QPSK RECEIVER BASED ON RECURRENT NEURAL NETWORKS

This paper describes the performance of a complete receiver based on R ecurrent Neural Networks (RNNs). The receiver uses a cascade of two RN Ns: one for equalization and one for demodulation. The simulations hav e been done by considering two cases: a single AWGN channel and it two paths model with the direct and the indirect rays interfering at the receiver. Two different types of digital filters have been used to sim ulate the effect of the transmitter-channel-receiver chain. QPSK digit al modulation has been analyzed. The neural system performs better tha n acoherent receiver (composed of a coherent demodulator and a Transve rsal equalizer) when the channel bandwidth is narrow, i.e. when inters ymbol interference increases. This is due to the fact that the Neural Receiver dynamically changes its Decision Regions for each symbol. Thi s dynamic behaviour has been investigated.