Abstract:

We compensate the biases of the proposed estimators which estimate the m parameter in Nakagamim fading since they all are not unbiased. We study the performances of these m estimators, with emphasis on the biases and the biases compensation of the estimators based on generalized linear model. We explore numerical linear fitting method to get experimental formula for each. According to our analysis and numerical results using Monte Carlo simulation, the bias is a linear function of m parameter and the number of i.i.d samples used in the estimator whenever the estimator is moment based or maximum likelihood based. Numerical results are presented to demonstrate that the linear compensating formulas work well for all the estimators. In real mobile environment, the fading channel may be time variant or m parameter should be estimated in time, which makes the i.i.d samples used in the estimation of m can’t be very large. In this situation, the compensated versions of m estimators will be useful.

Key words: Nakagamim fading channel, m parameter estimation, moment based estimator, maximum likelihood estimator, bias compensation