Sammendrag
Rapid boosted digitalization through the last decade, also during 2021 in addition to a worldwide orientation towards smarter and greener environments has resulted in many cities becoming more intelligent when it comes to automation and natural resources utilization. This includes faster development of 5G networks, Wi-Fi 6, blockchain and versatile application of AI. Smart Environment is a technology-enabled phenomenon that offers better, user-friendly and more efficient IoT-based environments with a specific focus on greener and more sustainable future both in rural and city-wise contexts.
IoT devices often lack the means for implementing necessary cybersecurity measures capable of handling adversarial actions both due to limited standardization and weak computation capabilities. Mirai IoT botnet back in 2016 was a frightening example of how such low-power devices can target large companies and cause DDoS attacks. The reason for this is mainly limited integration of data-in-transit, data-at-rest encryption and security practices as shown in 2017 in \#WatchOut study by the Norwegian Consumer Council. Machine Learning has proved efficiency in detecting cyber-attacks before, however, deploying Machine Learning on the Edge devices (like AVR and ARM) under the current state of the art is considered to be highly inefficient. Hence, the functionality of IoT devices might get potentially compromised due to their need for more power and time to complete their tasks. This potentially will lead to compromising the main function of the IoT devices requiring more power and time to complete the task the device was developed for. The overall vision of ENViSEC project is to enhance Smart Environments cybersecurity by introducing intelligent multi-agent data handling, cyber threats sharing, situational awareness and data streams aggregation from Edge devices. The project will utilise insights into novel solutions from recent developments in not only edge computing, but also look to combined approaches merging cloudlets, edge, fog and mist computing. Our ambition is to offer a resilient response to cyber-attacks as well as to ensure early detection of adversarial actions. Our new method enables multi-level data collection and off-chip Machine Learning model training to reduce the overhead and latency of the Internet of Things (IoT) components. It will contribute towards hardening cybersecurity and building an efficient infrastructure in a resource-constrained environment.
Vis fullstendig beskrivelse
