Answers are grounded in this paper and cite their sources.
Recently, the Genetic Algorithm (GA) has been developed to produce Labeling Diversity (LD) mapper designs irrespective of constellation size or shape. However, the parameter space of the GA has not been investigated thoroughly. This paper investigates the parameter space of a GA by parameter varying the parameter K. Tuning results show that the parameter K should be set such that K ≤ 4 to achieve close-to-optimal mapper designs in significantly less time. Monte Carlo simulations illustrate that when K = 1, 2, 3 the 16-APSK constellation exhibits a ≈ 1dB gain over all other values of K. The 64-APSK constellation is a peculiar case such that close-to-optimal mapper designs are achieved in terms of fitness values but perform equally to other values of K. Thus, in order to design a close-to-optimal mapper design in the least possible time, the value of K should be K ≤ 4. Furthermore, this study provides a deeper insight into developing more accurate mapper design GAs for future works.
No takes yet. Share an insight, caveat, or question.
Solwa et al. (2022) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: