Heartbeat design for energy-aware IoT: are your sensors alive?

Sarfo Gyamfi, James Brusey, Elena Gaura, Ross Wilkins

Research output: Contribution to journalArticle

1 Citation (Scopus)
2 Downloads (Pure)

Abstract

A number of algorithms now exist for using model-based prediction at the sensor node of a wireless sensor network (WSN) to enable a dramatic reduction in transmission rates, and thus save energy at the sensor node. These approaches, however, sometimes reduce the rate so substantially as to make the health state of the network opaque. One solution is to include a regular heartbeat transmission whose receipt or otherwise informs the sink about the health state of the node. However, given that a large period increases the probability that dead nodes go unnoticed at the sink, while a small period likely increases the energy cost of communication, what should be the period of the heartbeat transmission? In this paper, we examine the use of heartbeats in WSN design. We derive a general protocol for optimal and dynamic heartbeat transmission by minimising the Bayes risk, which is the expected cost of missing data from dead nodes plus the energy cost of heartbeat transmissions. Our proposed algorithm is dynamic in the sense that the heartbeat period is updated as time goes on and node failures become more probable. We validate our design experimentally using three real-world datasets, and show a 36% reduction in the total heartbeat operational cost over a heartbeat transmission with a fixed period; the results also highlight the superiority of our algorithm over arbitrarily chosen heartbeat periods in different WSN settings, thus promising significant cost savings in WSN applications.

Original languageEnglish
Pages (from-to)124-139
Number of pages16
JournalExpert Systems with Applications
Volume128
Early online date24 Mar 2019
DOIs
Publication statusPublished - 15 Aug 2019

    Fingerprint

Keywords

  • Edge mining
  • Failure detection
  • Heartbeat transmission
  • Internet of things
  • Wireless sensor networks

ASJC Scopus subject areas

  • Engineering(all)
  • Computer Science Applications
  • Artificial Intelligence

Cite this