The performance analysis of a nonlinear detector

In this work a nonlinear detector for burst signals in background Gaussian noise is proposed. The detector performance is theoretically derived and compared with both a quadratic detector and a fourth-moment detector. The asymptotic performance analysis has been carried out by invoking the asymptotic normality of the statistics and using analytical results from Scarano et al. (1993) to evaluate the mean and the variance of the nonlinear detector. The analytical results, confirmed by Monte Carlo simulations, show that when the signal occupies less than half of the observation window, the nonlinear detector outperforms both the quadratic and the fourth order detector for a wide range of SNR values.