Distributed hash tables are designed to provide reliable distributed data management, but present challenges for networks in which nodes have varying characteristics such as battery or computing power. Assuming that nodes are aware of their resource availability and relative network positions, this paper presents a novel distributed hash table protocol which uses nodes' resource levels to remove load from weak nodes, whose overuse may cause delays or failure, while using nodes' positions to reduce cross-network traffic, which may cause unwanted network load and delays. This protocol provides nodes with links that are physically near with high resource availability, and simultaneously provides scalability and an O(log(N)) routing complexity with N network nodes. Theoretical analysis and simulated evaluation show significant decreases in the routing and maintenance overhead for weak nodes, the physical distances that lookups traverse, and unwanted node failures, as well as an increase node lifetime.
展开▼
