Advances in the Analysis of Adaptive OFDM Systems
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This thesis analyzes and quantifes the performance of adaptive orthogonal frequency division multiplexing (OFDM) systems in conjunction with single and multi antenna systems operating over frequency selective Rayleigh and Rician fading environments. We introduce a generalized transceiver model for adaptive cyclic prefixed (CP)-OFDM and isotropic orthogonal transform algorithm (IOTA)-OFDM. Fundamental issues concerning the BER performance of CP-OFDM and IOTA-OFDM are investigated under typical system imperfections. Furthermore, a throughput comparison of these two types of adaptive systems is given. Next, the peak-to-average power ratio (PAPR) problem of adaptive OFDM is considered. Focusing on wideband channel variations in the frequency domain, we have developed a novel statistical analysis for adaptive multiple input multiple output (MIMO)-OFDM systems. In addition, a central limit theorem (CLT) was developed for a wide range of block-based performance metrics. This thesis aims to present a systematic study of channel variation with a statistical analysis of the MIMO-OFDM channel and system performance. In particular, we focus on the behavior of block based performance measurements by considering the correlation across the frequency bins of the OFDM block. In addition, we investigate the eigenvalue variation of MIMO-OFDM systems across frequency. We show that eigenvalue and link gain changes in frequency can be analyzed and have presented novel results on the distributions and moments of such changes. We have also given expressions for the autocorrelationfunctions (ACFs) of the maximum eigenvalue and the link gain. Of particular interest is the very simple approximation to the ACF of the maximum eigenvalue. This leads to accurate closed-form approximations to the variance and CDF of the eigenvalue differences. We consider three types of adaptive MIMO-OFDM systems; one is based on (diversity mode) maximal ratio transmission-maximal ratio combining (MRT- MRC), while the others are spatial multiplexing techniques using singular value decomposition (SVD) or minimum mean square error (MMSE) receivers with linear precoding. We derive closed-form expressions for the joint cumulative distribution function (CDF) of arbitrarily selected eigenvalues in the same bin and in different bins. Furthermore, for MIMO with MRT-MRC, and MIMO- SVD, the exact mean and variance of the number of bits transmitted per OFDM block has been computed analytically and veried with Monte Carlo simulations.