Sparse recovery methodologies for quasi-distributed dynamic strain sensing

Lihi Shiloh*, Roy Shen-Tzur, Avishay Eyal, Raja Giryes

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Quasi-distributed measurement of strain and/or temperature is often implemented using arrays of weakly reflecting fiber Bragg gratings (FBGs) whose reflection peaks are centered at the same nominal wavelength. The signals are obtained by measuring the phase difference between the reflections of consecutive FBGs. Typically, in such a system, the spatial resolution of the interrogator must be compatible with the spatial separation between consecutive FBGs. Insufficient resolution leads to an overlap of reflection peaks, a decrease in the differential-phase signal and poor sensitivity. In this paper, we study the use of two different sparsity based methodologies for improving the sensitivity of such quasi distributed acoustic sensing systems in the case where traditional signal processing approaches do not provide sufficient spatial resolution. These methods enable relaxing the requirements regarding the interrogator or, alternatively, reducing the needed separation between reflectors. Experimentally, these techniques were used to measure 1 kHz dynamic strain induced in a fiber segment between two discrete reflectors, located at the end of a 4 km long fiber. The separation between the reflectors was 18 m while the pulse (spatial) width was intentionally chosen bigger than that. It yielded approximately 5 dB increase in the measured signal compared to the traditional processing approach and an order of magnitude improvement in the sensitivity, ~0.9 μrad/ √Hz .

Original languageEnglish
Article number024002
JournalJPhys Photonics
Volume2
Issue number2
DOIs
StatePublished - 27 Feb 2020

Keywords

  • Compressed sensing
  • Dynamic quasi-DAS
  • Fiber Bragg grating
  • Optic fiber sensors
  • Sparse recovery

Fingerprint

Dive into the research topics of 'Sparse recovery methodologies for quasi-distributed dynamic strain sensing'. Together they form a unique fingerprint.

Cite this