Research On QoS Anycast Routing Algorithm

Anycast is a new type of network service. Users can visit the"nearest"server in a group of servers by an anycast address. The use of anycast can enhance the availability of network services and improve the quality of network services. At the same time, QoS routing is also considered as an important routing technology, which can guarantee the quality of service. The goal of QoS routing is to find the path which can meet the restrictive conditions and be effective in using the network resources. QoS anycast routing has attracted the interest of researchers when it was proposed at first.In this paper, we first introduce the background and significance of this project, and then make a profound study about QoS anycast routing. After that, we introduce the basic theoretical knowledge about anycast, QoS routing and genetic algorithm. Based on the analysis of existing domestic and international QoS anycast routing algorithm, a delay-constrained and least-cost anycast routing algorithm based on genetic algorithm is improved. In this algorithm, Depth_First search with weight and the selection method of roulette are used to make sure the original colony diversity. A new fitness function is proposed, in which the relationship between the cost and the delay are taken full account. The path which can meet the delay and has the smallest delay and cost has the biggest fitness value. And conversely, has the smaller value. The method of instructional aberrance is brought forward, and the convergence rate can be accelerated through instructional aberrance operation. In order to balance network resources and reduce the occurrence of network congestion, a load balanced anycast routing algorithm based on random method is improved. In this algorithm, the two metrics of the network balancing degree and server load are used as the basis for the choice of routing. That is, the metric which is got by converting the network balancing degree and server load into one through linear weighted was used as a basis for selection nodes, then routing selection and balancing of network resources are achieved by random selecting strategy.Finally, some simulations are given out. The experimental results show that the first algorithm can ensure the delay and the cost converging to the global optimal solution with a little genetic manipulation. The second algorithm has better performance in balancing the network resources, reducing the occurrence of network congestion and request receiving rate, although it has a little worse performance than Anycast-Routing algorithm in transmission delay. So, both of the algorithms are effective and feasible.