Performance Measure of Turbo Code Using LLR Histogram

Turbo codes have been deployed in many cutting-edge technologies because they can achieve very high coding gains. Turbo decoders deploy at least two Soft-Input-Soft-Out (SISO) decoders, which operate iteratively to incorporate their results to conclude the output. The soft outputs from the used constituent SISO decoders develop gradually along the iterations. This development is studied and analyzed in this work to understand the dynamics leading to the results. Histograms statistically group and visualize the soft results for further analysis and study. A method is proposed to evaluate the decoding performance based on the density of the values of the soft outputs within the histogram. Results show that the performance is inversely related to the ratio of the values of the soft outputs within the near-zero bins within the histogram. The proposed method can be deployed at the decoder to provide an early indication of the reception and whether it has the potential to be correctly decoded or not. This early decision can save the decoding resources.