On the exponential twisting in efficient Monte Carlo simulation

Various variance reduction techniques known as importance sampling methods have been employed successfully in the communications and statistical literature to reduce significantly the computational burden of brute-force Monte Carlo. This paper illustrates the use of large deviations theory as a powerful tool for designing highly computationally efficient and flexible importance sampling schemes. As an application, the simulation of fiber optic transmission systems, is considered.<
>