The availability of low-cost devices with sensing and networking capabilities has opened up an exciting opportunity to observe the physical world like never before. Many envisioned applications of sensor networks involve monitoring of mobile objects, such as vehicles or animals. This project designs, analyzes and implements lightweight sensor-aided distributed directories, which can guide users to current locations of (mobile) objects, and answer individual or aggregate queries about them. Directories are designed using locality as a primary criterion: the effect of a perturbation in an object's position or in the network structure must be local, and should not cascade beyond a restricted region that is close to the perturbation. Directories are designed to simultaneously manage multiple mobile objects or groups of objects moving together.
The project investigates the construction of novel directory schemes based on various distributed computation techniques, including link reversal, path reversal, hierarchical methods, and sketching of distributed data streams. The resulting directories are scalable, energy-efficient, and resilient to network failures. The performances of the directories are evaluated through a combination of theoretical analysis, simulation, and implementation on a sensor network testbed that has been built at Iowa State University. It is expected that the research will enable sensor-aided object monitoring at a much larger scale than is now possible, and will benefit any application of sensor networks that involves tracking mobile objects. The research will involve students at various levels of the educational system. The results of this research will be widely disseminated through publications in journals and conferences, and will be posted online at CoRR.
