Asynchronous implementation of an event-driven adaptive FIR filter

Abstract

This work aims at implementing an asynchronous FIR adaptive filter, based on the Recursive Inverse (RI) adaptive algorithm. Previous work has presented the proposed adaptive filter algorithm and has shown that the algorithm's performance is similar to that of the Recursive Least Squares (RLS) algorithm. Moreover, it offers better performance than the Transform Domain (TD) algorithms, i.e. the TD LMS with Variable Step-Size (TDVSS) in stationary environments. The asynchronous logic has been chosen because of its unique low-power characteristic towards stationary events. The asynchronous-based architecture has been designed to be fast enough to accommodate the iterative computation of the filter coefficients while being accurate to ensure a minimum number of iteration, and a fast convergence. This paper presents an overview of the proposed architecture, as well as performance comparison between the RI and the RLS algorithm. Preliminary test shows promising results, nevertheless some optimization is required to reduce the complexity of the design and to increase the accuracy of the computation.