基于遗传算法的分组调度

Apply GA in input-queued switch

现代通信网络中的大规模并行传输,意味着在交换机中的一个时间片内,可能会有不止一个分组申请抵达同样的目的端,即引起了无冲突传输的问题。运用改进遗传算法成功实现了分组调度,并在适应度函数的定义中,设计了两个新的策略,即队首分组等待时间越长,其对应的个体竞争力越强;分组队列越长,其队首分组对应的个体竞争力越强。仿真结果表明,运用遗传算法调度,分组的丢失率为0,吞吐量达到了100%。

in modem communication networks, the parallel use of multiple channels means in one time slot, there may be more than one packet that may have the same destination, thus cause the problem of conflict-free traffic assignment. The problem of the conflict- free packet switch is successfully solved by using genetic algorithm, during the process of evolution, two new strategies are devised, the first is that the longer the first packet in any queue waited, the greater competence the relevent individual had; the second strategy is that the longer any queue became, the greater competence the relevent individual of the first packet of the queue had. The result of the simulation suggest that by the use of GA, the rate of lost packets is zeros, and the throughput rate can reach 100 %.