Outlier Discrimination and Correction in Intelligent Transportation Systems

Abstract In order to choose the quickest route, drivers are used to checking the traffic conditions via navigation applications before going out. Malicious drivers can hack into intelligent transportation systems and implant misleading traffic information to build an imitatively congested path, so that they can monopolize the unobstructed path, and others would give up on such a “congested” path. This selfish behavior results in chaotic traffic conditions and reduces the efficiency of transportation systems. In addition traffic information is sometimes missing due to hardware heterogeneity and failure. So it is important to correct the outliers and reconstruct missing values in traffic information to guarantee an efficient transportation system. Nevertheless, it is challenging to detect outliers due to the fact that they are short-term, small in scale, and random. To tackle this challenge, we propose a novel method called outlier discrimination and correction (ODC), which takes advantage of the flow relationship between adjacent roads and compressive sensing theory to distinguish the misleading traffic information (i.e., outliers) and reconstruct the abnormal data. Extensive simulations have been conducted using real urban traffic datasets. Performance results demonstrate that ODC outperforms existing methods in terms of accuracy and time consumption.