8 October 2015 A source number estimation method for single optical fiber sensor
Author Affiliations +
Proceedings Volume 9679, AOPC 2015: Optical Fiber Sensors and Applications; 96790T (2015) https://doi.org/10.1117/12.2199696
Event: Applied Optics and Photonics China (AOPC2015), 2015, Beijing, China
Abstract
The single-channel blind source separation (SCBSS) technique makes great significance in many fields, such as optical fiber communication, sensor detection, image processing and so on. It is a wide range application to realize blind source separation (BSS) from a single optical fiber sensor received data. The performance of many BSS algorithms and signal process methods will be worsened with inaccurate source number estimation. Many excellent algorithms have been proposed to deal with the source number estimation in array signal process which consists of multiple sensors, but they can not be applied directly to the single sensor condition. This paper presents a source number estimation method dealing with the single optical fiber sensor received data. By delay process, this paper converts the single sensor received data to multi-dimension form. And the data covariance matrix is constructed. Then the estimation algorithms used in array signal processing can be utilized. The information theoretic criteria (ITC) based methods, presented by AIC and MDL, Gerschgorin’s disk estimation (GDE) are introduced to estimate the source number of the single optical fiber sensor’s received signal. To improve the performance of these estimation methods at low signal noise ratio (SNR), this paper make a smooth process to the data covariance matrix. By the smooth process, the fluctuation and uncertainty of the eigenvalues of the covariance matrix are reduced. Simulation results prove that ITC base methods can not estimate the source number effectively under colored noise. The GDE method, although gets a poor performance at low SNR, but it is able to accurately estimate the number of sources with colored noise. The experiments also show that the proposed method can be applied to estimate the source number of single sensor received data.
© (2015) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Junpeng Hu, Zhiping Huang, Shaojing Su, Yimeng Zhang, Chunwu Liu, "A source number estimation method for single optical fiber sensor", Proc. SPIE 9679, AOPC 2015: Optical Fiber Sensors and Applications, 96790T (8 October 2015); doi: 10.1117/12.2199696; https://doi.org/10.1117/12.2199696
PROCEEDINGS
7 PAGES


SHARE
Back to Top