Reliable Communication in a Computer Vision Enhanced Free-Space-Optical Mobile Network

Ensuring robust and efficient link discovery and maintenance is critical for optimal performance in Free-space-optical (FSO) communication networks. This paper presents a novel approach leveraging computer vision (CV) techniques to enhance an FSO networking system. Each node in the network utilizes a CV subsystem that includes multiple cameras to dynamically identify and track neighboring nodes, facilitating rapid line-of-sight (LOS) discovery and robust link maintenance. Our system combines CV with an FSO communication (FSOC) array consisting of multiple transceivers. The FSOC array is equipped with electronic beam steering, enabling multiple simultaneous optical wireless communication links. The proposed solution uses object detection models to analyze visual data from network environments, enabling real-time processing of environmental changes and automatic reconfiguration of network links. We present a preliminary prototype and evaluate its performance through testbed experiments. Results demonstrate that our CV-enhanced system reliably supports communication among multiple mobile nodes, and enables faster and error-free data transfer with reduced overhead over a legacy FSO system.