), calculating the output of the equalizer at the next moment, and entering an iteration process until correct data is recovered.