Speckle wavemeter based on a multi-core fiber and compressive imaging.

Random speckle patterns contain valuable information about the incident light. Researchers have successfully constructed spectrometers and wavemeters by utilizing the speckles generated by inter-mode interferences of a multimode fiber (MMF). However, cameras were often employed to record the speckle data in previous reports. The camera's high cost (especially in the near-infrared range), large size, and low response speed limit the applications in optical communications, metrology, and optical sensing. A seven-core fiber (SCF) was fused with an MMF to capture the speckle pattern, where each core coupled part of the speckle field. Furthermore, we take advantage of the space division multiplexing capability of the SCF by incorporating an optical switch. This allows the variety of speckles generated by the incidence of different cores into the MMF. A convolutional neural network (CNN) regression algorithm was designed to analyze the complicated speckle data. The experimental results show that the proposed wavemeter can resolve adjacent wavelengths of 1 pm with an error of about 0.2 pm. We also discussed how different lengths of MMF influence the wavelength resolution. In conclusion, our research presents a robust and cost-effective approach to a wavelength measurement device by use of a seven-core optical fiber.