A Novel Stopping Criterion for Turbo Decoding

Turbo decoder estimates message values sent from transmitter by computing iteratively to get the maximum posteriori probability. It will be inefficient if the number of iteration is set irrationally. To reduce decoding process latency, lots of stopping criteria for turbo decoding were proposed. Those methods also increase computation complexity at each iteration. The computation complexity may become issue in the real implementation in terms of gate count, power consumption, and computing latency in the end. In this paper, we propose a novel stopping criterion called convolution-sum (CS) stopping criterion with lower computation complexity. Turbo decoding achieves more efficiency with CS stopping criterion. The result of simulation indicates that the number of iteration with CS stopping criterion will be reduced by 30% at most when the signal-to-noise ratio is between 0.5 dB and 2 dB while maintaining turbo decoding performance