Accurate Peak Detection for Optical Sensing with Reduced Sampling Rate and Calculation Complexity

Abstract

Fiber Bragg gratings (FBGs) are widely applied in optical sensing systems due to their advantages including being simple to use, high sensitivity, and having great potential for integration into optical communication systems. A common method used for FBG sensing systems is wavelength interrogation. The performance of interrogation based sensing systems is significantly determined by the accuracy of the wavelength peak detection processing. Direct maximum value readout (DMVR) is the simplest peak detection method. However, the detection accuracy of DMVR is sensitive to noise and the sampling resolution. Many modified peak detection methods, such as filtering and curve fitting schemes, have been studied in recent decades. Though these methods are less sensitive to noise and have better sensing accuracy at lower sampling resolutions, they also confer increased processing complexity. As massive sensors may be deployed for applications such as the Internet of things (IoT) and artificial intelligence (AI), lower levels of processing complexity are required. In this paper, an efficient scheme applying a three-point peak detection estimator is proposed and studied, which shows a performance that is close to the curve fitting methods along with reduced complexity. A proof-of-concept experiment for temperature sensing is performed. 34% accuracy improvement compared to the DMVR is demonstrated.