Abstract:

The amplitude probability distribution (APD) of impulse noise can be changed when input single of communication receiver is filtered using a narrow-band filter. It is verified that the Middleton’s Class B model can be made to fit the APD of narrow bandwidth non-Gaussian noise. In this paper, the efficient estimation of the Class B model parameters based on characteristic function with least square gradient method is derived. The considered estimator has fast convergence and low computational complexity, with performance approaching theoretical up-limit for large data samples. Derivation of the estimator for the noise model parameters, and simulation results using this estimator with three unknown parameters are presented.

Key words: non-Gaussian noise, class B noise, parameters estimation