Application Of Random Matrix Theory In Mimo Communication System

Advanced wireless communication systems are characterized by an increasingly dense deployment of different types of wireless access points. Since these systems are primarily limited by interference, multiple-input multiple-output (MIMO) techniques as well as coordinated transmission and detection schemes are necessary to mitigate this limitation. As a consequence, wireless communication systems become more and more complex which requires that the useful mathematical tools for their theoretical analysis must evolve. In particular, these must be able to take the most important system characteristics into account, such as fading, path loss, interference, and imperfect channel state information (CSI).The aim of this thesis is to develop such tools based on large random matrix theory (RMT) and to demonstrate their effectiveness with the help of several practical applications. First of all, the random matrix theory (RMT), some relevant conclusions and the common MIMO wireless channel model are introduced briefly. Then, the characteristics of this common channel are analyzed with the approximations of the system performance (e.g.. in terms of mutual information. achievable rates, and signal-to-interference-plus-noise ratio (SINR)).Secondly, two applications of the random matrix in more complex wireless system model are introduced. These include the performance analysis of network MIMO and large-scale MIMO systems. The methods introduced in this paper provide deterministic approximations of the system performance which become arbitrarily tight in the large system regime with an unlimited number of transmitting and receiving devices. Although the analytical result is based on large system limit, simulation results are used to show that the random matrix method provides very accurate approximations for even small system dimensions. This leads in many cases to surprisingly simple and close approximations of the finite-size system performance and allows one to draw relevant conclusions about the most significant parameters. These methods could be a way to provide a deterministic abstraction of the physical layer which substantially reduces the system complexity.