Linear-prediction-aided differential detection of CPM signals transmitted over Rayleigh flat fading channels

A receiver structure, called linear-prediction-aided differential detection (LPADD), for detection of continuous phase modulation (CPM) signals with Nyquist III pulse shaping in fast Rayleigh flat-fading channels is presented. This technique was found to be capable of significantly reducing the bit error rate of conventional differential detection (CDD) at high signal-to-noise ratios. The algorithm reduces the phase noise power from the fading process at the differential detector. Therefore, performance benefits are only seen when the phase noise becomes the dominant source of degradation. Simulations showed that with practical predictor lengths the error floor of conventional differential detection can be reduced by more than a factor of 30