Solving TSP with Distributed Genetic Algorithm and CORBA

Distributed Genetic Algorithm (DGA) has been used to parallel computing and, Common Object Request Broker Architecture (CORBA) is, popular in software integration. In order to reduce the complexity of DGA software development and to minimize the maintenance costs, DGA software with CORBA and Java (DGAandCORBA) is proposed. The advantages and software architecture of DGAandCORBA are illustrated in detail. China Traveling Salesman Problem (CTSP) has been computed with different parameters and three rules about parameters selections are exposed. It's shown that the solution quality with best-migration policy is the same as that of random-migration policy, the good solution is easily available both in the case when migration rate is about 20parts per thousand. and reproduction generation is between 25,000, and 30,000. Finally, Several experiments are carried out to verify these rules. Experiments show the efficiency of our approach in solving TSP.