BLIND JOINT IDENTIFICATION AND EQUALIZATION OFWIENER-HAMMERSTEIN COMMUNICATION CHANNELS USING PARATUCK-2TENSOR DECOMPOSITION

摘要

In this paper, we consider the blind joint identification andrnequalization of Wiener-Hammerstein nonlinear communicationrnchannels. By considering a special design of the input signal, wernshow that output data can be organized into a third-order tensor. Wernshow that the obtained tensor has a PARATUCK-2 representation.rnWe derive new results on uniqueness of the PARATUCK-2 modelrnby considering structural constraints such as Toeplitz and Vandermondernforms for some of its matrix factors. We also constrain therninput signal to belong to a finite alphabet. Then an AlternatingrnLeast Squares (ALS) algorithm is proposed for estimating the factorsrnof the PARATUCK-2 model and therefore the parameters of thernWiener-Hammerstein channel and the unknown input signal. Thernperformances of the proposed joint identification and equalizationrnmethod are illustrated by means of simulation results.