| With the wide applications of the group communication in many fields, such as network video conference, video on demand, stock information distribution and remote education, multicast technology has become an important foundation for supporting these applications. As these real-time operations are sensitive to delay, delay-jitter, packet-loss and cost in the network transmission, it is liable for the real-time operations to be affected by the operation portfolios of multimedia business. So it demands QoS to assure real-time communications. Traditional routing protocols just offer"best effert"service rather than QoS. With the rapid development of new high-powered network technologies , the research on multicast routing algorithm and protocols based on QoS has become very essential.Compared with the unicast applications, the multicasting applications need more strict QoS requirements. Multicast QoS routing is how to select multicast routes trees with sufficient resources to meet the requirement of QoS parameters. It is proved to be a NP complete problem. Although many solutions have been proposed to solve QoS problem for network applications, such as admission control, routing selection, traffic shaping, congest control and negotiation mechanism, there still exists many disadvantages. For example, the algorithms are too difficult to realize and can not fit the specialty of multicasting applications.In the past few years, researchers have solved multicast routing problems by the means of computation intelligence. But only one computation intelligence scheme still leads to some shortcommings in solving the problems. It is the elite in the computation intelligence field to complex cooperative optimization scheme and construct the different integrating computation intelligence technologies. The major objective of this dissertation is to study theory and method of computation intelligence integration and optimize the Multicast QoS routing based on computational intelligence integration technology . The author does the utmost to develop multicast QoS routing algorithm with high call accepted success ratio, rapid connection speed and good scalability. The research mainly concentrates on designing algorithms for constructing multicast routing trees with QoS constraints. In order to overcome the deficiency in other algorithms, the problem of multicast trees with QoS constraints is converted into a multiobjective optimization one. Then the computational intelligence technology is used to solve the problem. This dissertation explores a new approach of comprehensive optimizing multicast routing based on computation intelligence. The main contents of the dissertation are as follows:According to multi-objective optimization idea, the centralized static Multicast QoS routing algorithm based improved GAs has been designed. The dissertation have presented two kinds of static multicast QoS routing algorithms. These algorithms are named as the following two algorithms:one is multicast routing algorithm based on heuristic fuzzy genetic algorithm.The other is multicast routing algorithm based on population adaptive immune genetic algorithm. Tree encoding and efficient genetic operation are used in these two algorithms. By these two algorithms does improve traditional genetic algorithm through different strategies. The first algorithm integrats GAs ,fuzzy selection and tabu search. Tabu search can guarantree diverse search. The fuzzy selection can be used to replace traditional selection operation. It can also determine adequate amount of colony to participate in evolution process. The second algorithm can adjust parameters of GAs to meet the requirement of population evolution process. Then the immune operator is introduced to guarantee the diversity of population and the pareto optimize solution. The above-mentioned two algorithms overcome the weakness of GAs, such as poor climbing capability and premature. At the same time, they can improve the searching capability of global optimal solution. Moreover, the second algorithm can simulate the process of competition, propagation and perish in ecosystem and improve the search capability of multiobjective GAs.In order to overcome the shortcoming of centralized algorithm, this dissertation presents decentralized multicast routing algorithm based on cooperative evolution ant colonial optimization. At first, every subpopulation is assigned a objection, such as searching optimal path to meet the need of QoS constraints. Next, these optimal paths can be integrated into a multicast tree to satisfy QoS constraints. At last, the multicast tree can be evaluated by the rule of pareto optimization. Based on the result, pheromone would be encouraged or punished. This algorithm is a decentralized one, it has high connection probability and better expansibility.To solve dynamic problem of multicast member or network topology, this dissertation presents dynamic multicast routing algorithm, which integrats decentralized and centralized algorithm. While a new node joining multicast session, decentralized algorithm can be used to search optimal path between a new node and current multicast trees. When a multicast member leaves multicast session, it can be decided that whether reconstruction operation is carried out or not , on the basis of the impact on the quality of multicast trees. Compared with the static multicast routing algorithm, dynamic multicast routing algorithm has better adaptability and flexibility.This dessertation also studies application-layer multicast technology thoroughly. It presents overlay network multicast routing algorithm based on chaotic immune algorithm, which integerats the advantage of network multicast with the advantage of application layer multicast. The main constraints in proposed algorithm are cost, balance capability and network portfolio. It provides an effective path for the research of multicast technology.All the multicast QoS routing algorithms presented in this dissertation are evaluated by using digital simulation. The simulation results verify their effectiveness.
|
PDF Full Text Request