Key points are not available for this paper at this time.
With the rapid advancement of artificial intelligence (AI) technologies, the stability of optical interconnects in data centers has become increasingly important. Vibration sensing integrated in optical interconnect systems is conducive to identifying external disturbances in optical interconnects and achieving intelligent operation and maintenance. This paper proposes an easy-integration vibration-sensing scheme based on the state of polarization (SOP) of the fiber link. This scheme combines photonic technology with low-complexity digital signal processing (DSP) to detect link vibrations, ensuring full compatibility with intensity-modulation direct-detection (IM/DD) optical interconnect systems while minimizing additional complexity. Experiments show that our proposed scheme effectively detects SOP variations across a wide frequency range (0.5 Hz to 159 kHz). Based on the sensing system, a recognition scheme leveraging the Gramian angular field analysis and convolutional neural network (CNN) is proposed to recognize four types of vibration events simulated by a robotic arm, achieving a classification accuracy of 98%. Furthermore, experimental results confirm that the sensing system can detect SOP variations even under conditions of extremely low received optical power (ROP), where the communication system becomes inoperative. The proposed scheme enables robust event detection with minimal hardware overhead, which is suitable for real-world deployment in pluggable optical modules.
Yang et al. (Mon,) studied this question.