Performance and modeling of communications in channels with memory.

Abstract

This dissertation considers the performance and modeling of communications in channels with memory. In particular, much of the work focuses on the mobile radio channel---a time-varying channel whose fading variation changes dramatically as a function of the channel correlation. In the first part of the thesis, an analytical framework is presented which characterizes the performance of a communication system in multipath fading as a function of mobile velocity. Closed loop power control, channel coding, and finite interleaving, common methods to mitigate the effect of the channel, are incorporated into the analysis. The analytical model is then used to investigate the sensitivity of these various schemes to the changing channel conditions. Next, the modeling of channels with memory is studied. A new performance criterion is presented which is directly related to the error probability of a representative communication system. The criterion, which is based on the mean-squared error of a minimum mean-squared error predictor, is used to measure the accuracy of Markov models of various order when there is a continuous state space and when the space is partitioned into a fixed number of states. In both cases, results suggest that employing a higher-order Markov model will improve the accuracy of model-based performance characterizations. Then, various design techniques for higher-order models are considered. Numerical results confirm that second-order Markov models significantly improve the accuracy of characterizing the performance of systems operating over fading channels. The third major focus of this thesis is the fundamental limits of communications in various channels of interest. The performance of a coded communication system using orthogonal modulation is considered in worst-case partial band jamming, the Rician fading channel, and the Nakagami fading channel. In each of these cases, as the size of the signal set grows asymptotically large, error-free communications can be achieved if sufficient signal energy is used.