提出一种估计最小均方误差的盲均衡算法。与RLS算法原理类似,该方法依据矩阵求逆引理逐步更新自相关矩阵及其伪逆,以达到快速收敛,且对迭代初始值不敏感。与非递归算法相比,该自适应在线算法无需直接计算相关矩阵的伪逆或引入奇异值分解,避免了估计相关矩阵的秩或信道阶数。快速收敛以及在线处理的特性使其可以应用到实时通信信号处理中。仿真结果证明算法具有很好的在线均衡性能。
This paper proposes an adaptive blind equalization algorithm for Minimum Mean Squared Error(MMSE). Like RLS algorithm, the method recursively updates correlation matrix and its inversion according to the matrix inversion lemma, thus ensures convergence and is not sensitive to initialization. Unlike many subspace methods, the algorithm does not require channel order estimation and it is robust to channel order mismatch. Fast convergence and online property enable it to be used in real-time communication applications. Simulation results show that the algorithm has good performance of online equalization.