Adaptive Modified Versoria Zero Attraction Least Mean Square Algorithms

Low complexity and ease of implementation provided by zero-attraction-based least mean square (LMS) algorithms have made them popular candidates for sparse system identification. In this brief, a new sparsity aware norm based on a modified Versoria function is proposed, and utilized to develop a novel Versoria zero-attraction LMS (VZA-LMS) algorithm. Moreover, to make the proposed algorithm independent of the zero-attraction parameter, an adaptive zero attractor VZA-LMS (A-VZA-LMS) algorithm is also derived. The bound on learning rate, which ensures convergence in the mean and mean square sense, is derived. Computational complexity of the proposed algorithms are also compared with the existing zero-attraction-based LMS algorithms. Simulation results show that the proposed algorithms provide improved performance over the state-of-the-art algorithms.