Reliability Support Protocol for Continuous Object Detection in Large-Scale Wireless Sensor Networks

Recently, detection schemes of specific phenomena, called continuous objects, such as wild fire and bio-chemical material have been actively studied. A continuous object might cover a wide area and it could dynamically change its own shape according to physical environments, e.g. wind and geographical features. For continuous object detection, the schemes have focused on detection of the current boundary shape of a continuous object at a time period. However, the schemes merely aim at reducing communication costs for boundary information collection, but they do not take into account reliable information transport. Hence, we propose a Reliable Boundary Information Transport (RBIT) protocol for continuous object detection. The protocol is composed of three major mechanisms: 1) explicit Representative Boundary Node (RBN) election, 2) cluster-based boundary information loss recovery, and 3) cluster-by-cluster packet loss recovery. The explicit RBN is considered in order to estimate boundary information loss due to the dynamic property of continuous objects. The protocol then constructs coarse-grained grid clusters, and it allocates a header node per each cluster. The header node collects location information of RBNs in its own cluster and revises lost boundary information. By the clustering the protocol distributively manages all RBNs of the large-scale continuous object. Finally, cluster-by-cluster packet loss recovery is provided for end-to-end reliable transmission from a header node to a sink. The cluster-by-cluster loss recovery fills up lost packets by data packet retransmission between clusters.