Scholarlink

This work describes the design and implementation of an open-source multithreaded event data recorder (EDR) and evaluates its performance on a Raspberry Pi 5 using a crash dataset from NHTSA. Results demonstrate that detection accuracy is good when using the Cumulative Sum (CUSUM) change point indicator. Moreover, lock-free queues used for communication among Broker, Logger, Postman, Safety, UI, and Cellular threads are faster and less CPU-intensive than lock-based alternatives, although they consume the same power. Bundling transactions for pre- and post-crash database storage provides a great speedup and lower CPU usage.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Design and Implementation of an Efficient Multithreaded Event Data Recorder on Raspberry Pi 5

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Vasilis Kavvalos,
  • Miltos D. Grammatikakis,
  • George Kornaros

摘要

This work describes the design and implementation of an open-source multithreaded event data recorder (EDR) and evaluates its performance on a Raspberry Pi 5 using a crash dataset from NHTSA. Results demonstrate that detection accuracy is good when using the Cumulative Sum (CUSUM) change point indicator. Moreover, lock-free queues used for communication among Broker, Logger, Postman, Safety, UI, and Cellular threads are faster and less CPU-intensive than lock-based alternatives, although they consume the same power. Bundling transactions for pre- and post-crash database storage provides a great speedup and lower CPU usage.