Synthesis of Algorithms of Adaptive Signal Processing Using of the Nonlinear Blocks with Approximation of Optimal Amplitude Transfer Characteristic

Abstract

The issues of synthesis of the algorithms of adaptive nonlinear signal processing were considered. These algorithms were implemented by using blocks of nonlinear transformation of discriminators with feed-forward and with approximation of the optimal amplitude transfer characteristic for broadband and narrowband channels to process signals influenced by non-Gaussian noise with unknown distribution of instantaneous values or with its envelope. The mathematical formulas were obtained, allowing us to determine both the values of optimal settings for blocks of nonlinear transformation with the amplitude transfer characteristic in the form of a generalized polynomial and the value of the maximum coefficient of noise suppression corresponding to them. It is shown that the proposed algorithms have energy losses resulting from the losses caused by the mismatch of the accepted model of optimal amplitude transfer characteristic and the losses within the mentioned model due to limiting the volume of learning sampling.