Firm-Vehicle: Trusted Communication Enabled Instruction Embedding Model for Resource-Constrained VANET Environments

Instruction embedding is an essential technique in firmware security research and serves as a fundamental component in many vulnerability detection and security analysis methods for VANET. Instruction embedding maps the semantic information of instructions into fixed-dimensional vectors. However, the current instruction embedding approaches often neglect the operational requirements of resource-constrained environments typical of VANET. Moreover, the dependency on third-party disassembly tools for the extraction of instruction call graphs presents challenges related to runtime environment, thereby complicating the feasible application of instruction embedding techniques. To address these limitations, we propose a novel cross-architecture firmware instruction embedding model called Firm-Vehicle, specifically tailored for resource-constrained environments in VANET. Primarily, we devise a lightweight algorithm for extracting instruction call graphs, eliminating the reliance on third-party disassembly tools and improving the efficiency of call graphs extraction, thereby enhancing the instruction embedding model. Through evaluations and comparison with other approaches, Firm-Vehicle not only reduces the required time for instruction call graphs extraction but also enhances the stability of the instruction embedding model, enabling secure and efficient operation within the VANET environment.