Decoding Code on a Sensor Node

  • Authors:
  • Pascal Rickenbach;Roger Wattenhofer

  • Affiliations:
  • Computer Engineering and Networks Laboratory, ETH Zurich, Switzerland;Computer Engineering and Networks Laboratory, ETH Zurich, Switzerland

  • Venue:
  • DCOSS '08 Proceedings of the 4th IEEE international conference on Distributed Computing in Sensor Systems
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Wireless sensor networks come of age and start moving out of the laboratory into the field. As the number of deployments is increasing the need for an efficient and reliable code update mechanism becomes pressing. Reasons for updates are manifold ranging from fixing software bugs to retasking the whole sensor network. The scale of deployments and the potential physical inaccessibility of individual nodes asks for a wireless software management scheme. In this paper we present an efficient code update strategy which utilizes the knowledge of former program versions to distribute mere incremental changes. Using a small set of instructions, a delta of minimal size is generated. This delta is then disseminated throughout the network allowing nodes to rebuild the new application based on their currently running code. The asymmetry of computational power available during the process of encoding (PC) and decoding (sensor node) necessitates a careful balancing of the decoder complexity to respect the limitations of today's sensor network hardware. We provide a seamless integration of our work into Deluge, the standard TinyOS code dissemination protocol. The efficiency of our approach is evaluated by means of testbed experiments showing a significant reduction in message complexity and thus faster updates.