Abstract: The convex combination of least mean square(CLMS) algorithm can overcome the contradiction between convergence rate, tracking performance and steady state error of traditional least mean square algorithm. However, in the normal adaptive CLMS algorithm, the rule for modifying mixing parameter is based on the steepest descent method. When the algorithm converges, it will generate zigzag phenomena, which can make the convergence speed become slowly. In order to solve this problem, a new rule based on the conjugate gradient method is proposed in this paper. At the same time, modified hyperbolic tangent function is used to reduce computational complexity. Meanwhile, instantaneous transfer scheme is used to further optimize the performance. Theoretical analysis and simulation results demonstrate that under different simulation environment, the proposed algorithm performs good property of mean square and tracking compared with the traditional CLMS and variable step-size CLMS algorithms.

Key words: adaptive filtering, system identification, least mean square algorithm, convex combination, conjugate gradient