Symbol Mapping Design for Bit-Interleaved Polar-Coded Modulation With Iterative Decoding

Abstract

In this letter, we design a bit-interleaved polar-coded modulation system with iterative decoding subject to constraints on complexity, error floor, and waterfall region performance. We show that the hill-like trajectory of polar code extrinsic information transfer curves allows us to predict the gap of the bottleneck region that a polar code achieves at a given received signal-to-noise ratio for a given labeling rule. Using the gap, we propose a 3-D linear objective function to optimize for error floor, waterfall region performance, and complexity. The objective function requires Monte Carlo simulations for only three points on the extrinsic information transfer curve of the demapper for a given labeling rule. Our simulations show that the proposed method designs labeling rule and polar code duos that trade off error-floor and waterfall region performance with complexity and latency.