A Secure Sensor Fusion Framework for Connected and Automated Vehicles under Sensor Attacks

Tianci Yang, Chen Lv

Abstract

Connected and automated vehicles (CAVs) utilize various sensors and wireless communication to enhance transportation safety and efficiency. However, this reliance increases vulnerability to sensor faults and adversarial attacks. This paper presents a secure sensor fusion framework that provides robust estimates of physical variables with bounded errors, independent of attack signals. By exploiting sensor redundancy, the proposed algorithm enables effective attack detection and isolation. Furthermore, an $H_{\infty}$ controller is integrated to stabilize vehicle platoons under Cooperative Adaptive Cruise Control (CACC), mitigating the impact of estimation errors and communication noise. Numerical simulations demonstrate the framework's effectiveness in maintaining string stability and safety despite active sensor spoofing.

Introduction: The Vulnerability of Connected Platoons

Modern Connected and Automated Vehicles (CAVs) rely on a suite of sensors—including LiDAR, Radar, and cameras—to perform Cooperative Adaptive Cruise Control (CACC). While these sensors enable vehicles to travel in tight, efficient platoons, they also introduce a significant "attack surface." Malicious actors can use inexpensive hardware to "spoof" physical signals, tricking a vehicle's sensors into perceiving obstacles that aren't there or miscalculating distances to the car ahead.

Existing security measures often focus on cryptographic protection of wireless messages, but these do not protect against the manipulation of the physical environment the sensors perceive. The challenge lies in maintaining safe operation even when a subset of sensors is providing intentionally misleading data.

Methodology: Resilience Through Sensor Redundancy

The proposed framework addresses this vulnerability by exploiting Sensor Redundancy. If multiple sensors measure the same physical variable (e.g., inter-vehicle distance), the system can establish a baseline for "truth" that is difficult for an attacker to completely subvert.

The proposed secure sensor fusion framework for CAVs under sensor attacks.

The framework consists of four key components:

  1. Reconstructability Analysis: The authors mathematically prove that the true state of the vehicle can be recovered as long as the number of compromised sensors ($q$) is less than half of the total sensors ($N/2$).
  2. The Fusion Algorithm: Rather than using a simple average—which is highly susceptible to outliers—the algorithm evaluates all possible subsets of sensors. It identifies the subset with the highest internal agreement (minimum maximum deviation) to generate a trusted estimate.
  3. Attack Detection and Isolation: Once a robust estimate is established, individual sensor readings are compared against it. Sensors deviating beyond a noise-calibrated threshold are flagged and isolated from the control loop.
  4. Robust $H_\infty$ Control: To ensure that remaining estimation errors do not lead to instability, a specialized $H_\infty$ controller manages the vehicle dynamics. This controller is designed to maintain "string stability," preventing small errors at the front of a platoon from amplifying into collisions at the rear.

Key Insights and Theoretical Guarantees

A critical contribution of this work is the formal proof that the fused measurement error is bounded by a constant factor (specifically, three times) of the maximum sensor noise. Crucially, this error bound is independent of the attack signal. This means that regardless of how extreme a spoofed signal might be, it cannot pull the fused estimate beyond a small, safe margin. Furthermore, the framework is designed to handle time-varying attacks where an adversary might switch targets between different sensors to bypass static filters.

Experimental Results

The effectiveness of the framework was evaluated through numerical simulations of a five-vehicle platoon using synthetic data modeled with Gaussian and Uniform noise distributions. In scenarios with three redundant sensors (e.g., LiDAR, Radar, and Ultrasonic), the isolation algorithm achieved a 100% success rate in identifying spoofed sensors.

Comparison between the actual distance $d_i$ and the fused estimate $\hat{d}_i$ during a sensor attack.

As shown in the simulation results, the fused estimate $\hat{d}_i$ closely tracks the actual distance $d_i$ even under active attack. The $H_\infty$ controller successfully maintained desired following distances across the platoon, effectively suppressing the "accordion effect" typically caused by sensor noise and communication delays.

Conclusion and Future Work

While the framework provides robust theoretical and practical defenses, it does require a minimum of three sensors per critical variable to survive a single attack ($N \geq 3$). For cost-sensitive automotive applications, this redundancy requirement may be a hurdle. Future research directions include optimizing the computational complexity for systems with high sensor counts and extending the model to handle non-linear vehicle dynamics and scenarios where more than half of the sensors might be compromised.

Reference: Yang, T., & Lv, C. (2021). A Secure Sensor Fusion Framework for Connected and Automated Vehicles under Sensor Attacks. arXiv:2103.00883v1.

Summary updated

Abstract

By using various sensors to measure the surroundings and sharing local sensor information with the surrounding vehicles through wireless networks, connected and automated vehicles (CAVs) are expected to increase safety, efficiency, and capacity of our transportation systems. However, the increasing usage of sensors has also increased the vulnerability of CAVs to sensor faults and adversarial attacks. Anomalous sensor values resulting from malicious cyberattacks or faulty sensors may cause severe consequences or even fatalities. In this paper, we increase the resilience of CAVs to faults and attacks by using multiple sensors for measuring the same physical variable to create redundancy. We exploit this redundancy and propose a sensor fusion algorithm for providing a robust estimate of the correct sensor information with bounded errors independent of the attack signals, and for attack detection and isolation. The proposed sensor fusion framework is applicable to a large class of security-critical Cyber-Physical Systems (CPSs). To minimize the performance degradation resulting from the usage of estimation for control, we provide an $H_{\infty}$ controller for CACC-equipped CAVs capable of stabilizing the closed-loop dynamics of each vehicle in the platoon while reducing the joint effect of estimation errors and communication channel noise on the tracking performance and string behavior of the vehicle platoon. Numerical examples are presented to illustrate the effectiveness of our methods.
BibTeX
@article{yang2022secure,
  author = {Tianci Yang and Chen Lv},
  title = {{A Secure Sensor Fusion Framework for Connected and Automated Vehicles under Sensor Attacks}},
  journal = {IEEE Internet of Things Journal},
  year = {2022},
  doi = {10.1109/jiot.2021.3101502},
}
Figure