A Comparative Study of the Family of Partial Update Adaptive Filter Algorithms in System Identification Application

The partial update adaptive filters are very useful in different applications of signal processing due to reduction in computational complexity. This paper compares the performance of various partial update adaptive filter algorithms in system identification application. These algorithms are periodic, sequential, and stochastic partial update version of the least mean squares (LMS), normalized LMS (NLMS), and affine projection algorithms (APA). Also, the M-max version of these algorithms is also presented. Simulation results show that the partial update adaptive algorithms have comparable performance with the full update adaptive filters. Furthermore, the computational complexity of this family of adaptive filters is lower than full update version of adaptive algorithms