This work addresses the problem of efficiently balancing the use of network resources when processing both spatially-constrained and (sensed) value based queries in Wireless Sensor Networks. To alleviate the drawbacks inherent to centralized approaches e.g., overheads in energy consumption and latency due to the transmission of the individual raw data/measurements to a dedicated sink, we propose in-network processing methodologies which unify the management of physical and data-space based queries. Since sensed data typically represents values that evolve over time, the distributed data management approaches need to be efficient in terms of communication cost and storage requirements. Furthermore, if the query processing paradigm(s) allows approximate answers, it may yield additional benefits if data abstractions based on higher-order statistics are integrated in the data management. The related challenges are further compounded if the nodes are mobile, for the purpose of adapting the quality of sensing/coverage to spatial changes in the data field. We present novel communication and storage efficient physical and data-space abstractions to facilitate in-network indexing of sensed data and processing of queries in WSNs consisting of mobile and static nodes. We also present novel algorithms to handle changes in the abstractions due to mobility of the nodes. To trade-off (im)precision vs. energy consumption, the proposed abstraction schemes combine rank order statistics, regular sampling, and bitmap representation. The proposed abstractions are generic, in the sense that they can be utilized in any hierarchical indexing structure that is based on binary space partitioning (BSP), such as k-d trees, Quadtrees and Octrees. Based on implementation in SIDnet-SWANS simulator, our experimental results demonstrate the effectiveness of the proposed abstractions under different mobility models, mobility speeds, and query streams.
|Original language||English (US)|
|Number of pages||33|
|Journal||International Journal of Next-Generation Computing|
|State||Published - 2011|