Adaptive asymptotic optimal algorithms for detecting signals in autoregressive noise

Abstract

Asymptotic optimal (AO) algorithms for detection of signals in additive autoregressive noise of order m (m-dependent Markov noise) are synthesized. The algorithms require the storage of m past data samples to achieve optimum performance. It is an AO memory discrete-time detector of a deterministic or quasideterministic signal in autoregressive noise. To assure the change of the detector's parameters as a result of learning the AO algorithm was modified to an adaptive one. Combining the AO algorithm with adaptation it is a powerful approach to overcome a priori uncertainty in information systems. The investigations are carried out by a common approach with many simulation results.