Set-Membership Affine Projection-Like Algorithm with Evolving Order

In this paper, we apply the framework of setmembership filtering (SM) to the Affine Projection-Like algorithm (APL), aiming to decreasing the update rates and the computational complexity. In addition, we propose an evolutionary method that adjust the projection order in each iteration. The resulting algorithm is the Set-Membership Affine Projection-Like algorithm with evolving order. The proposed algorithm retains the convergence speed and final misadjustment of the APL, furthermore, as demonstrated by simulation results, the number of updates and processing time can be reduced almost 80% and 50% respectively, in system-identification and acoustic echo cancellation applications, thus achieving meaningful computational savings.