Transmission policies for energy harvesting sensors with time-correlated energy supply

Nicolo Michelusi, Kostas Stamatiou, Michele Zorzi

Research output: Contribution to journalArticlepeer-review

94 Scopus citations

Abstract

This paper considers a wireless sensor powered by an energy harvesting device, which reports data of varying importance to its receiver. Modeling the ambient energy supply by a two-state Markov chain («GOOD» and «BAD»), assuming a finite battery capacity constraint, and associating data transmission with a given energy cost, we propose low-complexity transmission policies, that achieve near-optimal performance in terms of the average long-term importance of the reported data. In particular, we derive the performance of the Balanced Policy (BP), which adapts the transmission probability to the harvesting state, such that energy harvesting and consumption are balanced. Our analysis demonstrates that the performance of the BP largely depends on the power-to-depletion, defined as the power that a fully charged battery can supply on average over a BAD period. Numerical results show that the optimal BP achieves near-optimal performance and that a BP which avoids energy overflow further reduces the gap with respect to the globally optimal policy. A heuristic BP, based on the analysis of a system with a deterministic and periodic energy supply, is also proposed, and the parallels between the deterministic system and its stochastic counterpart are discussed.

Original languageEnglish (US)
Article number6522422
Pages (from-to)2988-3001
Number of pages14
JournalIEEE Transactions on Communications
Volume61
Issue number7
DOIs
StatePublished - 2013
Externally publishedYes

Keywords

  • Energy harvesting
  • Markov decision processes
  • Wireless sensor networks

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Transmission policies for energy harvesting sensors with time-correlated energy supply'. Together they form a unique fingerprint.

Cite this