Signal Recognition and Location of Distributed Optical Fiber Vibration Sensing System Using Reservoir Computing Based on Mutually Injected Semiconductor Lasers

A distributed optical fiber vibration sensing (DOFVS) system for signal recognition and location is proposed and implemented by using a reservoir computing (RC) system based on mutually injected semiconductor lasers (SLs). The sensitivity of the RC to the environment is used to detect an external vibration, that is, the mutually injected channel of the SLs is the sensing fiber of the DOFVS system. The phase changes induced by the vibration are converted into the intensity fluctuations through the mutually injected effects of the SLs. Both the signal recognition and location are realized by the training and testing process of the RC classifier model, especially building multiple linear classifiers of the RC and adopting a winner-takes-all method. It is verified numerically and experimentally that the system can detect well the external vibration and meets the basic properties as a reservoir in the injection locking (IL) state. The recognition and location accuracies of two types of vibration signals and 20 sensing positions can reach 99.99% and 100% in an experimental system with a 1.12-km-long sensing fiber. The experimental results show that the proposed system can simultaneously realize the signal identification and location with the simple method, fast processing speed, and high accuracy.