Sensing the living airspace: city-wide bird flight patterns revealed by communication networks

Jiachen Wei, Zhiyong Feng, Dingyou Ma, Qixun Zhang, Kan Yu, Ningyan Guo, Zheng Jiang, Jie Wang, Fan Liu, Zongqi Mo, Ping ZhangResearch Square 2025 · DOI

Abstract

Monitoring low-altitude bird movement is crucial for ecological research and aviation safety, yet a critical data gap exists due to the limitations of conventional radar and sparse sensor coverage in urban environments. Here, we demonstrate that existing cellular communication networks can be repurposed as a dense, city-scale biosensor network to fill this gap. We deploy a sub-6 GHz integrated sensing and communication (ISAC) system using two base stations (BSs), featuring a novel frame design that reserves only 1.4% of resources for sensing, thus maintaining uninterrupted communication services. To extract the weak micro-Doppler signal of birds from heavy urban interference, we develop a two-stage signal processing pipeline that recovers their characteristic micro-Doppler signatures. Using this system, we collect and analyze a large-scale dataset of 110 GB from hundreds of free-flying pigeons. Analysis of these signals first reveals significant variability in the time-frequency domain across individuals, a direct result of differences in their body size, flight posture, and angle relative to the BS. Despite this individual-level diversity, we extract a unifying biomechanical signature that represents a population-level commonality, providing a new layer of ecological information: a stable and consistent wingbeat frequency distribution for pigeons, with a mean of 6.42 Hz and a variance of 1.55 Hz². We show that this signature serves as a robust feature to distinguish birds from other aerial targets, such as unmanned aerial vehicles (UAVs), and we integrate it into a multi-station framework to map continuous, city-scale flight trajectories. Our work provides a practical pathway for transforming ubiquitous cellular infrastructure into a global tool for large-scale ecological monitoring and the mitigation of human-wildlife conflicts.

Introduction

Urban environments present a significant "blind spot" for monitoring low-altitude bird movements. While such data is vital for aviation safety, pathogen surveillance, and urban ecology, current methods are limited. Dedicated bird radars are expensive and localized, meteorological radars lack fine-grained ground-level resolution, and GPS tagging is unscalable for large populations. This research addresses the gap by repurposing ubiquitous cellular infrastructure—specifically base stations—into a distributed radar network capable of "sensing the living airspace."

Methodology: Integrated Sensing and Communication (ISAC)

The core of this approach is the deployment of a sub-6 GHz ISAC system. ISAC allows 5G-like networks to perform radio sensing (radar) simultaneously with data transmission. To minimize the impact on user services, the team developed a communication-centric frame design. A 5ms frame was engineered to dedicate only 1.4% of its resources to sensing, ensuring that high-bandwidth activities like video streaming remain uninterrupted while the base station tracks biological targets.

Overview of the ISAC system for bird sensing and communication in an urban environment.

Extracting weak biological signals in noisy urban environments required advanced signal processing. The researchers implemented a Null Space Pursuit (NSP) algorithm to isolate the faint "micro-Doppler" shifts caused by flapping wings from the heavy clutter of buildings and vehicles. Furthermore, a dual-gating Kalman filter was integrated to track both the physical position and the specific wingbeat frequency of the birds, preventing confusion with other aerial objects like drones.

Results and Discussion

The system was validated using a 110 GB dataset collected from hundreds of free-flying pigeons in Nanjing, China.

The Biomechanical "Biosignature"

A key discovery was the identification of a stable population-level biomechanical constant. Despite individual variations in flight posture and size, pigeons exhibited a consistent mean wingbeat frequency of 6.42 Hz. This "biosignature" allows the system to reliably identify species.

Discriminating Birds from UAVs

The system effectively distinguishes birds from Unmanned Aerial Vehicles (UAVs) by exploiting physical differences in their motion. UAV rotors spin at high speeds (44–91 Hz), which are "undersampled" by the 2.5 ms sensing interval of the ISAC model, causing their signal to appear as noise. In contrast, slower bird wingbeats remain clear and periodic.

Bird tracking results in an urban environment, showing real-world detection and mapped trajectories.

City-Scale Tracking Performance

The multi-station framework demonstrated successful "handover" capabilities, where a bird's trajectory is continuously mapped as it moves from the coverage area of one base station to another. The system achieved a detection range exceeding 160 meters with minimal communication overhead, proving the feasibility of using existing infrastructure for large-scale ecological monitoring.

Conclusion and Future Work

This work demonstrates a practical pathway for transforming cellular networks into a global scientific instrument. While currently focused on pigeons, future efforts will aim to build a library of biosignatures for various urban species and improve signal extraction in extremely dense urban "canyons." As 5G and 6G networks continue to densify, this technology offers a scalable, cost-effective solution for mitigating human-wildlife conflicts and advancing urban ecology.

Summary updated

Abstract

<title>Abstract</title> Monitoring low-altitude bird movement is crucial for ecological research and aviation safety, yet a critical data gap exists due to the limitations of conventional radar and sparse sensor coverage in urban environments. Here, we demonstrate that existing cellular communication networks can be repurposed as a dense, city-scale biosensor network to fill this gap. We deploy a sub-6 GHz integrated sensing and communication (ISAC) system using two base stations (BSs), featuring a novel frame design that reserves only 1.4% of resources for sensing, thus maintaining uninterrupted communication services. To extract the weak micro-Doppler signal of birds from heavy urban interference, we develop a two-stage signal processing pipeline that recovers their characteristic micro-Doppler signatures. Using this system, we collect and analyze a large-scale dataset of \(\:110\hspace{0.25em}GB\) from hundreds of free-flying pigeons. Analysis of these signals first reveals significant variability in the time-frequency domain across individuals, a direct result of differences in their body size, flight posture, and angle relative to the BS. Despite this individual-level diversity, we extract a unifying biomechanical signature that represents a population-level commonality, providing a new layer of ecological information: a stable and consistent wingbeat frequency distribution for pigeons, with a mean of \(\:6.42\hspace{0.25em}Hz\) and a variance of \(\:1.55\hspace{0.25em}H{z}^{2}\). We show that this signature serves as a robust feature to distinguish birds from other aerial targets, such as unmanned aerial vehicles (UAVs), and we integrate it into a multi-station framework to map continuous, city-scale flight trajectories. Our work provides a practical pathway for transforming ubiquitous cellular infrastructure into a global tool for large-scale ecological monitoring and the mitigation of human-wildlife conflicts.
BibTeX
@inproceedings{wei2025sensing,
  author = {Jiachen Wei and Zhiyong Feng and Dingyou Ma and Qixun Zhang and Kan Yu and Ningyan Guo and Zheng Jiang and Jie Wang and Fan Liu and Zongqi Mo and Ping Zhang},
  title = {{Sensing the living airspace: city-wide bird flight patterns revealed by communication networks}},
  booktitle = {Research Square},
  year = {2025},
  doi = {10.21203/rs.3.rs-7964150/v1},
}
Figure