🔢 mathematics

Shared Sky, Shared Spectrum: Coordinated Satellite-5G Networks for Low-Altitude Economy

This paper proposes a cost-effective framework for partially-integrated satellite-5G networks that utilizes coarse synchronization and link-feature-aided clustering to enable coordinated spectrum sharing for low-altitude aircraft, achieving significant performance gains with low overhead.

Original authors: Yanmin Wang, Wei Feng, Yunfei Chen, Baoquan Ren, Qingqing Wu, Cheng-Xiang Wang

Published 2026-03-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yanmin Wang, Wei Feng, Yunfei Chen, Baoquan Ren, Qingqing Wu, Cheng-Xiang Wang

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ 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

The Big Picture: A Traffic Jam in the Sky

Imagine the "Low-Altitude Economy" as a brand new, bustling city in the sky. It's filled with drones, delivery robots, and small planes (called Low-Altitude Aircrafts or LAAs) zipping around to deliver packages, film movies, and help in emergencies.

To keep this sky-city running, these aircraft need a constant, reliable internet connection. But there's a problem:

Ground 5G Networks are like a dense forest of cell towers. They work great for people on the ground, but their signals get blocked by buildings and don't reach high up well. Plus, if you try to aim them up, they cause a chaotic mess of interference.
Satellites are like giant lighthouses in space. They can see everything, but their signals are slow (high latency) and the data speed is often too slow for fast-moving drones.

The Solution? Combine them! But trying to perfectly sync a satellite (moving fast in space) with a ground tower (stationary) is like trying to dance a perfect waltz with a partner who is on a moving train while you are on a treadmill. It's too hard, too expensive, and takes too long to set up.

The Paper's Idea: "Good Enough" Synchronization

Instead of demanding a perfect, millisecond-perfect dance, the authors propose a "Coarse Synchronization" approach.

Think of it like a neighborhood potluck.

The Old Way (Full Integration): Everyone agrees to bring a dish at exactly 12:00:00 PM. If you are 1 second late, the party is ruined. This requires everyone to have atomic clocks.
The New Way (Coarse Sync): Everyone agrees to arrive "around lunchtime." As long as you show up within a 10-minute window, you can still share food and enjoy the party. You don't need perfect timing; you just need a general agreement on the schedule.

In this paper, the "10-minute window" is a time scale (let's call it T) that is much longer than the split-second timing usually required for 5G. This makes the system much easier and cheaper to build.

The Secret Sauce: "Interference Similarity" Clustering

Even with a relaxed schedule, if a satellite and a ground tower try to talk at the same time on the same frequency, they will shout over each other (interference).

The authors use a clever trick called Link-Feature-Aided Clustering. Imagine you are organizing a massive party with 100 guests (the aircraft and users). You want to seat them so people who talk loudly don't sit next to people who need to whisper.

The Problem: You can't calculate the perfect seating chart for 100 people instantly; it's a math nightmare (an "NP-hard" problem).
The Solution: Instead of looking at every individual, you group them by how they affect others.
- Group A: Drones that fly over the ocean (they don't bother the city towers).
- Group B: Drones that fly over the city center (they might bother the towers).
- Group C: Ground users in the suburbs.

The paper suggests using Location Awareness (since drones know exactly where they are) to create a "Radio Map." This map tells the system: "If Drone #1 is here, it won't bother Tower #5, but it will bother Tower #6."

By grouping drones that cause similar interference patterns together, the system can schedule them efficiently without needing to solve the impossible math problem for every single second.

How It Works in Practice

The Map: The system uses a pre-made map of the sky and ground to know where everyone is and how their signals travel.
The Grouping: It groups the drones and ground users into "clusters" based on who interferes with whom.
The Schedule: It assigns these clusters to different time slots and frequencies. Because the groups are well-separated, they can share the same "airwaves" without shouting over each other.
The Trade-off: The system adjusts how long these time slots last. If the network is busy, it might use shorter slots for better speed. If it's simple, it uses longer slots to save energy and computing power.

Why This Matters

It's Fast to Deploy: You don't need to wait years to build a perfect satellite-ground network. You can start using this "coarse" method right now.
It's Cheap: It doesn't require expensive, high-precision clocks or massive computing power.
It Works: The paper proves through simulations that this method gets almost as good a result as the "perfect" method, but with a fraction of the headache.

The Takeaway

This paper is about making the sky-city of drones possible today. Instead of trying to build a perfect, high-tech bridge between space and earth, they built a sturdy, practical footbridge that gets the job done efficiently, allowing our future delivery drones to fly safely and stay connected.

1. Problem Statement

The rapid growth of the Low-Altitude Economy (LAE), driven by Low-Altitude Aircraft (LAAs) such as drones and eVTOLs, demands ubiquitous and reliable communication coverage. However, existing infrastructure faces critical limitations:

Terrestrial 5G Networks: Originally designed for ground users, they suffer from coverage blind spots at low altitudes and complex interference issues when extended upward.
Satellite Systems: While offering wide-area coverage, they are limited by high latency and relatively low data rates, making them insufficient for standalone LAA operations.
Full Integration Challenges: A fully integrated satellite-terrestrial network is theoretically ideal but practically difficult due to massive differences in propagation delays, operational protocols, and the high overhead required for fine-grained time synchronization (millisecond-level) between satellite and terrestrial links.

Core Challenge: How to achieve efficient joint time-frequency spectrum sharing between satellites and 5G networks to serve LAAs without incurring the prohibitive complexity and overhead of fine synchronization.

2. Methodology

The paper proposes a Partially-Integrated Network Framework based on Coarse Synchronization and Link-Feature-Aided Clustering.

A. Coarse Satellite-Terrestrial Synchronization

Instead of requiring millisecond-level time-slot synchronization, the authors propose a coarse synchronization time scale ( $T$ ) (e.g., several seconds).

Mechanism: Satellite and terrestrial links operate with independent time-slot divisions but are aligned at the macro level of time scale $T$ .
Benefit: This avoids the need for complex timing advance mechanisms to compensate for rapidly varying satellite propagation delays (which range from 1.7ms to 4ms for Low Earth Orbit satellites).
CSI Utilization: The system relies on Large-Scale Channel State Information (CSI) (e.g., path loss, shadowing) derived from the known positions of LAAs and a pre-constructed Radio Map, rather than instantaneous small-scale fading.

B. Divide-and-Conquer Strategy: Link Clustering

To solve the NP-hard optimization problem of joint spectrum sharing, the authors employ a divide-and-conquer approach using Interference-Similarity-Based Link Clustering.

Feature Vector Design: Links are characterized by feature vectors representing the impact of their interference on the network's Quality of Service (QoS).
- For an LAA-satellite link, the feature vector quantifies how its transmission degrades the performance of various Terrestrial Base Station (TBS)-Terrestrial User (TU) links.
- This incorporates transmission parameters (power, beamforming) and large-scale CSI.
Clustering Algorithm: Using algorithms like K-means, links with similar interference profiles are grouped into clusters.
- Links in the same cluster are scheduled on the same carrier.
- This preserves "opportunistic spatial separations" where interference is naturally low.
Scheduling:
- Satellite Links: Clustered first based on interference similarity.
- Terrestrial Links: Scheduled opportunistically within the remaining resources using assignment subproblems (which have polynomial complexity).

C. Adaptive Cooperation Time Scale

The framework includes an adaptive mechanism to select the time scale $T$ . A larger $T$ reduces overhead and complexity but may slightly reduce performance; a smaller $T$ improves performance but increases signaling load. The system balances this trade-off based on specific network scenarios.

3. Key Contributions

Novel Framework: Proposes a joint time-frequency spectrum sharing framework for coordinated satellite-5G networks that operates under coarse synchronization, making it immediately deployable compared to full integration.
Position-Aware Clustering: Leverages the inherent position-awareness of LAAs to construct a Radio Map and derive large-scale CSI, enabling efficient link clustering without real-time instantaneous channel feedback.
Low-Complexity Optimization: Transforms a complex, NP-hard mixed-integer programming problem into a manageable process via link-feature-aided clustering, effectively decoupling the scheduling of satellite and terrestrial links.
Adaptive Trade-off: Introduces an adaptive time scale $T$ that allows network operators to dynamically balance performance gains against implementation overhead.

4. Results

The authors validated the framework through a case study involving:

Scenario: One satellite (500km altitude), 28 TBSs, 96 LAAs, and 672 TUs.
Comparison: The proposed framework was compared against:
- FineSync: A benchmark requiring millisecond-level fine synchronization.
- RandScheme: A random scheduling scheme with no power control.
Performance Metrics: Average sum rate improvement over a non-sharing baseline.
Findings:
- The proposed framework achieved significant performance gains in spectrum efficiency compared to non-sharing scenarios.
- The performance was comparable to the "FineSync" benchmark, despite avoiding the high overhead of fine synchronization.
- It significantly outperformed the "RandScheme," demonstrating the efficacy of interference-aware clustering and power control.

5. Significance

Economic Viability: By avoiding the need for expensive, high-complexity synchronization infrastructure, this approach offers an economically efficient solution for the immediate needs of the Low-Altitude Economy.
Scalability: The use of large-scale CSI and position-based clustering makes the system scalable for dynamic LAA traffic without overwhelming signaling overhead.
Future-Proofing: The framework serves as a pragmatic stepping stone, enabling current deployment while laying the technical foundation for future fully integrated satellite-terrestrial networks.
Spectrum Efficiency: It effectively addresses spectrum scarcity by enabling dynamic, opportunistic sharing between satellite and terrestrial domains in both time and frequency.