A Family of Blocked Proportionate Hybrid Error Criterion Algorithms

Abstract

A family of blocked proportionate normalized least mean mixed-norm (BPNLMMN) algorithms using hybrid error criterion are devised and investigated for estimating block-sparse channels that are always considered in networks and satellite echo channels. The derived BPNLMMN algorithms are realized by considering two different errors in the least mean algorithm and incorporating hybrid-norm constraints into the constructed cost function to fully use the prior cluster-sparse information in the systems to be estimated. We carefully derive the BPNLMMN algorithms and provide a completed analysis of the BPNLMMN algorithms for cluster channel estimations. The obtained results from various simulations illustrate that the BPNLMMN algorithms provide superior performance under various inputs and perform better than existing least mean mixed-norm algorithms.

Keywords

• Adaptive filter
• least mean mixed-norm
• proportional updating method
• hybrid norm
• block-sparse techniques