Research And Comparison Of Unicast Routing Algorithm Based On Replica Control In Delayed Network

Delay Tolerant Networks(DTN) refer to a kind of mobile wireless networks where there are break state communication links because of mobile nodes and the intermittent connectivity. Its main features are dynamic network topology, extremely long delay, so there is not a stable complete end-to-end delivery path between the source node and the destination node in DTN. By means of the store-carry-forward pattern, application data are spread among nodes and are eventually transmitted to their destination. Routing is the main function of the network layer. Therefore, it has become one of the core issues of DTN research that how to design routing protocols to improve the network connectivity, reduce consumption of energy and delay, and increase the rate of message transmission. There are many scholars have studies and analysed the routing algorithms in DTNs.The purpose of this study is to optimize the strategy of multi-replica selection. We can adopt the network node mobility or other auxiliary information to select less and better relay nodes rather than the global knowledge or the summary of link states of the network topology. This method can improve the routing efficiency to some extent. The solution can handle different DTN application scenarios by adjusting network parameters. It is flexible and practical. Extensive simulations have been conducted to verify that this solution performances well.The research work of this thesis is mainly focused on two points: the optimization of message copies distribution mechanism in the spray phase based on the Spray and Wait routing algorithm, and the improvement of utility metric by citing the node similarity. In the NPSW algorithm, the source node and the relay node choose the next hop according to the node performance. The node performance is evaluated by a new utility that combines node encounter times with destination and the last encounter time. This value illustrates the contact frequency with other nodes and the speed of spreading messages. The RABNS algorithm introduces the concept of the triangle similarity theorem to redesign the similarity of nodes. This method can reduce the number of message copies and the network overhead. The ASW algorithm mixes the node similarity and the delivery probability defined in PROPHET to design the message copies allocation function. It ensures to forward the message to the next hop potential node. In the end, the work of this paper is summarized.