Machine Learning-Based Cybersecurity in Advanced Autonomous and Connected Vehicles

Abstract

The rapid evolution of Advanced Autonomous and Connected Vehicles (AACVs) has redefinedthe future intelligenttransportation,bringingforthsignificantbenefitsintermsofsafety,efficiency,anduserconvenience.However,thisincreasedinterconnectivityhasalsointroducedawiderangeofcybersecuritychallenges.AACVsdependheavilyoncomplexarchitecturesinvolvingElectronic Control Units (ECUs), onboard sensors, and communication protocols such as Vehicle-to-Vehicle (V2V), Vehicle-to-Infrastructure (V2I), and Vehicle-to-Grid (V2G) networks. These components make the system vulnerable to numerous cyberattacks, including GPS spoofing, Replay Attacks, Man-in-the- Middle (MITM) Attacks, Denial-of-Service (DoS) attacks, and unauthorized remote access via Software Defined Radio (SDR) devices like HackRF One. Inthisstudy,wepresentarobustmachinelearning(ML)-drivencybersecurityframeworkspecifically designed to safeguard AACVs against these sophisticated threats. Our approach integrates a multi- layered Intrusion Detection System (IDS) combining rule-based filtering with real-time anomaly detection using decision trees, ensemble methods, and Generative Adversarial Networks (GANs). To address data privacy concerns, we employ federated learning techniques that facilitate decentralized modeltrainingwithoutexposingsensitivevehiculardata. Further,oursystemarchitectureincorporatessecurediagnostics,biometricauthenticationprotocols,and advanced encryption mechanisms to defend against zero-day vulnerabilities and internal threats. Experimentalsimulationsconductedonsynthetic CANbustrafficdemonstratetheproposedmodel\'sabilitytodetectandrespondtothreatswithhighaccuracyandlowlatency. Byleveragingartificialintelligence,thisresearchaimstoestablishanadaptive,scalable,andresilientcybersecurityframeworkthatevolveswithemergingthreats.Ourfindingsunderlinethecriticalroleof ML in enhancing vehicular cybersecurity and serve as a foundation for future innovations in safe, intelligent transportation systems.

Introduction

The rapid development of Machine Learning (ML) and Information and Communication Technology (ICT) has greatly advanced Connected Autonomous Vehicles (CAVs), enhancing driving efficiency, road safety, and user convenience. However, these benefits come with serious cybersecurity challenges due to increased reliance on wireless communication and autonomous systems.

Cybersecurity Threats in CAVs

Traditional security measures (e.g., mechanical locks, basic encryption) are no longer adequate. Modern vehicles are vulnerable to a wide range of cyberattacks, including:

• Replay Attacks and Man-in-the-Middle (MITM) Attacks, exploiting flaws in authentication.

• Sybil Attacks, manipulating traffic systems with fake identities.

• Attacks via HackRF One, a Software Defined Radio used to intercept and replay smart key signals.

• Sensor attacks (LiDAR, radar, cameras) via laser interference, distorting vehicle perception.

• GPS spoofing and jamming, misleading vehicle navigation.

• Exploits through Wi-Fi, Bluetooth, RF, and vehicle communication systems like CAN bus and V2X.

Literature Review Highlights

• Multiple studies reveal vulnerabilities in authentication systems, wireless communication, and smart key transmissions.

• Sensor-based attacks can affect navigation and perception accuracy.

• AI-based security mechanisms like Intrusion Detection Systems (IDS) and anomaly detection outperform traditional rule-based methods.

• Threats also affect infotainment systems, ECUs, and charging infrastructures in Electric Vehicles (EVs).

• Researchers propose solutions like encrypted communication, biometric authentication, blockchain, and RF shielding.

• There is a growing interest in edge/cloud computing, fuzzy logic, and trust-based models to ensure real-time, scalable, and secure systems.

Modeling and Simulation Framework

The study includes a robust simulation platform to evaluate cyber threats in real-time CAV environments using:

• Python-based CAN simulation (python-can) to emulate in-vehicle communication and inject attack scenarios like Replay, DoS, and ID spoofing.

• Deep Learning with TensorFlow using:

  • LSTM networks to model time-series patterns.

  • Autoencoders for unsupervised anomaly detection.

  • Convolutional layers for spatial feature extraction.

• Feature engineering with pandas and NumPy to analyze:

  • Message frequencies

  • Timing intervals

  • Payload entropy

  • Bit-level abnormalities

• WSL with Ubuntu enables seamless integration of Linux-native tools and cross-platform development without full virtualization, ensuring efficient development and testing of cybersecurity mechanisms.

Conclusion

The integration ofMachine Learning (ML) and Information and Communication Technology (ICT) in Connected Autonomous Vehicles (CAVs) has significantly revolutionized the transportation industry by enhancing operational efficiency, improving safety, and offering user-friendly solutions. However, these technological advancements have also brought forth critical cybersecurity challenges, leaving autonomous vehicles increasingly vulnerable to cyberattacks such as unauthorized access, signalspoofing,sensormanipulation,andremotehacking.Thisstudyunderscorestheurgentneed for robust and intelligent security frameworks to mitigate such threats. Through comprehensive analysis, this research has examined multiple vulnerabilities within CAV systems, including weaknesses in authentication mechanisms, exploitation of smart key systems, security gaps in wireless communications, and GPS spoofing attacks. It has also highlighted the risks associated with the compromise of sensor-based perception systems, emphasizing the importance of securing LiDAR, radar, and camera inputs against adversarial manipulation. Addressing these risks, the study aims to propose advanced cybersecurity models that ensure reliable and safe functioning of autonomous vehicles. The contributions of this work include the development of ML-powered intrusion detection systems (IDS), the application of encryption techniques, the exploration of blockchain-based authentication, and the implementation of anomaly detection using AI models. These mechanisms enable real-time threat monitoring and proactive defense against potential intrusions in vehicular networks. The deployment of intelligent security layers allows CAVs to dynamically identify, assess, and neutralize attacks, reducing the likelihood of system failure or accidents caused by cyber threats. Lookingahead,thefindingspresentedinthispaperprovideafoundationforfutureresearchfocusedonadaptiveAIbasedsecuritysolutions,quantumresistantencryptionprotocols,andblockchainintegratedidentityverification.Furthermore,advancinghybridlocalizationtechniquesthatcombine GPS with alternative systems may reduce vulnerability to spoofing and jamming. Collaborationamongautomotivemanufacturers,cybersecurityspecialists,andpolicymakerswillbeessentialtoestablishgloballyacceptedsecuritystandardsforautonomousvehicles. In summary, this research supports the long-term vision of secure, dependable, and scalable autonomous mobility. By advancing intelligent cybersecurity frameworks and integrating real-time AI-based threat detection, the industry can ensure the responsible deployment ofautonomoussystems while building public trust in smart transportation technologies. The continued evolution of defense mechanisms, coupled with proactive regulatory involvement, will be crucial in protecting future mobility ecosystems from emerging cyber risks.

References

[1] S. I. Jung and S. D. Ho, \"A Study on Hacking Attacks and Vulnerabilities in Self-Driving Cars with Artificial Intelligence,\" in Proceedings of the PAUL Math School Conference, Goesan-gun, Chungcheongbuk-do, Republic of Korea, 2023. [2] I. Durlik, T. Miller, A. ?obodzi?ska, E. Kostecka, and Z. Zwierzewicz,\"Cybersecurity in Autonomous Vehicles—Are We Ready for the Challenge?\" Maritime University of Szczecin, Szczecin, Poland, 2023. [3] P. Sharma and J. Gillanders, \"Cybersecurity and Forensics in Connected Autonomous Vehicles: A Review of the State-of-the-Art,\" University at Albany – State University of New York, Albany, NY, USA, 2023. [4] M. D. Mwanje, O. Kaiwartya, M. Aljaidi, Y. Cao, S. Kumar, D. N. Jha, A. Naser, and J. Lloret, \"Cybersecurity Analysis of Connected Vehicles,\" Nottingham Trent University, Nottingham, UK. [5] P. S. Devanandanan and R. B. Rengarajan,\"Cybersecurity in an Electric Vehicle: A Comprehensive Review Paper,\" International Research Journal of Modernization in Engineering, Technology and Science. DOI: 10.56726/IRJMETS56581. [6] R. S. Rathore, C. Hewage, O. Kaiwartya, and J. Lloret,\"In-Vehicle Communication Cybersecurity: Challenges and Solutions,\" Sensors, vol. 22, no. 16, p. 6679, Sep. 2022. DOI: 10.3390/s22176679.

Copyright

Copyright © 2025 Nikita Hatwar , Tejal Borkar , Prashik Lamsoge , Vedant Kapgate , Anurag Yamnurwar , Ashay Wanjari . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.