Side Effect Mitigation Algorithm for Cache Maintenance in Opportunistic Networks
Satoshi Fukumoto, Masayuki Arai, Kazuya Sakai · 2015
Cooperative caching with a data access tree (DAT) in opportunistic networks, where the connectivity among nodes are intermittently disconnected and the transmission opportunities are very limited, significantly improves routing performance, since a node can download data from the source node of the data as well as any caching nodes. With a DAT, cached data is updated in two different ways, intentionally and opportunistically. While intentional updates are performed along the tree structure, opportunistic updates are performed between two nodes without a parent-child relationship in the tree when they have a contact. Although opportunistic updates contribute the cache data to be fresh, it introduces the side effect. That is, a legitimate parent node in the tree cannot updates its child's cache by the delta update (patching only diff data) due to data version inconsistency introduced an opportunistic update. Therefore, in this paper, we first build mathematical model of the cooperative caching in opportunistic networks, and then, propose a side effect mitigation (SEM) algorithm to alleviate the side effect. The key idea of the proposed scheme is that having the nodes in the same levels try to keep the cache version be consistent, and as a result, the DAT with SEM leads cached data to be as fresh as possible with less side effect. The performance of the proposed scheme is proved by mathematical analysis, and we demonstrated that the DAT with SEM keeps cached data fresher than the original DAT.