Research On DTN Congestion Control Routing Algorithm Based On The Regional Characterisric Of Node Movement

With the advanced development of network technology and human’s exploration tothe unknown world, traditional network architecture is unable to satisfy therequirements of sophisticated application environments such as deep space/seadetection and data collection of remote districts. DTN (Delay/Disruption TolerantNetwork) has been proposed as a new network architecture to realize normalcommunications of heterogeneous network in these extreme environments, whichmakes it a prevailing research topic for researchers of network and communications athome and abroad.Existing work of DTN focuses on dynamic changes of network topology andimprovement of message successful delivery ratio, reduction of delivery overhead,decrease of delivery delay in order to guarantee the normal communication amongnodes in the lack of persistent end-to-end connection. However, in extreme networkenvironments, network storage resource is relatively limited. The persistent growth ofthe node communication demand would further aggravate the conflicts between the highdemand and short supply of storage, which could result in network congestion. Thespreading of network congestion would seriously affect major DTN performanceindexes like message successful delivery ratio. This thesis mainly focuses on how toreasonably tune the message transfer and control the network congestion in the case ofrestricted network storage resource. Those aspects are crucial to the improvement of theoverall performance of DTN.The main contribution of this thesis include: for the routing algorithm PROPHETbased on encounter probability history records, we proposed congestion control routingalgorithms from the different perspectives of sender, receiver and cooperation amongnodes with regard to the regional characteristics of node’s movement in the real DTNsuch as communication network of remote district. The experiments verify thecorrectness and effectiveness in resolving the network congestion problem of real-worldsettings. The major contribution and innovation are concluded as follows:1. Congestion control from the perspective of senderIn this thesis, a congestion control routing algorithm ACD based on Ant ColonyOptimization is proposed for DTN. Firstly, we leverage pheromone conception in AntColony Optimization to map node’s hops in a certain message delivery direction to thetransfer effect it has in this direction. After several times of message delivery, thetransfer effect (pheromone) of every node in each message delivery direction isevaluated.Meanwhile, inspired by the concept of heuristic in Ant Colony Optimization, wecombine current node storage’s congestion situation, i.e., proportion of the spare storage space (heuristic) with pheromone to present the integrated assessment value thatdetermines as the selection of the transfer node in the different message deliverydirections. It’s rule that when nodes meet, the node with larger integrated assessmentvalue is responsible for the message transfer task in the corresponding direction.Our experiment demonstrates that this algorithm can expand the option range ofmessage transfer nodes, improve message successful delivery ratio and distribute theprevious message transfer pressure in this delivery direction towards the node with ahigher probability of meeting the destination node. Meanwhile, algorithm effectivelycontrols network congestion by taking full account of each node’s storage usage.2. Congestion control from the perspective of receiverIn this thesis, a congestion control routing algorithm RMD by deleting messagebased on node regional movement is proposed for DTN. If we consider regionalcharacteristics of node movement, when nodes encounter and the receiver had noenough spare storage space to receive more new messages, the message which hasinconsistent message delivery region with node movement region is deleted.Consequently, congestion is alleviated and invalid occupancy of limited networkstorage resource by these messages is reduced as much as possible. It effectively avoidscongestion from happening again for the same reason and enhances the recycling usageratio of node storage space.Based on this, another congestion control routing algorithm PRMD based onPigeonhole Principle and node regional movement is further proposed. Firstly,Pigeonhole Principle is leveraged to explain the reasons for congestion and its controlmechanism. By introducing the concept of power and reducing its computationcomplexity, node’s encounter power and transfer power related to a certain messagedelivery direction is simplified into encounter weight and transfer weight. We ensurethat only when encounter weight is larger than transfer weight, receiver takes theresponsibility of message transfer task in this delivery direction. The algorithm achievesthe share occupancy of node storage space among tasks of the different transferdirections and actively avoids the occurrence of congestion.PRMD algorithm fully considers actual delivery and storage release ability in eachdelivery direction of every node. It proposes the standard that receiver passively deletesmessage and actively selects message to receive. Experiments demonstrate that thealgorithm can effectively improve message successful delivery ratio and reduce networkcongestion.3. Congestion control from the perspective of node cooperationIn this thesis, a congestion control routing algorithm CRSG based on SocialPsychology and Game Theory is proposed for DTN. Firstly, by leveraging socialrelationship and group theory in social psychology, we analyze how to balance thecontradiction between guaranteeing node’s normal communication and inhibiting its arbitrary transfer behavior.Then, according to the principle of responsibility even partition in NashEquilibrium Theory from Game Theory, the balance point of network storage shared byboth sides parties is computed based on the same responsibility of node congestiontaken by the sender and receiver. Then this balance point is used as the threshold foreach node’s network storage occupancy. It standardizes the behavior of each node fromthe institution and avoids the recognition of node’s selfish attribute. It reduces extracomputing and the storage burden caused by attribute recognition as well as the demandof professional recognition device applied in network.The experiment proves that our algorithm can effectively inhibit the selfishbehavior of node, promote the communication cooperation among nodes to controlnetwork congestion and improve message successful delivery ratio.