NEURAL-NETWORK MODELING AND IDENTIFICATION OF NONLINEAR CHANNELS WITHMEMORY - ALGORITHMS, APPLICATIONS, AND ANALYTIC MODELS

This paper proposes a neural network (NN) approach for modeling nonlin ear channels with memory, Two main examples are given: 1) modeling dig ital satellite channels and 2) modeling solid slate power amplifiers ( SSPA's). NN models provide good generalization(1) performance (in term s of output signal-to-error ratio). NN modeling of digital satellite c hannels allows the: characterization of each channel component, Neural net models represent the SSPA as a system composed of a linear comple x filter followed by a nonlinear memoryless neural net followed bg a l inear complex filter. If the new algorithms are to be used in real sys tems, it is impost-ant that the algorithm designer understand their le arning behavior and performance capabilities. Some simplified neural n et models are analyzed in support of the simulation results. The analy sis provides some theoretical basis for the usefulness of NN's Tor mod eling satellite channels and amplifiers. The analysis or the simplifie d adaptive models explains;the simulation results qualitatively but no t quantitatively. The analysis proceeds in several steps and involves several novel ideas to avoid solving the more difficult general nonlin ear problem.