---
title: Multi-Segment Architecture of LEO IoT Constellations
url: https://www.emergentmind.com/topics/multi-segment-architecture-of-leo-iot-constellations
type: topic
---

# Multi-Segment Architecture of LEO IoT Constellations

A multi-segment architecture for Low Earth Orbit (LEO) Internet of Things (IoT) constellations is a system-level design that partitions the end-to-end connectivity into functionally distinct segments, each optimized for specific roles in the overall network. Such architectures are characterized by the integration of terrestrial, aerial, and space segments, each with specialized interface logic, transmission technologies, and routing strategies, to deliver scalable, reliable, and cost-efficient IoT connectivity on a global scale. Recent research highlights several principal architectural instantiations, such as the terrestrial–LEO–GEO relay chain, hierarchical Rosette-based network embeddings, and all-optical space–HAPS–ground relays [2103.13348][2406.19677][2105.05560][2403.15659]. Each multi-segment scheme presents distinct trade-offs in terms of availability, scalability, addressability, routing stability, spectral efficiency, and resilience to atmospheric and mobility-induced impairments.

## 1. Segment Types and Functional Roles

The multi-segment paradigm decomposes the LEO IoT connectivity problem into a set of logically and physically separate components:

1. **Terrestrial Segment:** Comprises end IoT devices, ground gateways (GWs), and ground stations (GSs). Devices may connect directly to LEO satellites or via local gateways. GWs also serve as core network ingress points, protocol stack hosts, and beamforming centers [2103.13348].
2. **LEO Satellite Segment:** Consists of constellations of LEO satellites (hundreds to thousands) acting as radio or optical relays, supporting user links, inter-satellite forwarding, and, in some architectures, local (onboard) protocol stack processing.
3. **Aerial Segment:** Realized as high-altitude platform stations (HAPS, or "HAGS"), typically deployed at ~20 km altitude. HAPS act as fixed or semi-mobile relay nodes with minimal exposure to tropospheric attenuation, serving as optical or hybrid feeder links between the LEO segment and the ground [2403.15659].
4. **Geosynchronous Segment (GSO/GEO):** GEO satellites provide backhaul or extension of the network core, especially when ground–LEO links are unavailable. This segment often supports both transparent and regenerative payloads and may serve as a persistence anchor for global routing [2103.13348][2406.19677].

The interactions between these segments occur over well-defined interfaces (e.g., NB-IoT air interfaces, free-space optical links, DVB-S2/S2X/RCS), and their organization enables flexibility in deployment, resilience to link outages, and support for massive populations of IoT endpoints.

## 2. Segment Coupling, Interfaces, and Routing Abstractions

### Inter-segment Interfaces

- **GW ↔ LEO (Transparent/Regenerative):** NB-IoT protocol stack may reside entirely at the gateway (transparent) or be split between ground and satellite (regenerative), using NG/S1 or F1 interfaces typical of 5G/6G [2103.13348].
- **LEO ↔ GSO:** Connectivity is often managed via DVB-S2/S2X/RCS channels, with protocol translation as needed for air interface compatibility [2103.13348].
- **LEO ↔ HAPS ↔ GS:** All-optical architectures rely on free-space optical (FSO) links from LEO to HAPS and from HAPS to ground, with variable buffering and delay-tolerant networking (DTN) at HAPS for weather mitigation [2403.15659].

### Routing and Addressing Schemes

The complexity of segment-motion and dynamic topology necessitates sophisticated routing and addressing schemes:
- The F-Rosette framework assigns time-invariant, hierarchical addresses to satellites and partitions Earth into stable geo-cells, enabling geo-to-topo and topo-only routing without reconvergence events [2105.05560].
- Segment-aware routing adapts to momentary segment (LEO/HAPS/GSO) availability, using shortest-hop or multi-path traversal per the underlying topology and interface logic.
- The need for minimal routing state in resource-constrained satellites is satisfied by localized FIB (Forwarding Information Base) entries and hierarchical lookups, typically requiring O(MB) of memory and sub-ms lookup times in practical emulations [2105.05560].

## 3. Segment-specific Transmission and Interference Management

### LEO Segment: Full-Frequency Reuse and MMSE Beamforming

LEO segments supporting massive NB-IoT populations require efficient frequency utilization and interference cancellation. The architecture of [2103.13348] employs aggressive full-frequency reuse—each NB-IoT carrier (180 kHz) is reused on all beams of all NGSO satellites in a given "swarm" (typically, seven satellites per cluster). Centralized Minimum Mean Square Error (MMSE) beamforming, computed at the gateway or cluster controller, is employed to suppress co-channel interference:

\[
\mathbf{W} = \Bigl(\mathbf{H}^{H}\mathbf{H} + \mathrm{diag}(\alpha)\,\mathbf{I}\Bigr)^{-1}\,\mathbf{H}^H
\]
with per-antenna constraint (PAC) or max-power constraint (MPC) normalization applied to the beamforming matrix.

Performance metrics:
- **90th percentile SINR:** Increases from <10 dB (no beamforming) to 23–25 dB (MMSE+PAC/MMSE+MPC).
- **Per-user capacity:** Up to 1.4 Mbit/s per NB-IoT user (with MMSE+MPC at 600 km, 30 MHz total bandwidth) [2103.13348].

### HAPS Segment: All-Optical Relaying

HAPS, operating at ~20 km, alleviate atmospheric attenuation and cloud-blocking for FSO links compared to ground stations. Link budgets incorporate transmit power, aperture gains, geometric loss, pointing errors, and atmospheric extinction (\(\alpha\)):

\[
Pr_{SA} = Pt \cdot Gt \cdot Gr \cdot (\lambda/4\pi d)^2 \cdot L_{point} \cdot L_{atm}
\]
with coverage radius \(R_{cov}(h) \approx \sqrt{2 R_e h}\), yielding ~500 km for h=20 km [2403.15659].

Simulation demonstrates:
- **Delivery reliability (DR):** Approaches 100% under high cloud block rates with HAPS—a decisive improvement over direct ground station connectivity.
- **Delivery delay (DD):** Reduced by 20–50% relative to GS-only architectures; 2 HAPS outperform 8–10 GS for equivalent weather statistics.

### GEO Segment: Uplink Availability Modeling

The GEO segment primarily serves as a persistent relay or backhaul point in multi-segment architectures. Analytical performance is modeled as a two-hop chain (ILL: IoT→LEO, LGL: LEO→GEO), where coverage and availability are functions of satellite density (\(N_{LEO}\)), relay altitude (\(h_{LEO}\)), and transmit power:

\[
P^A = \int_0^{\theta^{max}_{IL}} f_{\theta_{IL}}(\theta)\,P_{LG}^A(\theta)\, d\theta
\]
where \(P^A\) is the end-to-end availability probability, encapsulating both segment visibilities and signal quality [2406.19677].

Key findings:
- Deployment of a few hundred LEO satellites at 600–800 km suffices for >99% coverage for typical SNR thresholds (\(\gamma_{IL}\approx10\,\mathrm{dB}\)), given IoT transmit power of 10–20 dBW.

## 4. Topological and Addressing Stability Under Segment Mobility

The architectural instability caused by non-stationary satellite ground tracks in LEO mega-constellations leads to frequent endpoint address churn and routing reconvergence under IP-based schemes (user address changes every 133–510 s; network usability drops below 20%) [2105.05560]. The F-Rosette framework addresses this by:
- **Recursive constellation construction:** Ground tracks and satellite footprints repeat via carefully phased orbits and inter-plane links.
- **Hierarchical addressing:** Each satellite receives a time-invariant address (s₀.s₁.…sₖ), and user addresses encode geographic cell locations. Address change frequency becomes a function of actual mobility between 2,000 km² cells, not satellite footpoint motion.
- **Routing embedding:** Topological routing algorithms incrementally navigate the hierarchy via layer permutations, producing paths of ≤(k+1)·N/2 hops and guaranteeing stability with unsaturated local table size (<1.3 MB/satellite).

Emulation confirms that such designs offer:
- Zero user IP churn (compared to >1,000/s in current architectures),
- Near-100% routing usability,
- Hop-optimal delays (<1.4% above policy minimum).

A plausible implication is that only architectures with explicit address and routing decoupling from motion dynamics can scale to very large IoT constellations without impeding network usability.

## 5. Performance Trade-offs, Scalability, and Design Guidelines

Comprehensive assessments across the referenced architectures yield the following implementation trade-offs and recommendations:

| *Aspect*           | *Key Findings*                                                                                     | *Architectural Implication*                           |
|--------------------|---------------------------------------------------------------------------------------------------|-------------------------------------------------------|
| Frequency reuse & interference | Full-frequency reuse maximizes throughput per swarm; requires centralized MMSE beamforming. | Best for low-power, massive NB-IoT; raises GW requirements. |
| Segment densities  | Hundreds of LEOs at 600–800 km suffice for global 99%+ availability.                              | Excessive density yields diminishing returns.         |
| Power budgeting    | Uplink SNR ≥10 dB at 95th percentile path-loss via modest (10–20 dBW) device Tx power.            | Optimizes end-device lifetime and coverage.           |
| HAPS deployment    | HAPS reduces required GS count 2–8× while hardening weather resilience.                           | Lowers CAPEX, improves DR to 100% in adverse weather.|
| Address/routing stability | F-Rosette scheme obviates IP churn and reconvergence, even at scale.                        | Minimizes control-plane load on constrained satellites.|

Best-practice guidelines include:
- Centralize beamforming and CSI processing at GWs or cluster heads.
- Use per-satellite power-constrained normalization (MPC) for higher fairness and capacity; PAC yields higher peak SINR but lower fairness [2103.13348].
- Optimize antenna design for the desired trade-off between footprint size (hand-over overhead) and link gain.
- For IoT-dense regions, place GEOs to minimize the IoT–GEO central angle; for global coverage, provision multiple GEOs and dynamic beam steering [2406.19677].
- Integrate HAPS where feasible to reduce terrestrial infrastructure, especially in regions prone to frequent tropospheric rain/cloud attenuation [2403.15659].

## 6. Implementation Challenges and Research Directions

Several persistent challenges in multi-segment LEO IoT deployments are identified:

- **Imperfect CSI:** Centralized beamforming is susceptible to outdated or imprecise channel state information, which can degrade interference suppression.
- **Segment Handover and Synchronization:** Dynamic coordination between moving LEO beams, swarms, and ground/relays requires low-latency, reliable signaling.
- **FSO Alignment and HAPS Stability:** Fast atmospheric fluctuations and platform drift at HAPS induce pointing errors, necessitating real-time corrections via fast-steering mirrors and adaptive optics [2403.15659].
- **Buffering and Congestion Control:** HAPS-based relays with delay-tolerant networking must efficiently manage large, bursty data offloads from multiple LEOs, demanding CGR and priority queuing [2403.15659].
- **Routing Table Scalability:** Even hierarchical/greedy schemes require O(MB) storage in high-index F-Rosette deployments, but this remains within resource bounds of modern satellites (<2 MB RAM).

Future research is directed towards:
- Joint optimization of satellite altitudes, platform positions (e.g., HAPS station-keeping), and routing for end-to-end delay and throughput.
- Hybrid segment integration (RF backup for FSO, multi-path over different segment routes).
- Standardization of protocol interfaces across space–aerial–terrestrial boundaries, particularly in the emerging 6G NTN context.

The cumulative evidence suggests that segment-aware, hierarchically addressed, and interference-managed architectures are essential to realize scalable, global IoT services over LEO constellations, with each segment’s technical choices dictating the overall system’s robustness, performance, and long-term operational cost.

Source: https://www.emergentmind.com/topics/multi-segment-architecture-of-leo-iot-constellations