An Adaptive Mediation Framework for Work ow Management in the Internet of Things

Emerging Internet of Things (IoT) systems consist of great numbers of heterogeneous physical entities that are interconnected via the Internet. These devices can continuously interact with the surrounding environment and be used for user applications that bene t human life in domains such as assisted living, e-health, transportation etc. In order to integrate the frontend physical things with IoT management systems, Work ow Management Systems (WfMS) have gained attention as a viable option. However, applying WfMS in IoT faces real-time service composition challenges such as continuous service discovery and invocation. Another question is how to integrate resource-contained IoT devices with the WfMS and execute work ows on the IoT devices. Tasks such as continuous device discovery can be taxing for IoT-involved devices with limited processing power and battery life such as smartphones. In order to overcome this, some tasks can be delegated to a utility Cloud instance. This thesis proposes a context-based framework for task mediation in Internet of Things applications. In the framework, tasks are modelled and executed as work ows. A case study carried out with a prototype of the framework showed that the proposed framework is able to decide when it is more cost-e cient to delegate discovery tasks to the cloud. However, sometimes embedding a work ow engine in an IoT device is not bene cial considering agility and energy conservation. This thesis compared two types of work ow execution: a) execution of work ow models using an embedded workow engine and b) execution of program code translations based on the work ow models. Based on experiments with real devices, the two methods were compared in terms of system resource and energy usage.