Context discovery in sensor networks

Context discovery refers loosely to the operations of extraction, aggregation, storage and deduction of environmental information. Such information can be delivered to context-aware applications so that they can adapt their behavior according to the discovered contexts. Context information may be sensed and extracted by various sensors deployed freely in the environment These sensors must be interconnected to allow sensed data be collected. They are often heterogeneous with different sensing and computing capabilities to allow multimodal sensing. To help the interaction among the heterogeneous sensors in a context-aware environment, we need an efficient data-centric communication substrate on top of these sensors. The substrate must be self-organizing, self-tuning, and self-healing. It also needs to exploit the ad hoc interactions among heterogeneous sensors for energy conservation and operation efficiency. In this paper, we introduce such a substrate called TRAILBLAZER to enable the discovery of high-level contexts. We evaluate TRAILBLAZER via simulation. The results show that the communication among sensors is efficient in terms of the consumed energy when compared with an ideal solution and it is also fault-resilient.