UC San Diego

Orthogonal Frequency DivisionMultiplexing (OFDM) is an efficient multicarrier modulation technique for high data rate broadband wireless communications. The ability of OFDM systems to combat the effects of multipath propagation with a comparatively simple receiver structure made it the modulation of choice for some of the most prominent wireless technologies such as the IEEE 802.11 wireless local area networks, the digital audio broadcasting (DAB), terrestrial digital video broadcasting (DVB-T) and the most recent IEEE 802.16 wireless metropolitan area networks. OFDM is also considered for the IEEE 802.20 mobile broadband wireless access standard currently under development and for the ultra wide band (UWB) standard under the multiband OFDM alliance. The strength of OFDMis undermined when operating in mobile environments. OFDM is sensitive to Doppler spreading as well as frequency synchronization errors and the orthogonality among the subcarriers of an OFDM symbol can no longer be maintained when the channel is time varying. This dissertation addresses the issue of OFDM systems operating in time varying channels, and investigates the impact of channel time variation on the performance of OFDM systems. A solution to minimize the performance degradation is also presented. First, a new analytical approach to investigate the performance of OFDM systems in time varying channels is presented. We show that the multiplicative distortion of the channel, and consequently, the signal-to-interference-plus-noise-ratio (SINR) is a function of many correlated Gaussian random variables. Averaging over all the random variables to obtain a closed form expression for the symbol error probability is a difficult task, and often lead to complicated and intractable expressions. By assuming that the channel is varying linearly across one OFDM symbol, it became possible to express the SINR as a function of only two correlated Gaussian random variables, and hence, a new closed form expression for the symbol error probability could then be derived taking into account the time correlation between subcarriers, which is an important parameter that was not accounted for in previous literature. We also show that under certain conditions, the intercarrier interference can be modeled as a Gaussian process. Secondly, we consider the case when the estimate of the channel at the receiver is either outdated or erroneous due to channel variations. A new model for the channel estimation error is proposed to facilitate the derivation of new closed form bit error probability expressions for OFDM systems in time varying channels and channel estimation error. Such expressions could not be derived in closed form using conventional models when quadrature amplitude modulation (QAM) is used. In the analysis, three types of receivers, namely, the zero forcing (ZF) receiver, the minimum mean square error (MMSE) receiver and the maximum likelihood (ML) receiver are analyzed and compared. Detailed derivation of new closed form bit error probability expressions for the three receivers is presented, along with a comprehensive comparison of the three receivers. Finally, we propose a receiver scheme to improve the performance of single- inputsingle- output (SISO) and space-time-block coded (STBC) OFDM systems in time varying channels. A suitable matrix-oriented system model representation is presented. From the system model, the causes of performance degradation can clearly be identified for both SISO and STBC OFDM systems. This identification will lead us to the design of a receiver which will first isolate, then systematically remove the terms causing most of the degradation. The proposed receiver only requires knowledge of the channel at the receiver and uses simple signal processing operations to improve system performance