Intelligent Fusion Manager

The goal of the Intelligent Fusion Manager is to reason for events and provide them in the form of collection files to be indexed by the search engine. The architecture of the Intelligent Fusion Manager and its configuration is as follows:

Metadata feeds stemming from the processing of sensor or social network streams are fused together with initialisation information that describes the setup of the sensors connected to the node. This initialisation information has sensor locations and any other calibration information.

Reasoning is applied to derive possible events of interest to the users. These events are reported following some configuration regarding:

• The frequency of the reports.
• The temporal window the report summarises.
• Any social networks involved.
• The rules that govern the fusion of the data feeds: What to fuse and how.
• The format of the report.

There are different sets of configurations. On the one hand there are configurations pertaining to some application or some user level. These are manual configurations.

On the other hand, SMART allows intelligent configuration by offering a mechanism to automaticaly change the configuration, based on reporting rule learning. Learning can for example utilise both event and query frequencies.