Joint distribution matching improves data rates in multiple communication channels

Divergence-Minimizing Distribution Matching for Parallel Channels and Channels with Memory

Information Theory

Summary

Sending information over several communication channels at once can be tricky because the channels have different properties. The paper explains a method to pick sequences of data that best fit the target distribution of these channels, even when the channels remember past data (have memory). Using a special shaped method across multiple channels at once helps reduce data rate losses, especially for short messages. The method was tested with simulations using real-world communication coding standards and showed clear benefits.

What this means in practice

Authors

Francesca Diedolo, Gerhard Kramer

Abstract

Theory for distribution matchers (DMs) is extended to distributions with memory. The divergence-minimizing DM is shown to select the sequences with the highest target probability, as in the memoryless case. A scaling law for divergence is extended to distributions driven by innovation processes. The theory is applied to parallel additive white Gaussian noise channels. A modified enumerative sphere-shaping (ESS) method with a weighted energy constraint is used in implementations. An illustrative example with three channels shows that joint ESS across channels reduces the rate loss by a large factor compared to product DMs at short blocklengths. The gains are confirmed by simulations with probabilistic amplitude shaping and a 5G-NR low-density parity-check code.