Flexible Adaptive FIR Filter Designs Using LMS Algorithm

Abstract

This paper proposes the least mean square (LMS) algorithm based versatile, vector, and fault tolerant adaptive finite impulse response (FIR) filter designs. Here, the M-taps versatile design is to perform the filter operation with the number of filter co-efficients varied from 2 to M. The M-taps vector design is to perform (formula presented) numbers of L-taps filter operations in parallel, where (formula presented). The fault tolerant M-taps filter is to perform the (formula presented) -taps fault free filter operation under the N numbers of faulty filter kernels, where (formula presented). All the existing and proposed designs are implemented with 45 nm CMOS technology. The proposed 16-taps vector adaptive filter design achieves 93% of improvement in throughput as compared with the distributed arithmetic based design.