A Multiple Constrained Routing Selection Of Mobile IP Based On Extended Distributed Genetic Algorithms

Mobile IP is a new recommended Internet protocol designed to support the mobility of a user, where the crux of mobile IP is the searching of routes. Comparing with other routing selection technology, the route of Mobile IP must pass the home agent of mobile user, viz. pass a fixed node, meanwhile the route should satisfy definite constraint of quality of service (QoS). This thesis focuses on the Multiple Constrained QoS Routing Selection of Mobile IP. The main purpose of this problem is searching a best route pass a fixed node which satisfy the request of bandwidth,delay,cost and so on in network.This thesis summarizes the objective,measurement,background and trend of research of QoS routing problem, expounds the fruit of research currently, The difficulties and the existent problems are also discussed.Then, we proposes Genetic Algorithms(GA) to solve the routing selection problems, the principle,limitation and improving method of GA are discussed. Distributed Genetic Algorithms(DGA) as an improvement of traditional GA has good performance in optimized routing problem, for making the search smarter, Combing with the Cascade GA Model, we design a central monitor adjusting the sizes of these sub-populations dynamically and making the migration more directive,the improved algorithms are denominated Extented Distributed Genetic Algorithms (EDGA). We give the basic model and the process of EDGA, the details of each step and the mathematical theories of EDGA are analyzed.After that, we give the network model based on the characteristics of Multiple Constrained QoS Routing Selection of mobile IP, and applied EDGA to this model. EDGA with matrix encoding, choosing the roulette wheel selection operator, assuring the best individuals participate in the crossover operator; choosing the uniform crossover operator, assuring the algorithms searching the whole defined space; choosing mutation operator based on matrix encoding, assuring the algorithms could skip the local minima.; choosing dynamic migration operator, the sub-populations which have better evolutional ability will have larger space and more diversiform genes searching for optimization by absorbing the best individuals from other poorer sub-populations, making the evolvement of the whole population more directive.Chapter 5 performs the results of Simulation. On condition of same controls parameters and optimized problems,the experimental results show that Comparing with the other routing algorithms, the EDGA has a significant superiority in both speed of convergence and precision of optimized results. The EDGA performs efficiently in multiple constrained QoS routing selection of mobile IP.Finally, the whole work of this thesis and the future research have been concluded and discussed.