A Genetic Algorithm for Designing Uncoded Space-Time Labelling Diversity Mappers

Abstract

The extent to which Uncoded Space-Time Labelling Diversity is able to improve the error performance of space-time block coded (STBC) systems is dependent on the binary mappers used to encode information. Existing design techniques are limited; as they either rely on symmetry-based heuristics, or constrain the size of the constellation due to high computational costs. This paper proposes a new genetic algorithm for labelling diversity (LD) mapper design which is applicable to constellations of any shape or size. The proposed algorithm is tested using 16QAM, 64QAM, 32PSK and three 16APSK constellations that do not display diagonal symmetry. The proposed LD mappers match the best heuristic designs for 16QAM and 64QAM. The 32PSK LD mapper produced achieves a diversity gain of ≈8dB when compared to the available heuristic-based LD mapper. In addition, the 16APSK mappers achieve a diversity gain of approximately 3 to 8dB compared to Alamouti-coded STBC systems for the three non-symmetric constellations considered.