Stochastic soft-input soft-output detection for intersymbol interference channels

We propose a fully-parallel low-complexity hardware architecture of a soft-input soft-output detector based on message passing on a factor graph. Conventionally, the message computation in the graph nodes consists of complex arithmetic operations on probabilities. Here, we represent probabilities by bit-serial streams and transform the arithmetic operations into simple bit-wise operations. This principle is known as stochastic computation and this is its first application to a detection algorithm. We evaluate the bit error rate (BER) performance of the stochastic detector by computer simulations of a communication system with intersymbol interference employing turbo equalization. The results show a performance loss of about 0.5 dB at a BER of 10^-3 compared to the floating-point implementation of the detector.