基于峰度的变步长恒模盲均衡技术

相对于基于训练序列的均衡方式,盲均衡能够提高信道利用率,更适于自适应组网和侦听,其中的恒模算法因实现简单而受到广泛关注,但是固定步长的恒模盲均衡器存在收敛速度和稳态误差上的矛盾。根据峰度定理,盲均衡器输出信号的峰度小于等于源信号的峰度,且只有达到收敛时才相等。文章采用均衡器输出信号峰度误差控制步长的变化,仿真表明变步长均衡器能够一定程度克服这一矛盾。

Blind Equalization with Variable Step Constant Modulus Algorithm Based on Kurtosis

In contrast to equalization based on training sequences, blind equalization can enhance the ratio of channel usage and are suitable for self-organizing network and interception. The constant modulus algorithm which is easy to be realized is very popular. But it suffers from the contradiction between the speed of convergence and steady error. According to the kurtosis theory, the kurtosis of output of blind equalizer is smaller than its source＇s before the convergence. In this paper, the step size was control according to the kurtosis of the output of blind equalization. Simulations given the proof that it conquered the contradiction to a certain extent.