Sensors provide unprecedented access to a wealth of information from the physical environment in real-time. However, they suffer from a variety of resource limitations, most importantly power consumption and communication bandwidth. Additionally, environmental conditions can contribute to sensor failures, disrupting the flow of query results. In this paper, we propose new techniques to deal with sensor failures based on the principles of partition and single path redundancy. Our experimental results confirm the efficiency of our techniques with respect to different performance metrics in general, and, in particular, high quality of data.
Andreea Berfield, Panos K. Chrysanthis, Alexandros