Convergence analysis of a correntropy induced metric constrained mixture error criterion algorithm

Correntropy induced metric (CIM) criterion has been extensively studied for measuring the sparsity property of the in-nature sparse signals. In this paper, a CIM constrained l 2 –l p (CIM-L2LP) adaptive filtering algorithm is proposed and its convergence analysis is given in detail. By using a CIM penalty, the CIM-L2LP algorithm achieves improved convergence speed while it maintains a lower steady state error floor, which is more useful for sparse channel estimation applications. Moreover, the convergence analysis of the CIM-L2LP algorithm is derived in detail. The studies of convergence analysis indicate that the CIM-L2LP algorithm can provide lower steady-state error and faster convergence speed than the l 2 –l p (L2LP), zero attracting L2LP (ZA-L2LP) and reweighted ZA-L2LP (RZA-L2LP) algorithms in the context of the sparse channel estimation. Simulation examples are set up to help to verify the excellent performance of the CIM-L2LP algorithm under different sparsity levels.