This work proposes a Cyber-Physical System (CPS) for protecting smart electric grid critical infrastructures using video surveillance while remotely monitoring them. Due to the critical nature of the smart grid, it is necessary to guarantee an adequate level of safety, security and reliability. Thus, this CPS is back-boned by a Time-Sensitive Network solution providing concurrent support for smart video surveillance and smart grid control over a single communication infrastructure. To this end, TSN delivers high-bandwidth communication for video surveillance and deterministic quality of service, latency and bandwidth guarantees, required by the time-critical revision smart grid control. On the one hand, the CPS utilizes High-availability Seamless Redundancy in the control subsystem via Remote Terminal Units (RTUs) guaranteeing seamless failover against failures in smart grid. On the other hand, the smart video surveillance subsystem applies machine learning to monitor secured perimeters and detect people around the smart grid critical infrastructure. Moreover, it is also able to directly interoperate with RTUs to send alarms in case of for example, an intrusion. The work evaluates the accuracy and performance of the detection using common metrics in surveillance field. An integrated monitoring dashboard has also been developed in which all CPS information is available in real time.

The pace of population ageing is increasing and is currently becoming one of the challenges our society faces. The introduction of Cyber-Physical Systems (CPS) has fostered the development of e-Health solutions that ease the associated economic and social burden. In this work, a CPS-based solution is presented to partially tackle the problem: a Deep Multimodal Habit Tracking system. The aim is to monitor daily life activities to alert in case of life-threatening situations improving their autonomy and supporting healthy lifestyles while living alone at home. Our approach combines video and heart rate cues to accurately identify indoor actions, running the processing locally in embedded edge nodes. Local processing provides inherent protection of data privacy since no image or vital signs are transmitted to the network, and reduces data bandwidth usage. Our solution achieves an accuracy of more than 80% in average, reaching up to a 95% for specific subjects after adapting the system. Adding heart-rate information improves F1-score by 2.4%. Additionally, the precision and recall for critical actions such as falls reaches up to 93.75%. Critical action detection is crucial due to their dramatic consequences, it helps to reduce false alarms, leading to building trust in the system and reducing economic cost. Also, the model is optimized and integrated in a Nvidia Jetson Nano embedded device, reaching real-time performance below 3.75 Watts. Finally, a dataset specifically designed for indoor action recognition using synchronized video and heart rate pulses has been collected.