Hierarchical Data Management for Spatial-Temporal Information in WSNs

Abstract

Wireless sensor networks (WSNs) comprise a large number sensor devices with limited power and resource. One of the key design challenges in the sensor networks is to efficiently utilize limited local storage capacity at individual sensor nodes to store historical data and produce fast and approximate responses to user queries. In this article, we propose a hierarchical data caching technique which consists of sensor nodes, cache nodes, and the base station to efficiently answer user queries with different data quality levels. In the proposed system, each sensor node forecasts and summarizes sensed data. Then cache nodes receive and merge the forecasting parameters from sensor nodes, and provide approximate answers to spatial-temporal user queries. Thus transmission cost can be significantly saved. When the accuracy of the response cannot satisfy the user-specified confidence bound, the query will be forwarded to lower levels of the proposed hierarchical architecture for more accurate answers. Besides, we also propose a cache management policy to prevent storage overflow by merging historical forecasting parameters for limited storage capacity of cache nodes. The simulation results illustrate that our caching system can further reduce the transmission overhead. Moreover, the proposed approach in our work consumes fewer computational resources and can still provide satisfactory responses to injected user queries.