Identification of nonlinear satellite mobile channels using Volterra filters

This paper addresses the problem of nonlinear Satellite Universal Mobile Telecommunication System (S-UMTS) identification. The satellite channel is structured as a time-invarying (TIV) filter, followed by a zero-memory nonlinearity (ZMNL) and a TIV linear filter. The non-linearity distortion is due to the on-board satellite amplifier. The mobile channel is modeled by a Time-varying (TV) structure which describes the multipath effects. This paper shows that a specific class of TV Volterra series (with odd kernels) can model this specific satellite mobile bandpass channel. A least squares estimate of the TV Volterra kernels with finite memory is derived for any arbitrary channel input.