IoT Edge Intelligence – EIG

Research Activity

The research of the IEIG group is focused on the development of new architectures, frameworks and algorithms to support distributed, and decentralized edge learning models for collections of embedded IoT smart devices capable of collecting and aggregating data in order to adapt to user needs, learn and implement control policies, adopt predictive maintenance techniques, recognize anomalous situations and possibly react to them.

The focus of the research is on the study of distributed/decentralized algorithms based on cognitive paradigms (consensus learning, federated learning, swarm learning, etc.) that allow the devices themselves to cooperate by learning from their experiences, adapt to changing situations and predict probable future situations by exchanging information, distributing activities and coordinating their actions; spatial-temporal aggregation models with the aim of significantly simplifying online aggregation and analysis operations in IoT contexts integrated with 5G/6G; a multi-agent edge cognitive platform for the design, development, scheduling, deployment and maintenance of complex and large-scale systems that supports co-simulation via digital twin in the context of cyber-physical and IoT systems; the use of blockchain techniques in IoT devices to reduce the risk of hacking by decreasing potential access points.

Goals

ICAR’s IoT Edge Intelligence (IEIG) group carries out research and development of cognitive IoT platforms to facilitate the execution of intelligent algorithms in Edge Computing environments.

The main purpose of the IEIG group is to move the AI algorithms from the Cloud to the peripheral ecosystem and vice versa (Continuum Computing) with the aim of exploiting new AI chips (Intel Movidius, Nvidia Jetson nano, Google Coral, etc.) that are able to support complex real-time applications of artificial intelligence in IoT devices.

This approach makes it possible for Edge Intelligence applications and services to operate locally and autonomously, with clear benefits that result from having systems with low latency, reduced bandwidth consumption and greater protection of privacy.

Application Fields

The main fields of application are: smart cities, smart streets, smart grids, cognitive buildings, smart water, urban intelligence, cognitive smart home, smart mobility, smart parking.