Low-Altitude Economy: Framework & Applications

• Low-Altitude Economy (LAE) is a paradigm using intelligent aerial agents for logistics, sensing, and urban mobility in sub-1km airspace.
• It employs hierarchical architectures that integrate ground, aerial, and cloud layers to support low-latency, mission-critical communications and data processing.
• Economic growth in LAE is driven by high sortie volumes and advanced ISAC/ICC techniques, achieving measurable improvements in latency, energy efficiency, and service continuity.

Low-Altitude Economy (LAE) represents an economic and technological paradigm built upon dense, intelligent operations in sub-1-kilometer airspace, leveraging fleets of manned and unmanned aerial agents such as UAVs and eVTOLs for tasks including logistics, sensing, mobility, emergency response, and public safety. The LAE is characterized by tightly coupled three-dimensional airspace utilization, hierarchical resource fabrics spanning ground, aerial, and cloud/satellite layers, and latency-/mission-critical service requirements. Economic projections indicate explosive growth, with sectors in China expected to exceed one trillion RMB by 2026 and sustained global investments from industrial actors and governments (Lyu et al., 28 May 2025, He et al., 7 May 2025).

1. Definition, Scope, and Economic Drivers

LAE is defined by commercial and social activities primarily executed in the sub-500–1000 m altitude regime using fleets of intelligent aerial agents (IAAs). The paradigm encompasses last-mile delivery, urban air mobility (UAM), precision agriculture, public-safety missions, and remote area connectivity. Key features include three-dimensional, dynamic airspaces; the integration of sensing, communication, and computation (SCC); and multi-layered infrastructure from resource-constrained ground nodes to edge platforms and satellite/cloud backends (Lyu et al., 28 May 2025, He et al., 7 May 2025).

Table: Typical LAE Vehicle Classes and Features

Vehicle Type	Altitude Range	Roles
UAV (Rotary)	0–500 m	Delivery, mapping, base station
UAV (Fixed)	<1000 m	Surveillance, offhaul, survey
eVTOL	50–800 m	Passenger transit, air taxi

Economic motivation arises from dense urban populations’ logistics needs, bypassing ground congestion, expanding agricultural and environmental monitoring, and serving disaster-affected or remote zones. High sortie volumes (projected 13B UAV sorties by 2030) drive requirements for ultra-reliable, low-latency communications, multi-service links, and robust positioning (He et al., 7 May 2025).

2. Hierarchical System Architectures for Deployment

LAE leverages hierarchical architectures for large AI model deployment (LAIMs), enabling scalable intelligence despite severe resource constraints. This comprises:

• Ground layer: Low-power nodes (sensors, user equipment, gateways) performing early feature extraction with ultra-light LAIM fragments.
• Aerial layer: UAVs and IAAs running pruned and segmented LAIM modules for local inference and participating in flying mesh networks with decentralized resource sharing.
• Cloud/satellite layer: Datacenters and LEO satellites host full-scale LAIMs for global inference, training, and cross-layer model orchestration (Lyu et al., 28 May 2025, He et al., 7 May 2025).

Resource allocation exploits pruning-aware co-inference, balancing quality $Q(\rho)$, latency $D(\rho,f,P)$, and energy $E(\rho,f,P)$, with real-time model partitioning according to available CPU frequency, transmit power, and channel gains.

3. Enabling SCC Technologies and Co-Evolution Mechanisms

Critical technical pillars include:

• Integrated Sensing and Communication (ISAC): Dual-purpose waveform design, joint channel estimation, and dynamic 3D beamforming enable simultaneous data transmission and environmental perception, supporting predictive multi-beam coverage and spatial spectrum access.
• Integrated Communication and Computation (ICC): Semantic communications transmit task-relevant embeddings, enabling over-the-air computation (AirComp) for rapid consensus in swarms, e.g., $y = \sum_{i=1}^N h_i\,z_i + n$.
• Integrated Sensing–Computation–Communication (ISCC): Task-oriented joint optimization across SCC domains, supporting split learning, mixture-of-expert routing, and parameter-efficient LAIM fine-tuning under continually adaptive real-world feedback loops (Lyu et al., 28 May 2025).

Table: Core SCC Integration Techniques

Mode	Key Enabler	Example Function
ISAC	Duplex waveform	Joint sensing/beamforming
ICC	Semantic comms	Distributed feature fusion
ISCC	Split learning	Co-inference, MoE routing

4. Representative Application Scenarios

LAE spans three major application classes:

• Communication-centric: UAV base stations for pop-up coverage, self-organized mesh, and LAIM-guided spectrum access; ISAC-enhanced beamforming for ad hoc networks.
• Sensing-centric: Multi-angle mapping for agriculture, urban air pollution detection, and dynamic obstacle avoidance in delivery or inspection missions.
• Computation-centric: Edge inference using split LAIMs for anomaly detection in grids, large-scale 3D mapping, traffic forecasting, and cloud-coordinated airspace management (Lyu et al., 28 May 2025, He et al., 7 May 2025).

Satellite-assisted architectures enable ubiquitous coverage and robust control, with distributed MIMO arrays at satellites and LAV swarms, two-timescale beam steering, and coherent combining for massive link gain and extended service continuity (He et al., 7 May 2025).

5. Joint SCC Optimization and Task Execution Frameworks

Task-oriented LAE execution pipelines consist of:

• Offline initialization: Digital twin generation, coarse placement optimization, and schedule creation for nodes and resources.
• Multi-modal sensing: Adaptive, task-driven sampling, prioritizing region-of-interest or event-triggered data collection.
• Information transmission: Horizontal consensus via AirComp and dynamic protocol selection for vertical transfers (bit-level versus semantic-level).
• Split/co-inference for tasks: Decoupled LAIM deployment, e.g., semantic heads on terrestrial gateways, transformer encoders on UAVs, global decoders on cloud.
• Closed-loop control adaptation: Real-time feedback for position adjustment, resource reallocation, and gradient/RL-based model updates (Lyu et al., 28 May 2025).

Empirical studies demonstrate that LAIM-guided UAV deployments achieve up to 20% higher sum-rates in realistic multi-beam environments, with resource-aware co-inference reducing latency by 30% and energy consumption by 25% at equal task quality compared to cloud-only approaches.

6. Open Challenges and Future Research Directions

Key open areas involve:

• Integration of Airspace Knowledge Graphs and LAIMs: Ensuring regulation compliance and risk-aware navigation through real-time knowledge grounding.
• Multi-agent gaming and coordination: Game-theoretic frameworks for strategic interaction under partial observability and privacy constraints.
• Sustainable deployment: Satellite-based WPT, design of energy-aware models, and ultra-robust UAV power management for persistent operations.
• Ecosystem standardization: Unified SCC APIs, cross-vendor interoperability, and collaborative data-sharing platforms.
• Security and resilience: Layered anti-jamming designs, combined physical/semantic control, and alignment with 6G standardization, including delay-tolerant non-terrestrial protocols and interoperable service meshes (Lyu et al., 28 May 2025).

The continual co-evolution mechanism—LAIMs adapt to dynamic LAE environments while LAE leverages evolving intelligence from LAIMs—drives both operational efficiency and robustness, shaping next-generation air–ground digital economies.

7. Real-World Implementations and Performance Highlights

Field prototypes have validated the LAE with:

• MmWave and sub-6 GHz ISAC BS networks for 1–2 Gbps links, sub-decimeter localization, and multi-domain spectrum monitoring (Luo et al., 10 Jun 2025).
• Multi-band ISAC fusion improves trajectory tracking in NLOS urban conditions.
• RAS-driven LAE boosts coverage probability from 0.65 (fixed) to 0.92 (RAS), with 7 dB power gain and substantial spatial clutter-to-noise ratio enhancement (Li et al., 1 Nov 2025).
• Scalable UAV-based logistical frameworks combine combinatorial LKH routing and DRL-based trajectory planning, reducing mission times by nearly 50% compared to baselines (Zhai et al., 10 Nov 2025).

These implementations demonstrate the feasibility of large-scale, intelligent, and resilient LAE deployments across diverse economic and technical environments.

---

Key References:

• "Empowering Intelligent Low-altitude Economy with Large AI Model Deployment" (Lyu et al., 28 May 2025)
• "Satellite-Assisted Low-Altitude Economy Networking: Concepts, Applications, and Opportunities" (He et al., 7 May 2025)
• "Rotatable Antenna System Empowered Low-Altitude Economy" (Li et al., 1 Nov 2025)
• "A Synergy of Computing Power Networks and Low-Altitude Economy Intelligent Communications" (Sun et al., 28 Sep 2025)