Next generation wireless systems are being designed to provide ubiquitous broadband link access to information infrastructure. Diversity techniques play a vital role in supporting such high speed connections over radio channels by mitigating the detrimental effects of multiuser interference and multipath fading. Equal gain combining (EGC) diversity receiver is of practical interest because of its reduced complexity relative to optimum maximal ratio combining scheme while achieving near-optimal performance. Despite this, the literature on EGC receiver performance is meager owing to difficulty in deriving the probability density function of the diversity combiner output. This problem is further compounded when the diversity paths are correlated.

Since spatial, pattern, or polarization diversity implementations at a mobile handset are usually limited to a small diversity order with closely spaced antenna elements (owing to cost and ergonomic constraints), any performance analysis must be revamped to account for the effects of branch correlation between the combined signals. This thesis presents a powerful characteristic function method for evaluating the performance of a two-branch EGC receiver in Nakagami-m channels with non-independent and non-identical fading statistics. The proposed framework facilitates efficient error probability analysis for a broad range of modulation/detection schemes in a unified manner.

The thesis also examines the efficacy of an average diversity combiner in slotted direct sequence spread-spectrum access packet radio networks. A two-dimensional EGC diversity combining scheme is introduced, wherein a corrupted packet is retained and combined with its retransmission at the bit level to produce a more reliable packet. The mathematical analysis of the average diversity combiner presented in this thesis is sufficiently general to handle generalized fading channel models with independent fading statistics for a myriad of digital modulation schemes.