Abstract
The emerging unmanned aerial vehicles (UAVs) have been widely exploited for sensing purposes due to the larger service coverage compared with the conventional fixed sensor nodes. However, due to the limited computation capability of UAVs, real-time sensory data needs to be transmitted to the BS/server for real-time data processing. In this regard, the cellular networks are necessary to support the data transmission for UAVs, which is called the Internet of UAVs. Very recently, 3GPP has approved a study item on the enhanced support to seamlessly integrate UAVs into future cellular networks.
Introduction
The "Internet of UAVs" represents a paradigm shift where unmanned aerial vehicles are no longer isolated tools but integrated nodes within a global cellular infrastructure. By leveraging existing and future cellular networks, researchers aim to overcome the inherent battery and computational limitations of drones, enabling ultra-reliable, real-time sensing and communication. This editorial introduces a curated collection of research focused on the technical hurdles and breakthroughs required to make this integration a reality, particularly as 3GPP moves toward standardized support for aerial nodes.
The Challenges of Aerial Connectivity
While UAVs offer exceptional mobility for wide-area sensing, they are constrained by limited on-board battery life and processing power. To perform complex real-time tasks, they must offload data to ground-based servers. However, traditional terrestrial cellular networks are optimized for 2D ground-level connectivity and are not inherently designed for the sky.
Aerial nodes introduce unique complexities that traditional protocols struggle to handle:
- 3D Mobility: High-speed movement in three-dimensional space creates dynamic network topologies.
- Link Instability: Aerial links are often "weakly connected" due to interference and varying altitudes.
- Operational Constraints: Navigation must account for no-fly zones and energy-efficient path planning.
This special issue addresses these hurdles through innovations in 3D placement, trajectory planning, and resource allocation to make UAVs a permanent fixture of our communication infrastructure.
Methodological Frameworks
The research presented utilizes a diverse toolbox of AI and optimization techniques to manage the complexity of UAV networks:
- Deep Reinforcement Learning (DRL): Applied to multi-UAV trajectory planning and autonomous flight control to minimize data loss and manage buffer overflows in dynamic environments.
- Federated Learning: Specifically used for air quality monitoring, allowing UAV swarms to train models on haze images locally, thereby reducing the need to transmit massive raw datasets.
- Optimization Theory: Techniques such as Dantzig–Wolfe decomposition and POMDP (Partially Observable Markov Decision Process) models solve joint optimization problems involving power, path loss, and data schedules.
- Geometric and Signal Processing: Development of 3D nonstationary geometry-based stochastic models for 6G channels and location-aware beamforming algorithms that remain robust despite GPS inaccuracies.
Key Insights and Strategic Results
The integration of UAVs into cellular networks yields several transformative insights:
- UAVs as Infrastructure: Drones are evolving from simple "users" to "Aerial Base Stations" that can provide emergency coverage or enhance capacity in high-demand zones.
- 6G and Satellite Synergy: The integration of Low-Earth Orbit (LEO) satellites with UAVs creates a multi-layered "Internet of Remote Things," ensuring global connectivity even in isolated regions.
- Robustness over Precision: Given the mobility of UAVs, research is prioritizing "location-robust" algorithms that maintain high-speed links (e.g., 60-GHz mmWave) even when the drone's exact position is slightly off.
Quantitative Performance
The studies report significant gains across several metrics:
- Throughput: A study on 60-GHz beamforming achieved a 260% bit-rate improvement over the 802.11ad standard by specifically accounting for aerial dynamics.
- Precision: Location-based beamforming designs reduced Direction-of-Arrival (DoA) estimation errors by over 90%, ensuring stable high-speed connections.
- Efficiency: New hierarchical machine learning frameworks (HMTD) successfully minimized the weighted-sum cost of UAV tasks while maintaining strict inference error constraints.
Security and Future Directions
As UAVs become integral network nodes, they also become targets for cyber threats. Research into DNS-based malware detection highlights the need for a specialized security architecture for the Internet of UAVs to protect against Advanced Persistent Threats (APT).
Looking forward, several challenges remain:
- Ultra-Reliability: High mobility continues to challenge the requirements of ultra-reliable low-latency communication (URLLC).
- Energy Bottlenecks: Battery life remains the primary constraint for long-term autonomous deployment.
- Scalability: Transitioning from small swarms to city-scale deployments will require managing the "curse of dimensionality" in AI models.
Ultimately, these foundational works pave the way for 6G, where UAVs will be seamlessly integrated into the architectural fabric of global communications.
Summary updated