Application of importance sampling to performance analysis of digital communication system

Abstract

Monte Carlo method is the most important tool for the designers in communication systems and data transmission networks. In fact, while designing a new system, it is too costly to build a prototype before being sure of its quality. In such situation, the performance of the new system can be analyzed by the simulation technique, where the Monte Carlo method helps simulating the system activities on computers. The most used parameter describing the performance of any data transmission is the bit error rate Pb (BER). When Pb is very small, to estimate Pb using Monte Carlo method is time consuming. To improve the quality of Monte Carlo technique, we will have to transform the problem with a high variance into a problem with much smaller variance. This transformation would drastically reduce the simulation time. This approach is called “Importance sampling technique”. My thesis work consists of doing literature research, trying to understand the “Importance sampling technique”, some important algorithms published in the open literature and to reproduce known advanced results in the domain of performance evaluation of communication systems. The main result of this thesis is the generalization of the importance technique to class A non Gaussian perturbation and, if the available time allows, to fading channels