Qos Multicast Routing Based On Nastural Computation

With the increasing demand for real-time multimedia communication, the routing algorithms which satisfy QoS guarantees become a hot issue to be studied. QoS guarantees mainly include delay, bandwidth, hop, bandwidth occupancy rate and so on. The key to achieve QoS is to build a multicast tree which satisfies constraint conditions with root as source node and covers all members of the tree nodes. So data could be transferred in parallel along the tree branch to save resources and reduce congestion.MAS is a surging brand-new branch of computer science in the last 20 years and is also the fastest-growing branch of computer science. Its goal is to change large and complex systems into small ones which are coordinated, easy to manage and can communicate with each other. This dissertation presents some researches into the multicast routing problem by combining Multi-Agent system and Immune Clone Selection algorithm.The main researches done are as follows:1. The existing heuristic algorithms and intelligent algorithms for Steiner tree problem are summarized, and the Artificial Immune system, the Multi-Agent system, and their evolutionary ways and studies are briefly introduced.2. Inspired by intelligent grid structure of MAGA(Multi-Agent genetic algorithm), the present author proposes a Multi-Agent Neighborhood Immune Clone Selection Algorithm basing on Multicast Routing (MNICSA) in this dissertation. Before the operation of competition of agents in the neighborhood, the methods at first operate the individual for Immune Clone Selection to guarantee the fixed position of the antibody in the intelligent grid. At the same time, as for the Qos multicast routing, vaccine strategy is introduced to guide evolution and a mutation operator is designed to increase the diversity of the population. Agent cloning ways and mutation are something innovative of this dissertation. Experimental results show that the proposed approach achieves performance in the speed of convergence and the quality of solving.3. To address the outstanding individual rapid proliferation problem, a multiple Multi-Agent multicast routing algorithm has been proposed. Population will be divided into several sub-populations which make intelligent grid structure and evolve separately and set up a mapping relation between sub-populations and memory individual through memory base. A transference operator is designed to speed up the dissemination of good individuals and an individual creation operator is devised to increase the diversity of the population in the algorithm. The establishing of sub-populations intelligent grid and the designing of individual creation operators are innovative in this dissertation. The algorithm improves high solving accuracy within the given times of fitness evaluation.