T-check: bug finding for sensor networks

  • Authors:

  • Peng Li;John Regehr

  • Affiliations:

  • University of Utah;University of Utah

  • Venue:
  • Proceedings of the 9th ACM/IEEE International Conference on Information Processing in Sensor Networks
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Sensor nodes are resource poor and failure-prone. Sensor networks are composed of many such nodes that are often hard to physically reach and that are connected by unreliable wireless links. Together, these factors make sensor network debugging into a challenging activity, and in fact it is not uncommon for a deployed sensornet to encounter sporadic faults that are effectively impossible to locate, reproduce, and fix. We developed T-Check, a tool that uses random walks and explicit state model checking to find safety and liveness errors in sensor network applications running on TinyOS. By building upon TOSSIM---an event-driven simulator that abstracts away interrupt-driven concurrency and other low-level hardware interaction---T-Check loses the ability to detect certain low-level errors, but gains enough scalability to detect distributed errors such as a collection tree protocol's failure to properly repair when a node dies. We have used T-Check to find previously unknown bugs in TinyOS.