RMIT University has published guidance on how Dr Abebe Diro’s research seeks to enhance Internet of Things (IoT) security for spatial information networks.
The Internet of Things has become a fast-growing technology field with great potential to transform various industries, including healthcare, transportation, manufacturing, and smart homes. However, this growth has also brought with it many security challenges and concerns.
Unlike traditional information technology (IT) systems, security in IoT environments is challenging due to resource constraints, heterogeneity, and the distributed nature of smart devices. Addressing security issues requires a multifaceted approach that involves not only improving security standards and protocols, but also implementing effective security measures at the device, network, and application levels.
Dr Abebe Diro, Lecturer in Cyber Security at RMIT University and member of RMIT’s Center for Cyber Security Research and Innovation (CCSRI) has demonstrated that traditional internet security is not up to scratch for IoT devices. Throughout his research, Diro has focused on rethinking and redesigning the architecture and algorithms of existing IoT security systems.
Using cryptography and machine learning, Diro has successfully tested and created models with improved IoT security by decentralizing aspects of the cloud and offloading intensive computation for IoT devices. With the recent success of deep learning in image recognition and language processing, Diro decided to apply this more layered and complex subset of machine learning to his field of research. In particular, he studies deep learning in the context of anomaly detection systems.
Ultimately, Diro has been successful in traditional IT systems reducing false positives, but when applied to IoT, deep learning algorithms are more vulnerable than traditional shallow machine learning models. These mixed results led Diro to delve deeper into anomaly detection, where he ventured beyond the clouds and into space.
“There is a real commonality between IoT anomaly detection and spatial anomaly detection,[namely]they focus on identifying and mitigating anomalous events that can have significant consequences,” Diro said.
“It’s not just the nature of these disciplines, but the physical ecosystem. Given the miniaturization of satellites, the increase in space hardware and activities will depend on corresponding advances in security, just like the rapid growth of the Internet of Things on Earth,” Di Luo said.
Spatial Information Networks (SINs) are networks of space-based assets, ground stations, and communication links, and detecting anomalies in these networks is challenging due to their inherent diversity. Seen by some as a new frontier in cybersecurity, SIN is an important element of national security and a prime target for cyberattacks.
“Spatial anomaly detection is an important aspect of spatial security protection by identifying unusual yet insightful patterns in SINs. Existing SIN anomaly detection methods, such as simple thresholding techniques, are inaccurate, inefficient and uninterpretable, misleading high reporting rate, high resource requirements, and low scalability,” Diro said.
Diro is now focused on building and applying his previous work on IoT anomaly detection to SIN, according to a statement. By devising new algorithms and methods, as he has done for the Internet of Things in the past, Diro hopes to improve the accuracy of detecting spatial anomalies, provide near real-time and scalable data, and ultimately enhance spatial security.