위치벡터 기반 인접투사 부호 알고리즘을 이용한 성긴 시스템 식별

Abstract

In this paper, an improved affine projection sign algorithm utilizing a position vector is proposed to achieve sparse system identification with fast convergence. Specifically, information on non-zero coefficients’ positions is utilized as the position vector to accelerate the convergence in sparse system identification. To update both a position vector and a weight vector, the proposed algorithm employs two adaptive filters that are optimized in an alternating way. Computer simulations show that the proposed algorithm outperforms the conventional algorithms in terms of convergence rate.