Optimal and near-optimal demodulation for frequency-selective Rayleigh fading channels
Abstract
This work is a systematic study of the optimal and near-optimal demodulation for frequency-selective Rayleigh fading channels with or without training sequences. When training sequences are allowed, the optimal sequence demodulator that only uses training sequences for channel estimation is analyzed and the union bound of performance is derived. Additionally, some properties of unit training sequences are obtained that can be used to optimize training sequence design. When the training sequences may not be available, an optimal soft output demodulator is derived for the channel modeled by a correlation function. The optimal demodulator, which is a joint channel and data estimator, consists of recursive likelihood computations for each possibly transmitted data sequence. Two methods of complexity reduction are explored: likelihood computation reduction techniques and survivor reduction techniques. For channel taps modeled as autoregressive (AR) processes, the optimum symbol-by-symbol demodulator is derived and shown to consist of joint data and Kalman filter (KF) channel estimator. Additionally, a method to model mobile channels as AR process is suggested. A demodulator with an extended KF is proposed that jointly identifies and tracks the channel and the unknown parameters in AR channel models. A comprehensive simulation study is presented which explores performance boundaries of these algorithms. These algorithms provide near optimal performance that can be used as benchmark for performance comparison. Although blind, these algorithms are also compatible to the schemes of transmitting known symbols. Additionally, they are robust to parameter mismatches in the channel model such as Doppler spread.
Degree
Ph.D.
Advisors
Gelfand, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.