Advances in clock synchronization techniques for sensor networks as well as wireless ad-hoc networks allow an approximated global time for an increasing number of configurations in ubiquitous and pervasive computing environments. This paper presents an event stream based on-line algorithm that fuses the data reported from the heterogenous processors in the network to detect predicates of interest. The algorithm detects properties that can be specified using predicates under a rich palette of time modalities. The algorithm has low space, time, and message complexities. The main idea used to design the algorithm is that the predicate is decomposed as a collection of predicates between pairs of system devices. The algorithm leverages the
between processes so as to incur a low overhead and hence be highly scalable.