节点文献

基于蚁群优化的OBS光网络多径路由保护算法研究

Multi-Route Protection in Optical Burst Switching Networks by Means of Ant Colony Optimization

【作者】 刘舒蕸

【导师】 杨春勇;

【作者基本信息】 中南民族大学 , 通信与信息系统, 2010, 硕士

【摘要】 生存性问题是光突发交换网络面临的关键问题。传统的光路交换网络常用1+N的保护和修复技术解决此问题,但其发送冗余数据会导致大量资源的消耗,难以及时应对故障和业务处理的要求,尚不能直接用于解决光突发交换网络的生存性问题。针对光突发交换网络的生存性问题,本文根据蚁群优化理论设计保护路由算法,实现多路径故障保护。一旦某条路由上的链路或者节点出现故障,整个网络仍然可以通过路由表的更新实现故障恢复。经典蚁群算法只是根据蚂蚁走过路径的历史信息来进行路由,而对于目标食物的信息却无法获取。考虑利用网络中链路和节点的负载动态变化特性作为选路依据,若直接利用经典蚁群算法实现路由过程会出现信息滞后,因此需要对算法进行改进。本文受自然蚂蚁可以靠嗅觉辨向的启发,在运用数据传输的历史信息来模拟路径信息素的基础上,对经典蚁群算法做出如下改进:增加目的节点泛洪负载信息来模拟食物向环境散发气味的过程,使得路径上的各节点都可以获得目的节点与路径的最新信息;节点根据链路上的信息素,目标节点信息,链路的可见度综合生成概率表,为后继蚂蚁提供选路依据。本论文最后运用NS对改进算法进行仿真,测试结果表明该算法不仅可以实现网络的多路径保护,而且在故障发生时能尽快更新路由,减少传输时延,降低网络负载的波动幅度,实现网络故障的快速恢复。

【Abstract】 A crucial issue in optical burst switching networks is survivability. Traditional optical circuit switching networks used 1 + N protection and restoration technology to solve this problem, but sending redundant data will lead to a great amount of resource consumption at the same time. Therefore it is difficult to deal with failure and operational requirements in time yet can not be directly used to solve optical burst switching networks survivable issue.Aiming at survivability of optical burst switching networks, routing protection algorithm design in this thesis is under the theory of ant colony optimization to achieve multi-path failure protection. Hence,the whole network achieves restoration by updating routing table in case the route failure occurs on a link or node.Classical ACO simply routing according to history path information which ants pass through, but can’t get the overall information of target food. Considering the dynamic load changing of links and nodes in the network, and routing is determined by the latest information, there will be lag error if we use ant algorithm directly to route path. Hereby the algorithm needs to be improved.Inspired by ants use olfactory information for navigational means, the thesis adds the food giving off scents in order to simulate the destination address flooding load information, when pheromone trail is simulated by the history information of the forwarded data bursts. So every node can receive the latest information of destination node and links, and consequently the classical ACO expression is improved. Tables of probabilities are created by nodes based on pheromone,target node information, and link visibility, then the following ants can choose the routes in terms of the updating routing table mentioned above.This routing algorithm is simulated by Network Simulator. The test result indicates that it can not only realize multi-path protection, but also update the routing table since failures take place, as well as to reduce the delay of transmission, decrease the fluctuating range of network loading, implement the load balancing of links and fast restoration.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络