On the Convergence Speed of Turbo Demodulation with Turbo Decoding

TL;DR

This paper analyzes the convergence speed of turbo demodulation combined with turbo decoding, proposing an optimized iteration schedule that reduces complexity and latency while maintaining near-optimal error performance.

Contribution

It introduces an original iteration scheduling method and uses EXIT charts to analyze convergence, significantly reducing demapping iterations with minimal performance loss.

Findings

Proposed iteration schedule reduces demapping iterations by half

Performance loss is less than 0.15 dB with the new schedule

Complexity and power consumption are significantly decreased

Abstract

Iterative processing is widely adopted nowadays in modern wireless receivers for advanced channel codes like turbo and LDPC codes. Extension of this principle with an additional iterative feedback loop to the demapping function has proven to provide substantial error performance gain. However, the adoption of iterative demodulation with turbo decoding is constrained by the additional implied implementation complexity, heavily impacting latency and power consumption. In this paper, we analyze the convergence speed of these combined two iterative processes in order to determine the exact required number of iterations at each level. Extrinsic information transfer (EXIT) charts are used for a thorough analysis at different modulation orders and code rates. An original iteration scheduling is proposed reducing two demapping iterations with reasonable performance loss of less than 0.15 dB. Analyzing and normalizing the computational and memory access complexity, which directly impact latency and power consumption, demonstrates the considerable gains of the proposed scheduling and the promising contributions of the proposed analysis.