STOCHASTIC GRADIENT OPTIMIZATION OF IMPORTANCE SAMPLING FOR THE EFFICIENT SIMULATION OF DIGITAL-COMMUNICATION SYSTEMS

Importance sampling (IS) techniques offer the potential for large spee d-up factors for bit error rate (BER) estimation using Monte Carlo (MC ) simulation. To obtain these speed-up factors, the IS parameters spec ifying the simulation probability density function (pdf) must be caref ully chosen. With the increased complexity in communication systems, a nalytical optimization of IS parameters can be virtually impossible. I n this paper, we present a new IS optimization algorithm based on stoc hastic gradient techniques. The formulation of the stochastic gradient descent (SGD) algorithm presented in this paper is more general and s ystem-independent than other existing IS methodologies, and its applic ability is not restricted to a specific pdf or biasing scheme. The eff ectiveness of the SGD algorithm is demonstrated by two examples of com munication systems where IS techniques have not been applied before. T he first example is a communication system with diversity combining, s low nonselective Rayleigh fading channel, and noncoherent envelope det ection. The second example is a binary baseband communication system w ith a static linear channel and a recursive least square (RLS) linear equalizer in the presence of additive white Gaussian noise (AWGN).