Sobre o Desempenho de Receptores à Taxa de Símbolos em Canais Contínuos

Abstract

Symbol-sampled receivers have had their performance evaluated via Forney's discrete-time transversal filter (DTfF) model for the equivalent channel, which contains a standard matched filter and thus assumes previous knowledge of the continuous-time channel impulse response. If the continuous channel is unknown then the DTfF model provides only an upper bound on the system performance. In this paper, we quantify theoretically the performance loss of an alternative, more realistic model for the equivalent discretetime channel relative to the DTTF model, in terms of information capacity. Moreover, the loss is calculated as a function of the parameters of the transmit filter, which is assumed to be a root-raised cosine filter. We then show by means of computer simulations that this performance loss for the case of an iterative, symbol-sampled, semi-blind receiver, with combined maximum-a-posteriori equalization and turbo decoding can be negligible, which agrees with the theoretical prediction.