Qiang Liu

Multicore fiber Bragg gratings (MCF-FBGs) enable vector bending sensing through the responses of spatially distributed cores. However, peak-based demodulation can be affected by peak localization errors and spectral distortion. This paper proposes a physical-prior-guided spectral token network (PST-Net) for simultaneous angle and curvature demodulation using seven-core FBGs (7-core FBGs). A multichannel composite input tensor is constructed from pre-bending and post-bending spectra, absolute wavelength coordinates, and differential spectra, and mapped into fused spectral tokens through overlapping spectral patch embedding. In parallel, a physical-prior branch encodes bending-related wavelength and intensity descriptors into a physical-prior token. Multi-head self-attention jointly processes these tokens, combining sensing-related prior information with cross-patch spectral feature learning. Experimental results show mean root-mean-square errors of 0.459° and 0.095 m ^-1 for angle and curvature, respectively, with corresponding mean coefficients of determination of 99.98% and 99.62%. The proposed physically driven framework integrates bending-response knowledge with spectral representation learning for MCF-FBG vector bending demodulation.

Fiber-optic acoustic sensors offer immunity to electromagnetic interference, passive sensing, and remote signal transmission. However, degradation or failure of a single acoustic sensor can substantially impair unmanned aerial vehicle (UAV) recognition. Here, we develop a UAV acoustic-recognition system based on a four-channel extrinsic Fabry–Perot interferometer (EFPI) sensor array and propose quality-aware reliability-weighted fusion (QARWF). The array preserves alternative discriminative evidence after local channel degradation, whereas QARWF jointly exploits channel-level task features and time–frequency quality descriptors to estimate reliability weights. Without prior knowledge of the fault type or affected channel, it suppresses low-quality observations and emphasizes contributions from healthy channels. Paired-seed experiments on event-level samples show that, under severe degradation of one channel, the accuracy of a single-channel convolutional neural network–long short-term memory (CNN–LSTM) model falls to 54.06% ± 3.77%, whereas the four-channel QARWF system retains 94.92% ±0.96%. Across four frozen unknown two-channel faults, QARWF improves the worst-case accuracy of fixed feature averaging from 68.08% to 76.30% while maintaining 95.58% accuracy under fault-free conditions. These results demonstrate that synchronous multipoint observations from an EFPI array, together with quality-aware fusion, improve information availability and the recognition floor under partial sensor degradation, providing a fiber-optic solution for continuous and reliable UAV acoustic monitoring in complex environments

Fiber-optic Fabry–Pérot sensors have already been applied to many applications that require fast dynamic measurement due to their superior advantages of electromagnetic interference immunity, high resolution, fast response time, etc. In order to maintain high measurement accuracy, a fast demodulation technique based on white light interferometry is essential for accurate Fabry–Pérot cavity length determination. In this work, we utilize a deep learning network incorporating a compressed sensing algorithm, achieving real-time cavity length demodulation using a Vernier-tuned distributed feedback laser whose output wavelength can be rapidly tuned within a very broad spectral bandwidth. Our method can, in principle, achieve a measurement sampling rate equivalent to the laser wavelength switching rate, up to 10 MHz, over 2 orders of magnitude enhancement compared to the existing techniques, which clearly has great potential in applications that require high-resolution and high-speed absolute measurement.

A microwave photonic (MWP) fiber-ring (FR) temperature sensor that enhances sensitivity with the assistance of the self-Vernier effect is proposed and demonstrated. By using a fiber-ring interferometer combined with a delay fiber, the Vernier effect can be generated through the superposition of the interferometer’s intrinsic spectrum with its time-delayed counterpart. The method eliminates the need to integrate two separate interferometers and to precisely control their slightly different optical path differences (OPDs). Microwave photonic technology converts spectrum shifts of the interferometer caused by temperature variations into frequency shifts of the notch point in the frequency domain, and the self-Vernier effect further amplifies this frequency changes, enhancing the sensitivity. The sensitivity of the proposed simple microwave photonic fiber-ring temperature sensor constructed in the experiment reached -392.317 kHz/℃, representing approximately 15 times higher than that without the self-Vernier effect.

Current optical carrier-based microwave interferometry sensing systems generally rely on indirect measurements of amplitude-frequency characteristics of microwave signals, which suffer from conversion errors and limited linearity, while the phase information in the microwave domain remains underexploited. To address these limitations, the 4-step Carré phase demodulation algorithm, widely used in the optical domain, is introduced to microwave signal processing. This approach requires only four equally spaced phase samples and is insensitive to phase shift steps, significantly reducing demands on hardware precision and environmental stability. The experimental results show that extracting phase directly from microwave interference signals is successful. Compared with the scanning frequency method, the sampling efficiency is improved by an order of magnitude, and highly sensitive temperature sensing is achieved. A temperature sensitivity coefficient of 2.5 rad/(GHz $\cdot ^{\circ }$ C) is achieved, confirming the feasibility and potential of this microwave phase demodulation method for constructing highly robust and cost-effective sensing systems.

A primary bottleneck in deploying Fiber Bragg Grating (FBG) sensors lies in their inherent dual sensitivity to thermal and mechanical variations, which mandates robust decoupling mechanisms for precise parameter extraction. To address this persistent cross-sensitivity issue, this study introduces a novel interrogation scheme that integrates a Convolutional Neural Network with a Bidirectional Long Short-Term Memory (CNN-BiLSTM) architecture. Instead of relying on conventional peak-tracking algorithms or isolated central wavelengths, our proposed data-driven strategy directly mines structural features from the full reflection spectra, thereby substantially mitigating cross-interference errors. The experimental results reveal that the coefficients of determination (R²) for strain and temperature prediction reach 99.37% and 99.75% each, while the root mean square errors (RMSEs) are 13.51 µε and 1.42 °C, respectively. The proposed method requires only a single FBG sensor, which reduces the sensor requirements, showing great potential in sensing applications requiring low costs and high adaptability. In addition, in some special environments, temperature information cannot be obtained, so we utilize another reference FBG to realize the temperature compensation. Meanwhile, we proposed a spectral differencing method (SDM) by differencing the spectra of the two FBGs to obtain the spectra containing only strain information and sent them as a dataset for model training, with a 4-times improvement in accuracy over traditional compensation methods. Finally, we also explored the application of the system for distributed FBGs, achieving an absolute peak wavelength interrogation precision of approximately ±0.02 nm. The system is expected to be applied in the field of structural health monitoring, which is promising even in harsh environments.

We propose a high-resolution curvature sensor utilizing seven-core fiber Bragg gratings (7-Core FBGs) interrogated by a fiber ring microwave photonic filter (FR-MPF). By mapping optical wavelength shifts to significant microwave frequency shifts, the system breaks the resolution limit of traditional spectral analysis, achieving a theoretical resolution of 0.26 × 10 ⁻² m ⁻¹ . Crucially, we optimized the system for high-speed interrogation, demonstrating a resolution of 0.65 × 10 ⁻² m ⁻¹ even at measurement speeds exceeding 100 Hz. The system’s multi-directional sensing capability and environmental stability were also experimentally validated, with average relative errors for curvature and direction of 1.09% and 1.91%, respectively. This work integrates high precision and dynamic capability into a compact sensing system, offering significant potential for real-time medical navigation and large-scale structural monitoring.

Fiber-optic acoustic sensors are vital in structural health monitoring, photoacoustic spectroscopy, and biomedical applications. Achieving simultaneous measurement of quasi-static temperature and dynamic acoustic vibration with effective decoupling remains challenging. In this work, we propose an integrated fiber-optic sensor probe with dual Fabry-Perot (FP) cavities: a silicon-based cavity for temperature sensing and a diaphragm-based air cavity for acoustic pressure sensing. A novel hybrid interferometry is introduced, employing programmable modulation of a tunable Modulated grating Y-branch (MG-Y) laser to combine coarse wavelength scanning (20 kHz) at 1530 nm for acoustic measurement and dense wavelength scanning (10 Hz) at 1550 nm for temperature measurement, using a single demodulation system. Experimental results demonstrate a linear temperature response ( R ² =0.996) from 27°C to 56°C. Crucially, by leveraging real-time temperature compensation, the maximum acoustic pressure measurement error was reduced from 28.4% to 4.6%, validating the system’s efficacy in high-precision, temperature-compensated acoustic metrology.

Microwave photonic frequency-to-phase-slope mapping (FTPSM) is susceptible to frequency measurement errors, primarily induced by delay-rate uncertainty and fluctuations. To address this limitation, we propose a novel frequency-to-phase-slope-ratio mapping (FTPSRM) method that incorporates a reference signal sharing the same time-varying delay path as the signal under test (SUT). By leveraging the phase-slope ratio, this approach eliminates the need for delay-rate calibration while mitigating the impact of delay-rate fluctuations on measurement accuracy. The FTPSRM is implemented in a microwave photonic system utilizing polarization-division multiplexing of time-varying delay. The experimental results demonstrate that the FTPSRM reduces the maximum frequency error from 602 to 234 MHz and the standard deviation from 743 to 301 MHz across the 1–10-GHz range, significantly improving both accuracy and stability under delay fluctuations.

Fiber optic sensor is capable of capturing tiny displacements and deformations, making them ideal for real-time monitoring of human limb movement, so it has great potential in human-computer interaction based on gesture recognition. In this paper, we propose a wearable, flexible polymer-embedded fiber Bragg grating sensor array to detect subtle changes in the brachialis and palmaris longus muscles during muscle contractions in order to recognize different dynamic gestures. Nine kinds of common dynamic gestures are considered. Four fiber optic sensors are firmly attached to the surface of the target muscle group with adhesive tape to detect tiny changes in the muscle. Changes in different muscle contractions cause some differences in the optical signals output by these fiber optic sensors. These optical signals are collected and segmented into signal segments corresponding to each dynamic gesture. We propose a method to convert multi-dimensional optical signal segments into 2D image. These images are fed to the Convolutional Neural Network (CNN) to classify different dynamic gestures. Traditional machine learning methods based on image feature extraction are compared. Experimental results show that CNN has a stronger ability to extract subtle differences in muscle changes from 2D image, which helps to classify different dynamic gestures, with an average recognition rate of 98%. This paper provides a new and valuable reference for hand motion monitoring.