The Research Of Key-techniques In Disruption-tolerant Network Routing

More and more new networks such as sensor networks and deep space communication networks are characterized with high link delays and intermittent connectivity, which be a result of high node mobility, low node density, intermittent power from energy management (on/off) schemes, short radio range, radio obstruction or malicious attacks and so on, We call the network consisting of these new networks DTN, interconnecting these new networks with TCP/IP encountered great difficulties. Research of DTN aims to provide reliable communicat- ion in the networks in which end-to-end connectivity cannot be assumed. Various limitations of TCP/IP in DTN are analyzed in this thesis, and DTN architecture are introduced, some key-topics are analyzed, and demonstrated by a simulator, for example, Epidemic routing with different drop methods, then two kinds of routing strategies had been studied: one is based on multi-replica of each message, the other is based on probability of each link's connectivity.Based on SNW routing, two phases such as spraying and waiting in SNW are studied. In Spraying phase, a new spraying method called proportional spraying is provided, which is suitable for dense regions. Besides, in waiting phase, a forwarding method uses a metric to weigh those next nodes, and forwards the message copy to the best node. A new routing with these improvements is compared with binary SNW and MaxProp, and simulation results show that the new routing has better performance of delivery ratio, average latency and overhead.The routing using link holding probability is a distributed calculation method, in order to weigh different link, it calculates the link holding probability and uses route holding probability, the hops of messages, and each link holding probability to sort the "message- link", then forward the message with higher priority first. The routing method can avoid invalid sending causing by link interruption, and make use of available knowledge of mobility enough. Lastly the performance of different routing algorithms under various network conditions are compared.