Modified Least-Mean Mixed-Norm Algorithms For Adaptive Sparse System Identification Under Impulsive Noise Environment

In this paper, new algorithms robust to a mix of Gaussian and impulsive noises that approximate an unknown sparse impulse response of an LTI system are proposed. They are using the sigmoid cost function and based on the Least-Mean Mixed-Norm (LMMN) adaptive algorithm. It is shown by simulations that the proposed sigmoid LMMN (SLMMN) algorithms that exploit sparsity-enforcing penalties achieve superior performance to other competing algorithms in the sparse system identification context.