Models and efficient receivers for communication channels with memory (Ph.D. Thesis abstr.)

Abstract : In the present work, some aspects of digital communications over channels with memory are studied. The basic objective is to model channels with memory and to develop more efficient and reliable communication techniques over such channels. In particular, the receiver design problem for channels with memory is considered and then the relationship of receivers and channel models is examined. In the receiver design problem, fading is assumed to be the source of channel memory. First, the idea of receivers with memory is investigated and a suboptimal receiver with one-bit memory is derived which performs better than the optimal receiver without memory. A receiver with large memory is devised which consists of an estimator and a detector. The decision rule adapts to the channel conditions based on the information provided by the estimator. The performance of the receiver is examined by computing the average probability of error. The estimator is a limited-memory decision-feedback estimator, the estimation criterion being the MMSE. Finally, a methodology is described which can be used to develop channel models based on the physical processes involved. Two measures are defined which quantitatively characterize the correlation of errors and then the relationship of the receiver design problem to channel modeling is examined. (Author)