A New Block-based NLMS Algorithm and Its Realization in Block Floating Point Format

We propose a new normalized LMS (NLMS) algorithm, which of convergence and chosen according to: 0 < ~ 1 < 2 for congives satisfactory performance in certain applications in comaprison with convergence [2] and a is an appropriate positive number introduced ventional NLMS recursion. This new algorithm can be treated as a block based to avoid divide-by-zero like situations which may arise when 2 2 simplification of NLMS algorithm with significantly reduced number of multi3⁄4 becomes very small. It may be noted that the parameter 3⁄4 n n ply and accumulate as well as division operations. It is also shown that such a is updated time-recursively in each nth index as recursion can be easily implemented in block floating point (BFP) arithmetic, treating the implementational issues much efficiently. In particular, the core