Fiber-optic sensing, especially distributed optical fiber vibration (DVS) sensing, is gaining importance in Internet of Things (IoT) applications, such as industrial safety monitoring and intrusion detection. Despite their wide application, existing post-processing methods that rely on deep learning models for event recognition in DVS systems face challenges with real-time processing of large sample data volumes, particularly in long-distance applications. To address this issue, we propose to use a four-layer convolutional neural network (CNN) as the student model with ResNet as the teacher model for knowledge distillation. Compared to the baseline CNN model, which achieves 83.41% accuracy on data from previously untrained environments, the distilled CNN improves accuracy significantly to 95.39%, demonstrating its superior generalizability and robustness. Additionally, we propose a novel hardware design based on field-programmable gate arrays to further accelerate model inference. This design replaces multiplication with binary shift operations and quantizes model weights, enabling high parallelism and low latency. Our implementation achieves an inference time of 0.083 ms for a spatial-temporal sample covering a 12.5 m fiber length and 0.256 s time frame. This performance enables real-time signal processing over approximately 38.55 km of fiber, about$2.14\times $the capability of an Nvidia GTX 4090 graphics processing unit. The proposed method greatly enhances the efficiency of vibration pattern recognition, promoting the use of DVS as a smart IoT system. The data and code are available athttps://github.com/HUST-IOF/Efficient-DVS.

In this paper, we report the first all-optical solution for multi-false-target range-velocity compound deception jamming based on frequency-comb-enabled photonic RF memory (PRFM). The operation frequency of current digital RF memory (DRFM) in electronic countermeasures is severely limited by low-speed digital devices. Since optical processing has a larger bandwidth and higher speed, it is valuable to develop advanced functions of PRFM. Our method skillfully applies a carrier-signal separation strategy to enhance the performance and extend the functions of PRFM. By bandwidth-optimized storage of the RF signal, the ASE noise accumulation and high-power optical self-oscillations can be effectively suppressed, contributing to a significant improvement of the signal SNR. On the other hand, by comb modulation of the optical carrier, the beaten signal contains multiple coherent false targets with different range-velocity deception information. In our experiment, the maximum storage time has exceeded$840~\mu $s with less than 10 dB SNR degradation, and the storage signal frequency has reached 16 GHz. Furthermore, more than 10 false targets have been obtained in the 10 dB bandwidth simultaneously, with the SNR consistently close to 40 dB. The outstanding storage and jamming performance allows our scheme to promote the deployment and upgrade of PRFM in electronic countermeasures.

Spatial resolution (SR) is one of the most important parameters of Brillouin optical time-domain analysis (BOTDA) sensors, which determines the minimum length that a perturbation event can be distinguished. In the field of Internet of Things (IoT), there is an urgent need for sensors with large-scale high-precision sensing capability for scenarios, such as intelligent monitoring of production lines and urban infrastructure. Conventionally, the SR is normally restricted to be longer than 1 m due to the$\sim 10$-ns acoustic lifetime limitation in silica optical fibers. For long-distance smart monitoring systems, the SR is generally on the order of several meters or even worse. However, it does not meet the needs of many applications. Therefore, there is an urgent need to achieve SR in the submeter magnitude. In this work, for the first time to the best of our knowledge, we propose a convolutional neural network (CNN) to process the data of conventional BOTDA sensors, which achieves unprecedented performance improvement that allows to directly retrieve submeter SR from the sensing system that use long pump pulses. By using the simulated Brillouin gain spectrums (BGSs) as the CNN input and the corresponding high SR Brillouin frequency shift (BFS) as the output target, the trained CNN is able to obtain an SR higher than the theoretical value determined by the pump pulse width. In the experiment, the CNN accurately retrieves 0.5-m hotspots from the measured BGS with pump pulses from 20 to 50 ns, and the acquired BFS is in great agreement with 45/40 ns differential pulse-width pair (DPP) measurement results. Compared with the DPP technique, the proposed CNN demonstrates a twofold improvement in BFS uncertainty with only half the measurement time. In addition, by changing the training data sets, the proposed CNN can obtain tunable high SR retrieval based on conventional BOTDA sensors that use long pulses without any requirement of hardware modifications. It is worth mentioning that the proposed method is also applicable to larger pulse widths to retrieve a submeter SR. The proposed data post-processing approach paves the way to enable novel high SR BOTDA sensors, which brings substantial improvement over the state-of-the-art techniques in terms of system complexity, measurement time, reliability, etc.