Inverse-Gaus ian Distribution: A New Shadowing Model and Its Application to Communication Systems

Real-time fading channels are affected by multipath fading as well as shadowing. In this thesis, we propose the Inverse-Gaussian distribution as a less complex alternative to the classical Log-normal model to describe shadowing effects in composite multipath fading/shadowing environments and to get closed-form solutions for the most important figures of merit. A main motivation for this selection has been the poor accuracy of the analytically friendlier Gamma distribution to approximate the Log-normal distribution, when the latter has a large variance or long tails. As such, we demonstrate that the Rayleigh/Inverse-Gaussian distribution can serve as a more efficient approximation to the prevalent Rayleigh/Log-normal distribution. Our study starts with the performance evaluation of distributed multiple-input multipleoutput (MIMO) systems in composite Rayleigh/Inverse-Gaussian fading channels. The potential of combining MIMO spatial multiplexing gains with macro-diversity gains is realized by distributed MIMO systems that promise to enhance the channel capacity and cell coverage. Capitalizing on some generic bounding techniques, we first derive new closed-form bounds on the ergodic capacity of optimal receivers. In order to gain useful insights into the impact of fading parameters on optimal receivers’ performance, a detailed characterization in the asymptotically high and low signal-to-noise ratio regimes is also provided. In addition, we explore the “large-system” regime and provide asymptotic expressions when the number of antennas grows very large. A similar performance analysis is performed for the achievable sum rate of distributed MIMO systems employing linear zero-forcing and minimum mean-square error receivers. Finally, we perform an effective rate analysis of multiple-input single-output systems over composite Nakagami-m/Inverse-Gaussian fading channels and in the presence of statistical queueing constraints. All the resulting closed-form expressions are validated via a set of Monte-Carlo simulations.