Adaptive importance sampling for bit error rate estimation over fading channels

Abstract

Computer simulation is an essential approach to access the performance of mobile and portable communications systems. However, in the case of a slowly fading channel (where the number of fading cycles dominantly determines the confidence interval of the simulation results), computer simulation time can be prohibitively long in order to obtain an accurate bit error rate (BER) estimate using the Monte Carlo (MC) method. This paper develops an adaptive importance sampling (AIS) technique for BER estimation over Rayleigh fading channels. The AIS simultaneously biases statistical properties of both channel fading process and input Gaussian noise and adaptively searches for the optimal biased density function during the course of simulation. The AIS technique is applied to analyze the BER performance of QPSK with multiple-symbol differential detection. Computer simulation results show that the AIS technique significantly reduces the simulation time compared with the conventional MC technique, and simplifies the procedure of selecting the optimal biased density function.