Multicast Routing Optimization And Simulation Based On Ant Colony Algorithm

The improvement of Internet network transmission and processing capabilities enables a substantial increase in web-based applications, such as remote education, video conference, datadata distribution and online gaming and so on. These multimedia applications have the characteristics of big data quantity, high latency, long duration and so on, so it needs to adopt forwarding technology that is different from traditional unicast and broadcast and quality of service assurance mechanisms. Multicast technology is the ideal solution to solve this problem. Multicast is a mechanism which can supports multi-point communication efficiently, and generally it refers to the communication that sends messages from one or more of the source node to multiple destination nodes. This approach makes IP multicast resource-efficient in delivering data to a group of members simultaneously and this subset is called multicast group. QoS is a demand set which is traffic flow to network service when network transmits the traffic flow, and the traffic flow is packet flow associated with a specific QoS from the source to the destination. IETF has proposed a variety of service models and mechanisms to meet the needs of a variety of QoS multicast. The real-time applications often make performance requirements to delay, delay jitter, bandwidth, lossrate, operational costs and other parameters at the same time. When these parameters are independent of each other, selecting the optimal multicast routing to meet multiple parameters restrictions becomes a NP complete problem.In this paper, based on in-depth study of the multicast routing problem and ant colony optimization algorithm, two algorithms are proposed to solve the QoS multicast routing problem. They are ant colony optimization algorithm based on forest growth and distributed ant colony optimization algorithm:(1) For the multicast tree cost minimization problem, namely the Steiner tree problem, which refers to construct the optimal multicast tree using tree optimization function, a new ant colony algorithm is proposed to solve the Steiner tree problem. That is ant colony optimization algorithm based on forest growth. In this algorithm, A forest is formed during the ant movement progress, and for every ant, each step it takes is to make current forest grow. The objective of ant movement is to connect all the trees in the forest to form a single tree, which contains all target nodes. B class topology in the SteinLab Repository and Waxman topology model generated by MRSIM topology generator are used as the test topology. Simulation experiments indicate that performance of the algorithm is superior to KMB,SPH,CDKS and traditional bio-heuristic algorithms. The algorithm is much faster than with the traditional ant colony algorithm, genetic algorithm in speed, and has better performance when the topology's size increases. This algorithm has solved the unconstrained multicast tree cost minimization problem effectively.(2) For the multi-constrained QoS multicast routing problem, a distributed ant colony optimization algorithm is proposed. The algorithm takes into account a variety of QoS constraints, including bandwidth, delay, delay jitter and packet lossrate. In this algorithm, A multicast tree's forest is formed during the ant movement progress. The objective of ant movement is to connect all the trees in the forest to form a single tree. The most important advantage of the algorithm is distributed implementation. Simulation experiments show that this algorithm performs better than ANTNET,traditional ant colony algorithm, genetic algorithm and other algorithms both in results and convergence speed and has better performance when the topology's size increases. The distributed ant colony optimization algorithm can be implemented parallel and distributed, but also converges to a desired result in a faster speed.