Mobile Ad Hoc Network(MANET)is composed of mobile nodes without infrastructures,and it is widely used in many fields such as emergency rescue,scientific investigation and military battlefield.However,MANET’s high dynamic topology and distributed paradigm lead to difficulties in network management,effective allocation and network resource utilization,limiting the development of MANET in large-scale networks.The centralized control of the Software-Defined Network(SDN)provides a new solution to the problems mentioned above in MANET.Recently,many researchers have proposed to adopt Software-Defined Mobile Ad-Hoc Network(SD-MANET)architecture in wireless ad hoc networks to manage and utilize the resources dynamically.However,due to the changes in network topology,the traditional routing method of SDMANET suffers from the short lifetime time of the flow entry and considerable control overhead.Therefore,it is challenging to deploy SD-MANET in large-scale wireless ad hoc networks now.Based on the SD-MANET network architecture,we propose a routing method,SDSR(Software Defined Segment Routing),for large-scale SD-MANET,which adopts the methods of source routing,segmented routing and the greedy algorithm.SD-SR leverages the advantages of source routing.The controller sends the path to the source node and attaches it to the packet header to reduce the overhead of frequently searching and updating flow table entries in the intermediate node.SD-SR’s segment forwarding mechanism aims to tackle the problem that any failures among nodes in the path lead to the loss of the whole path,thus improving the routing path’s survival time,reducing the number of routing requests after the loss of the route.The routing mechanism within segments is the location-based GPSR(Greedy Perimeter Stateless Routing)method,which causes lower forwarding overhead and enables the intermediate node to cope with the dynamic changes of topology.We also propose a mathematical model of the SD-SR segment path selection problem and design a greedy algorithm for a quick segment path selection.The segment path selection model offers several constraints for the optimization model,including feasible paths,key intermediate nodes,traffic load balancing,and multiple network centrality indicators.The segment path selection algorithm SPSA(Segment Path Selection Algorithm)performs a greedy search iteratively,calculating the near-optimal segmented path within the limited response time of the controller.We implement a network simulation tool with discrete event simulation mechanism,and implement SD-SR’s switch and controller model on this platform,evaluating our proposed methods by simulation.The numerical simulation results show that the accuracy between the segment path nodes selected by the SPSA algorithm and the optimal solution is more than 85%.The average distance between the segment path nodes and the optimal solution is less than 0.15 hops.The discrete-event-simulation-based experiments results show that the SD-SR routing method proposed in this article outperforms the traditional MANET routing methods in network delivery rate,the lifetime of the flow entry,and control overhead.The proposed method is more suitable for complex,large-scale mobile networks. |