Digital transmission of information over a radio communication channel suffers from distortions arising from the time-varying multipath propagation of signal, and there has been a large number of research activities for reliable communication over such channels around the world. Among them, the orthogonal frequency division multiplexing (OFDM) technique has been attracting great attention in mobile radio communication fields because of its inherent advantages over other modulation schemes in fading multipath channels.
OFDM is one of the possible realizations of multi-carrier modulation (MCM) scheme. In OFDM, information is transmitted over several narrowband subchannels by frequency division multiplexing. It allows spectral overlapping between adjacent subchannels but maintains the orthogonality between them. The modulation and demodulation of OFDM can be easily implemented with the discrete Fourier transform (DFT). OFDM can achieve high spectral efficiency even though it has a long symbol duration compared to single carrier modulation. The long symbol duration and narrow frequency band occupied by each subchannel result in great advantages to the communication over multipath channels.
This dissertation investigates the maximum-likelihood (ML) receivers to improve the performance of OFDM when it is employed in the transmission of information over time-varying multipath fading channels. In this dissertation, we derive the distortions resulting from these channels and propose several compensation and detection techniques that can improve the performance of OFDM.
First, we examine the detection problem of signal from unknown random channel. We derive the ML receiver for OFDM for combating the effects of multipath propagation. Under the assumption of slow fading condition, the demodulated OFDM signal at each subchannel experiences the multiplicative fading, that is, frequency nonselective fading distortion. Based on this type of distortion, differential modulati...