Recursive Least Squares Fractionally-spaced Blind Equalization Algorithm for Underwater Acoustic Communication

Abstract

Constant Modulus Algorithm (CMA), Fractionally-spaced Equalizer (FSE) and Recursive Least Squares (RLS)-CMA are investigated, and a low computational complexity RLS-CMA based on FSE with the simplified cost function of CMA (LRLS-FSE-SCMA) was proposed in this paper. To overcome the instability of RLS-CMA, the cost function of CMA is simplified to meet second norm form and an improved RLS-CMA (RLS-SCMA) is obtained. For the input signals of FSE are obtained by oversampling the received signal, the input signals of each sub-equalizer are correlative, and the autocorrelation matrixes of the input signals of each sub-equalizer are little difference. To reduce the computational complexity, we use one of the autocorrelation matrixes instead of the others in RLS-FSESCMA, and then LRLS-FSE-SCMA is obtained. The simulation results under the underwater acoustic channel condition show that LRLS-FSE-SCMA has faster convergence rate and lower steady state error compared with Stochastic Gradient Descent (SGD)-CMA and RLS-SCMA with transversal equalizer, while LRLS-FSE-SCMA has similar performance as RLS-FSE-SCMA with lower computational complexity.