Optimizing network-side queries with timestamp-join in wireless sensor networks

This paper proposes a new method for optimizing in-network distributed queries that perform join of data produced simultaneously by different sensors in a wireless sensor network. We adopt a modified version of the standard join operator that relates tuples having the same timestamp, and an optimized version of it, which provides on-demand, pull-mode, data acquisition from sensors. The optimizer uses an algebraic approach based on transformation rules and ordering of operators to generate and chose a query plan that reduces the query execution cost in terms of consumed energy. We implemented these join operations in a query processor for mica-class sensors and we performed extensive tests to prove that our approach may reduce energy required to process a long running query, by order of magnitudes, with respect to non optimized query plans.