Research Of Low-Consumption Epidemic Routing In Delay-Tolerant Network

Internet is composed of different computers and networks. It becomes the most widely used network through the use of TCP/IP stack. In the Internet, the stable operation of TCP/IP protocol depends on some specific assumptions: there is an end to end path between the source and the destination; the end-to-end loss and the TTL value are relatively small. With the development of wireless network, the movement of hosts is becoming an important part of the network.Recently, Delay-tolerant network (DTN) is becoming popular. Hosts in DTN always move in different speed and can store messages for a long time. So no end to end path in certain time is one of the DTN's characters. Most algorithms in DTNs are based on"carry-move-forward"mechanism, and using the flooding method. The typical algorithms are Epidemic and Spray and Wait. The former algorithm transfers the messages to all the nodes in the network, leading to large network redundant, it is a waste of resources; the latter one limits the size of flooding, but all the messages have the same number of copies, and the delivery ratio have much relation to the nodes'movement.In this paper, we propose a new protocol called location based epidemic routing. We use the idea of Spray and Wait algorithm to achieve better performance. We make several assumptions for our protocol: each node can get its location; and destination of the message is known to the source node. When a node generates a new message, it determines the message's max number of copies in the whole network, the number is different from message to message. And when a node wants to transfer a message, it calculates the utilities of all its neighbors based on the distance between local host and the message's destination, and also based on the history information, then, it transfers the message to the best one, and reduces the number of copies in local host. When there is only one copy in the node, it transfers to the best neighbor, and delete the message in local buffer. If the best next hop is the node itself, the node just carries the message until it meets a better next hop or the message's destination. Finally, the new algorithm is simulated on the ONE(Opportunistic Network Environment) simulator. The results of the simulation show that the new algorithm has a lower use of buffers, and has a higher deliver ratio.