A study of LMS and SER algorithms for FIR filtering (Q1092868)

A thorough comparative analysis of the algorithms used in finite impulse response (FIR) adaptive filters is presented. The discussed algorithms are the following: the least mean squares (LMS), the sequential regression (SER), and a new proposed one, called Dvoretzky LMS (DLMS). In comparing the discussed algorithms discrete model techniques are used and the corresponding learning curves versus time are graphically represented. The resulting conclusion is that the DLMS algorithm offers some computing advantages over the other ones.