归一化自适应共轭梯度恒模盲均衡仿真研究

为提高恒模盲均衡算法收敛速度,提出了一种归一化自适应共轭梯度恒模盲均衡算法并进行了仿真研究。利用级联滤波方法对恒模盲均衡器输入进行了重新定义,并在此基础上采用自适应共轭梯度算法对均衡器参数进行更新,对均衡器输入采用归一化进行处理,以保证算法的稳定性。共轭梯度算法计算复杂度介于LMS和RLS算法之间,与LMS算法相比较具有更快的收敛速度,仿真结果证明归一化自适应共轭梯度算法恒模盲均衡与传统恒模盲均衡算法相比具有更好的均衡性能,复杂信道条件下剩余码间干扰可降低约10 dB,均衡系统中引入的级联滤波器可视为时变信道的一部分,表明算法对于时变信道同样有效。

Simulation of Blind Equalization of Constant Modulus Algorithm with Normalized Adaptive Conjugate Gradient Algorithm

In order to improve the CMA blind equalization algorithm convergence rate, a kind of normalized conjugate gradient CMA blind equalization algorithm has been proposed and simulations have been performed. Used the cascade filtering, the input of blind equalizer has been redefined which can transfer the cost function of CMA to satisfy second normal form. And on this basis used adaptive conjugate gradient algorithm the equalizer weights have been updated. Meanwhile, the equalizer update input adopts power normalization process to ensure the stability of the algorithm. The computational complexity of conjugate gradient algorithm is lower than RLS and higher than LMS, and has faster convergence rate than LMS. The simulation results prove that normalization adaptive conjugate gradient algorithm has better performance than the traditional CMA blind equalization, and under complicated channel condition residue intersymbol interference can reduce 10dB, for the cascade filter adds to the equalization system can be taken as time - varying channel, which shows that the algorithm for time - varying channel is also effective.