Repeat-Accumulate Signal Codes

TL;DR

This paper introduces repeat-accumulate signal codes (RASCs), a new class of state-constrained signal codes, with efficient filter optimization via Monte Carlo density evolution and a low-complexity decoding algorithm, achieving near-Shannon limit performance.

Contribution

The paper proposes RASCs, analyzes their asymptotic behavior with MC-DE, and develops an EMS decoder that reduces complexity significantly while maintaining performance.

Findings

Noise threshold within 0.8 dB of Shannon limit

EMS decoder reduces complexity to less than 25%

Performance degradation less than 1 dB with reduced complexity

Abstract

State-constrained signal codes directly encode modulation signals using signal processing filters, the coefficients of which are constrained over the rings of formal power series. Although the performance of signal codes is defined by these signal filters, optimal filters must be found by brute-force search in terms of symbol error rate because the asymptotic behavior with different filters has not been investigated. Moreover, computational complexity of the conventional BCJR used in the decoder increases exponentially as the number of output constellations increase. We hence propose a new class of state-constrained signal codes called repeat-accumulate signal codes (RASCs). To analyze the asymptotic behavior of these codes, we employ Monte Carlo density evolution (MC-DE). As a result, the optimum filters can be efficiently found for given parameters of the encoder. We also introduce a low-complexity decoding algorithm for RASCs called the extended min-sum (EMS) decoder. The MC-DE analysis shows that the difference between noise thresholds of RASC and the Shannon limit is within 0.8 dB. Simulation results moreover show that the EMS decoder can reduce the computational complexity to less than 25 % of that of conventional decoder without degrading the performance by more than 1 dB.