Weight classifier using optical time domain reflectometry and a long period grating sensor

The capabilities of signal processing tools to extract high-level features of interest from fiber sensors measurements have not been properly explored for interrogation purposes. In this context, this paper advances an interrogation system for long-period grating sensors by means of a novel low-cost method that combines an optical time-domain reflectometer with suitable descriptors to classify the sample weight under test. The operation principle is based on the attenuation bands induced by cladding modes of a long-period grating, which are affected by the applied mechanical strain. The proposed interrogation system does not require an optical spectrum analyzer, since the measured reflectometry traces directly feed a feature extractor tailored for the intended classification. Experimental results show, for example, that this method is capable of successfully differentiating a load whose mass is 7.2 g from another one whose mass is 14.1 g, with the interrogation equipment located 10 km from the loads. This implies a high sensitivity that is quite unexpected to attain with the commercially available off-the-shelf devices required by the proposed scheme.