New Simple Evaluation of Error Probability of Bit-Interleaved Coded Modulation using the Saddlepoint Approximation

This paper presents a simple and very accurate method to evaluate the error probability of bit-interleaved coded modulation. Modelling the channel as a binaryinput continuous-output channel with a non-Gaussian transition density allows for the application of standard bounding techniques for Gaussian channels. The pairwise error probability is equal to the tail probability of a sum of random variables; this probability is then calculated with a saddlepoint approximation. Its precision is numerically validated for coded transmission over standard Gaussian noise and fully-interleaved fading channels for both convolutional and turbo-like codes. The proposed approximation is much tighter and simpler to compute than the existing techniques and reveals as a powerful tool for practical use.