🔢 mathematics

Modeling and Analysis of Sensing Assisted UAV Networks for Urban Vehicular Communications

This paper models and analyzes a sensing-assisted UAV network for urban vehicular communications using stochastic geometry to derive detection probability, coverage probability, and rate coverage, revealing that while increased altitude degrades sensing performance, rate coverage exhibits a non-monotonic trend.

Original authors: Kaushlendra Pandey, Nithin V Sabu, Abhishek K. Gupta

Published 2026-06-30

📖 4 min read🧠 Deep dive

CC0 1.0

Original authors: Kaushlendra Pandey, Nithin V Sabu, Abhishek K. Gupta

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a busy city where cars are constantly moving, and the streets are lined with tall buildings that block signals, much like how walls block sound in a house. To keep these cars connected and aware of their surroundings, the city deploys a fleet of drones (UAVs) hovering in the sky.

This paper is like a blueprint for figuring out exactly how high these drones should fly and how they should "look" to do two jobs at once: spotting the cars (sensing) and talking to them (communicating).

Here is the breakdown of their findings using simple analogies:

1. The Setup: A Grid of Roads and a Sky of Drones

The researchers modeled the city like a giant Manhattan grid (perfectly straight streets running North-South and East-West).

The Cars: They are scattered randomly along these streets, like beads on a string.
The Drones: They are floating at a fixed height, scattered randomly across the sky, like fireflies.
The Mission: Each drone spins around in a circle, scanning the ground in slices (like a pizza cutter). If it spots a car, it immediately locks onto it to send data. If it sees nothing, it stays silent to avoid causing noise (interference) for others.

2. The Two Big Hurdles: "Seeing" vs. "Hearing"

The paper explains that a drone's ability to work is limited by two different things, depending on how high it flies:

The "Building Wall" Problem (Low Altitude):
If the drone flies too low, the tall buildings act like a curtain, blocking its view. Even if the drone has a super-powerful radar, it can't "see" a car if a skyscraper is in the way. The drone is limited by the geometry of the city, not its power.
- Analogy: It's like trying to see a friend in a crowded hallway; no matter how loud you shout, the people in front of you block the view.
The "Distance" Problem (High Altitude):
If the drone flies very high to avoid the buildings, the signal has to travel a much longer distance. As it travels, it gets weaker (like a flashlight beam spreading out and dimming) and gets absorbed by the air.
- Analogy: It's like trying to hear a whisper from the top of a mountain; the distance is just too great, even if nothing is blocking the path.

3. The "Sweet Spot" Discovery

The most interesting finding is that there is a Goldilocks Zone for the drone's altitude.

Too Low: The buildings block the view.
Too High: The signal gets too weak to hear.
Just Right: There is a specific height where the drone sees enough cars without the signal dying out.

The paper also found that this "sweet spot" changes depending on the city. In a city with huge skyscrapers, the drone needs to fly higher to peek over the tops. In a city with smaller buildings, it can fly lower.

4. The Trade-Off: Seeing More vs. Talking Faster

There is a tricky balance between how many cars the drone sees and how fast it can talk to them.

If the drone uses a very narrow "flashlight" beam, it can see further and talk louder to a specific car.
However, if it sees too many cars at once, it has to split its attention (and bandwidth) among all of them.
The Result: Seeing more cars doesn't always mean better performance. Sometimes, seeing fewer cars allows the drone to talk to each one much faster. The paper shows that the overall "speed coverage" (Rate Coverage) goes up and down like a rollercoaster as the drone changes altitude, peaking at that "Goldilocks" height.

5. The "Silent" Drones

A key part of the system is efficiency. The drones only turn on their communication radios if they actually spot a car.

If a drone scans a whole sector and sees no cars, it stays silent.
This is crucial because if every drone talked all the time, the airwaves would be so noisy (interference) that no one could hear anything. By staying silent when there's no one to talk to, the drones keep the network clean.

Summary

The paper concludes that you cannot just pick a random height for these drones. You have to carefully tune the altitude and the width of the drone's "flashlight" beam together.

In dense cities with tall buildings, you need to fly higher to see over the walls.
In open areas, flying lower is better to keep the signal strong.
The best performance happens when the drone flies high enough to see over the buildings but low enough to keep the signal strong, creating a perfect balance between "seeing" the cars and "talking" to them.

Technical Summary: Modeling and Analysis of Sensing Assisted UAV Networks for Urban Vehicular Communications

Problem Statement
Emerging urban vehicular networks (VNs), particularly those supporting 6G applications like 3D navigation and cooperative driving, require seamless connectivity and situational awareness. However, existing ground infrastructure struggles to support these services due to high vehicular mobility and severe urban blockage, which degrade link reliability. While Unmanned Aerial Vehicles (UAVs) are proposed as flexible aerial base stations to establish Line-of-Sight (LoS) links, current models often treat UAVs purely as communication nodes. They fail to account for Integrated Sensing and Communication (ISAC), where sensing information (e.g., vehicle location and speed) is used to proactively align communication beams, thereby eliminating the overhead of conventional beam training. Furthermore, existing studies lack a systematic analytical treatment that jointly models UAV sensing reliability, altitude-dependent blockage, and the resulting communication performance in structured urban road networks.

Methodology
The paper proposes a stochastic geometry (SG) framework to model a sensing-assisted Low-Altitude Wireless Network (LAWN) in an urban environment. The key modeling components include:

Spatial Models: The urban road network is modeled as a Manhattan Poisson Line Process (MPLP) to capture the grid-like structure of cities. Vehicles are distributed on these roads as one-dimensional Poisson Point Processes (PPPs), forming a Cox process. UAVs are distributed at a fixed altitude $H$ as a two-dimensional homogeneous PPP.
Sensing Protocol: UAVs operate in a "sense-then-communicate" cycle. They perform a sequential angular scan using a directional radar beam of width $\gamma$ . Upon detecting a vehicle, the UAV immediately reuses the beam alignment to establish a communication link. UAVs that detect no vehicles remain idle to reduce interference.
Channel and Blockage Models: The system incorporates a Swerling I target model for radar cross-section fluctuations and a path-loss model including geometric spreading and molecular absorption (relevant for mmWave frequencies). Crucially, an elevation-dependent sigmoid LoS blockage model is used, which transitions from near-certain LoS at high elevation angles to high blockage at shallow angles.
Analytical Derivations: The authors derive the Detection Probability (DP), the maximum reliable sensing radius (limited jointly by Signal-to-Noise Ratio and blockage), and the distribution of the number of detected vehicles. They further derive the Laplace Transform (LT) of aggregate interference, accounting for the fact that only active UAVs (those that successfully sensed a vehicle) and only those whose beams align with the victim vehicle contribute to interference. Finally, they characterize the Coverage Probability (CP) and Rate Coverage (RC).

Key Contributions

Analytical Framework: The paper establishes a tractable SG framework combining MPLP for roads, MPLP-PPP for vehicles, and 2D PPP for UAVs, specifically tailored for ISAC-enabled urban vehicular networks.
Sensing Characterization: It derives the DP under directional sensing with altitude-dependent path loss and blockage. The work identifies two distinct operating regimes for the sensing radius: a blockage-dominant regime (where urban obstructions limit range regardless of power) and an SNR-dominant regime (where path loss and absorption limit range).
Activation and Load Analysis: The study characterizes the UAV activation probability and the distribution of the sensing load (number of detected vehicles per sector). It shows that in high-traffic environments, the activation probability saturates, indicating that connectivity is limited by sensing geometry (road density) rather than vehicular density.
Communication Performance: The paper derives the LT of aggregate interference considering sensing-driven activity and directional patterns. It provides closed-form expressions for CP and RC, quantifying the trade-off between sensing coverage and per-user throughput (as detecting more vehicles increases the load $L$ , reducing the bandwidth $B/L$ per vehicle).

Results and Insights
Numerical results, validated by Monte Carlo simulations, reveal several critical insights:

Altitude Trade-offs: Increasing UAV altitude improves LoS probability (reducing blockage) but increases path loss and slant distance. Consequently, the sensing radius and detection probability exhibit a non-monotonic trend with altitude, first increasing and then decreasing. An optimal altitude exists that balances these effects.
Coverage Degradation: Communication coverage probability (CP) generally degrades monotonically with increasing altitude due to signal attenuation, despite the reduction in aggregate interference from fewer active UAVs.
Rate Coverage (RC): The typical RC (accounting for both sensing success and communication rate) exhibits a non-monotonic trend with altitude, peaking at a moderate altitude before declining.
Beamwidth Impact: Narrower beamwidths improve both CP and RC. This is achieved by increasing the main-lobe antenna gain (strengthening the desired signal) and reducing the angular fraction of interfering UAVs (thinning the interference).
Environmental Impact: High-rise urban environments maintain better coverage at higher altitudes compared to suburban areas because building-induced blockage suppresses interfering links more effectively than it suppresses the desired LoS link.

Significance
The paper claims that its framework provides the first scalable analytical characterization of sensing-assisted UAV networks in structured urban environments. It highlights that UAV altitude and antenna beamwidth are tightly coupled design variables that cannot be optimized independently. The results offer system-level insights for designing antenna patterns, deployment densities, and altitude selection for future 6G-enabled urban vehicular networks, demonstrating that sensing reliability fundamentally dictates communication performance in these ISAC systems.