Signal constellation design for symbol-level rate adaptation in AWGN channels

This paper considers rate adaptation for the additive white Gaussian noise (AWGN)-channel with unknown channel capacity. We show how multi-layer signal constellations can be optimized layer-wise for particular signal-to-noise ratio (SNR)- points, to provide simple and efficient ARQ-based symbol-level rate adaptation with fine granularity. A turbo-coded system is used as an example showing that the proposed design provides actual communication rates that are close to the channel capacity.