UC Riverside

Automated and autonomous systems are emerging new technologies that promise to revolutionize transportation and traffic applications. Connected vehicles (CV) applications can improve safety, efficiency, and capacity of transportation systems while reducing their environmental footprints. A large number of CV applications have been proposed towards these goals, with the US Department of Transportation (US DOT) recently initiating three deployment sites. Unfortunately, the security of these protocols has not been considered carefully, and due to the fact that they affect the control of vehicles, vulnerabilities can lead to breakdowns in safety (causing accidents), performance (causing congestion and reducing capacity), or fairness (vehicles cheating the intersection management system). In our work, we perform a detailed analysis of a recently published CV-based application protocol, Cooperative Adaptive Cruise Control (CACC), and use this analysis to classify the types of vulnerabilities that occur in the context of connected Cyber-physical systems such as CV. We show, using simulations, that these attacks can be extremely dangerous: we illustrate attacks that can cause crashes or stall emergency vehicles. We also carry out a more systematic analysis of the impact of the attacks, showing that an individual attacker can even have substantial effects on traffic flow and safety, even in the presence of message security standard developed by US DOT. We believe that these attacks can be carried over to other CV applications if they are not carefully designed. We also explore various defense frameworks to mitigate these classes of vulnerabilities in CV applications.At the same time, autonomous systems AVs are vulnerable to physical attacks that manipulate their sensors through spoofing or other adversarial inputs. If the sensor values are incorrect, an autonomous system that acts on them can be made to malfunction or even controlled to perform an adversary’s chosen actions, making this a critical threat to the success of these systems. To counter these attacks, recent works propose developing physics- based detectors that estimate the future state of an autonomous vehicle and use this estimate to detect anomalous sensor inputs. The accuracy and responsiveness of this detection algorithm are important to the security and robustness of autonomous systems. State of the art solutions that are based on Kalman filters face challenges in terms of configuration parameters and the limitations of the algorithm: we show that, while they constrain some attacks, an attacker is still able to bypass them. We focus on the security of CVs and AVs in terms of application- level attacks and defenses. First, we demonstrate scenarios where the vulnerabilities can be exploited to cause safety breakdowns or to interfere with an emergency vehicle. We define metrics for evaluating the attack impact that measures mobility (traffic throughput) and safety (average separation between cars). We show that attacks can substantially interfere with the operation of CVs, leading to increased vehicular speeds and reduced safety margins. Having established these attacks on the CVs application- level, we need to consider a mitigation framework where we use the classification of the five vulnerability types we introduce to guide the design of the mitigation steps that either eliminate or interfere with them. For example, one class of vulnerabilities occurs when the application logic does not check whether the data in the messages are consistent with other information it has about the system. Some of the message values are impossible to verify due to the lack of independent information to confirm it. Thus, we propose having an alternative source of data (specifically, data from cameras) to validate information in CVs application messages. We show that the defense indeed mitigates the vulnerabilities we identified in CVs without substantially harming performance. We also use \textit{blockchain}, which is traditionally used in applications from cryptocurrencies to smart contracts, as a potential solution to CV security. The BlockChain technology has the potential to revolutionize connected vehicles. It is far more secure than other record keeping systems because each new message transaction is encrypted and linked to the previous transaction. Specifically, we exploit the immutability of BlockChain to ensure safety from falsified information and attacks. Therefore, we propose a BlockChain -based scheme to protect the vehicular ecosystem and increase its security. We demonstrate these properties by developing an algorithm that uses BlockChain to maintain trusted communications between vehicles in the context of a cooperative ramp merging application. Next, we propose a new system to defend against physical attacks on AVs through: (1) Training the Kalman Filter to improve its ability to operate within its target environment; (2) Introducing a residual machine learning- based algorithm to capture non-linear dynamics of the system; and (3) Incorporating a change detection model to detect anomalies in the temporal behavior of the sensor data, to improve the assessment of deviation between the predicted and measured data. Our framework combines components that track a number of non-linear physical invariants and derives additional learned invariants coefficients through an optimization algorithm. It also uses an optimized residual prediction module based on a neural network, followed by a change detection algorithm, for keeping track of the historical anomalies. Taken together, these ideas enable for high accuracy to estimate the physical state of the vehicle, detecting a number of attacks that bypass state of the art defenses, with low overhead compatible with real-time implementations.