A set-membership mixed-tone binormalised LMS-based per-tone DMT equalisation

This paper presents a novel structure of the set-membership filtering adaptive algorithm with the framework of a mixed-tone per-tone equalisation in discrete multitone systems. The idea of two mixed-tone constraint sets of interfering adjacent tones are extended to the concept of set-membership binormalised least mean square algorithm. The proposed algorithm provides the fast convergence and low computational complexity with regard to the normalised data reusing algorithm. Simulation results show the performance of proposed algorithm in terms of faster convergence rate and robustness.