Random field reconstruction with quantization in wireless sensor networks

Ido Nevat, Gareth W. Peters, Iain B. Collings

Research output: Contribution to journalArticlepeer-review

42 Citations (Scopus)

Abstract

We develop new and novel algorithms for spatial field reconstruction and spatial exceedance level estimation. We consider spatial physical phenomena which are partially observed by a wireless sensor network. We focus on the practical scenario in which, due to bandwidth and power constraints, the observations at the sensors are quantized and transmitted over imperfect wireless channels. We first develop the Spatial-Best Linear Unbiased Estimator (S-BLUE) algorithm for the spatial field reconstruction estimation. We then develop an algorithm that is based on a multivariate series expansion approach resulting in a Saddle-point type approximation to solve both problems of spatial field reconstruction and exceedance level estimation. We derive the Posterior Cramér Rao Lower Bound (PCRLB) and quantify the achievable MSE in the estimation. The estimation accuracy of the proposed algorithms is quantified via numerical comparisons between the two algorithms and the PCRLB.

Original languageEnglish
Article number6588606
Pages (from-to)6020-6033
Number of pages14
JournalIEEE Transactions on Signal Processing
Volume61
Issue number23
DOIs
Publication statusPublished - 2013
Externally publishedYes

Keywords

  • Bayesian inference
  • Gaussian processes
  • Kernel methods
  • saddle point approximation
  • wireless channels
  • Wireless sensor networks

Fingerprint

Dive into the research topics of 'Random field reconstruction with quantization in wireless sensor networks'. Together they form a unique fingerprint.

Cite this