EXIT Chart Analysis of Block Markov Superposition Transmission of Short Codes

TL;DR

This paper introduces a modified EXIT chart analysis for BMST of short codes, revealing how decoding thresholds vary with SNR and encoding memory, and confirming results with simulations.

Contribution

It presents a new EXIT chart method that accounts for MI-BER relations, providing insights into the convergence and threshold behavior of BMST codes.

Findings

Threshold curve achieves a lower bound at high SNR

Error propagation affects thresholds at low SNR with increased memory

Finite-length simulations confirm threshold analysis

Abstract

In this paper, a modified extrinsic information transfer (EXIT) chart analysis that takes into account the relation between mutual information (MI) and bit-error-rate (BER) is presented to study the convergence behavior of block Markov superposition transmission (BMST) of short codes (referred to as basic codes). We show that the threshold curve of BMST codes using an iterative sliding window decoding algorithm with a fixed decoding delay achieves a lower bound in the high signal-to-noise ratio (SNR) region, while in the low SNR region, due to error propagation, the thresholds of BMST codes become slightly worse as the encoding memory increases. We also demonstrate that the threshold results are consistent with finite-length performance simulations.