CPFSK Signals Detection in Bursty Impulsive Noise

In some communication scenarios, non-Gaussian impulsive noise is dominant, for which conventional Gaussian noise based signal detection algorithm cannot achieve desirable performance due to the mismatch of noise model. Therefore, it is necessary to develop novel signal detection algorithm for communication receiver in non-Gaussian impulsive noise. In this paper, we focus on the detection of continuous phase frequency shift keying (CPFSK) signals for bursty impulsive noise, which is modeled as the stationary m-order α-sub-Gaussian (αSG(m)) process. For coherent detection, a sequence detection algorithm is proposed based on Viterbi algorithm by utilizing the Markov property of αSG(m) noise. To reduce the computational complexity, a multidimensional myriad branch measure is proposed to replace the complicated Maximum Likelihood(ML) branch measure. For non-coherent detection, a multiple symbols aided method is applied to minimize the detection error. The multidimensional myriad measure is also used. For the sake of comparison, the Gaussian measure derived from Gaussian distributed noise is also considered in both coherent and non-coherent detection. The simulation results show that the performance of the myriad measure based algorithm can closely approach that of the ML measure based algorithm, and is much better than that of the Gaussian measure based algorithm.