Pipelined LMS adaptive filter using a new look-ahead transformation

Abstract

This paper proposes an adaptive algorithm that can be pipelined and an efficient architecture for hardware implementation of the least mean square (LMS) adaptive algorithm. The proposed algorithm provides the possibility of the pipelined processing of the LMS algorithm without degradation of the convergence characteristic. The architecture based on the proposed algorithm is considered and the efficient implementation of the architecture to reduce the amount of hardware required is also considered.