Realtime Big Data Analytics For Event Detection In Highways

The unprecedented growing and accessibility of traffic data has materialized various opportunities to improve the quality of transportation related services. In this paper, we extend Sipresk, a big data analytic platform, to ingest real-time traffic data and provide online analytics for various stakeholders. We leverage extended Sipresk to detect and classify events in Ontario highways. More specifically, we designed and implemented a set of algorithms that detect four traffic patterns each of which detects a specific type of event. The platform reports the event specifications no later than 20 seconds that provides the opportunity to react promptly to incidents in highways. Also, we configure Sipresk to process traffic sensor data from January 2015 to June 2016 for detection and classification of events for every single day. In less than 2 minutes, Sipresk is able to process 24 hours of data and generate a list of events occurred during that specific date.