A generic and scalable IoT data fusion infrastructure

Abstract

Applications in the IoT domain are in need of information coming from many different sources. To mitigate the processing overhead for increased volumes of raw data (seen at the application layer) on vast pervasive networks (a significant pillar of 5G networks), we introduce a customizable middleware platform. Such platform allows the treatment of incoming data flows through complex, yet fully specifiable/controllable data processing workflows. The derivation of new information through this high level processing task is termed data fusion, hence, the presented architecture is named “Fusion Box”. The middleware platform is customizable through a Domain Specific Language that allows the domain (and not the IT) expert to easily specify the needed processing and automatically transform such specification to executable workflows. The coupling between the Domain Specific Language and the Fusion Box is based on the concept of contextors, a versatile processing unit that can be instantiated and managed in many different ways as needs dictate.