Internet of Things (IoT) refers to the ever-growing network of objects connected to the Internet. Cisco predicted that by 2020 the Internet will contain over 50 billion smart objects. Intelligently processing the data produced by these objects will lead to a wealth of advanced applications in domains like smart cities, pervasive health & environmental sensing. The generated data is voluminous, heterogeneous, time-varying and possibly noisy or incomplete. It is challenging to integrate & analyze this data on the fly to derive actionable insights. Semantic Web technologies have a proven track record in consolidating data from heterogeneous sources. Semantic reasoning is able to derive high-level knowledge out of the combined streams. However, reasoning is computationally intensive and slow compared to the velocity of IoT data.

The research objective is to devise a reasoning framework able to derive actionable insights in a timely manner by performing efficient & scalable complex reasoning on IoT data streams. Based on available domain knowledge, network properties and data stream characteristics, the framework can automatically distribute the reasoning across the IoT network as a hierarchy of cascading reasoners, in which low-level reasoners close to the data sources are combined with high-level reasoners positioned deeper in the network. To deal with the unpredictability of IoT, the self-learning framework changes its configuration at run-time to optimize its performance.