Geng Yu

The Internet of Things (IoT) and edge computing have enabled the utilization of IoT for environmental monitoring and energy optimization. The Internet of Vehicles (IoV), with its extensive coverage and rich sensing capabilities, is particularly well-suited for these purposes, including traffic resource optimization. However, the openness of the IoV has introduced security and privacy concerns. Vulnerabilities in IoV systems could allow malicious adversaries to cause environmental contamination and to potentially compromise critical transportation infrastructure through targeted attacks. To address these issues, a solution based on secure multi-party computation (SMC) is proposed. First, an efficient anonymous authentication scheme is designed using linear equation system theory and secure multi-party computation, which prevents illegal vehicles from accessing the Green IoV. Second, a verifiable secure multi-party computation scheme is proposed based on replicated secret sharing, allowing vehicles to submit data to the Green IoV for environmental monitoring and resource optimization while maintaining privacy. Security analysis reveals that the proposed scheme possesses anonymity, verifiability, privacy preserving, and dynamic joining capabilities. Furthermore, experimental analysis demonstrates the proposed scheme achieving a runtime of only 3.7 ms each vehicle for authentication and the lightweight multi-party computation reduces once multiplication latency to 1.2 ms, 71% faster than similar methods.

Vehicular ad-hoc networks (VANETs) enable communication among vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) with a dedicated short range communications (DSRC). In VANETs, the real-time trajectory prediction service can be provided by roadside units (RSUs) using data collected from vehicles. This service is important for the prevention of traffic accidents. Due to the openness of the communication in VANETs, the transmitted data is vulnerable to illegal users' attacks that results in the leakage of vehicles' data. To address the above concerns, a privacy-preserving multi-party computation scheme with secret sharing is proposed for vehicle trajectory prediction in VANETs. The proposed scheme is designed to protect the privacy of transmitted vehicle data during trajectory prediction. Note that four distributed interactive protocols of $\boldsymbol{(\cdot)}$ , $\boldsymbol{SecExp( \cdot)}$ , $\boldsymbol{SecRec(\cdot)}$ and $\boldsymbol{SecCmp(\cdot)}$ are introduced to carry out the computation of nonlinear logarithm, exponent, reciprocal and comparison functions in the proposed scheme. The aforementioned interactive protocols facilitate the distributed deployment of the Adaboost ensemble learning algorithm among multiple RSUs, while ensuring the privacy of vehicle data using secret sharing simultaneously. Security analysis proves that the proposed scheme can effectively protect the privacy of vehicle data. Performance analysis demonstrates that the proposed scheme has a minimal impact on the prediction accuracy of the Adaboost algorithm compared to the differential privacy (DP)-based scheme. Furthermore, the low latency of the proposed scheme allows for efficient trajectory prediction, as demonstrated by the runtime of the proposed scheme on test set.