SAGIN: Space-Air-Ground Integrated Network

• SAGIN is a multi-layered network integrating satellites, HAPs, and terrestrial systems to provide global-scale, resilient, and low-latency communications.
• It leverages advanced technologies like network slicing, AI-driven orchestration, and quantum communications to optimize resource management across heterogeneous segments.
• The framework addresses challenges such as dynamic resource allocation, interference mitigation, and secure multi-domain orchestration for diverse applications.

Space-Air-Ground Integrated Network (SAGIN) is a multilayered, cross-domain networking paradigm that unifies spaceborne, airborne, and terrestrial communication segments into a single, flexible 3D network fabric. The goal is seamless, global‐scale, high-capacity, low-latency, and resilient connectivity for next-generation wireless systems, supporting both classical and quantum communications for diverse applications including ubiquitous mobile broadband, critical emergency services, intelligent transportation, industrial IoT, and quantum internet (Trinh et al., 2 Mar 2025, Chen et al., 2023).

1. Multilayer SAGIN Architecture

SAGIN comprises three tightly coupled segments, each with distinct technologies and propagation characteristics:

• Space Segment: Low-Earth-Orbit (LEO), Medium-Earth-Orbit (MEO), and Geostationary (GEO) satellites form multilayered satellite networks (MLSN). Functions include relaying, multicasting, quantum key distribution (QKD), and global backhaul. Modern mega-constellations (e.g., Starlink, Kuiper) employ microwave and optical inter-satellite links (ISLs) with rates exceeding 100 Gbps (Lan, 25 Oct 2025).
• Air Segment: High-Altitude Platforms (HAPs, ~20 km), manned aircraft (~9 km), and swarms of battery-powered Low-Altitude Platforms (LAPs, <5 km) serve as mobile relays, base stations, mobile edge computing (MEC) nodes, and on-demand coverage enhancers. HAPs bridge satellite-ground gaps, and UAV swarms facilitate multi-hop relays and airborne sensing (Chen et al., 2023).
• Ground Segment: Terrestrial networks (4G/5G/6G, MANET, Wi-Fi) are integrated with optical ground stations, base stations, edge/cloud servers, and mobile agents. Software-defined networking (SDN) and network function virtualization (NFV) unify management across radio access and core networks.

Interconnections: Satellite–satellite and satellite–air links primarily use free-space optical (FSO) or millimeter-wave (mmWave) communications, subject to stringent pointing, tracking, and atmospheric constraints. Air–air and air–ground links face intense dynamics, blockage, and fading effects. Non-line-of-sight (NLoS) connectivity is enabled via relay nodes and emerging reconfigurable intelligent surfaces (RIS/ORIS) (Trinh et al., 2 Mar 2025).

2. Enabling Technologies and Control Paradigms

SAGIN exploits several cross-domain technologies for unified connectivity and orchestration:

• Network Slicing, NFV, and SDN: Virtualized service functions are mapped flexibly onto nodes across all layers, realizing end-to-end slices tailored to varying QoS, bandwidth, and latency requirements. SFC (service function chaining) and associated resource allocation problems are solved via mixed-integer programming, matching-theoretic algorithms, and distributed reinforcement learning (Cao et al., 2023, Zhou et al., 2019).
• Quantum Communications: LEO satellites, HAPs, and ground nodes implement optical QKD and entanglement distribution using FSO links. Challenges such as LoS blockage, diffraction, pointing errors, and atmospheric loss are addressed with optical RIS/ORISs, which adaptively steer, refocus, and recover beams via metasurfaces (Trinh et al., 2 Mar 2025, Xu et al., 2022).
• AI-Native Orchestration: Deep reinforcement learning, generative diffusion models, and transformer-based architectures optimize resource allocation, routing, handover, and channel state prediction. AI-driven controllers operate at multiple levels—local (edge/cloud), intra-domain, and inter-domain—enabling predictive, multi-objective control (Wu et al., 14 May 2025, Zhang et al., 2023, Wang et al., 2024).
• Blockchain and Symbiotic Radio: Secure multi-party trading and resource exchange among heterogeneous radios are governed by DAG-based blockchains and smart contracts. ML agents optimize policy in this distributed marketplace, improving service rationality, resource utilization, and fairness (Cheng et al., 2022).
• Joint Sensing-Communication-Computing: Integration of radar sensing (particularly in the THz band), joint waveform design (OTFS, AFDM), and JCRS protocols augments environmental awareness, resilience, and localization (Han et al., 25 Feb 2025).

3. Channel Models, Propagation, and Performance Metrics

SAGIN channels are highly heterogeneous and span several critical dimensions:

• Propagation Environments: Space–air and space–ground links are LoS-dominated with negligible multipath above 50 m altitude (K-factors >50 dB). Atmospheric losses include molecular absorption (exponential in altitude), rain, scattering, and pointing/jitter effects. Urban ground–air links are NLoS-prone (Zhang et al., 2024, Han et al., 25 Feb 2025, Chen et al., 2023).
• Path-Loss Models: For FSO/THz, total path loss is composed of free-space loss (FSPL), absorption (Beer-Lambert law), and weather attenuation. Bending rays due to refraction and elevation angle dependencies are non-negligible at low elevation (Zhang et al., 2024).
• Antenna & Coverage Geometry: Unified spherical-dome coverage models analytically capture the footprint of ground-to-air, air-to-space, and cross-layer beams under varying frequency, altitude, and antenna parameters, simplifying planning and interference analysis (Liu et al., 30 Apr 2025).
• Key Performance Metrics:
  • Link capacity: $C = B \log_2(1+\text{SNR})$ per layer and interface.
  • End-to-end latency: $$T_{e2e} = \sum_{\text{hops}}(\text{propagation} + \text{tx} + \text{queue} + \text{proc delay})$$.
  • Outage and coverage probability: $\mathrm{P}_{out} = 1-\mathrm{P}_{cov}$, with $$\mathrm{P}_{cov} = \Pr(\mathrm{SINR} \geq \gamma_{th})$$.
  • Goodput, ergodic and outage rates for quantum FSO channels (Zhang et al., 2024, Trinh et al., 2 Mar 2025).

4. Resource Management, Orchestration, and Optimization

Resource orchestration in SAGIN requires solving complex, cross-layer optimization problems:

• Bi-directional Mission Offloading: Missions are flexibly offloaded between ground and air/space segments. Compute-intensive air/space tasks can be processed on ground edge/cloud, while communication-intensive ground tasks utilize HAPs/UAVs for robustness and backhaul (Zhou et al., 2019).
• Virtualization and SFC Embedding: Missions are abstracted as ordered SFCs; placement of VNFs and paths is optimized under constraints of CPU, storage, spectrum, energy, and delay. Multi-objective formulations are solved by distributed, scalable algorithms, e.g., matching-game, alternating optimization, GA, or reinforcement learning approaches (Cao et al., 2023, Geddam et al., 16 Sep 2025, Wang et al., 2024, Wang et al., 2022).
• Task Scheduling: In environments with signal-propagation delays (“ripple effect”), scheduling is formulated as a high-dimensional, partially observed Markov game and solved using multi-agent PPO with local observation filtering (Huang et al., 17 May 2025).
• Caching and Segment Cooperation: Satellite caching, ISL-based content retrieval, and bandwidth allocation across air–space–ground links are scheduled via exact-penalty and alternating optimization, minimizing download latency for both cached and non-cached content for aviation IoT (Chen et al., 2024).
• User Association and Handover: Decisions are based on SINR, load, priority, and proximity, with dynamic reassignment under overload/failure scenarios. SDN orchestrators provide <10 ms cycle latency for resource allocation (Geddam et al., 16 Sep 2025).

5. Advanced Applications: Quantum, AI, and Massive Connectivity

As SAGIN evolves, it is being extended into several advanced domains:

• Quantum-Secured SAGIN: Quantum Internet services, QKD, and secure entanglement distribution are achieved by harmonizing fiber-based, satellite-based, and UAV-based QKD under centralized orchestration, using two-stage stochastic programming to minimize key provisioning cost subject to uncertain demand (Xu et al., 2022). Optical RIS and ORIS technologies further enable non-LoS quantum links and adaptive beam shaping (Trinh et al., 2 Mar 2025).
• AI-Native and Generative Frameworks: Generative AI models (GAN, VAE, diffusion, transformer) augment resource prediction, channel map construction (GDM), semantic communications, image denoising, anomaly detection, and topology optimization (Zhang et al., 2023, Wu et al., 14 May 2025).
• Joint Communication and Sensing: THz-band JCRS protocols simultaneously deliver Gbps-class throughput and high-resolution radar sensing in LoS-dominated links, enabling functions such as beam-tracking and agile handover (Han et al., 25 Feb 2025).
• Symbiotic Radio & Blockchain: Resource trading and function exchange among heterogeneous SAGIN radios are implemented via blockchain-backed, incentive-aligned service marketplaces, with ML-based strategy adaptation (Cheng et al., 2022).

6. Key Challenges and Future Directions

Critical technical hurdles and research avenues identified include:

• Blockage Mitigation and ORIS Fabrication: Mass deployment of meter-scale metasurface ORIS with precise, low-loss phase control for quantum and classical non-LoS beam routing; vertical transmissive ORIS enabling upward/downward links and high-dimensional holography (Trinh et al., 2 Mar 2025).
• Dynamic Resource Allocation and Handover: Cross-domain spectrum/compute/power management, interference avoidance, mobility-aware handover, and predictive routing in rapid-topology environments (Lan, 25 Oct 2025, Chen et al., 2023). Deep RL and federated learning approaches are under development (Wu et al., 14 May 2025, Yin et al., 2022).
• Security and Privacy: Secure, AI-aided orchestration, quantum-resilient control, and end-to-end privacy across administrative domains with possibly untrusted relays (Xu et al., 2022, Cheng et al., 2022, Chen et al., 2023).
• Energy Efficiency and Green Operations: Optimization of air/space vehicle endurance, wireless power transfer to UAVs, energy-aware networking, and sustainable AI (Zhang et al., 2023, Trinh et al., 2 Mar 2025).
• Unified Protocol Stacks and Digital Twin Integration: End-to-end 3D protocol stacks, integration with digital twins (cybertwin-plane), and simulation-driven design for unified control and anomaly detection (Yin et al., 2022, Wu et al., 14 May 2025, Lan, 25 Oct 2025).
• Massive Scaling and Real-World Prototyping: Challenges remain in scaling distributed control, cross-layer learning, and security to mega-constellations, UAV swarms, and IoT endpoints; standardized frameworks and large-scale pilots are being pursued (Liu et al., 30 Apr 2025, Zhang et al., 2023, Lan, 25 Oct 2025).

7. Representative System Use Cases and Performance Benchmarks

SAGIN abstractions have been validated in scenarios such as:

• Drone–ORIS–Drone Entanglement Share: With λ=810 nm, beam angle 500 μrad, and total path 1.35 km, ORIS beam shaping yields fidelity F = 0.4–0.8 under daytime noise (Trinh et al., 2 Mar 2025).
• Bi-Directional Task Offloading: Enabling up to 40% air node endurance increase and up to 30% reduction in per-mission computational cost (Zhou et al., 2019).
• Resource Scheduling under Ripple Effect: Multi-AP MAPPO achieves up to 20% reduction in average age-of-information (AoI) at users and 15% AP energy savings (Huang et al., 17 May 2025).
• Cluster-Based Multi-Agent Scheduling: Clustering-based MADDPG achieves ≥25% improvement in overall system profit and reduced training complexity vs. conventional multi-agent RL (Wang et al., 2024).
• Interference Management with UAV-RIS: DoF improvement up to 133% vs. benchmarks with limited satellite antennas, enabling full-stream interference elimination in heterogeneous scenarios (Li et al., 2023).

In summary, SAGIN defines a programmable, vertically integrated network capable of supporting advanced communication, sensing, and quantum services globally, underpinned by a rich suite of cross-layer models, virtualization and orchestration mechanisms, AI and blockchain toolchains, and emerging metasurface hardware (Trinh et al., 2 Mar 2025, Lan, 25 Oct 2025, Wu et al., 14 May 2025, Liu et al., 30 Apr 2025).