Performance analysis of modulation with unequal power allocations over fading channels: A genetic algorithm approach

Modulation with unequal power allocation (MUPA) is an unequal error protection technique (UEP) used when selective channel coding is barely suitable. This is the case of bandwidth constrained wireless communications where the overhead introduced by error control coding often turns out inefficient systems. On the other hand, MUPA protects data which exhibit different sensitivity to channel errors by distributing different transmission power to the modulation symbols according to the individual bit error sensitivities. MUPA has been studied in case of AWGN channel. In this work we propose an optimal power allocation scheme for the transmission of sensible data over fading channels. As to the optimization we rely on genetic algorithms (GA) which guarantee fast approximate solution with reduced complexity. Performance is evaluated on the Gilbert-Elliot (GE) model, extensively used in literature to reproduce the memory associated with radio channels.