Start Date: January 1, 2008
Alexandropoulos Georgios
The rapidly increasing demands for wireless wideband services and the recent advances in the design and implementation of mobile terminal devices with Internet-based service providing capabilities expedite the introduction of fourth generation (4G) wireless communications systems in the international wireless market. These systems are expected to ensure increased data rates and quality of service in an anytime anywhere basis. Wireless systems that utilize multiple antennas at the transmitter and/or receiver as well as space-time signal processing techniques play a fundamental role in accomplishing the demands imposed by 4G wireless communications systems. Well-known multiple-antenna systems that enable simple implementations are space diversity receivers (SDRs). By properly combining the multiple received replicas of the transmitted signal, SDRs are capable of effectively mitigating the detrimental effects of multipath fading, known as small-scale fading (SSF), that is inherent in wireless channels. SDRs are theoretically known to improve wireless system’s performance compared with conventional single-antenna receivers. This improvement requires that the SSF channels among multiple receiver’s antennas are statistically independent. However, in practical implementations, due to several parameters such as for example the small distance among the receiver’s multiple branches, SSF channels are arbitrarily correlated. This doctoral dissertation presents a theoretical performance study of SDRs operating over arbitrarily correlated SSF channels. Although numerous scientific papers deal with correlated SSF channel modeling and the impact of correlated SSF on the performance of SDRs, their vast majority, which utilizes the statistical properties of multivariate distributions for studying SDRs’ performance, is restricted to special forms of fading correlation and conventional SDR techniques. This happens mainly due the fact that there is a lack of simple mathematical expressions for the statistical properties of multivariate distributions with arbitrarily correlated random variables (RVs) in the literature. Within the framework of this dissertation, firstly, the previously proposed mathematical expressions for the most prevalent statistical properties of the multivariate Rayleigh, Nakagami-m, Weibull and generalized Gamma (ΓG) distributions with various forms of correlation are summarized. Moreover, their capabilities of being utilized for the performance study of SDRs operating over correlated SSF are described. Next, by presenting a new methodology for generating arbitrarily correlated and not necessarily identically distributed (ID) ΓG RVs that is based on arbitrarily correlated Gaussian RVs and the special class of Householder matrices for tridiagonalizing the correlation matrix (CM) of Gaussian RVs, a closed-form upper bound expression for the joint probability density function (PDF) and an analytical upper bound expression in infinite series form for the joint cumulative distribution function (CDF) of arbitrarily correlated and not necessarily ID ΓG RVs are derived. The proposed upper bounds contain several known mathematical expressions for the joint PDF and CDF as special cases. In addition, by approximating the CM of arbitrarily correlated Gaussian RVs with the special class of Green’s matrices, a closed-form approximate expression for the joint PDF and an analytical approximate expression in infinite series form for the joint CDF of arbitrarily correlated and not necessarily ID ΓG RVs are obtained. Furthermore, analytical expressions in infinite series form for the most prevalent statistical properties of the trivariate ΓG distribution with an arbitrary CM and not necessarily ID RVs as well as of the multivariate ΓG distribution with a constant CM and not necessarily ID RVs are presented. The proposed analytic mathematical expressions of all forms for the PDF and CDF of the multivariate ΓG distribution are used for the performance study of selection diversity (SD), maximal-ratio diversity (MRD), and switch-and-examine diversity (SED) receivers over various arbitrarily correlated SSF channels. Firstly, analytical upper bound expressions for the outage probability (OP), average symbol error probability (ASEP) for several modulation formats, and average channel capacity (ACC) in Shannon’s sense of SD receivers operating over arbitrarily correlated and not necessarily ID ΓG fading are derived. Moreover, analytical expressions for the same performance criteria of triple-branch SD receivers as well as analytical approximate expressions for the performance criteria of multibranch SD receivers are presented. Next, by obtaining new analytic mathematical expressions in infinite series form for the most prevalent statistical properties of the sum of any number of arbitrarily correlated and ID Gamma RVs, analytical expressions for the OP, ASEP for several modulation formats, and ACC in Shannon’s sense of multibranch MRD receivers operating over arbitrarily correlated and ID Nakagami-m fading are derived. For the same fading conditions, analytical expressions in infinite series form for the OP and ASEP for several modulation formats of multibranch SED receivers are presented. The tightness of the proposed upper bounds for the performance criteria of multibranch SD, MRD, and SED receivers in various arbitrarily correlated SSF environments, the correctness of the analytical expressions for the same criteria, and the accuracy of the proposed approximations for them are studied in depth through comparisons between numerically evaluated results for the expressions and equivalent results obtained by means of computer simulations that were implemented for this purpose.