Diversity and equalization in digital cellular radio.
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This thesis analyses and quantifies the performance of a range of important diversity combining receivers operating in a digital cellular radio environment with linear modulation, frequency-selective Rayleigh multipath fading, and co-channel interference (CO). The following three types of diversity receiver are treated: (i) the maximum likelihood receiver (the best-possible receiver); (U) optimum linear receivers; and (iii) memoryless linear combining receivers with and without post-combiner equalization. The frequency-selective multipath fading channels are statistically characterized by a delay spectrum with an associated root-mean-square delay spread. The co-channel interference on each diversity branch is assumed to be composed of (i) a small number (possibly zero) of dominant interferers; and (ii) a large number of weak interferer; the sum of which is modelled as independent gaussian noise, or noise-like CCI. The potential bit-error-rate (BER) performance of the maxim likelihood receiver is analysed in the case of noise-like CCI alone (i.e., no dominant interferers), using the matched filter bound. The thesis presents a general. exact, and totally analytical solution of the matched filter bound on BER performance taken over the ensemble of multipath channel responses. The key to the solution is the application of the Karhunen-Loeve representation of these channel responses. The BER performance of an optimum linear receiver. optimized according to the minimum mean-square error (MMSE) criterion, is estimated using (i) a range of BER computation methods, including Metzger's algorithm and Saltzberg's bound; and (ii) random generation of the channel responses, using the efficient Karhunen-Loeve method. Five sub-classes of the memoryless linear combining receiver, with and without post-combiner equalization, are studied. They include arrangements of the following combining and equalization schemes: maximal ratio combining, maximal power combining, MMSE combining, MMSE linear equalization, and ideal intersymbol interference (lSI) cancellation. The BER performances of the five receiver sub-classes are estimated using similar techniques to those used for the optimum (MMSE) linear receiver. For quaternary phase shift keying (QPSK) and all the above receivers, the thesis presents a set of numerical results that show the influence of the diversity order. the number of dominant interferers (set to zero for matched filter bounds), the delay spectrum shape, the delay spread, and the signal-to-interference ratio. This extensive set of data shows the power of diversity and equalization at combating lSI and eel in frequency-selective fading environments, and important tradeoffs between receiver performance and complexity.