Authors: Pasteur Poda ¹; Samir Saoudi ^2,3; Thierry Chonavel ^2,4; Frédéric GUILLOUD ; Théodore Tapsoba ¹

• 1 Université Polytechnique Nazi Boni Bobo-Dioulasso
• 2 Département Signal et Communications
• 3 Lab-STICC_IMTA_CACS_COM
• 4 Lab-STICC_IMTA_CID_TOMS

The standard Monte Carlo estimations of rare events probabilities suffer from too much computational time. To make estimations faster, kernel-based estimators proved to be more efficient for binary systems whilst appearing to be more suitable in situations where the probability density function of the samples is unknown. We propose a kernel-based Bit Error Probability (BEP) estimator for coded M-ary Quadrature Amplitude Modulation (QAM) systems. We defined soft real bits upon which an Epanechnikov kernel-based estimator is designed. Simulation results showed, compared to the standard Monte Carlo simulation technique, accurate, reliable and efficient BEP estimates for 4-QAM and 16-QAM symbols transmissions over the additive white Gaussian noise channel and over a frequency-selective Rayleigh fading channel.