RNCA Boosts Performance of a 7-Core FBG Sensor Based on the Microwave Photonic Self-Vernier Effect

In this paper, we proposed a 7-core fiber Bragg grating (7-core FBG) curvature sensing system based on the microwave photonic self-Vernier effect, and combined it with a residual network with channel attention (RNCA) to achieve intelligent demodulation. The system achieved the sensitivity amplification through the superposition of the raw response from a single interferometric structure and its delayed replica in the microwave domain. Concurrently, the RNCA model eliminated the cumbersome traditional envelope fitting process by learning the complex nonlinear mapping between the raw microwave spectrum and curvature end-to-end. Experimental results demonstrated that the sensor achieved the sensitivity of up to −3.693 MHz/m⁻¹ in curvature measurement. Under sparse sampling conditions, the RNCA model maintained R²=99.95% and RMSE=0.044 m⁻¹ with the single inference time of only 0.2 s, where the RMSE is the root mean square error. This research established a novel technical paradigm for realizing optical fiber curvature sensing systems characterized by structural simplicity, intelligent demodulation, and superior performance.