SNEE-A: A Sensor Network Query Engine for In Network Data Analysis
University:Uni. of Athens
Time:4:00pm (coffee: 3:30)
Sensor networks have become ubiquitous, supporting applications such as environmental monitoring and precision agriculture, and their proliferation provides new research challenges. Main goals of in-network processing include network longevity, proper adjustment to changing conditions while achieving some (user-defined) quality of service. We present our prototype system SNEE-A, that supports in-network data analysis and mining tasks, building on top of a SOTA sensor network query processing engine, SNEE. We describe the necessary language extensions to support data analysis, from the language to the implementation of additional components. Our approach has the advantage that data analysis tasks are not black boxes, but are refactored to equivalent algebraic operators which are optimizable by SNEE. We also provide an experimental evaluation of SNEE-A against custom implementations, destined to run on real motes.