Receiver structures for wireless mobile channels
This thesis deals with the receiver structures for signals transmitted over time-varying channels. Two problems are addressed. One is the simplified receiver design for Rayleigh fading channels when the modulation has a large constellation (such as the quadrature amplitude modulation-QAM) or when coding with interleaving is applied. The other is the optimum receiver design for general time-varying channels modelled as a finite state Markov channel (FSMC). A simplified maximum likelihood sequence estimation (MLSE) receiver for QAM signals over the frequency nonselective Rayleigh fading channel is proposed first. Adaptive FIR (finite impulse response) prediction filters are applied based on the per-survivor processing. This avoids the computational complexity due to the prediction filter design that the original MLSE receiver proposed in (78) needs. In addition, a measure to determine the memory length of the received signal is proposed based on the Kullback-Liebler "distance" for probability density functions. This measure determines the complexity of the MLSE receiver in (78) and can be evaluated easily for the Rayleigh fading channels. Suboptimal receivers for coded signals with interleaving over both frequency nonselective and selective channels are also studied. First a two-stage receiver structure is proposed for frequency nonselective Rayleigh fading channels, where the first stage compensates the channel by generating asymptotically optimum channel estimates and the second stage makes soft-decision decoding using the channel estimates from the first stage. Iterative joint detection/decoding receivers are then developed based on the two-stage receiver to further improve the error performance. The decoded sequence from the second stage is iterated to further improve the channel compensation in the first stage. Both frequency nonselective and frequency selective channels are considered. Computer simulations are done to evaluate the performance of the receivers. The design of MLSE receivers requires the knowledge of multidimensional distribution of the received signal. For general time-varying channels, this is not possible. This thesis considers modelling such channels by a finite state Markov channel (FSMC) model and the corresponding receiver design. Optimal sequence detection and channel state estimation over frequency nonselective FSMCs is developed.