Efficient Application-Level Causal Multicasting for Pre-Planned Sensor Networks Using Resynchronization Method
In this paper, we present a causal order guaranteeing multicast protocol based on sensor brokers as publishers that aggregate the information of results in sensor networks, periodically gossiping about the result messages to the query nodes according to their interesting topics. Although subscribers’ join and leave are highly perturbed in such sensor networks, some sensors can form overlapping multicast groups and query nodes as subscribers receive the results of the queries based on gossip-style dissemination. Each sensor broker manages a vector for each overlapping multicast group that represents its knowledge for every member of the group, and uses a vector whose dimension is the number of groups to time-stamp message. Each sensor broker needs piggybacking only one value per group for each message, which carries only one vector. Also, the broker makes information of the causally ordered delivery list and attaches the list to all messages. It gossips about them to their subscribers in overlapping multicast groups Therefore, this protocol guarantees causally ordered delivery of messages in a highly reliable manner using gossip-style dissemination technique.
Keywords: Sensor Network, Reliable Group Communication, Overlapping Groups, Scalability, Resynchronization
Download Full-Text
