Iterative EM channel estimation for turbo-coded DS CDMA receiver under different communication scenarios

Abstract

This thesis proposes a scheme of obtaining an estimate of channel coefficients and noise power spectral density (PSD) using iterative expectation maximization based on a turbo-coded code-division multiple-access (CDMA) receiver under different communication scenarios such as time-varying interference and pulse-band jamming. At the receiver, an initial estimate is obtained with the aid of pilot symbols. The subsequent values of channel coefficient and noise PSD are updated by soft feedback from the turbo decoder. The updated channel coefficient and noise PSD are iteratively passed to the turbo decoder, which results in improved decoding accuracy. The proposed systems are verified through simulations using a structure similar to the Third Generation Partnership Project Long-Term Evolution (3GPP LTE) system under Jakes and Rayleigh fading environments. In addition, this thesis also proposes the scheme of obtaining an estimate of channel coefficients and noise PSD without sending any pilots under a single-user environment. At the receiver, the initial estimate of channel coefficient and noise PSD are obtained without pilots using blind estimation, and then the further estimations are done using expectation maximization. The estimated values are updated iteratively by feedback from the turbo decoder. The updated channel coefficient and noise PSD are iteratively passed to the turbo decoder, which yields improved decoding results. The elimination of pilot symbols sacrifices performance but allows increased energy per transmitted symbol, increased information throughput, or the inclusion of additional parity bits.