关键词: 标准遗传算法, 多种群遗传算法, 比例速率约束, 收敛速度, 早熟收敛

Abstract: Aiming at the slow convergence and premature convergence problem of standard genetic algorithm in solving power distribution of multiuser OFDM with proportional rate constrains, a power distribution algorithm based on multiple population genetic strategy is proposed in this paper. The proposed algorithm takes service fairness index for fitness value and ends with the retainable generations of the best individual. All population different control parameters are used and contract with one another through immigrant operator. Simulation results show that the proposed algorithm’s convergence speed （almost 100 generation） is faster than standard genetic algorithm （almost 300 generation） and the result is steadier（almost tend to be zero）. It can maintain the fairness of proportional rate while maximizing the sum capacity.

Key words: standard genetic algorithm, multiple population genetic algorithm, proportional rate constrains, convergence speed, premature convergence