Transition from Traditional Intelligent Transportation Systems to Connected and Automated Vehicles: Modeling Potential Benefits and Cybersecurity Risks

With increasing traffic demand, new approaches are being utilized to manage traffic in an effort to enable efficient utilization of the transportation network. Intelligent transportation system (ITS) applications such as Active Traffic Management (ATM) systems, relying on interconnectivity of components are widely deployed by transportation agencies to manage recurrent and non-recurrent congestion. One example of such traditional systems is the lane control signals (LCS), which alerts drivers of upcoming lane closures and then provides merging information through individual dynamic lane signs installed on roadway gantries. Since LCSs are mounted on overhead gantries, this approach has several limitations. First, the physical infrastructure required to deploy LCS applications can be costly and the effectiveness of these systems also ultimately relies on driver’s compliance with the displayed information. The rapid development of connected and automated vehicles (CAVs) provides a potential to address these limitations. CAVs connect the distinct entities of transportation systems through vehicle to vehicle (V2V) and vehicle to infrastructure (V2I) communication. The opportunities provided by this technology can be leveraged to develop an in-vehicle CAV-enabled lane control (LCS) application, providing direct control inputs to CAVs instead of displaying signs on gantries. Unlike the existing LCS applications, the vehicles will be able to perform merge actions immediately when a merge or lane closure advisory is received by the CAVs, without any intervention by the driver.

Although such applications, relying on interconnectivity of components and communication between vehicles and infrastructure have great potential to be effective in managing traffic and improving the efficiency of transportation systems, it’s unfortunate that this connectivity that supports these systems also provides potential system access points that results in vulnerability to cyberattacks. This is becoming more pronounced as these systems begin to integrate internet of things (IoT) devices. Hence, there is a need to rigorously evaluate the traditional ATM systems and CAV environment for cyberattack vulnerabilities, and explore design concepts that provide stability and graceful degradation in the face of cyberattacks. With the above background, this research focuses on three major thrusts.

First, the concept of transition from traditional ITS applications to CAVs for direct digital connectivity between vehicles and infrastructure was explored by developing a prototype CAV-enabled lane control (LCS) application. The CAV-enabled LCS provides direct control inputs to CAVs instead of displaying signals on gantries. The performance comparison showed that the CAV-enabled LCS outperformed real-world LCS with increased throughput of 18.4%, 9.6% and 12.8% for three selected scenarios under the best case of a 1 sec headway. Furthermore, the CAV-enabled LCS application was also found to reduce volatility, represented by variation in acceleration and deceleration regimes, by an average of 25.6% and 49.6%. These results reveal that CAV-enabled LCS application represents the future of Active Traffic Management and shows the potential of CAV-enabled LCS to improve both the operations and safety of traffic.

Second, the cyber risks of Active Traffic Management (ATM) system, which is a component of ITS system were explored by developing a prototype model of ATM in a simulation environment. Further, vulnerabilities of these systems to cyberattacks were identified and various types of cyberattacks (including denial of service, spoofing and sensor tempering/message falsification) were emulated within the simulation environment. It was observed that on average, the cyberattacks reduced the network performance, represented by 16% reduction in speeds. The research then developed a concept of prototype monitoring system capable of detecting anomalies within the ATM system and in case of anomaly detection, reverting the system back to a safe state of operation. The results revealed that the monitoring system was able to help the ATM system perform resiliently, shown through improvement in performance by 15% increase in network speeds. This study will enable researchers and agencies to realize the cybersecurity concerns in ATM systems and it illustrates that monitoring is a mean for such smart mobility systems to perform resiliently under cyberattacks.

Finally, the CAV environment was explored for its vulnerability to cyberattacks in a realistic traffic environment with multiple platoons under lane change events. The CAV-enabled LCS application developed by the author was utilized for this analysis. Initially, various access points for hackers were identified within the attack tree and three types of cyberattacks (message falsification, denial of service and spoofing attack) were emulated within the simulation environment to quantify their impact in CAV environment from traffic stream stability and surrogate safety perspective. Both traffic stream stability and surrogate safety were impacted under all three cases of cyberattacks. Message falsification was observed to have the most pronounced effect in reducing the stability of traffic shown by volatility, representing increase in acceleration and deceleration regimes by an average of 30.4% and 34.7%. Similarly, from safety perspective, the worst case was represented by message falsification attack, showing increase of crash conflicts by an average of 3000 conflicts. Another important finding was that lane change crash conflicts were more severe compared to rear end crash conflicts. This also indicates a potential increased crash severity since lane change crashes are equivalent to sideswipe crashes. These results indicate the necessity of considering multiple platoons in realistic traffic environment for cyber risk assessment and the critical nature of lane change events in CAV environment. These results also pave the way for future design of resilient systems from a monitoring perspective.