Safe microcontroller with code block fingerprint analysis accelerator for embedded-software integrity verification

Daejin Park, Jeonghun Cho

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

On-the-fly instruction code fail-detection in microcontrollers requires an additional code area, causing the periodic interruption of user software execution to scan the code memory. In this paper, newly designed instruction read-path architecture is proposed in order to provide safety-conscious execution of the user-programmed software. The proposed binary code conversion method generates the executable target code with additional padding data, enabling fast detection of abnormal code executions with small computation resources by representing the intended code execution flow with the corresponding fingerprint digest message. The code integrity verification based on the fingerprint comparison is silently performed with a time-multiplexed inspection method in the background mode. The implementation results show that three times analysis speed up and seven times the current reduction by the proposed read-path architecture, which could be achieved with the hardware overhead of about 0.18um CMOS-based 4500 NAND gates and only additional 1KB fingerprint table in the case of 16 protection sectors.

Original languageEnglish
Pages (from-to)5027-5036
Number of pages10
JournalInformation
Volume18
Issue number12
StatePublished - Dec 2015

Keywords

  • Binary translation
  • Dependable computing
  • Fault robust MCU design
  • Flash memory

Fingerprint

Dive into the research topics of 'Safe microcontroller with code block fingerprint analysis accelerator for embedded-software integrity verification'. Together they form a unique fingerprint.

Cite this