Decision Tree Regression Supervised Machine Learning Assisted Large Dynamic Range Refractive Index Detection Using MMI Coreless Multimode Fiber Sensor

A supervised machine learning (ML) algorithm is proposed for measuring refractive index (RI) values both below and above the RI of the fiber material using a multimode interference (MMI) fiber sensor. The sensor is constructed by splicing a coreless multimode fiber (CMF) segment between two single-mode fiber (SMF) leads. Measurement of low and high RI regimes is accomplished through the decision tree (DT) regression algorithm. The trained model algorithm demonstrates a wide dynamic range in RI measurement, covering the ranges of 1.30–1.39 (low RI regime) and 1.46–1.55 (high RI regime) without any RI ambiguity, achieving model accuracy of 99.77%. The guided and leaky modes mechanisms within the all-silica-based structure of the CMF are fundamentally insensitive to temperature, making it highly practical for deployment in conditions with varying temperatures without the need for any compensating scheme.