Extended distributed genetic algorithm for channel routing

In this paper, we propose a new parallel genetic algorithm (GA), called extended distributed genetic algorithm (EDGA) for channel routing problem. The EDGA combines the advantages of previous parallel GA models, viz., master/slave GA model and distributed GA model. In EDGA, the root processor executes the conventional genetic algorithm with global selection on total population and the remaining processors execute conventional genetic algorithm with local selection on subpopulations. After certain number of generations, the total population on the root processor and the subpopulations on the remaining processors are unchanged, and the process is repeated till terminating conditions are reached. This incorporates features of both global and local selection in the proposed EDGA. The EDGA is designed to obtain good speedup, global optimal solution, and full utilization of the parallel system. We have implemented master/slave GA, distributed GA, and the proposed EDGA in C on a transputer-based parallel MIMD machine and compared their performance. It is found that the EDGA achieves higher speedup than both master/slave GA, and distributed GA