Internet of Things

Cluster

Distributed Systems, Internet of Things, and Cybersecurity

Description

Thanks to technological advances, in particular the advent of wireless communication technologies, the miniaturisation of computers and the low-cost spreading of a broad range of sensors, an evolution of the use of the web is taking place, which is known as "Internet of Things". This essentially picks up and implements what was theorised by Peter T. Lewis as far back as 1985 (well before the appearance of Internet in the public domain). Lewis prophesised the advent of a "network of things" (Internet of Things), which would allow the connection of computers and devices equipped with sensors through the existing communication infrastructures, so as to generate direct interactions among the machines, that is, the sending of control commands and the management of data flows. Therefore, the IoT and the machine-to-machine communication paradigm (M2M) have existed for a long time (if we take into account the very fast times of IT evolution). However, thanks to the spreading of wireless technologies, embedded systems, a large variety of mobile devices (e.g., smartphones, tablets) and wearable devices (e.g., smart watches, virtual video glasses) equipped with a broad range of sensors, the IoT is going through a new development stage and achieving large-scale diffusion, thus becoming a new frontier for information technologies and industry. Just think of applications in the domotics, automotive and social network fields: by now, the IoT is no longer a simple subject of academic research, but it is so pervasive that it is often called "Ubiquitous Computing". Nowadays, our smartphones, always online, are virtually inexhaustible data mines (data derived from the logging of our activities on social networks, but also from the sensors of our smartphones and the sensors of the systems or devices we control remotely through our smartphones or our PCs). However, extracting useful information from this data flow is not simple. This is the reason why a good part of the academic research and also of the ICT industry's research is dedicated to designing models and techniques for extracting and managing information from such sources.

The research in this field also provides for the designing and experimenting with networks of sensors and systems for the acquisition, storing and data mining of data flows. Also, the research provides for the study of interfaces that will be suitable for facilitating the visualisation of raw data, and the consequent inference of higher-level patterns and information. In particular, there is an ongoing collaboration with the Polytechnic Department for the experimenting with IoT technologies in the nautical field (within the scope of the research of the UniUD Sailing Laboratory), for the purposes of monitoring and improving the navigation of sailing boats.

The group is involved in a US project for smart power grids, in particular for the development of distributed algorithms, to be performed by agents located within the electrical network.

Each agent is sited in the vicinity of a production station, and it communicates with other agents close to it. The stations are mostly, although not exclusively, small in size, often they are household stations, and the objective is that of dynamically routing electrical power across the network, thus minimising the dispersion due to dissipation.

The experience and the results of the research are expected to be used for drawing useful didactic ideas for the namesake course within the scope of the Bachelor's Degree course in Internet of Things, Big Data and Web. Students will thus be able to start concretely experimenting, through internships and dissertations, a reality that they will then find in the work environment.

Research subjects

Development of systems for the acquisition and M2M distribution of data from sensors
Algorithms for an efficient management of data flows
Algorithms for the management of networks of distributed sensors
Algorithms for the automated adjustment of intrinsic and extrinsic parameters of active sensors
Distributed algorithms for solving optimisation problems (DCOP)
Application of machine-learning techniques to raw data in order to extract "hidden" patterns and information
Design and development of interfaces for monitoring data and information extracted from the sensors