Constant Modulus Algorithm with Reduced Complexity Employing DFT Domain Fast Filtering

Abstract

In this paper, a novel CMA (constant modulus algorithm) algorithm employing fast convolution in the DFT (discrete Fourier transform) domain is proposed. We propose a non-linear adaptation algorithm that minimizes CMA cost function in the DFT domain. The proposed algorithm is completely new one as compared to the recently introduced similar DFT domain CMA algorithm in that, the original CMA cost function has not been changed to develop DFT domain algorithm, resulting improved convergence properties. Using the proposed approach, we can reduce the number of multiplications to O(N log ₂ N), whereas the conventional CMA has the computation order of O(N²). Simulation results show that the proposed algorithm provides a comparable performance to the conventional CMA.