Abstract
The main concern of pervasive computing applications is to integrate as smoothly as possible into everyday activities of the individual, to the extent that their use and interaction with humans is natural, while they should be able to operate in the background without being perceived and be in a position to make decisions based on the information they hold.
A new architectural approach, Edge Computing, could be the solution to various problems of pervasive computing applications. Edge Computing refers to the use of technologies to enable data processing at the end nodes of a network. The principles of Edge Computing architecture show that it has the potential to address issues such as the response time of pervasive computing applications, the limitations of battery usage as the energy source, the volume of data collected by the sensors and the cost of their transmission, as well as issues concerning the security and protection of personal data.
However, problems remain, such as the increased volume of data and the need for each artifact to make decisions based on user wishes, context awareness and information received. For this reason, we will examine the possibility to apply data processing techniques to the Edge Computing model in order to facilitate decision making. Such techniques are Machine Learning, Deep Learning (supervised and unsupervised), Predictive Analytics and Ontologies, with the aim of better categorizing the data received, as well as facilitating recognition of the context of use and decision making..
Advisory Committee
Achilles Kameas (supervisor), Professor, School of Science and Technology, HOU
Sotiris Kotsiantis (member), Assist. Professor, Department of Mathematics, University of Patras
Christos Goumopoulos (member), Assist. Professor Univ. of the Aegean