Modified Sparsity-aware Set-Membership Affine Projection algorithm

Abstract

Recently, a Sparsity-aware Set-Membership Affine Projection (SSM-AP) algorithm has been developed, which presents lower Mean-Squared Error (MSE), lower misalignment, and lower computational complexity, as compared to other sparsity-aware algorithms under the same conditions. The SSM-AP updating rule is governed by a vector parameter, called the Constraint Vector (CV). Currently, there are two main choices for the CV: one leads to faster convergence, whereas the other yields lower MSE and complexity. This paper proposes an alternative to those choices, which can improve both convergence speed and steady-state MSE of the SSM-AP algorithm with a given CV, while also decreasing the overall number of updates.