Geometric Methods for Monitoring Distributed Streams

Monitoring distributed data streams is the focus of much research in recent years, with applications in dynamic, distributed database systems and sensor networks. Most solutions, however, deal with monitoring simple aggregated values, such as the frequency of appearance of items in the streams. More involved challenges, such as dynamic feature selection, or generally monitoring non-linear functions, are yet to be solved. In this talk I will present a general approach to this problem, based on studying the geometric properties of the domain of the monitored functions.