跳到主要导航 跳到搜索 跳到主要内容

FirmCAN: Sensitive CAN Knowledge Leakage from Automotive ECUs

科研成果: 期刊稿件文章同行评审

摘要

As de facto standards of in-vehicle network communications among various ECUs (Electronic Control Units), CAN (Controller Area Network) protocols invented by Bosch rely on the privately defined unique identifiers CAN ID in CAN messages that do not convey any destination address. However, sticking to error-handling mechanisms in CAN protocols, an ECU with an amount of transmission errors would enter bus-off state (i.e., go offline thereupon) and shall be recovered according to some prescribed bus-off recovery mode (BOM). All this sensitive CAN knowledge concealed inside an ECU by OEMs (Original Equipment Manufacturers) shall not be revealed; however, it could have extensive practical applications, both for adversarial behavior (e.g., target ECU attacks) and for security enhancement mechanisms (e.g., intrusion detection system designs). The article presents FirmCAN, the first automatic analysis framework to dope out sensitive CAN knowledge (CAN IDs and bus-off recovery mode, in particular) compiled in automotive ECU firmwares. FirmCAN first identifies base address (using accurate absolute function entry addresses) and then performs CAN module API positioning and sensitive configuration information resolution. We buckle down to automotive ECUs that resort to fixed-address mailboxes as CAN module transmission buffers (e.g., Renesas SuperH/RA series) and present concrete algorithms for each analysis phase. Our experimental evaluations first investigate firmwares extracted from real automotive ECUs. We then develop our own applications using RA6M4 development boards, which not only produce required firmwares to evaluate FirmCAN but also create ground truth through hardware debugging. All evaluations demonstrate that FirmCAN can accurately garner above-mentioned sensitive CAN knowledge. FirmCAN can be trivially generalized to engage in extended frames and CAN modules with similar transmission logic, e.g., TI (Texas Instruments), ST (ST Microelectronics), and so on.

源语言英语
文章编号76
期刊ACM Transactions on Embedded Computing Systems
24
5
DOI
出版状态已出版 - 12 9月 2025

指纹

探究 'FirmCAN: Sensitive CAN Knowledge Leakage from Automotive ECUs' 的科研主题。它们共同构成独一无二的指纹。

引用此