Resource optimization in subband coding and discrete multitone transmission systems.

摘要

A certain class of resource optimization problems encountered in the subband coding of wide-sense cyclostationary (WSCS) signals and discrete multitone (DMT) transmission systems supporting multiple services is considered in this dissertation. Several common features in the problem formulation and solution methodology to both problems lead to this unified treatment.; The first part of the thesis focuses on optimum subband coding of WSCS signals. Specifically, the problem considered is to design the uniform orthonormal bank of filters employed as the subband coder such that the mean-squared distortion is minimized under optimum bit allocation subject to the quantizer bit rate constraints, given the WSCS statistics of the input signal. Optimum bit allocation criteria are derived by finding the optimum distribution of bits to each quantizer of the subband coder. These problems are solved under three practical bit allocation schemes, and two cases wherein the filter bank is maximally decimated and overdecimated. It is shown that in all the cases, a common optimizing solution is a principal component filter bank. The design of such optimal subband coders based on compaction filters is also considered.; The other part of the thesis focusses on DMT systems. The problem of transceiver design is considered when a single DMT system supports different user services endowed with varying quality of service (QoS) requirements. The goal is to minimize the total transmitted power subject to the bit rates and symbol error rates, which constitute the QoS requirements of each user. This minimization is done by determining the optimum distribution of bits/symbol to be assigned to each user subchannel, and the optimum transmitter/receiver pair that achieves the transmission power minimum. It is shown that an optimum receiver transform is a unitary matrix that diagonalizes the autocorrelation matrix of a certain noise vector at the receiver input, and the optimum transmitter transform is its inverse. An issue related to the characterization of channel resistant DMT systems is also discussed.