A Markov Decision Based Routing Algorithm In Hybrid Air-space DTN

With the continuous advance of information technology and the constant improvement of human demand for information,a single network has been unable to meet the communication needs of today.T he space-air information network can realize the complementary advant ages of each load platform in space-based and air-based,so as to realize efficient transmission of information.Considering frequent intermittent connections between high-speed moving nodes in the hybrid networks,the network communication architecture ba sed on the Delay/Disrupt-Tolerant Network(DTN)is proposed.As DTN has a good custody and transfer mechanism,it is able to tolerate long delays and abrupt interruption for data transmission.Most existing DTN routing strategies are based on prior knowle dge of the network topology,among them,the Contact Graph Routing(CGR)algorithm is the typical routing algorithm.However,in the hybrid space/air network,there are uncertain connection nodes whose contact information is not known in advance.In addition,CGR algorithm does not have itself optimal path decision method,but borrows Dijkstra algorithm to find the best path.The traditional Dijkstra algorithm is a globally optimal static decision-making method.In the hybrid space-air network,the connection between nodes is constantly changing,and it is difficult to have an end-to-end complete path at every moment.In this case,Dijkstra's algorithm can not take advantage of global optimality and can not even find the complete path at some moment.Therefore,this paper designed a best path selection strategy for the above problems.According to the hybrid space-air network model,the various connection characteristics in the network are analyzed,and the connection probability is introduced.The contact graph of this hybrid DTN is designed.The limitations of CGR algorithm in air-space hybrid DTN are pointed out and the improvement strategy is proposed.According to the UAV motion characteristics,a model of relative motion between two nodes based on semi-Markov model is proposed,which further deduces the connection probability model between two UAVs and obtains the connection probability of uncertain connection in this network.The relative motion model was simulated in MATLAB,and the change of connection probability was analyzed by changing the parameters of UAVs.Under the contact graph of this network,an optimal routing algorithm of CGR based on Markov decision process is d esigned.At the same time,the derivation of bundle transmission delay model is simplified,and the connection waiting time is considered to get the earliest arrival time model.Markov decision process,which is defined for the hybrid space-air DTN routing problem,includes behavior space,state space,state transition probability an d objective function.In this paper,aiming at the network characteristics,a routing algorithm is designed in two parts.Using Matlab,Satellite Tool Kit(STK)and Interplanetary Overlay Network(ION),the routing algorithm based on Markov Decision Routin g The traditional Dijkstra algorithm is more suitable for the best path selection in the hybrid space-air DTN.