Coding for multiple-access and fading channels

Sammanfattning: This thesis treats some information-theoretical and coding aspects of communication over multiple-access and fading channels. It consists of a general introduction and seven papers. In code-division multiple-access (CDMA) communication, many users have access to the channel simultaneously. Since the users interfere with one another, error correcting coding is necessary in order to obtain good performance. In Paper~I, formulas for the channel capacity as a function of the channel parameters and the number of users is given for some idealized noiseless multiple-access channels. Furthermore, two asymptotic results are derived, from which the optimum number of users is easily seen. Paper~II presents and analyzes a novel coding strategy for one of the channels considered in Paper~I. In r~III, frequency-hopped CDMA over a Rayleigh fading channel is treated. An expression for the symbol error probability is derived and used to find the optimum code rate and alphabet size. Direct-sequence CDMA is considered in Paper IV and Paper V. Paper~IV is concerned with comparisons of different coding schemes for the reverse-link. Both frequency non-selective and frequency selective fading are considered. In Paper V, which deals with the forward-link, the concept of pre-coding is introduced. The improvements possible to achieve by using pre-coding are studied for both the additive Gaussian noise channel and the Rayleigh fading channel. When information is transmitted over a channel that is slowly fading and there is a constraint on the acceptable delay, the performance of a coded system can be substantially degraded. In Paper VI, the Gilbert-Elliott (GE) channel model is considered. First, a closed form expression for the codeword error probability is derived. Then, it is shown that the GE channel can be used to accurately evaluate the error probability for block coded transmission over the land mobile channel. Finally, by using the GE channel model, it is found that the performance is not necessarily improved with increased block length, because the interleaver works better for shorter codes. Paper VII is concerned with convolutional coding with imperfect interleaving for the land mobile channel. Analogously to the results found in Paper~VI, the gain by using codes of large complexity when compared to less complex codes is found to be reduced because of the imperfect interleaving.