Scalable Pythagorean Mean-based Incident Detection in Smart Transportation Systems

• DOI: 10.1145/3603381
• Publication Date: 2023-06-05T10:02:54Z
• Authors: Md. Jaminur Islam; Jose Paolo Talusan; Shameek Bhattacha...; Francis Tiausas; Abhishek Dubey; Keiichi Yasumoto; Sajal K. Das
• Keywords: 11. Sustainability; 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

Modern smart cities need transportation solutions to quickly detect various traffic emergencies and incidents in the city avoid cascading disruptions. To materialize this, roadside units ambient sensors are being deployed collect speed data that enables monitoring of conditions on each road segment. In this article, we first propose a scalable data-driven anomaly-based incident detection framework for city-scale system. Specifically, an incremental region growing approximation algorithm optimal Spatio-temporal clustering segments their data; such strategically divided into highly correlated clusters. The clusters enable identifying Pythagorean Mean-based invariant as anomaly metric is stable under no but shows deviation presence incidents. We learn bounds invariants robust manner can generalize unseen events, even when learning from real noisy data. Second, using cluster-level detection, folded Gaussian classifier pinpoint particular segment cluster where happened automated manner. perform extensive experimental validation mobility collected four Tennessee compare with state-of-the-art ML methods prove our method within real-time outperforms known methods.