Effects of uncertain anomalous information on traffic flow of automated vehicles with V2V communication

Abstract

Automated vehicles (AVs) equipped with vehicle-to-vehicle (V2V) communication can operate by sensing real-time status information through onboard sensors and wireless connections. Nevertheless, under the influence of multifarious random factors in real traffic, this critical information that support the normal movement of such vehicles may be anomalous, raising concerns on their mobility and traffic security. Due to the lack of appropriate analytical model, previous studies have not comprehensively uncovered the effects of uncertain anomalous information on traffic flow of AVs with V2V communication. Therefore, this study aims to bridge this critical gap. Firstly, by introducing a probabilistic parameter (i.e., information anomaly probability), we propose a general model that integrates the normal and compromised models, thereby capturing the longitudinal dynamics of AVs featuring V2V communication in the presence of uncertain anomalous information. To enable the detailed theoretical and experimental analyses, we specify it through the cooperative adaptive cruise control model calibrated with real-car data. Subsequently, we define the concept of pseudo string stability and parameterize the stability condition based on the characteristic equation method, so as to demonstrate the relationship between traffic flow stability and the parameters and probability of information anomaly. Finally, we refine the proposed probabilistic model and conduct extensive numerical experiments. The findings show that uncertain anomalous information could result in sudden or even frequent acceleration and deceleration of AVs, causing traffic oscillation, reduced traffic efficiency, and even collision accidents. In particular, the greater the information anomaly probability, the larger the disturbances experienced by traffic flow. Meanwhile, at the same level of anomaly, the combined impacts of various anomalous information could lead to more severe consequences than the singular impact of any individual anomalous information. Furthermore, the duration of anomalous information directly affects the time it takes for traffic flow to return to normal.