A similarity query system for road traffic data based on a NoSQL document store

Abstract

Model road traffic data as traffic matrices.Propose a traffic similarity query processing method based on singular value decompositions (SVDs) of traffic matrices.Implement the proposed system on a NoSQL document to handle large amount of road traffic data.Utilize an incremental computation of SVD for traffic matrices.Use real-world traffic datasets for justifying the performance of the proposed system. Advancements in sensing and communication technologies are enabling intelligent transportation systems (ITS) to easily acquire large volumes of road traffic big data. Querying road traffic data is a crucial task for providing citizens with more insightful information on traffic conditions. In this paper, we have developed a similarity query system for road traffic big data, called SigTrac, that runs on top of an existing MongoDB document store. The SigTrac system represents road traffic sensor data having spatio-temporal characteristics into traffic matrices and stores them into a MongoDB NoSQL document store by exploiting map-reduce operations of MongoDB. In addition, SigTrac efficiently processes similarity queries for traffic data with singular value decomposition (SVD)-based and incremental SVD-based algorithms. Our experimental studies with real traffic data demonstrate the efficiency of SiqTrac for similarity query processing for road traffic big data.