Associated Load Shedding Strategies for Computing Multi-joins in Sensor Networks

In sensor networks, multi-join queries are processed by joining sensor data generated at different sensor nodes. Due to the energy constraint, however, sometimes only partial sensor data can be transmitted to the join site. In handling the energy constraint on each sensor node, the load shedding strategy under the MAX-subset principle has been considered for shedding selected data at each node so that transmitted data may yield the maximal possible results in a multi-join. Existing load shedding approaches, however, isolate sensor data to be joined, which often do not yield the maximal results. To obtain as many results as possible, we propose two load shedding strategies, the basic local associated shedding strategy and the global associated shedding strategy, based on different shedding constraints in this paper. We also present the max-loss-first associated shedding strategy and the multi-round associated shedding strategy for improving the basic local associated shedding strategy. Experimental results show that the proposed strategies generate as many results as possible on multi-joins.