PhD student of Dept. of Software Science Jaanus Kaugerand is defending his PhD thesis "Mediated Interactions for Collection and Exchange of Situational Information in Smart Environments" on Tuesday, Aug 18, 2020 at 10 AM in U05-210 (Ehitajate tee 5) and via zoom.

Embedded computation and communication technologies for IoT and Smart Environments have developed very rapidly in recent years and have become capable to perform complex computational tasks needed for collecting and processing situational information already within the network. The applications for such tasks, implemented already at the edge within Smart Environments, are characterised by persistent dynamical update and stream-like processing of data. This in turn requires methodical consideration of contextual (e.g. temporal and spatial) validity and consistency. In order to ensure the quality of the situational information, this thesis utilises a concept of mediated interactions. This concept allows on-line data validation and novel selective delivery and processing to improve consistency of exchanged data. The thesis points out that in Internet of Things solutions, which may rely on distributed wireless ad hoc and mesh sensor networks, especially those, where global clock synchronisation is not either feasible or even possible, the data that is used for in-network data processing algorithms may not always be temporally consistent. This problem is particularly clear in systems where, for simplicity, data is used in the order of arrival. The different communication paths may have different and time variable delays, for example due to changing topology, changing network load or different network nodes getting access to the common communication channel at different times. Applying concept of mediated interactions allows viewing each communication channel at an abstract level as an intelligent mediator agent with a task to ensure data validity and that only mutually consistent and compatible data is selected from the streams as inputs for situational information processing. This intelligent mediator agent is implemented as a middleware software layer at each network node. In this way, the rules for validity and consistency can be configured separately for each data stream according to the situational information requirements. The corresponding validation and consistency checking tasks are carried out online and information is conveyed to the consumer only through valid situation parameters. In software, the channel, or smart mediator agent, is implemented as a middleware called ProWare.
The theoretical contribution of the dissertation is researching and supplementing the concept of mediated interactions with the purpose of applying it in Smart Environments for collecting situational information. The main contributions are the introduction of methodology for achieving context-based consistency of input data from distributed sources and a technical solution for on-line checking of the validity and consistency of data required for processing situational information. The dissertation presents the results of a number of experiments described also in articles published by the author of the dissertation which demonstrate the practical application of the aforementioned theoretical work. The experiments have been performed using wireless sensor network technologies in military and urban environments. The results of the experiments show both the applicability of the developed methods and a significant improvement in the quality of produced situational information.

Supervisor Jürgo-Sören Preden, PhD

Opponents:

• Prof Michael Henshaw (Loughborough University, UK)
• Senior Researcher Gabriel Jakobson (CyberGem Consulting, USA)

The PhD thesis is available in Tallinn University of Technology digital library.