Thresholds of Braided Convolutional Codes on the AWGN Channel

TL;DR

This paper analyzes the decoding thresholds of braided convolutional codes on the AWGN channel, demonstrating that spatial coupling improves threshold accuracy and approaches channel capacity as coupling memory increases.

Contribution

It introduces a threshold analysis method for BCCs on AWGN channels using Monte-Carlo density evolution and compares it with erasure channel predictions, highlighting the benefits of spatial coupling.

Findings

Spatial coupling improves threshold prediction accuracy.

Predicted thresholds approach capacity with increased coupling memory.

Unpunctured ensemble thresholds aid in predicting punctured ensemble performance.

Abstract

In this paper, we perform a threshold analysis of braided convolutional codes (BCCs) on the additive white Gaussian noise (AWGN) channel. The decoding thresholds are estimated by Monte-Carlo density evolution (MC-DE) techniques and compared with approximate thresholds from an erasure channel prediction. The results show that, with spatial coupling, the predicted thresholds are very accurate and quickly approach capacity if the coupling memory is increased. For uncoupled ensembles with random puncturing, the prediction can be improved with help of the AWGN threshold of the unpunctured ensemble.