Modification of Theoretical Fixed-point LMS Algorithm for Implementation in Hardware

Abstract

In order to realize the adaptive signal processing in real time, a modified fixed-point LMS algorithm for implementation in hardware was proposed in this paper. Expressions of this modified fixed-point LMS algorithm were deduced from the theoretical fixed-point LMS algorithm according to rounding errors. Data problems were discussed and efficient settling methods were suggested. A modified 8-weight fixed-point LMS algorithm was implemented on FPGA. The correctness and quality of this modified fixed-point LMS algorithm can be confirmed through comparing simulation results between the implemented 8-weight fixed-point LMS algorithm and 8-weight theoretical fixed-point LMS algorithm. Rounding errors were inevitable for the reason of limited word-length. Values of rounding errors can be reduced to the minimum through adjusting the amplitude of input signal, step or others variables. When the range of signal- to-noise ratio can be allowed, this modified fixed-point LMS algorithm was very efficient and useful.