Efficient Window-Based Stopping Technique for Double-Binary Turbo Decoding

In order to reduce the number of computations for convergent values, a window-based stopping criterion is proposed for double-binary convolutional turbo decoding. The proposed technique can detect and tag convergent windows by detecting the a posteriori values of each window with a threshold. Thus, the number of computations for convergent windows can be reduced in further iterations. Additionally, the proposed algorithm employs the hard bits to provide approximate values for extrinsic value exchanging. From simulation results obtained using the WiMAX convolutional turbo code, we found that the proposed stopping technique can reduce around 47% of the convergent windows at E_b/N₀ of 1.4 dB with a small coding gain loss.