Performance analysis of digital lightwave systems using efficient computer simulation techniques

Analytically-based methods for evaluating the performance of digital lightwave systems in terms of bit error rates (BERs) are extremely difficult to develop without making restrictive assumptions. A Monte Carlo simulation approach can offer an attractive alternative. However, for typical optical systems, this approach would require an excessive amount of computer time. Importance sampling (IS) is a variance reduction method which can substantially increase the computational efficiency of Monte Carlo simulations. This paper presents an IS method to efficiently evaluate the BERs of direct-detection optical systems employing avalanche photodiode (APD) receivers. Specifically, using a heuristic argument based on large deviations theory, a large class /spl Fscr/ of exponentially twisted sampling distributions for the APD-based receiver is developed. It is then demonstrated that when used as a sampling distribution, the "optimized" exponentially twisted distribution of large deviations theory is the most efficient distribution among the sampling distributions in /spl Fscr/. Further, it is demonstrated that such a distribution would estimate the performance of optical systems with a high degree of accuracy to warrant its possible use as a powerful and flexible tool in computer-aided design, analysis and modeling of fiber-optic transmission systems.<>