Back to Main Page, Next: Mobility in Publish/Subscribe Systems

Related

Semantic Sensor Data Fusion

The aim of Semantic Sensor Data Fusion is to investigate automatic MicroToPSS application synthesis given some knowledge of the application domain, as expressed in an application ontology. For example, an application may simply be interested in whether a conference room is "crowded" and not the specific audio, temperature, and motion sensor readings used to determine the status of the room.

So far, there has been little emphasis on how this data should be presented to applications as higher-level phenomena, thereby enabling the kind of utility envisioned by grid and pervasive sensor networks. In order to infer these high-level events, sensor data needs to be filtered, aggregated, and correlated from heterogeneous (and possibly geographically dispersed) sensor networks. In other words, data fusion must occur across independent sensor networks.

The challenge is to give applications the ability to exploit the full utility of sensor networks without being burdened with sensor level data. This is difficult because it requires fusing data across heterogeneous networks under semantics that differ between application domains. Even within a domain, applications may need to access sensor networks in a way that reflects their own context without affecting the interfaces seen by coexisting applications.

We have developed a novel system to address these challenges. Our approach allows applications to express interest in semantic events which are transformed into sensor network level events. The semantic events are "defined" independently of the application and sensor network interfaces. Our main contribution is a system that allows efficient semantic event evaluation to occur within the sensor network while providing flexible high-level interfaces to applications. We also use a publish/subscribe messaging model, which is appropriate for the event-driven nature of these scenarios.

This project draws from our work on Semantic Filtering of Web Data. More details about this project are available in this technical report.


Architecture of Semantic Sensor Fusion system using MicroToPSS
Page last modified on September 11, 2006, at 02:51 AM