Implementation of the Hartley-Transform-Based Block LMS Algorithm

Abstract

Abstract This paper presents a cost-effective The Frequency-domain Least-Mean-Square (FLMS) adaptive algorithm (or more generally the Transform-domain LMS adaptive algorithm) [12], [13] has mainly two advantages over the conventional LMS algorithm [19]. The first is that it overcomes the slow convergence of the LMS algorithm by orthogonalizing the input (thereby performing better than the LMS for correlated input signals) and the second advantage is that it can be used for implementing the time-domain Block LMS (BLMS) algorithms as well [18]. The Hartley transform is a newly introduced real-to-real transform that is a suitable replacement for the complex Fourier transform [1] and [2] in several adaptive filtering applications such as adaptive interference cancellation that has wide applicability to problems in telecommunications, biomedical engineering, etc. The realization of the Transform-domain BLMS adaptive algorithm based on the Discrete Hartley Transforms (DHT) and its implementation on the TMS320C30 digital signal processor chip is described.