Multimode fiber (MMF) sensors based on intermodal interferences have been widely studied due to their advantages of easy manufacture and high sensitivity. We introduced the scheme of spatial multiplexing to MMF sensors for improvement of resolution. A seven-core fiber (SCF) was fused with the NCF sensing head with a slight misalignment. Therefore, each core of the SCF has different coupling coefficients with the NCF due to its particular location at the cross section of the fusion point. The NCF sensor head was then shared by multiple cores and thus can be considered to be multiplexing. In addition, traditional demodulation method based on transmission spectrum mainly depends on the tracking spectral peaks or dips, where only a portion of spectrum data were utilized. To address this limitation, we further proposed a deep convolutional network to interrogate the spectrum that can exploit the whole spectrum data. The measurements of temperature and liquid refractive index were demonstrated in experiments. With the help of spatial multiplexing and deep learning algorithm, the resolution of temperature and refractive index of 0.0054°C and 2.25×10-5 RIU were obtained, about ten times improvement of the traditional NCF sensor. Our proposed spatial multiplexing offers a simple way to enhance the resolution without extra processing of the NCF.
Read full abstract