2020/06/09 by Miki E. Verma, Robert A. Bridges, Verma, Miki E. +7 · 1 citation
Computer Science · Engineering · #FOS: Computer and information sciences #FOS: Electrical engineering #Other Computer Science (cs.OH) #Real-Time Systems Scheduling #Signal Processing (eess.SP) #Vehicular Ad Hoc Networks (VANETs) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2006.05993
openalex publication_date 2020/06/09 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
CANs are a broadcast protocol for real-time communication of critical vehicle\nsubsystems. Original equipment manufacturers of passenger vehicles hold secret\ntheir mappings of CAN data to vehicle signals, and these definitions vary\naccording to make, model, and year. Without these mappings, the wealth of\nreal-time vehicle information hidden in the CAN packets is uninterpretable,\nimpeding vehicle-related research. Guided by the 4-part CAN signal definition,\nwe present CAN-D (CAN-Decoder), a modular, 4-step pipeline for identifying each\nsignal's boundaries (start bit, length), endianness (byte order), signedness\n(bit-to-integer encoding), and by leveraging diagnostic standards, augmenting a\nsubset of the extracted signals with physical interpretation. We provide a\ncomprehensive review of the CAN signal reverse engineering research. Previous\nmethods ignore endianness and signedness, rendering them incapable of decoding\nmany standard CAN signal definitions. Incorporating endianness grows the search\nspace from 128 to 4.72E21 signal tokenizations and introduces a web of changing\ndependencies. We formulate, formally analyze, and provide an efficient solution\nto an optimization problem, allowing identification of the optimal set of\nsignal boundaries and byte orderings. We provide two novel, state-of-the-art\nsignal boundary classifiers-both superior to previous approaches in precision\nand recall in three different test scenarios-and the first signedness\nclassification algorithm which exhibits a >97 % F-score. CAN-D is the only\nsolution with the potential to extract any CAN signal. In evaluation on 10\nvehicles, CAN-D's average \ℓ1 error is 5x better than all previous methods\nand exhibits lower ave. error, even when considering only signals that meet\nprior methods' assumptions. CAN-D is implemented in lightweight hardware,\nallowing for an OBD-II plugin for real-time in-vehicle CAN decoding.\n