All-Optical Space–HAPS–Ground Relays

• The paper demonstrates that all-optical space–HAPS–ground relays decouple LEO-to-ground links from tropospheric cloud outages, enhancing availability and reducing ground infrastructure.
• It employs advanced PAT, multi-beam HAPS optimization, and optional OIRS integration to mitigate alignment issues and extend optical wireless coverage.
• Simulation and analytical models reveal up to a 30% increase in effective capacity and significant cost reductions compared to traditional ground station networks.

All-optical space–HAPS–ground relays constitute a three-segment optical communication architecture leveraging free-space optical (FSO) links among low Earth orbit (LEO) satellites, high-altitude platform stations (HAPS), ground stations (GS), and optionally optical intelligent reflecting surfaces (OIRS). By elevating the atmospheric relay from the terrestrial layer to the stratosphere, these systems decouple the critical LEO–ground segment from tropospheric cloud-induced outages, enabling higher availability, greater link efficiency, and a reduction in ground infrastructure footprint. This architecture underpins emerging non-terrestrial network (NTN) solutions with applications in broadband satellite Internet, disaster-resilient communications, cloud-immune backhaul, and urban optical wireless access.

1. System Architecture and Topology

All-optical space–HAPS–ground relay networks are characterized by a layered topology comprising:

• LEO Satellites: Each LEO node is equipped exclusively with an FSO terminal used for both inter-satellite and LEO–HAPS (uplink/downlink) optical links. No RF fallback exists on the spacecraft; all data transfer—including contact initiation—is executed using optical transceivers. LEO satellites operate on a deterministic contact plan derived from orbital mechanics and ground relay geometry (Madoery et al., 2024).
• High-Altitude Platform Stations (HAPS)/High-Altitude Ground Stations (HAGS): HAPS platforms at altitudes near 20 km are outfitted with:
  • One FSO terminal pointed upward for LEO link acquisition/tracking.
  • One or more downward-pointed FSO terminals dedicated to ground coverage.
  • Buffering for store-carry-forward operation during cloud-induced ground link outages (Madoery et al., 2024, Truong et al., 2023).
• Terrestrial Ground Stations (GS): Traditional optical relay endpoints at sea level, receiving traffic from HAPS via FSO downlink. The availability of this segment is primarily constrained by tropospheric weather.
• OIRS-Assisted Segment (optional): Introduction of an optical intelligent reflecting surface (OIRS) provides a mechanism for bypassing urban non-line-of-sight obstacles, extending reach to NLOS users (Shang et al., 3 Nov 2025).

The network supports several relay paradigms: direct LEO–HAPS–GS (all-optical relaying), LEO–HAPS–OIRS–User (OIRS-assisted relay), and multi-hop HAPS mesh for wide-area aggregation.

2. FSO Link Budget and Channel Modeling

Each FSO hop's link budget is modeled as an aggregate of transmitter power, geometric gain, free-space loss, atmospheric attenuation, turbulence loss, and system margin: $$P_r = P_t + G_t + G_r - L_{\text{space}}(d) - L_{\text{atm}}(\lambda,h,\theta) + M_{\text{margin}}$$ where:

• $P_t$ = optical transmit power [dBm];
• $G_t, G_r$ = transmitter/receiver aperture gains, $$G_t = 10\log_{10} \left(\frac{\pi D_t}{\lambda} \right)^2$$;
• $$L_{\text{space}}(d) = 20\log_{10}\left( \frac{4\pi d}{\lambda} \right)$$ = free-space path loss at range $d$;
• $L_{\text{atm}}(\lambda, h, \theta)$ = atmospheric losses (molecular, aerosol, turbulence), encapsulated by empirical models (e.g., MODTRAN) or

$$L_{\rm atm}(\lambda,h,\theta) = \alpha(\lambda)\,d_{\rm atm}(h,\theta) + L_{\rm turb}(\theta)$$

where $\alpha$ is the attenuation coefficient and $L_{\rm turb}$ is the turbulence margin;

• $M_{\text{margin}}$ = engineering margin for pointing, aging, noise (Madoery et al., 2024).

Channel models incorporate:

• Atmospheric absorption and scattering: Modeled via Beer–Lambert law or empirical fits (Shang et al., 3 Nov 2025).
• Turbulence: Gamma–Gamma or log-normal models parameterized by Rytov variance, with exact CDFs given in Meijer-G or Fox-H functions (Shang et al., 3 Nov 2025).
• Pointing and misalignment: Statistically modeled using the Farid–Hranilovic or truncated Hoyt series approaches, with explicit closed-form PDFs (Shang et al., 3 Nov 2025).
• Composite links (e.g., HAP–OIRS–User): Net gain and impairment factors are derived for cascaded FSO hops using unified SNR/statistics (Shang et al., 3 Nov 2025).

To guarantee a target BER, the received power $P_r$ must exceed receiver sensitivity by the stipulated link margin.

3. Coverage, Capacity, and Multi-Beam HAPS Optimization

HAPS-based relays critically depend on geometric line-of-sight and beam divergence. For a single FSO downlink: $$R_\alpha(h,\alpha) = h\,\tan\left( \frac{\alpha}{2} \right)$$ with divergence $\alpha$ constrained by edge-of-footprint power ($P^{\rm rx}_{\rm edge} \geq P_{\min}$). For $h=20\,\mathrm{km},\ \alpha_{\max}=37^\circ$, $R_\alpha \approx 6.69\,\mathrm{km}$ (Truong et al., 2023).

To expand coverage, multi-beam ("mFSO", Editor's term) arrays are used:

• One principal beam ($\alpha$) and $m$ supplementary beams ($\beta$), azimuthally distributed, cover an extended footprint:

$R_{\rm ext} = h\,\tan(2\phi)$

where $\phi$ is analytically calculated as a function of $\alpha, \beta, m$ (Truong et al., 2023).

Optimization problem:

• Objective: Minimize total cost $\hat C(N, \theta)$ (amortization, maintenance, and energy) for area $S$, given CAPEX/OPEX, mass, power, solar constraint ($E_{\rm cons}(N,\theta) \leq E_{\rm solar}$), and coverage requirements.
• Multi-beam designs can double ground reach; e.g., $m=13, \alpha^*=37^\circ, \beta^*=16^\circ$ yields $R_{\rm ext} \approx 11.93\,\mathrm{km}$.
• Multi-beam HAPS configurations reduce network cost by 54–87% versus single-beam for large ground node clusters, contingent on energy budgets above $\sim50\,\mathrm{kWh}$/day (Truong et al., 2023).

With OIRS-assisted topologies, coverage is further extendable into obstructed urban scenarios, constrained primarily by the geometric/misalignment loss exponent and OIRS aperture/element size (Shang et al., 3 Nov 2025).

4. Performance Metrics and Comparative Analysis

Critical system-level metrics include delivery ratio (DR), delivery delay (DD), buffer occupation (BO), SNR/outage probability, BER, and capacity.

Key findings from simulation and analytical modeling:

• Delivery Ratio: With TCS = 5 h, a single HAPS achieves $100\%$ file delivery for TCC down to 0.1 h, whereas 10 GS reach only ~80%. Under longer clear times (TCS = 25 h), single HAPS still achieves near-100% while GS-only systems can fall below 60% with few GS (Madoery et al., 2024).
• Delivery Delay: At high cloud rates (TCC = 0.5 h), 2 HAPS yields DD ≈ 200 min vs. 10 GS ≈ 250 min; HAPS consistently outperforms GS in all analyzed configurations.
• Capacity: Effective pass duration (aggregate link time) increases by ~30% with HAPS, weeklong cumulative capacity can improve by 25% (1 HAPS vs. 5 GS, moderate cloud) (Madoery et al., 2024).
• Equivalency Model: Under heavy cloud (TCC = 0.1 h, TCS = 5 h), 2 HAPS equates to 8 GS; fewer GS suffice as weather improves.
• OIRS-assisted Link: Outage probability is reduced by up to two orders of magnitude versus direct HAP-to-user, and heterodyne detection outperforms IM/DD in all regimes. Maximum capacity approaches 1.7 nats/s/Hz at 30 dB SNR (Shang et al., 3 Nov 2025).

5. Design Guidelines and Engineering Constraints

System integration and deployment require adherence to several design recommendations:

• HAPS/GS Siting: One HAPS can replace ~2–5 GS, depending on local cloud climatology. Site selection is weather-driven for maximal benefit (Madoery et al., 2024).
• Aperture Sizing: Use apertures ≥30 cm on HAPS to sustain high link margin, compensate for stratospheric turbulence, and ensure SWaP is within platform constraints.
• Buffering: Provision ≥200 GB onboard cache per HAPS to absorb outage intervals of up to 5 h, enabling lossless store-carry-forward (Madoery et al., 2024).
• Multi-Beam Arrays: Optimal beam count, divergence, and transceiver configuration determined by explicit minimization of cost under energy and coverage constraints; dense WDM in the FSO transceivers enables further reduction in the number of required HAPS platforms (Truong et al., 2023).
• Pointing, Acquisition, and Tracking (PAT): Mitigate HAPS motion and wind-induced jitter via dual-stage PAT (wide FOV acquisition, narrow FSO beam tracking), adaptive optics, and inertial reference for rapid retargeting (Madoery et al., 2024).
• Network Synchronization and Routing: Employ GPS-disciplined timing, contact-graph routing (CGR), and (optionally) inter-HAPS optical ISLs for network-wide clock alignment and optimal delay-tolerant networking (DTN) function (Madoery et al., 2024).

6. Channel Impairments, Reliability, and OIRS-Assisted Extensions

Atmospheric and geometrical impairments are the fundamental limiting factors in FSO-based space–HAPS–ground operation:

• Atmospheric/Weather: Turbulence in the HAPS domain and lower troposphere drive both scintillation and beam wander; main mitigation includes adaptive margin control, link adaptation, and site diversity (Madoery et al., 2024).
• PAT and Misalignment: Closed-loop beaconing and steerable mirrors are essential, with the Farid–Hranilovic and truncated Hoyt models enabling accurate BER/outage statistics as a function of jitter/pointing error (Shang et al., 3 Nov 2025).
• OIRS Integration: For OIRS-assisted relay, main channel impairments—atmospheric turbulence, geometric and misalignment loss, OIRS reflection efficiency—are analytically modeled, and their statistics specified in closed form. Increasing OIRS element size raises the geometric loss exponent, which increases diversity gain but imposes tighter alignment demands (Shang et al., 3 Nov 2025).

Amplify-and-forward (AF) versus decode-and-forward (DF) paradigms yield trade-offs in complexity and outage probability; DF provides only modest outage improvements relative to AF at the cost of increased relay complexity (Shang et al., 3 Nov 2025).

7. Prospects, Limitations, and Future Research Directions

All-optical space–HAPS–ground relays offer several architectural advantages:

• Decoupling of cloud/weather from space-to-ground bottleneck.
• Simplified and lighter LEO payloads using optical-only terminals.
• Scalability via clustering and mesh networking of HAPS for mega-constellation scenarios (Madoery et al., 2024).

Nevertheless, several limitations remain:

• Static beam footprints in current models do not account for dynamic node distribution or real-time weather variability (Truong et al., 2023).
• Omits end-to-end modeling of atmospheric outage along the full LEO–HAPS–GS/OIRS–user relay chain.
• Designs are sensitive to stratospheric wind/turbulence, requiring further development of robust PAT and adaptive optics systems.
• Optimal system dimensioning under mixed urban–rural and temporally variable ground user distributions is not fully explored.

Future research is anticipated on adaptive beam shaping, real-time weather-integrated routing, satellite–HAP atmospheric modeling, and cross-layer optimization for integrated optical wireless NTNs (Truong et al., 2023, Madoery et al., 2024, Shang et al., 3 Nov 2025).