Improving Opportunistic Data Dissemination via Known Vector

Abstract

An opportunistic network is a network where each node only has connectiviy with other nodes opportunistically. To disseminate data in this kind of network is a challenging task and had drawn lots of research effort. When two nodes have connectivity with each other, the data bundles stored in their memory will be either replicated or forwarded to the other node in the hope that the data eventually reaches the destination node. The protocols that has only one copy of each data bundle is catagorized as forwarding protocol while the ones with multiple copies called replicating protocols. In opportunistic network, the replicating protocols are preferred over forwarding protocols. Yet, a big overhead is to be solved for replicating routing protocols, i.e.the exchanging of metadata. To avoid sending data bundles the other party already has, replicating protocols usually exchange metadata at the beginning of the connecting period. The naive method, sending the data bundles index summary in the memory, seems reasonable. But over time, this metadata exchange will become a big overhead. This paper propose a new scheme of indexing data bundles in the memory and can also reduces the metadata exchanged by large extent.