AI-Assisted Next-Gen Outdoor Optical Networks: Camera Sensing for Monitoring and User Localization
Meysam Ghanbari, Mohammad Taghi Dabiri, Rula Ammuri · 2025
Read AI-Assisted Next-Gen Outdoor Optical Networks: Camera Sensing for Monitoring and User Localization by Meysam Ghanbari, Mohammad Taghi Dabiri, Rula Ammuri (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-18087.txt.
Introduction
Meysam Ghanbari leads a five author team adding a watch layer to outdoor light based networks. Rooftop optical links beam data through open air at high speed, but open beams can be tapped or abused. Their system locates users by reading the light pattern that lands on a grid of small detectors.
A physics based model does the first pass, and a light learning step cleans up device quirks. In matched tests the hybrid beat a plain deep learning method on accuracy while training faster and cheaper. Adding more detector cells helped up to a grid near 60 by 60 or 80 by 80. Live position checks against network logs could flag odd use, like a hidden listener.
Catalog record
| CREATOR | Meysam Ghanbari, Mohammad Taghi Dabiri, Rula Ammuri |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-18087
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
AI-Assisted Next-Gen Outdoor Optical Networks: Camera Sensing for Monitoring and User Localization Thanks: M. Ghanbari and M.T. Dabiri are with the Qatar Center for Quantum Computing, College of Science and Engineering, Hamad Bin Khalifa University, Doha, Qatar. email: ([email protected]; [email protected]). Thanks: Rula Ammuri is with Professionals for Smart Technology (PST), Amman, Jordan (email: [email protected]). Thanks: Mazen Hasna is with the Department of Electrical Engineering, Qatar University, Doha, Qatar (email: [email protected]). Thanks: Khalid A. Qaraqe is with the College of Science and Engineering, Hamad Bin Khalifa University, Doha, Qatar, and also with the Department of Electrical Engineering, Texas A&M University at Qatar, Doha, Qatar (email: [email protected]). Thanks: This publication was made possible by NPRP14C-0909-210008 from the Qatar Research, Development and Innovation (QRDI) Fund (a member of Qatar Foundation), Texas A&M University at Qatar, and Hamad Bin Khalifa University, which supported this publication. Meysam Ghanbari, Mohammad Taghi Dabiri, Rula Ammuri Abstract We consider outdoor optical access points (OAPs) , which, enabled by recent advances in metasurface technology, have attracted growing interest. While OAPs promise high data rates and strong physical-layer security, practical deployments still expose vulnerabilities and misuse patterns that necessitate a dedicated monitoring layer—the focus of this work. We therefore propose a user positioning and monitoring system that infers locations from spatial intensity measurements on a photodetector (PD) array. Specifically, our hybrid approach couples an optics-informed forward model and sparse, model-based inversion with a lightweight data-driven calibration stage, yielding high accuracy at low computational cost. This design preserves the interpretability and stability of model-based reconstruction while leveraging learning to absorb residual nonidealities and device-specific distortions. Under identical hardware and training conditions (both with 5 × 10 5 5\times 10^{5} samples), the hybrid method attains consistently lower mean-squared error than a generic deep-learning baseline while using substantially less training time and compute. Accuracy improves with array resolution and saturates around 60 × 60 60\times 60 – 80 × 80 80\times 80 , indicating a favorable accuracy–complexity trade-off for real-time deployment. The resulting position estimates can be cross-checked with real-time network logs to enable continuous monitoring, anomaly detection (e.g., potential eavesdropping), and access control in outdoor optical access networks. I Introduction Free-space optical (FSO) communication has emerged as a promising technology for next-generation wireless access owing to its vast unlicensed bandwidth, immunity to radio-frequency interference, and high link directionality [ 1 ] . In recent years, the use of intelligent reflecting surfaces (IRSs) and modulating retroreflectors (MRRs) has facilitated the realization of lightweight and power-efficient optical access points that can extend network coverage without complex alignment requirements [ 2 ] . However, these systems remain highly sensitive to physical and environmental disturbances. The bidirectional optical path magnifies angle-of-arrival (AoA) fluctuations, atmospheric turbulence, and geometric pointing errors; even milliradian-level offsets can disrupt the link [ 1 ] , [ 3 ] . Consistent with this sensitivity, [ 4 ] developed a comprehensive unmanned aerial vehicle (UAV)-to-ground MRR channel model incorporating tracking errors, UAV attitude fluctuations, turbulence, and range, showing that sub-milliradian pointing can interrupt the double-pass link. To improve robustness, [ 5 ] employed MRR arrays on small UAVs and derived closed-form outage and capacity expressions under realistic turbulence and vibration, and [ 6 ] introduced a network-coded MRR relay that demonstrated superior angular stability compared with amplify-and-forward (AF) and IRS relays. Parallel progress in the IRS domain has targeted FSO resilience and flexibility. In [ 7 ] , an IRS-assisted FSO link under random misalignment was modeled via a unified geometric–misalignment framework capturing incidence and reflection angles; [ 8 ] analyzed mixed FSO/active-IRS Multiple Input, Single Output (MISO)-Non-Orthogonal Multiple Access (NOMA) with imperfect channel state information (CSI), showing active IRS can mitigate turbulence; [ 9 ] optimized multilink IRS placement while accounting for turbulence and pointing errors; and [ 10 ] examined optical-IRS multiuser space-division multiple access (SDMA), quantifying performance impacts of atmospheric turbulence and interference optical power. Although FSO links are inherently secure given their narrow beams, this advantage diminishes in MRR and IRS-assisted architectures. Here, bidirectional/reflective paths widen the effective field of view (FoV), enabling in-FoV unauthorized users to access or infer channel information without disturbing the legitimate link. In [ 11 ] , this vulnerability was quantified for autonomous aerial vehicle (AAV)-based MRR links, deriving eavesdropping probability as a function of FoV and eavesdropper geometry, and revealing a trade-off: narrowing the FoV increases confidentiality but reduces availability under motion and turbulence. However, none of the existing approaches provides an in-situ optical situational-awareness layer for real-time localization of unauthorized interrogators at FSO access points. We close this gap with a unified, camera-inspired monitoring framework for IRS/MRR-based outdoor access points: a wide-FoV lens focuses residual interrogator energy onto a focal-plane photodetector array to form spatial intensity maps. A geometric–radiometric forward model solved via sparse, model-based inversion, complemented by a lightweight data-driven calibration, yields interpretable, stable position estimates under turbulence and nonidealities. Cross-checking these estimates with network logs enables continuous monitoring, anomaly/eavesdropper indication, and access control—providing the missing optical situational-awareness layer for next-generation FSO access networks. Fig. 1: Illustration of the proposed monitoring system: A wide-FoV lens maps incident Gaussian beams from both legitimate users and eavesdroppers to a focal-plane photodetector array, enabling joint detection and localization. II System Model II-A Motivation and Monitoring Framework In recent optical access networks based on IRSs and MRRs, accurate localization of legitimate users as well as potential eavesdroppers has become a critical requirement [ 12 , 11 ] . Although such systems offer unique advantages including robustness to UAV dynamics and simplified user terminals, they are inherently exposed to a security challenge: an unauthorized interrogator located within the FoV of the IRS/MRR can illuminate the UAV and receive the modulated reflection without interfering with the legitimate link. Each ground transmitter (legitimate or eavesdropper) communicates with the UAV access point via an IRS/MRR link; however, for monitoring purposes we exploit the residual portion of their interrogator beams that naturally extends beyond the IRS/MRR aperture. At the UAV, this residual optical field impinges on the monitoring lens and can be modeled as a narrowband Gaussian beam with a planar wavefront at the lens surface due to its large spot size relative to the lens radius. To capture and analyze these residual beams, we propose a camera-inspired sensing module based on a focal-plane photodetector (PD) array with dimension N_{\mathrm{ar}}=N_{x}\times N_{y} . The array is co-located with the UAV-mounted IRS/MRR access point and placed at the focal distance of a circular lens of radius r_{a} . This monitoring plane enables two primary functions: i ) accurate estimation of legitimate users’ positions through AoA and range estimation of their interrogator beams, and ii ) detection and localization of eavesdroppers attempting to exploit the link [ 11 ] . As depicted in Fig. 1 , the incident Gaussian beams are mapped to distinct focal-plane spots over the PD array, enabling joint monitoring and classification of all active transmitters within the coverage area. II-B Received-Power Model for User/Eavesdropper Residual Beams We deploy a camera-like monitoring module (a lens co-located with the IRS/MRR access point) that focuses the portion of the optical field impinging on its lens onto a focal-plane array (Fig. 1 ). Different user locations and ranges induce distinct AoAs and received