Métodos eficientes para a estimação de desempenho em sistemas de comunicação digital.

Abstract

Monte Carlo Trials is the main method used since the 70's to perform communication systems simulation. Nowadays, a variance-reducing method has been used to improve the simulation performance. This method is called Importance Sampling (IS) and it allows to perform faster simulations with same accuracy. In order to use IS, it is necessary to know in detail the underlying stochastic process and to choose specific simulation parameters, which have a strong performance impact. In this work, IS is applied to encoded communication systems, using the Stochastic Gradient Descent (SGD) method to optimize the parameters search. The performance gain sped up the simulation process ten times when compared to the original simulation with IS. Experiments were performed with Hamming encoded system over additive white Gaussian noise (AWGN) channels.