Parallel linear-constrained RLS-algorithms for adaptive filtering

Аuthors

Djigan V. I.

Company Research and Production Center "Elvis", pass 4922, building 4, p. 2, Moscow, Zelenograd, 124498, Russia

Abstract

Computational procedures are presented for new sliding window linear-constrained regularized recursive least squares (RLS) adaptive filtering algorithms. These procedures are fitted to a parallel implementations by means of four processors. The algorithms are obtained for a general case of multi-channel adaptive filters with unequal number of complex-valued weights in the channels. Special cases of the algorithms can be used for single-channel adaptive filters or for filters with real-valued weights. Complexity estimations and simulation results are considered also for the algorithms. These simulation results demonstrate a computational efficiency of the algorithms in case of non-stationary signal processing. The algorithms can be used to solve various problems related to linear-constrained adaptive filtering for non-stationary signals.