Adaptive ML neural network based receiver for Q2PSK modulated data-transmission systems

The maximum-likelihood (ML) receiver structure is derived for the detection of digital data symbols transmitted using a four-dimensional signalling scheme in the presence of intersymbol interference and additive white Gaussian noise. The transmitter employs quadrature-quadrature phase-shift keying (Q²PSK) which is a relatively new spectrally efficient modulation scheme and can be considered as two minimum-shift-keying type signalling schemes in parallel. The implementation of this receiver is discussed and an artificial neural network capable of computing the maximum-likelihood estimate of the transmitted sequence is described. Mapping of the derived objective function onto the artificial neural network structure is carried out and a neural network based receiver is presented. Preliminary simulation results are also presented showing promising error rates for several intersymbol interference durations with and without fading