Wireless sensor networks (WSN) have been widely used in natural environment monitoring, military, health care, human life and other fields. Distributed snapshot, i.e. the global view/state of the whole system, is a fundamental block for many network control and data processing operations in WSNs. Although there have been quite a number of snapshot algorithms proposed, most of them are designed for traditional network environments. The characteristics of WSN in resource constraints, communication link and network topology require delicately new design. In this paper, we propose a new snapshot algorithm, named RES, which can tolerate packet loss with efficient communication cost. Basically, RES is executed based on the level of a node, which is determined by the hop count to the sink node. The snapshot query and snapshot data are propagated level by level. To tolerate packet loss, a node sends a packet by local broadcasting to all neighbors, to reduce communication cost, the feedback packet from lower level nodes will be merged with replicated data removed. Our design even does not require the knowledge of node constitution and neighborhood in prior, which makes RES more practical, especially for networks with high dynamics. We conduct extensive simulations via OMNET++ to evaluate the performance of RES and compare it with similar algorithms. The results show that RES can save both communication and time cost, and it can tolerate packet losses efficiently.
展开▼
