Research On Anycast Routing And Its Key Technologies

Anycast is one of three communication models of the future network defined by IPv6, and may play vital role in the research on aeronautical manufacturing concurrent engineering as well as many network techniques. The prime problem of anycast is the design for highly effective secure anycast route, which is requested to determine the optimal path of multiple hosts in same address according to its definition. The implementation of routing protocol is the precondition and foundation of anycast research, and it can satisify the request of every application after combined with key technologies such as QoS, routing lookup, group management. The main contributions of this dissertation are summerized as follows:(1) The anycast addresses are not much in quantity, high frequency in use, and multiple hosts. Two anycast routing algorithms are proposed based the former characteristics. The first algorithm is based on the expanding method. The router joins the domain of the group member accordings to the computed metric that integrates the Expand message sent by anycast server and the capability of network. The second algorithm is based on improved ant colony algorithm. Tables of probabilities are created by routers accordings to the pheromone, smell and visibility on the path. Tables act as the basis that the succeeding packets choose the routes. This two routing algorithms are simulated by NS-2. The test result indicates that the first algorithm is the least average delay of transmission on the condition of light network load. The test result also indicates that the second algorithm is the least average delay of transmission, implements the load-balancings of links, increase the capability of networks transmission, and improve the scalability of anycast service on the condition of heavy network load.(2) The problem of cross service is intensively studied, and the architecture and implementation of routing optimization system for application-layer QoS anycast flows on DiffServ network are proposed. The optimization server is designed. The optimization system is divided into collecting informations, computating optimal paths and adjusting flows three kinds of mechanisms. This method can reduce the network resource wasted by Cross Service. The best result is reached when the adjusting period is equal to the average transmission time of anycast flows.(3) Two routing optimization algorithm for QoS anycast flows based on GA and hybrid strategy are proposed. Under constructing the question model of routing optimization, the implementation process of routing optimization is obtained about the optimal objective such as end-to-end delay, server load, networks traffic, and routes adjustment. Two algorithms can implement single-objective optimization or multi-objective combinatorial optimization under QoS constraint. The solution algorithms based on hybrid strategy has a stronger stability and higher accuracy in contrast to the solution algorithms based on GA. (4) An anycast routing lookup algorithm with LC-Trie is proposed. An anycast routing table is divided into an address table and a weight table. These two tables complete the routing lookup and the weight update together. It can reduce the average search length of routing and the space of the routing table, and satisfy each requirement of anycast routing lookup.(5) A secure distributed anycast group management scheme is proposed. The operations of group member such as joining and exiting should be authenticated by group other members. The monitoring between group members adopts the method which divides members into smaller groups. Cipher text is used in the transmission of contol datagram. This scheme can satisfy the delay requirements of anycast group management, and has the strong security, robustness and scalability.