A novel buffer management policy based on prediction in Delay Tolerant Networks

Delay Tolerant Networks are the kind of networks in which an end-to-end path between any source and destination pair may never exist. In these networks, nodes use store-carryforward mechanism for data transmission. In this mechanism, nodes store the messages in their buffers and carry them to reach the appropriate forwarding chance. Most of the previous efforts on improving the performance of routing in these networks are focused on modification of routing protocols and little attention has been paid on buffer management, which is the main purpose of this paper. In this paper, a new policy has been proposed to remove the messages in the node's buffer. In this policy, it is not necessary to have extensive knowledge about network situation or large storage space. Utilizing time series analysis, the proposed policy estimates the time of the next contact. Based on this estimation, the message, whose destination node will be observed in further estimated time relative to the destination of other messages, will be removed from the node's buffer. According to the performed evaluations, the proposed policy outperforms the conventional policies in terms of average number of delivered messages and average forwarding hops in the buffer management of Delay Tolerant Networks.