In this article, a new method for performance evaluation of turbo-like codes is presented. This is based on estimating the probability density function (pdf) of the bit log-likelihood-ratio (LLR) using higher order statistics. We do not restrict ourselves to any specific model for thepdfand try to estimate it directly using a cumulant matching method. Numerical results show a close agreement between the proposed method and simulations. The complexity of this method is similar to the Monte-Carlo simulation with the advantage of providing similar accuracy using significantly fewer samples.

In this paper, a new method for performance evaluation of turbo-like codes is presented. This is based on estimating the probability density function of the bit log-likelihood-ratio using higher order statistics. We do not restrict ourselves to any specific model for the pdf and try to estimate it directly using a cumulant matching method. Numerical results show a close agreement between the proposed method and the simulations. The complexity of this method is similar to the Monte-Carlo simulation with the advantage of providing similar accuracy using significantly fewer samples.