Communication Efficient Multi-Vehicle Private Information Retrieval for VANETs

Internet of Vehicles (IoV) and Vehicular Micro Clouds (VMC) advance the services of Vehicular Ad-hoc Net works (VANETs) by offering flawless data sharing, enhancing the travel experience and convenience. However, this seamless data access also raises serious privacy concerns, as querying data may inadvertently reveal vehicle locations and behavior patterns. Solutions to protect the privacy of vehicles while accessing Location-Based Services (LBS) in VANETs are mostly based on cryptographic primitives, which become inefficient when the volume of accessed data is huge. Private Information Retrieval (PIR) techniques help to improve information retrieval rate. But PIR problem becomes very challenging when data is stored in single-server or colluding servers of the same service provider. However, the VANET scenario allows enhancement in the utilization of costly cellular spectrum for Network-to-Vehicle (N2V) communication, leveraging the spectrum available for Vehicle to-Vehicle (V2V) communication. In this work, we address the problem of single-server multi-vehicle private information retrieval (MVPIR) for VANETs with multi-server PIR solutions. The system consists of U vehicles, each requesting one or more of the K available messages from a centralized database, without revealing their data demands. We consider two privacy models: (i) full vehicle privacy, where the data demands of each vehicle remain hidden from all others (other vehicles and server), and (ii) group-private PIR, where the collective demand set of a group of vehicles is revealed within the group; but not individual data demands. We present capacity expressions for both these models with the corresponding designs of achievable schemes. Since the capacity-achieving solution for privacy model (ii) comes with a huge subpacketization cost, this work proceeds to analyze the trade-off between subpacketization level and achievable rate for various sub-optimal PIR schemes. Based on this analysis, a sub optimal scheme with a very low subpacketization level is proposed as a promising solution for highly dynamic VANET application scenarios. Despite its lower subpacketization, this scheme maintains information-theoretic privacy guarantees and demonstrates strong performance under realistic VANET parameters. Overall, our results extend classical multi-server PIR frameworks to single-server multi-vehicle domain, offering scalable, efficient, and deployable solutions that strike a balance between privacy, communication cost, and implementation feasibility.