Exploring Cyber-Physical Anomaly Detection: Insights from NOMS 2024
The Network Operations and Management Symposium (NOMS) 2024 commenced with Lars WĂĽstrich’s presentation on “Shells Bells: Cyber-Physical Anomaly Detection in Data Centers,” marking the inauguration of the technical track. This talk, based on collaborative research between Lars WĂĽstrich, Sebastian GallenmĂĽller, Stephan GĂĽnther, Georg Carle, and Marc-Oliver Pahl, delved into a novel approach to anomaly detection within data center environments.
The research posits that monitoring side-channel sound holds potential for enhancing anomaly detection in data centers. However, the complex acoustic environment within data centers poses a challenge, as the dense setup results in a composite soundscape, making it difficult to isolate individual device sounds. In response, the authors propose a cyber-physical anomaly detection approach that validates device activity within realistic composite audio signals.
This approach leverages information from management network traffic to predict changes in the data center’s soundscape. By employing a convolutional neural network, the researchers compare these predictions with real observations to validate correct device activity and identify anomalies. Notably, the evaluation conducted using data from a real data center environment achieved an accuracy of 98.62% in identifying spoofed and masqueraded activity.
Lars WĂĽstrich’s presentation stimulated insightful discussions among attendees, focusing on the technical intricacies of cyber-physical anomaly detection and its implications for data center security. The collaboration between Lars WĂĽstrich and his colleagues exemplifies the importance of interdisciplinary collaboration within the realm of cybersecurity and network operations.
For those seeking a deeper understanding of the research presented at NOMS 2024, we encourage exploration of the paper authored by Lars WĂĽstrich and his colleagues. This research represents a significant advancement in anomaly detection methodologies within critical infrastructure environments, with potential implications for enhancing cybersecurity and operational resilience.
Marc-Oliver is an adjunct professor of Carleton University in Canada. Marc-Oliver’s research focus is on a holistic approach to cybersecurity. He is an experienced teacher and an eLearning pioneer, holding several teaching awards.
Marc-Oliver holds a Diploma from University of TĂĽbingen and a PhD from Technical University of Munich.
Marc-Oliver is Vice President of the German Chapter of the Association for Computing Machinery (ACM) (https://germany.acm.org/). He heads the Future Education activities of the German-French Academy for the Industry of the Future (https://future-industry.org/).
Marc-Oliver’s research focus is on a holistic approach to cybersecurity. He works on security-by-design, anomaly detection, human-in-the-loop, and automation. His goal is making cybersecurity manageable, resulting in highly resilient and reliable systems. Marc-Oliver publishes regularly in the network and service management and security communities.
