Persistent transportation traffic volume estimation with differential privacy

Abstract

Traffic volume estimation is critical to the transportation engineering. Persistent traffic volume reveals the amount of core, stable traffic at locations of interest, which is meaningful to many transportation applications, such as traffic flow guidance system. Unfortunately, most of the existing state-of-the-art studies that concentrate on the persistent traffic estimation issue only provide limited privacy preservation. To tackle this challenge, we first present two schemes with differential privacy respectively for estimating the persistent point traffic volume and the persistent point-to-point traffic volume in this work. Then, we further propose a general scheme with differential privacy for estimating the persistent multi-point traffic volume. We encode the passing vehicles in privacy-preserving data structures by using the random communications between vehicles and Road-Side Units (RSUs). Then, we derive the persistent traffic estimators through mathematical analysis and bitwise operations. We also prove that the proposed privacy-preserving mechanism can achieve $$\epsilon$$-differential privacy for protecting the location and trajectory privacy of vehicles through rigorous theoretical analysis. The experimental results based on the real transportation traffic traces data demonstrate the effectiveness of the proposed schemes.