Abstract: |
Optimal constellation design is important in military digital communications for Quadrature Amplitude Modulation (QAM). Optimization realizes a reduced probability of bit error (Pb) while keeping the same bandwidth and power for transmitting the signal. Constellation shapes currently used in QAM include rectangle, triangle, hexagonal, and concentric circles. In this study, two dimensional (4, 8,..., and 256)-ary constellations at specific normalized signal to noise ratios Eb/No having lower Pb are sought. Various models are designed to provide a Multi-objective Evolutionary Algorithm (MOEA) with a near exact model to utilize as a fitness function. MOEA found solutions are tested for merit using a Monte Carlo simulator. Comparisons of Eb/No vs. Pb between the rectangular constellation and new designs are illustrated. New designs are shown to be different and comparable to the standard constellations used today |
