According to a study published in the IEEE Internet of Things Journal and co-authored by scientists from the University of Sharjah in the United Arab Emirates, the University of Maryland in the United States, and Abdul Wali Khan University Mardan in Pakistan, scientists have created an artificial intelligence tool to consolidate vehicle and driver privacy.
Architecture of a secured distributed edge intelligence for IoV. Image Credit: IEEE Internet of Things Journal (2024): DOI: 10.1109/JIOT.2024.3483275
According to scientists, preserving the privacy of the so-called Internet of Vehicles (IoV) has emerged as a major challenge due to vehicles' geographical mobility and limited resources.
The scientists contend that the issue has gotten worse because of the “limited resources of onboard units (OBUs)” and the inadequacies of embedded sensors in cars, which “lure the adversaries to launch various types of attacks.
The scientists stated, “Thus, lightweight but reliable authentication schemes need to be designed to combat these attacks.”
Thanks to a network known as the Internet of Vehicles (IoV), vehicles can communicate with one another, as well as with intelligent communication devices in parking lots, pedestrians, and road infrastructure. According to the authors, this technology “has transformed cities around the globe by providing real-time communication.”
IoV-connected vehicles have also embedded sensors and units that gather valuable data and transmit it to the nearest server modules or roadside units (RSUs). According to the researchers, “The operational capabilities of these vehicles are further augmented by artificial intelligence, particularly machine learning and deep learning, which analyze and interpret data in real-time.”
In the era of the IoV, vehicle security is susceptible to cyberattacks that could intercept or even change vehicle-infrastructure communication, leading to unfortunate events. The authors’ AI tool is marketed as a machine learning solution, which has been proposed as a remedy.
Autonomous vehicles now have an onboard unit device, or OBU, as part of their intelligence transportation system, or ITS.
The authors assert, however, that the communication system installed in the cars still faces difficulties, mainly those pertaining to limited bandwidth and delays in receiving responses from cloud-based services within a predetermined window of time.
The authors stress that even with the addition of machine learning (ML) and deep learning (DL) algorithms, the current cloud servers are still unreliable because they are “unable to provide swift responses to the vehicles that can lead to catastrophic circumstances at the roads.”
Similarly, embedded sensors on-board units (OBUs) and RSUs “are resource-constrained and are unable to support computationally complex security and privacy preservation schemes. It would require ample resources for these devices to securely communicate with the cloud servers,” the authors emphasized.
To overcome these obstacles, the researchers suggest “an ML-based authentication scheme that trains and classifies the vehicles at the edge servers in a distributed manner, preserves the privacy of communicating entities, and minimizes the bandwidth consumption and delay experienced by the vehicles.”
For this purpose, the authors create a new machine learning-based authentication mechanism to address privacy and security concerns that the emerging Internet of Vehicles (IoV) ecosystem is currently dealing with.
The researchers conducted their experiments in a simulated environment, comparing the proposed scheme to “the existing state-of-the-art schemes in terms of communication, processing, and storage overheads.”
The authors noted, “Simulation results have concluded that the proposed scheme is not only pruned against well-known intruder attacks, but it is equally lightweight and effective concerning various performance evaluation metrics such as computation, communication, and storage overheads.”
The authors emphasize that the scheme they have developed addresses the issue of bandwidth scarcity and excessive delays that vehicles currently face when communicating with cloud servers.
The authors also noted, “The ML-based approach extends the decision power of vehicles and edge servers to identify adversaries. Our scheme requires that each vehicle participates in an offline phase, where a trusted authority shares a list of MaskIDs and secret keys of legitimate vehicles and edge servers.”
The proposed scheme requires each vehicle to participate in an offline phase in which a trusted authority shares a list of masked identities, or MaskIDs, as well as secret keys for legitimate vehicles and edge servers.
Once they have their own list of masked identities, vehicles and servers can authenticate one another without relying on cloud servers, guaranteeing quicker and more effective communication.
To lessen the computational load on the vehicle, the nearest edge server uses the MaskIDs and secret keys to confirm its identity when it begins to communicate.
The scientists explained, “In our scheme, each vehicle and edge server (via RSU) is equipped with an ML algorithm to classify adversaries from legitimate ones.”
The machine learning algorithm strengthens security against common cyberattacks, such as impersonation and man-in-the-middle attacks, by analyzing and validating communication patterns in real time.
Including a timespan in the payload of each encrypted message to prune the proposed scheme against well-known adversarial attacks distinguishes the method from other currently available tools.
The researchers concluded, “The simulation results verify the exceptional performance of our scheme in terms of computational overhead, communication overhead, and storage overhead.”
Journal Reference:
jan, M. A. et. al. (2024) An ML-Based Authentication for Privacy-Preservation in a Distributed Edge-Enabled Internet of Vehicles. IEEE Internet of Things Journal. doi.org/10.1109/JIOT.2024.3483275