Genetically-enhanced Multi-dimensional Constellation Design for Wireless Systems

Abstract

In wireless communications, the design of digital constellations and the associated bit labeling methods have a large impact on the robustness of transmissions. With the aid of increased minimum Euclidean distance (MED), high-dimension constellations have shown notable performance improvements against their conventional two-dimensional (2-D) counterparts, such as conventional quadrature amplitude modulation (QAM). In this paper, we propose a new method that exploits genetic algorithm (GA) to construct multi-dimensional constellations with optimized MED and bit labeling. More specifically, the GA-based design method is capable of producing both new constellations that have greater MED values and near-optimum labeling that yields improved performance, when compared with conventional schemes. A few design examples and simulation results are provided to demonstrate the benefits of the proposed constellation and labeling design method.