Data-Centric Filter Management and Routing Optimization for Distributed Monitoring Systems^*

[Abstract]

Emerging technologies such as sensor networks, RFID, and WiMAX have led to a new class of applications that continuously monitor data streams of interest in a distributed fashion. For example, in environmental monitoring, a large number of sensor nodes collaboratively keep track of the highest pollution level of the region. To save the network cost for distributed monitoring, existing studies have explored in-network query processing and data approximation techniques. However, most of the prior approaches did not incorporate data semantics in their algorithm designs. This research project proposes to take data-centric designs to improve performance. We focus on two issues, namely filter management for approximate monitoring and routing optimization for in-network data aggregation. We propose an asymmetric filtering technique to suppress data update traffic by exploiting the correlations of stream readings. Under asymmetric filtering, we will investigate adaptive filter allocation and dynamic filter migration mechanisms. We also propose to optimize the routing performance for in-network data aggregation based on data changing patterns and communication costs. Since this optimization problem is NP-hard, we will develop suboptimal algorithms and analyze their performance against the optimal solution. Finally, we will evaluate the proposed data-centric approaches by combining simulation studies and testbed experiments.