Performance analysis of M-QAM in a non-Gaussian environment

| This paper considers a class of nonGaussian noise modeled as a combination of a white Gaussian process and a shot noise process as suggested by Bello and Esposito [1]. With this noise model as input of a quadrature demodulator, the noise output vector is shown to have circular symmetry. In the particular case of Rayleigh distribution for the noise envelope amplitude, the symbol error probability for M-QAM systems is exactly obtained and numerically computed with any desired precision to cover a large range of parameters values. General results show that the system performance depends on two parameters: the average number of impulses occurring in the detection interval, and the ratio of impulsive noise power to Gaussian noise power. Furthermore, in all situations, there is a threshold phenomena which separates into two distinct regions: for high SNR, it is mainly the impulse noise which a ects the performance of the systems; for low SNR, white Gaussian noise dominates.