Data Acquisition in Sensor Networks with Large Memories

Sensor Networks will soon become ubiquitous, making them essential tools for monitoring the activity and evolution of our surrounding environment. However such environments are expected to generate vast amounts of temporal data that needs to be processed in a power-effective manner. To this date sensor nodes feature small amounts of memory which mostly limits their capabilities to queries that only refer to the current point in time. In this paper we initiate a study on the deployment of large memories at sensor nodes. Such an approach gives birth to an array of new temporal and top-k queries which have been extensively studied by the database community. Our discussion is in the context of the RISE (RIverside SEnsor) hardware platform, in which sensor nodes feature external flash card memories that provides them several Megabytes of storage.