Fiber bend position sensor enhanced by twisted light

In this presentation, we will outline a method to detect the position of fiber bends in both few-mode and multi-mode fiber by measuring the distortion of OAM modes during propagation. The complex mode coupling that occurs can make it difficult to directly infer what caused the optical distortions one observes. Therefore, we implement a novel optical element enhanced Machine Learning approach that combines all-optical mode transformation with a Support Vector Machine Learning algorithm to accurately determine the location of a fiber bend with greater than 87% accuracy in MM fiber and 68% in FM optical fiber.