Multimode Microstructured Optical Fiber Hydrophone

We demonstrate a fiber-optic hydrophone using a multimode microstructured pressure-sensitive optical fiber, which has a sensitivity comparable to a commercial electric hydrophone and 14 to 17 dB improvement over conventional single mode and multimode optical fiber. We then configure the pressure-sensitive fiber with a multi-fiber fan-out output where each output channel encodes different coupling from the pressure-sensitive fiber’s modal interference pairs. Thus, each channel has a different response to both the acoustic wave and environmental noise. We use a neural network model as the signal processing module to combine the acoustic signals from the multiple fiber channels to reconstruct the original signal. By combining the channels, the neural network model can effectively reduce crosstalk from environmental noise. We then demonstrate the ability to reconstruct complex audio signals using several classical music symphonies.