A system for semantic data fusion in sensor networks

Abstract

Emerging sensor network technologies are expected to substantially augment applications such as environmental monitoring, health-care, and home/commercial automation. However, much of the existing work focuses mainly on collecting and using sensor level data from isolated sensor networks directly, which still burdens applications with the task of interpreting the context and meaning of sensor data. In order to infer high-level phenomena, sensor data needs to be filtered, aggregated, correlated, and translated from many heterogeneous and dispersed sensor networks. In this paper, we present a novel system for decoupling the process of semantic data fusion from application logic based on semantic Content-based Publish/Subscribe techniques. Our main contribution is an integrated system that allows efficient semantic event detection to occur both within and across sensor networks by translating events using ontologies.

Research on energy efficient fusion-driven routing in wireless multimedia sensor networks

“…Anandkumar et al presented a novel formulation for optimal sensor selection and in-network fusion for distributed inference known as the prize-collecting data fusion in terms of optimal trade-off between the costs of aggregating the selected set of sensor measurements and the resulting inference performance at the fusion center in [11]. Wun et al presented a novel system for decoupling the process of semantic data fusion from application logic based on semantic Content-based Publish/ Subscribe techniques [12].…”

The NRL Geospatial Hub for ocean sensor processing and collaboration

“…As an alternative to stove piped processing chains, an ontology of processing steps could assist in service orchestration [LWP + ], [WPJ07].…”

Modeling and optimizing content-based publish/subscribe systems

“…For example, the notification containing three attributes {(id, nobody), (status, ill), (time, 6 : 10am)}, can be used to indicate that in a social network the status of the member nobody was set to "ill" at 6 : 10am. In this thesis, we abstract from the semantics of notifications [104] and do not cover aspects like event hierarchies or scopes [33]. As already mentioned, filters are normally defined as conjunctions of attribute filters, where each attribute filter places a constraint on the value of a given attribute.…”