Identification of Nonlinear Channels in Bandpass Communication Systems with OFDM Inputs

Abstract

A general approach to the derivation of simple yet accurate formulas for identifying bandpass nonlinear channels in orthogonal frequency-division multiplexing (OFDM) systems is proposed in this paper. The OFDM is an widely adopted technique for wideband communication applications. An OFDM signal could be highly vulnerable to nonlinearities in a communication link. This makes proper estimation of the nonlinear channel essential for an OFDM system. A nonlinear bandpass communication channel is commonly described by a complex Voletrra series, in which the task of nonlinear channel identification is to determine the Volterra kernels. This task can become very difficult when the order of the nonlinear system increases. In this paper, we explore higher-order statistical properties of the OFDM signal to develop a universal scheme for solving the Volterra kernel identification problem. Based on this scheme, a simple yet accurate formula for identifying Volterra kernels of 5th-order nonlinear OFDM systems is derived. The resulting solution is shown to attain the minimum mean square error (MMSE) in both theory and computer simulation.