Predicate Signaling in Distributed Sensor Networks

Nodes in a sensor network can generate messages periodically, or when anomalies are detected, or when queried by other nodes. In this paper we propose a strategy called predicate signaling that generalizes these schemes by generating messages when specified predicates - that can deal with both time and anomalies - hold. We show how power consumption, message generation rates and estimation errors can be controlled by choosing predicates appropriately. We compare predicate signaling with other schemes. We derive formulas based on stochastic differential equations to estimate performance measures in predicate signaling. We analyze measurement data, and we compare simulations based on measured data with results predicted by our theory.