An efficient block floating point implementation of the LMS algorithm

An efficient scheme is presented for implementing the LMS-based transversal adaptive filter in block floating point (BFP) format which permits processing of data over a wide dynamic range at a processor cost marginally higher than that of a fixed point processor. Appropriate BFP formats for both the data and the filter coefficients have been adopted and adjustments made in filtering as well as weight updating operations in order to sustain the adopted format and also to prevent overflow in both these operations jointly. For the presented method to work properly, the algorithm step size is to be chosen below an upper limit, which is, however, not very restrictive when compared with the upper bound for convergence, thereby having marginal effect on convergence speed.