To further improve the performance of CMA blind equalization, an improved algorithm with variable momentum controlled by equalizer weights energy was proposed. The momentum gradient algorithm can improve the convergence rate effectively. However the additional gradient noise would produce by momentum, which results in high steady state residual error after convergence. Hereby, a variable momentum algorithm is designed according to the equalizer weights energy, which the momentum factor adjusted needs no empirical parameters during the iterative process. Therefore, the proposed algorithm is more universal than the other adaptive momentum algorithms. The variable momentum controlled by equalizer weights energy not only can take advantage of the momentum gradient algorithm to improve the convergence rate, but also can avoid the additional gradient noise produce by momentum for CMA blind equalization, and the effectiveness is shown by computer simulation under underwater acoustic channel condition.
展开▼