Simulation On Anycast Routing Algorithm Based On NS2

With the rapid development of Internet and the increasing connections between network hosts, the demands for network service have exceeded its capacity, which has brought great impacts on the application of QoS (Quality of Service) service such as VOD (video on demand) and IP telephone etc. In order to enhance the usability of service and improve the flow distribution of IP network, a new network model - anycast is proposed in IPv6 to support the QoS service, and the service quality depends much on the choosing of QoS routing.To know whether or not the designed routing algorithm is practical, besides the careful theoretical amylases and reasoning, a great many experimental verification and performance tests on the network are necessary. It is often costly to realize the real network system due to various reasons. Therefore, the network simulation has been widely applied in the design and research of network thanks to its advantages in controllability, recurrence and extensibility.This paper, under the Linux environment, mainly studies the simulation and evaluation of the genetic algorithm-based anycast QoS routing algorithm in NS2. It first introduces the NS2 simulation platform and analyzes in details the principle of its simulation modeling, and it further states the approaches of conducting modeling simulation and expanding component library. Finally, the paper analyzes in details the system structure of NS2-based routing simulation module and its simulation flow. On the bases of these as well as the researches on anycast network related protocols, this paper points out the shortages of this simulation module in supporting the genetic algorithm-based anycast QoS routing algorithm, and it revises and expands the simulation module. By means of script simulation programming, the paper carries out simulation experiments on the well-designed anycast routing algorithm and then conducts analyses and evaluation on various parameters of this algorithm.During the experiment process, considering the complicated and changing network environments as well as the various terminals in real practice, it is hard to reflect the actual changes of performance parameters under various network environments by experimenting on only several nodes. Therefore, this paper conducts simulation experiments on 10 nodes to 100 nodes respectively and then compares various performance parameters under the same simulation condition. The results show: the genetic algorithm-based anycast QoS routing algorithm can choose the one with the best QoS performance from many service nodes to provide service for the users within certain time for different QoS demands, and as the amount of node keeps increasing, this algorithm becomes more efficient. Furthermore, the simulation tests can be conducted in other network routing protocols by revising slightly the scrip program, and on this base this module can be expanded by applying tools such as C++ etc.