A Complete Guide to Satellite Constellations
From GPS to Starlink, from Inmarsat to Sentinel: a structured reference covering existing operational constellations and the engineering principles that govern their space segment and ground segment design — Walker geometry, orbital regime selection, ground station networks, frequency allocation, and the trade-offs that shape every architecture.
What Is a Satellite Constellation?
A satellite constellation is a coordinated group of satellites operating together as a single system to provide a service that no individual satellite could deliver alone — typically continuous global coverage, redundancy, or revisit performance impossible to achieve from a single orbital slot.
A single satellite in Low Earth Orbit (LEO) sees only a small footprint and passes over any given point only briefly. To deliver continuous coverage, persistent communications, or rapid revisit imaging, multiple satellites must be distributed in a geometry that ensures the required service availability with the smallest fleet size that meets mission cost and complexity targets.
- Coordinated orbital geometry across multiple satellites
- A unified ground segment serving the full fleet
- Service continuity guaranteed by the system, not any single asset
- Designed-in redundancy and graceful degradation
- Constellation-level operations (replenishment, station-keeping, maneuvers)
The two-segment view
Every operational constellation can be analysed through two coupled architectures: the space segment — the satellites themselves and how they are arranged in orbit — and the ground segment — the infrastructure on Earth that operates them and delivers value to users. A third element, the user segment, is sometimes treated separately (terminals, receivers, handsets), but for design purposes it is generally an interface specification driven by the choices made in the other two.
Satellites, orbital geometry, inter-satellite links, payloads, attitude and power subsystems, propulsion, on-orbit redundancy and spares.
TT&C stations, mission operations, payload data downlink, gateways, network operations centre, mission control, scheduling, calibration.
Receivers, terminals, handsets, antennas — the equipment that consumes the service. Driven by choices in the space and ground segments.
Orbital Regimes & Selection
The choice of orbital regime is the single most consequential decision in constellation design. It determines latency, coverage, satellite count, launch cost, ground station network, and lifetime.
Regime comparison
| Regime | Altitude | Period | One-way latency | Sat. footprint | Sats for global | Typical use |
|---|---|---|---|---|---|---|
| LEO | 300 – 2,000 km | 90 – 120 min | ~1 – 7 ms | ~5 – 15° lat | 40 – 4,000+ | Broadband, EO, SAR, science |
| MEO | 2,000 – 35,786 km | 2 – 24 h | ~50 – 130 ms | ~30 – 60° lat | 8 – 30 | GNSS, regional broadband |
| GEO | 35,786 km | 23h 56m 4s | ~120 – 280 ms | ~1/3 of Earth | 3 – 4 | TV broadcast, FSS, MSS, weather |
| HEO | Variable (apogee high) | ~12 h (Molniya) | Variable | High-lat dwell | 2 – 4 | High-latitude comms, SBIRS |
| SSO | 600 – 800 km (typ.) | ~98 min | ~2 – 4 ms | ~5 – 10° lat | Mission-specific | EO, weather (polar), mapping |
| IGSO | 35,786 km, i ≠ 0° | 23h 56m 4s | ~120 ms | Figure-8 trace | 3+ for region | Regional GNSS (BeiDou, QZSS) |
| cislunar / L-points | > 384,000 km | Days+ | ~1.3 s+ | N/A | N/A | Lunar comms (LCRNS, Lunar Pathfinder) |
A special case of LEO selected for the J2 nodal regression of the orbital plane to match Earth's mean motion around the Sun (≈0.9856°/day). This holds the local solar time of the descending node (LTDN) constant, giving each pass over a given latitude the same illumination conditions — essential for time-series Earth observation.
Choosing a regime
Voice, real-time gaming, financial trading, tactical edge — these demand LEO. Speed-of-light round-trip to GEO (≈540 ms) breaks interactive applications. Starlink, OneWeb, and Iridium exist because GEO physics cannot serve them.
Broadcast TV, regional MSS, fixed VSAT — GEO is unbeatable. One satellite illuminates ~1/3 of the Earth's surface from a fixed sky position, so the user antenna does not need to track. The economics are simple: very few satellites, very simple terminals.
GNSS uses MEO (~20,000 km) because it is high enough for one satellite to be seen from a wide area (good geometric dilution of precision when many sats are in view), low enough for acceptable signal strength, and out of the heaviest debris environment of LEO.
GEO disappears below the horizon above ~70° latitude. Russia historically used Molniya HEO (12-hour, ≈63.4° inclined, ~600 × 39,800 km elliptical) so that satellites dwell over the apogee for hours. Today, polar LEO megaconstellations and IGSO satellites serve the same need.
Walker Constellations
Most modern constellations follow one of two geometric patterns developed in the 1970s by John G. Walker at the UK Royal Aircraft Establishment: Walker Delta (inclined) and Walker Star (near-polar).
i = inclination (deg) · T = total satellites · P = number of orbital planes · F = phasing factor (0 ≤ F ≤ P−1)
Phase offset between satellites in adjacent planes:
Δϕ = F × 360° / T. Satellites per plane: S = T / P. RAAN spacing between planes: 360° / P (Delta) or 180° / P (Star).
Planes spread evenly across 360° of RAAN. Inclination < 90°. Excellent for coverage of mid- and low-latitude bands. No coverage of the geographic poles unless inclination is raised. Examples:
- GPS — 55° : 24/6/2 (24 sats in 6 planes)
- Galileo — 56° : 24/3/1
- O3b mPOWER — 0° equatorial MEO
- Globalstar — 52° : 48/8/1
Planes spread across 180° of RAAN (one hemisphere); satellites in adjacent planes travel in opposite directions across the equator. Inclination ≈ 90°. Provides true global coverage including the poles. Examples:
- Iridium NEXT — 86.4° : 66/6/2
- OneWeb Gen-1 — 87.9° : 588/12/1 (operational)
- Starlink polar shells — 97.6° SSO sub-shell
Phasing factor & relative geometry
The phasing factor F defines the relative true anomaly of satellites in adjacent planes at the moment of plane intersection. It is a key tuning parameter: it controls the maximum and minimum satellite-to-satellite distances, the geometry of inter-satellite links, and — for navigation constellations — the geometric dilution of precision (GDOP) seen by users on the ground.
24 satellites in 6 planes (4 per plane). Plane RAAN spacing: 60°. Phase offset between adjacent planes: F × 360° / T = 2 × 360° / 24 = 30°. Within a plane, satellites are 90° apart in true anomaly. The result is the famous "honeycomb" sky-coverage pattern that guarantees ≥4 satellites in view almost everywhere on Earth at all times.
Global Navigation Satellite Systems (GNSS)
Six independent GNSS systems are operational or building toward operational status. Four are global (GPS, GLONASS, Galileo, BeiDou); two are regional (NavIC, QZSS). All operate in MEO or GEO/IGSO/HEO regimes.
| System | Operator | Coverage | Walker | Altitude | Inclination | Period | Sats (operational) | Status |
|---|---|---|---|---|---|---|---|---|
| GPS USA | USSF / 2SOPS | Global | 55° : 24/6/2 | 20,180 km | 55° | 11h 58m | 31 | FOC since 1995 |
| GLONASS RUS | Roscosmos | Global | 64.8° : 24/3/1 | 19,140 km | 64.8° | 11h 16m | 24 | FOC since 2011 |
| Galileo EU | EUSPA / ESA | Global | 56° : 24/3/1 (+spares) | 23,222 km | 56° | 14h 04m | 24 + 6 spares | FOC since 2022 |
| BeiDou-3 CHN | CSNO | Global | Mixed MEO/IGSO/GEO | 21,528 / 35,786 km | 55° / 55° / 0° | 12h 53m / 24h | 24 MEO + 3 IGSO + 3 GEO | FOC since 2020 |
| NavIC IND | ISRO | Regional (IN + ~1500 km) | IGSO + GEO | 35,786 km | 29° / 0° | 23h 56m | 7 (4 IGSO + 3 GEO) | L1 expansion in progress |
| QZSS JPN | QSS / Cabinet Office | Regional (Japan + APAC) | QZO + GEO | ~32,000 – 39,000 km | ~43° (QZO) / 0° (GEO) | 23h 56m | 4 (expanding to 7) | Expansion to 7 by ~2026 |
GPS — the architectural reference
- Master Control Station (MCS) — Schriever SFB, CO, USA
- Alternate MCS — Vandenberg SFB, CA, USA
- 11 Monitor Stations globally distributed (USAF + NGA)
- 4 Ground Antennas for S-band TT&C uplink
- OCX — next-gen Operational Control Segment for GPS III
- Pre-launch Compatibility Station (Cape Canaveral)
The MCS computes the navigation message (clock and ephemeris) from monitor station observations and uplinks updates to each satellite typically once per day. Modernised satellites support autonomous navigation via inter-satellite cross-links.
Why GNSS chose MEO
MEO (~20,000 km) balances three competing goals: (1) enough visibility from any point on Earth to provide good geometric dilution of precision (GDOP) — typically ≥4 satellites visible; (2) acceptable signal strength at the user receiver; (3) avoidance of the dense LEO debris environment. The 12-hour orbital period of GPS produces a repeating ground track every two revs, which simplifies operations.
Each pseudorange equation has 4 unknowns (user x, y, z and clock bias δt), so a minimum of 4 simultaneous satellite observations is required for a solution.
SBAS & augmentation
Several Satellite-Based Augmentation Systems (SBAS) use GEO satellites to broadcast corrections improving GNSS accuracy and integrity for safety-of-life applications (notably aviation):
| System | Region | Operator | Status |
|---|---|---|---|
| WAAS | North America | FAA | Operational |
| EGNOS | Europe | EUSPA | Operational (V2 → V3) |
| MSAS | Japan | JCAB | Operational |
| GAGAN | India | AAI / ISRO | Operational |
| SDCM | Russia | Roscosmos | Operational |
| BDSBAS | China & APAC | CSNO | Coming online |
| KASS | South Korea | MOLIT / KARI | Operational since 2023 |
| SouthPAN | Australia / NZ | Geoscience Australia / Toitū Te Whenua | In implementation |
GEO Communications Constellations
The geostationary belt — a thin equatorial ring at 35,786 km — is the workhorse of satellite communications, hosting fixed satellite services (FSS), broadcast (BSS), and mobile satellite services (MSS). A single GEO satellite covers ~1/3 of Earth's surface from a fixed sky position, so user antennas can be small and pointed once.
- Satellite appears stationary in the sky — no tracking needed
- Single satellite covers ~120° of longitude (sub-tropics & tropics)
- 3 sats spaced 120° apart = near-global coverage (excl. polar)
- Mature regulatory regime: ITU orbital slot allocation
- Long lifetimes (typ. 15 yrs); large platforms (5 – 6,000+ kg, 15+ kW)
- ~280 ms one-way latency (~540 ms round-trip)
- No coverage above ~70° latitude (satellite below horizon)
- High path loss (large gateways, large user dishes for high data rate)
- Crowded longitudes — slot rivalry, coordination complexity
- End-of-life: graveyard orbit ≥300 km above GEO mandatory
Major operational GEO operators
| Operator | Fleet (GEO) | Services | Frequencies | HQ |
|---|---|---|---|---|
| Intelsat FSS | ~50+ | FSS, video, gov, mobility, IFC | C, Ku, Ka | McLean, USA |
| SES FSS | ~40 GEO + O3b mPOWER MEO | FSS, video, gov, cloud connectivity | C, Ku, Ka | Betzdorf, Luxembourg |
| Eutelsat FSS | ~35 GEO (+ OneWeb LEO) | Video, broadband, mobility, gov | C, Ku, Ka | Paris, France |
| Inmarsat MSS | ~14 (after Viasat merger) | Maritime, aero, land mobile, gov | L, Ka (Global Xpress) | London, UK (Viasat group) |
| Viasat FSS | 7+ (incl. ViaSat-3 Ka HTS) | Broadband, IFC, mobility, gov | Ka HTS | Carlsbad, USA |
| Hughes (EchoStar) FSS | ~10+ | Consumer broadband (Jupiter), gov, ent. | Ka HTS, Ku | Englewood, USA |
| Arabsat FSS | ~8 | Video, broadband, gov (Arab League) | C, Ku, Ka | Riyadh, KSA |
| Yahsat MSS/FSS | 5 (Y1A, Y1B, Al Yah 3, T4-NGS, T4-NGS-2) | Gov, broadband, mobile (Thuraya) | C, Ku, Ka, L (Thuraya) | Abu Dhabi, UAE |
| Es'hailSat FSS | 2 (Es'hail-1, Es'hail-2) | Video, broadband, gov | Ku, Ka | Doha, Qatar |
| Nilesat BSS | 3 | DTH video (MENA) | Ku | Cairo, Egypt |
| Telesat FSS | ~15 (+ Lightspeed LEO) | Video, broadband, gov, enterprise | C, Ku, Ka | Ottawa, Canada |
| Sky Perfect JSAT FSS | ~17 | Video, broadband, gov (APAC) | C, Ku, Ka | Tokyo, Japan |
Inmarsat: a textbook MSS architecture
Inmarsat operates a distributed GEO MSS system focused on maritime, aeronautical and land-mobile users. The fleet typically includes 3 – 4 prime satellites at separated longitudes for global L-band coverage (60° W, 25° E, 64° E, 178° E nominal slots), with the Global Xpress (GX) Ka-band HTS fleet providing higher data-rate broadband.
High Throughput Satellites (HTS)
Modern GEO comsats are HTS: instead of a few wide beams, they project tens to hundreds of narrow spot beams, each reusing the same frequency on a regular pattern (typ. 4-colour reuse). This multiplies effective bandwidth by 10× – 100× over a wide-beam satellite.
A wide-beam C-band transponder might deliver ~500 Mbps over a 1/3-Earth footprint. A modern Ka HTS with 200+ spot beams delivers 1 Tbps+ over the same footprint, at the cost of a far more complex payload, larger antenna apertures, and a denser gateway network. The number of gateways on the ground often becomes the system bottleneck — see Section 11.
LEO & MEO Communications — Megaconstellations
The economics of small launch and mass-production of small satellites have unlocked a new class of constellation: hundreds to tens of thousands of LEO satellites delivering low-latency broadband globally. The result is a fundamental shift from "satellite as a single node" to "constellation as a distributed network".
| System | Operator | Regime | Walker | Altitude | Total sats | Status |
|---|---|---|---|---|---|---|
| Iridium NEXT | Iridium | Polar LEO | 86.4° : 66/6/2 | 781 km | 66 + 9 spares | Operational (since 2019) |
| Globalstar | Globalstar | Inclined LEO | 52° : 48/8/1 | 1,414 km | 48 | Operational |
| Orbcomm | Orbcomm Inc. | Inclined LEO | ~45° / 50° mixed | ~750 km | ~30 | Operational (M2M / IoT) |
| O3b mPOWER | SES | Equatorial MEO | 0° : 11+ / 1 / 0 | 8,063 km | 11+ (target 13) | Operational |
| Starlink | SpaceX | LEO multi-shell | 53° / 70° / 97.6° (mix) | 340 – 614 km | ~7,000 deployed (target 12,000 → 42,000) | Operational, expanding |
| OneWeb | Eutelsat OneWeb | Polar LEO | 87.9° : 588/12/1 | 1,200 km | ~636 deployed | Gen-1 service complete |
| Project Kuiper | Amazon | LEO | ~51.9° / 42° / 33° (3 shells) | 590 – 630 km | 3,236 planned | Deployment underway |
| Telesat Lightspeed | Telesat | LEO | ~50° + polar | 1,015 km | 198 planned (Gen-1) | Build phase |
| Guowang (SatNet) | China SatNet | LEO | Multiple shells | 500 – 1,145 km | ~13,000 planned | Early deployment |
| Qianfan (SpaceSail) | SSST | LEO | Polar | ~1,160 km | ~14,000 planned | Early deployment |
| AST SpaceMobile | AST | LEO direct-to-cell | ~50° (BlueWalker / BlueBird) | ~700 km | ~5 ops (target 168) | Deploying |
| Lynk Global | Lynk | LEO direct-to-cell | Polar | ~520 km | ~10 ops (target ~5,000) | Service expanding |
Iridium NEXT — the polar gold standard
Iridium was the first constellation to use inter-satellite links (ISLs) operationally. A user call is routed through the constellation in space, hopping between satellites until it reaches a satellite over a ground gateway — only then does it descend to Earth. This means the system needs far fewer gateways than a "bent-pipe" architecture, and it can serve users in oceans, polar regions, and conflict zones where ground infrastructure is unavailable.
Iridium's Aireon hosted payload provides the world's only space-based ADS-B aircraft surveillance, a textbook example of constellation-as-platform.
Starlink — scale & rate
Starlink is the largest operational constellation in history. Its design philosophy is the opposite of Iridium: many cheap satellites, low altitude, short lifetime, continuous replacement.
- Multiple shells at 53°, 70°, 97.6° (SSO)
- 340 – 614 km altitude
- Optical ISLs on v1.5 / v2 satellites
- Phased-array Ku/Ka user payload
- Krypton ion propulsion, atmospheric demise EOL
- ~5-year operational life by design
- Continuous launch cadence (Falcon 9, Starship)
- Autonomous collision avoidance
- Atmospheric drag-assist deorbit at EOL
- Software-defined radio, OTA updates
- ~25 – 250 Mbps download (residential)
- ~25 – 60 ms latency (LEO physics)
- Self-aligning phased-array terminal (Dishy)
- Direct-to-cell variant (D2C / "Direct to Cell")
Inter-satellite links (ISLs)
ISLs let satellites talk directly to each other in orbit, removing the requirement that user, satellite, and gateway all lie in a single line of sight. Modern ISLs are typically optical (free-space laser, ~1,550 nm) delivering 100+ Gbps per link with very low interference and excellent security properties. Starlink, OneWeb (V2), and Telesat Lightspeed are all ISL-equipped.
A constellation with ISLs is a routed mesh network in motion. Each satellite has 4 – 6 ISL neighbours (typically 2 in-plane fore/aft + 2 cross-plane), and packets are forwarded in the same way as on a terrestrial backbone. The topology changes continuously as the orbits move, so routing is a time-varying graph problem — algorithms must handle predictable handovers and unpredictable outages (e.g. eclipses, anomalies).
Earth Observation Constellations
Earth Observation (EO) constellations are organised around revisit — the time between consecutive observations of any given point. A single SSO satellite revisits a given location every 5 – 16 days; a constellation can deliver sub-daily, hourly, or even minute-scale revisit.
| Programme | Operator | Sensor type | Sats / orbit | Resolution (GSD) | Revisit | Status |
|---|---|---|---|---|---|---|
| Landsat PUB | USGS / NASA | Multispectral optical + thermal | 2 ops (8, 9), SSO 705 km | 15/30/100 m | 8 days (combined) | Operational since 1972 |
| Sentinel-1 PUB | ESA / Copernicus | C-band SAR | 1 op (1A), SSO 693 km | 5 × 20 m IW mode | ~6 days (was 12 with 1A+1B) | 1B lost 2022, 1C launched 2024 |
| Sentinel-2 PUB | ESA / Copernicus | Multispectral optical (13 bands) | 2 ops (2A, 2B), SSO 786 km | 10 / 20 / 60 m | 5 days (combined) | Operational |
| Sentinel-3 PUB | ESA / EUMETSAT | OLCI + SLSTR + altimeter | 2 ops, SSO 815 km | 300 m / 500 m / 1 km | ~1 day | Operational |
| Planet Dove / SuperDove COM | Planet | Multispectral optical (3U/6U cubesats) | ~150+, SSO 475 – 525 km | 3 – 4 m | Daily (entire landmass) | Operational ("Flock") |
| Planet SkySat COM | Planet | Sub-metre optical + video | 21, SSO + 53° inclined | 0.5 m | Up to 12× daily | Operational |
| Maxar WorldView Legion COM | Maxar | Very-high-resolution optical | 6, mid-inclination + SSO | ~30 cm | Up to 15× daily over ROIs | Deploying / operational |
| Airbus Pléiades Neo COM | Airbus DS | Very-high-resolution optical | 2 ops (Neo 3, Neo 4) | 30 cm | 2× daily any point | Operational |
| Capella Space COM | Capella | X-band SAR | ~7 | 0.5 m (Spotlight) | Hourly tasking goal | Operational |
| ICEYE COM | ICEYE | X-band SAR | ~30+ | 0.25 – 1 m | Sub-hourly persistent ROI | Operational |
| BlackSky COM | BlackSky | Optical (~1 m) | ~16 | ~1 m | ~15× daily ROI | Operational |
| HawkEye 360 COM | HawkEye 360 | RF geolocation (clusters of 3) | ~30 (10 clusters) | RF geo (km-class) | Sub-hourly | Operational |
| MBZ-SAT UAE | MBRSC | Sub-metre optical | 1 | ~0.7 m | N/A (single sat) | Launched 2024 |
| KhalifaSat UAE | MBRSC | Optical | 1 | 0.7 m PAN | N/A | Operational since 2018 |
Copernicus Sentinel — the public reference architecture
The Sentinel programme operates a family of complementary missions rather than a single homogeneous constellation. Each pair of identical satellites flies in the same SSO at 180° in true anomaly to halve revisit time. All raw data is free and open under the Copernicus data policy.
- Sentinel-1: C-band SAR, all-weather imaging
- Sentinel-2: 10 m optical, agriculture & land use
- Sentinel-3: ocean / land / atmosphere wide-swath
- Sentinel-5P: tropospheric chemistry (TROPOMI)
- Sentinel-6: high-precision sea-level altimetry
- Core ground stations: Svalbard (SG1, SG2), Matera (IT), Maspalomas (ES), Inuvik (CA), Neustrelitz (DE)
- Mission Performance Centre (MPC) per mission
- Processing & Archiving Centres (PAC) distributed in Europe
- Copernicus Data Space Ecosystem — free user access portal
- Flight Operations Segment (FOS) at ESOC, Darmstadt
Why polar / SSO dominates EO
For systematic Earth observation, sun-synchronous polar LEO is overwhelmingly preferred because:
- Global coverage — every point on Earth is observed (vs. inclined orbits which exclude high latitudes)
- Consistent illumination — every pass at same local solar time, allowing time-series comparison
- Repeating ground tracks — selectable repeat cycle (e.g. Landsat 16-day, Sentinel-2 10-day) to revisit exactly the same scene
- Low altitude — small GSD (ground sample distance) for given aperture; 700 km ≈ standard sweet spot
where λ = wavelength, h = altitude, D = aperture diameter. At 700 km, λ = 550 nm, D = 1.5 m → GSD ≈ 0.31 m.
Weather & Environmental Constellations
Operational meteorology relies on a coordinated, internationally pooled space segment combining geostationary imagers (continuous regional observation) and polar-orbiting sounders (global vertical profiles).
Geostationary weather satellites
| System | Operator | Region | Active series | Slot |
|---|---|---|---|---|
| GOES USA | NOAA / NASA | Americas | GOES-16/18/19 (R series) | 75°W (East), 137°W (West) |
| Meteosat EU | EUMETSAT / ESA | EU, Africa, Atlantic, IO | MSG & MTG-I | 0°, 9.5°E, 41.5°E (IODC) |
| Himawari JPN | JMA | APAC, W Pacific | Himawari-8/9 | 140.7°E |
| FY-2 / FY-4 CHN | CMA / CNSA | Asia | FY-2H, FY-4A/B/C | 79°E – 123°E |
| INSAT / GSAT IND | ISRO | India / IO | INSAT-3DR/3DS | 74°E, 82°E |
| GEO-KOMPSAT-2A KOR | KMA / KARI | E Asia / W Pacific | GK-2A | 128.2°E |
| Elektro-L RUS | Roshydromet / Roscosmos | EU, Africa, MENA, Asia | Elektro-L 2/3/4 | 14.5°W, 76°E, 165.8°E |
Polar weather satellites
- NOAA-15/18/19, NOAA-20, NOAA-21 (JPSS) — USA, SSO ~833 km
- Suomi NPP — joint NASA/NOAA, SSO 824 km, VIIRS imager
- MetOp-A/B/C — EUMETSAT, SSO 817 km (MetOp-SG follows)
- Meteor-M N2-2/N2-3 — Russia, SSO ~830 km
- FY-3 — China, SSO ~830 km
Geostationary satellites deliver high-frequency (10-min full disk) imagery for nowcasting and severe-weather monitoring, but they have weak performance at high latitudes (severe oblique angles) and cannot perform high-quality vertical sounding. Polar SSO satellites pass over each location twice a day at consistent local times and carry sounders (microwave + IR) that produce the temperature and humidity profiles that numerical weather prediction models depend on. Both regimes are essential and complementary — neither replaces the other.
The WMO Integrated Global Observing System (WIGOS) coordinates the international weather satellite fleet, including the CGMS (Coordination Group for Meteorological Satellites) which manages backup arrangements (e.g. Meteosat-IODC over the Indian Ocean during NOAA gaps, cross-calibration of imagers, common data formats and dissemination).
Specialised environmental constellations
- GPM (Global Precipitation Measurement) — NASA/JAXA core observatory + 9 partner radiometers
- A-Train / Afternoon Constellation — historically Aqua, Aura, CALIPSO, CloudSat (legacy formation flying)
- COSMIC-2 / FORMOSAT-7 — 6 sats for GNSS radio-occultation soundings
- Spire — commercial GNSS-RO + AIS/ADS-B (~100+ cubesats)
- GeoCarb / Carbon Mapper / MethaneSAT — greenhouse-gas focused missions
Military, ISR & Space Domain Awareness
Many military constellations operate in the same regimes and use the same architectures as civilian systems but with hardened links, classified payloads, and dedicated control segments. Public information is necessarily limited.
| System | Mission | Operator | Regime | Status |
|---|---|---|---|---|
| WGS | Wideband mil-comms (X & Ka) | USSF | GEO | 10+ ops |
| AEHF / Milstar | Protected EHF strategic comms | USSF | GEO | Ops (AEHF replacing Milstar) |
| MUOS | UHF tactical narrowband | US Navy / USSF | GEO + IGSO | 5 ops |
| SBIRS | Missile warning IR | USSF | GEO + HEO | Operational |
| Next-Gen OPIR | Missile warning successor | USSF | GEO + Polar | Deploying |
| SDA Tranche 0/1/2 | Proliferated LEO transport & tracking | SDA / USSF | LEO | Tranche 1 deploying |
| NRO ELINT/IMINT | Various reconnaissance | NRO | LEO/GEO | Classified |
| Lotos / Pion-NKS | SIGINT / ELINT | VKS / Roscosmos | LEO | Operational |
| Yaogan | Reconnaissance / SIGINT | PLASSF | LEO | Operational |
| Skynet | Mil X-band | UK MoD / Babcock | GEO | Operational |
| Syracuse | Mil X / Ka comms | French DGA | GEO | Operational |
| SATCOMBw | Mil X / Ka | Bundeswehr | GEO | Operational |
| Cosmo-SkyMed | X-SAR (dual-use) | ASI / IT MoD | SSO | 2nd-gen ops |
| SAR-Lupe / SARah | X-SAR (mil) | Bundeswehr | SSO | SARah deployed |
Proliferated LEO — the SDA model
The US Space Development Agency's Proliferated Warfighter Space Architecture (PWSA) represents a doctrinal shift: instead of a few exquisite, high-value satellites (which become single points of failure and high-value adversary targets), the PWSA fields hundreds of smaller, cheaper LEO satellites organised in functional layers:
- Transport Layer — Link 16 + optical mesh routing
- Tracking Layer — Missile warning & missile tracking IR
- Custody Layer — Maritime / mobile target tracking
- Battle Management Layer — On-board processing & decision support
- Navigation Layer — PNT alternative
- Deterrence Layer — SDA / cislunar awareness
Space Domain Awareness (SDA / SSA)
Tracking the space environment — debris, active satellites, anomalous behaviour — is itself a constellation problem. Sensors are split between ground (radars, telescopes) and space:
- Space Surveillance Network (SSN) — US (~30 sites)
- Space Fence — S-band phased array, Kwajalein
- GEODSS — optical tracking (deep space)
- EU SST — European consortium (radars + telescopes)
- LeoLabs — commercial S-band radar network
- ExoAnalytic, NorthStar, Slingshot — commercial SDA
- GSSAP — USSF GEO surveillance (4+ sats, near-GEO drift)
- NEOSSat — Canadian asteroid + RSO tracking
- Sapphire — Canadian DND deep-space tracker
- USA-326 (NROL-87) — speculated SDA mission
- Inspection / RPO demonstrators (commercial & mil)
Space Segment Design
Designing the space segment of a constellation is a coupled optimisation of orbital geometry, payload capability, satellite bus design, and operational concept. The objective is to deliver a defined level of service (coverage, capacity, latency, revisit) with the smallest, simplest fleet that closes the mission and business case.
10.1 Coverage analysis
Coverage is computed by simulating the constellation over a representative time window and counting how often each point on Earth is visible above the minimum elevation angle to at least the required number of satellites. The key metrics:
Fraction of time any given point sees ≥N satellites. ≥1 for comms, ≥4 for GNSS positioning, ≥2 for redundancy / handover.
Maximum gap between satellite passes over a target. Drives EO constellation sizing. Function of altitude, inclination, swath width, and number of sats.
For GNSS: GDOP, geometric dilution of precision. For comms: link margin and minimum elevation angle (typ. 20° – 40°).
10.2 Constellation geometry — the design knobs
| Parameter | Symbol | Typical drivers | Effect |
|---|---|---|---|
| Total satellites | T | Coverage, capacity, redundancy | Cost ↑↑ |
| Number of planes | P | Latitude coverage, launch strategy | Replenishment cost ↑ |
| Sats per plane | S = T/P | In-plane gap, ISL connectivity | Drives in-plane handover rate |
| Inclination | i | Latitude band coverage | ≈90° for global, low-i for tropics-only |
| Altitude | h | Latency, footprint, lifetime, drag | Determines required T for given coverage |
| Phasing factor | F | Inter-plane geometry | Affects max/min sat-sat distance |
| Eccentricity | e | HEO dwell vs. circular orbit | Non-zero only for HEO/Molniya/Tundra |
| Argument of periapsis | ω | For HEO: sets apogee latitude | Frozen at +270° or +90° for stability |
10.3 Spare strategy
Extra satellites pre-positioned in (or near) operational planes so that a failure can be replaced in days, not months. Used by GPS (1 spare per plane), Galileo (2 spares per plane), Iridium (1 spare per plane), Inmarsat (1 hot spare in orbit).
- Pro: rapid recovery, predictable service availability
- Con: higher fleet cost, ageing while idle
Satellites ready on the ground for launch within weeks of a failure. Suited to systems with flexible launch access and lower service-availability requirements (most EO constellations, cubesat fleets).
- Pro: lower idle inventory cost
- Con: replenishment latency = launch lead time
10.4 The satellite bus
A constellation is only viable if the satellite bus is reproducible at scale. Constellation buses share a few defining characteristics:
- Series production — Starlink: 1,000s/yr; OneWeb: 2/day at peak
- Standardised mechanical & electrical interfaces for stack launch
- Electric / Hall-effect propulsion (krypton or xenon) for station-keeping & maneuvering
- Software-defined radio payloads for over-the-air capability updates
- Autonomous collision avoidance using on-board ephemeris & ground-supplied conjunction data
- Atmospheric demise EOL (LEO) — passive deorbit within 5 yrs
10.5 Inter-satellite links (ISLs)
ISLs change the constellation from a "satellite + gateway" architecture to a routed network in space. The trade-offs:
| Property | Optical (laser) ISL | RF (Ka / V / W) ISL |
|---|---|---|
| Data rate | 10 – 100+ Gbps | 100 Mbps – 10 Gbps |
| Power efficiency | Excellent (narrow beam) | Moderate |
| Pointing tolerance | Tight (μrad) | Loose (mrad) |
| Acquisition time | Seconds (challenging) | <1 s typical |
| Regulatory (spectrum) | None | ITU coordination required |
| Examples | Starlink v1.5+, OneWeb v2, SDA | Iridium, Globalstar, Spaceway-3 |
10.6 Station-keeping & collision avoidance
Constellations operate within tight orbital tolerances (slot boxes) to maintain coverage and avoid mutual interference. Drivers and mitigations:
- Atmospheric drag (LEO) — strongest below 600 km, varies with solar activity
- J2 oblateness — drives RAAN regression and apsidal precession
- Solar & lunar gravity — dominant at GEO
- Solar radiation pressure — relevant for high-area/mass satellites
- Earth tides & non-spherical gravity — small but cumulative
- LEO: regular drag-make-up burns, electric propulsion, autonomous CAM
- GEO: NS station-keeping (lunar/solar), EW (longitude drift), figure-8 box
- MEO/GNSS: less aggressive, mainly clock & ephemeris updates
- Conjunction screening: 18 SDS, EU SST, commercial (LeoLabs, Slingshot)
- End-of-life disposal: 25-yr (now 5-yr) LEO rule, GEO graveyard +300 km
Ground Segment Design
The ground segment is the half of the constellation that pays for the rest. It commands and controls the satellites, ingests payload data, distributes products to users, and meets the regulatory and cyber-security obligations of operating a space asset. A weak ground segment will compromise even an excellent space segment.
11.1 Functional architecture
11.2 Ground station network — TT&C
Telemetry, Tracking and Command (TT&C) stations are the lifeline to the satellite. They downlink housekeeping telemetry, perform two-way ranging and Doppler tracking for orbit determination, and uplink commands. Most operate in S-band (2 GHz region) for compatibility with CCSDS standards.
For a LEO mission you need ground stations distributed in latitude and longitude to bound the maximum time between contacts. Polar stations (Svalbard 78°N, McMurdo 78°S, Inuvik 68°N, Troll 72°S) see every SSO orbit, so they are heavily favoured for EO and weather data downlink. Equatorial stations serve LEO inclined orbits and GEO uplinks.
- Parabolic dish, motorised mount (Az/El or X/Y) — TT&C, payload data
- Phased array — multiple simultaneous beams, expensive but scalable (Starlink, AWS Ground Station alternatives)
- Large reflector + multi-feed — DSN-style multi-mission
- Small flat-panel ESA — user terminals (Dishy, OneWeb terminals)
- Optical ground station (OGS) — laser comms downlink, weather-dependent
11.3 Mission Operations Centre (MOC)
The MOC is where the satellites are flown. Core functions include:
- Flight Dynamics — orbit determination, manoeuvre planning, station-keeping, conjunction analysis
- Mission Planning — pass scheduling, payload tasking, contact-window management
- Real-time Operations — TM monitoring, anomaly response, command execution
- Spacecraft Engineering — subsystem health, trending, FMECA, software upload
- Configuration Management — flight software baselines, command databases (MIB)
11.4 Network Operations Centre (NOC)
For comms constellations, the NOC sits between the MOC (which flies the satellites) and the customers (who consume the service). The NOC manages:
- End-to-end service availability and quality (QoS, KPIs, SLA tracking)
- Capacity allocation across beams / spot beams / gateways
- Customer activation, terminal registration, billing data
- Inter-gateway handover and routing in ISL networks
- Cyber security monitoring and incident response
11.5 Gateway architecture for HTS & LEO comms
Modern HTS GEO satellites and LEO comms constellations have many gateway sites — often dozens — because:
- The Ka/V-band feeder links require line of sight in good weather (rain attenuation is severe)
- Ka/V terrestrial spectrum is heavily reused — gateway must be sited where coordination allows
- Throughput per gateway is finite, so total system capacity scales with the number of gateways
- Geographic diversity provides site diversity against weather (a Ka gateway is paired with another 30 – 100 km away to swap beams during local storms)
A modern Ka HTS GEO satellite delivering 1 Tbps may use 20 – 40 gateway sites; a global LEO constellation may use 50 – 200+ gateway sites worldwide.
11.6 Ground-Station-as-a-Service (GSaaS)
A new class of provider rents access to globally distributed antennas on demand, removing the capital burden of building a TT&C network for small operators:
| Provider | Stations | Bands | Notes |
|---|---|---|---|
| KSAT (Kongsberg) | ~270 antennas across 26+ sites | VHF, UHF, S, X, Ka | Largest commercial network; Svalbard core |
| Viasat Real-Time Earth (ex-RBC Signals + Inmarsat) | 30+ sites | UHF, S, X, Ka | Pay-per-pass model |
| AWS Ground Station | 10+ regions, AWS-integrated | UHF (limited), S, X | Cloud-native pipeline |
| Microsoft Azure Orbital | Partner network (KSAT, Viasat) | S, X, Ka | Azure-integrated |
| Atlas Space Operations | ~30 sites (Freedom network) | S, X, Ka | Software-defined ground |
| Leaf Space | ~20 sites (Europe + global) | S, X | EU-focused, smallsat heritage |
| SSC (Swedish Space Corp) | 10+ sites | S, X, Ka | Esrange (SE) + Inuvik, Punta Arenas, etc. |
| Goonhilly (UK) | Goonhilly Earth Station + partners | L, S, C, X, Ku, Ka | Deep-space & lunar comms support |
11.7 Standards & interoperability
The Consultative Committee for Space Data Systems (CCSDS) publishes the standards that make cross-support possible — i.e. one operator's satellite can be tracked or downlinked by another's ground station with minimal friction. Key documents:
- CCSDS 401.0-B — RF and Modulation systems (S/X/Ka)
- CCSDS 131.0-B — TM Synchronization & Channel Coding
- CCSDS 132.0-B — TM Space Data Link Protocol
- CCSDS 232.0-B — TC Space Data Link Protocol
- CCSDS 727.0-B — CFDP (file transfer)
- CCSDS 401.0-B-31 — Optical communications (recent)
- SLE (Space Link Extension) — cross-support interface for ground stations
- ECSS — European space engineering standards (E-ST-50 series for comms)
11.8 Cybersecurity considerations
The ground segment is the dominant attack surface for a constellation. Recent doctrine (NIST SP 800-series, ESA SSE, CCSDS 350-series Security Architectures) emphasises:
- Authenticated & encrypted commanding — mandatory for all military and increasingly civil missions
- Segregated networks for command, telemetry, payload data, and corporate IT
- Zero-trust architecture for cloud-integrated ground stations
- Insider threat mitigations (two-person rule for commanding, command audit)
- Supply chain security for ground software and FPGA payloads
- Resilience to RF interference (jamming, spoofing — especially of GNSS uplinks & downlinks)
Frequency Allocation & Regulatory Framework
Spectrum is the most contested resource in space systems. Every constellation depends on an ITU filing that secures the right to transmit on a specific frequency band, in a specific orbit, under a specific service category. The choice of band drives antenna size, weather resilience, throughput, and ultimately the entire architecture.
12.1 Frequency bands & their applications
The following are the principal radio bands used by operational constellations. Higher frequency generally means more bandwidth (capacity) but greater rain attenuation and tighter pointing requirements.
| Band | Frequency | Wavelength | Primary Uses | Trade-offs |
|---|---|---|---|---|
| VHF | 30–300 MHz | 1–10 m | Orbcomm, AIS, amateur (AMSAT), legacy TT&C | Penetrates buildings; very low data rate; ionospheric effects |
| UHF | 300 MHz–3 GHz | 10 cm–1 m | Tactical milcom (MUOS UHF follow-on), CubeSat TT&C, Iridium gateway feeders, Globalstar | Reliable through foliage/weather; congested; limited bandwidth |
| L-band | 1–2 GHz | 15–30 cm | GNSS (GPS L1/L2/L5, Galileo E1/E5/E6), Inmarsat, Iridium user link, Thuraya, MSS | Excellent rain immunity; very limited bandwidth; ITU-protected |
| S-band | 2–4 GHz | 7.5–15 cm | TT&C (NASA Near Earth Network), Globalstar feeders, weather radar, ISS comms | Robust; ground antennas modest size; limited capacity |
| C-band | 4–8 GHz | 3.75–7.5 cm | Legacy GEO TV/data (Intelsat, SES), tropical broadcast, VSAT in equatorial regions | Strong rain immunity; large dishes (1.8–3.7 m); 5G interference pressure |
| X-band | 8–12 GHz | 2.5–3.75 cm | Military comms, NASA Deep Space Network, EO downlink (Sentinel, Pleiades, ICEYE), SAR | Reserved for government/military & EO; good rain performance; high-rate downlink |
| Ku-band | 12–18 GHz | 1.7–2.5 cm | DTH TV, VSAT, OneWeb user links, Starlink user links, maritime/aero broadband | Smaller dishes (60–90 cm); moderate rain fade; very congested orbital arc |
| Ka-band | 26.5–40 GHz | 0.75–1.1 cm | Viasat-3, Jupiter, Inmarsat GX, Starlink gateway, Kuiper, HTS spot beams, ISL feeders | Wide bandwidth (HTS economics); significant rain fade (UPC required); precise pointing |
| Q/V-band | 33–75 GHz | 4–9 mm | Future HTS feeder links, gateway diversity (Q/V offload from Ka) | Massive bandwidth; severe rain attenuation; experimental on most platforms |
| W-band | 75–110 GHz | 2.7–4 mm | Inter-satellite links (some prototypes), atmospheric science radars | Nearly opaque to rain; only used for ISLs or short atmospheric paths |
| Optical | ~193 THz (1550 nm) | 1.55 µm | Inter-satellite laser links (Starlink, SDA, EDRS), high-rate downlink (TBIRD, OPALS) | Massive capacity; no ITU filing; cloud-blocked downlinks; precision tracking |
At Ka-band, a heavy tropical rain cell can introduce 15–25 dB of attenuation. Operators counter this with Uplink Power Control (UPC) on gateways, Adaptive Coding & Modulation (ACM, DVB-S2X), and gateway diversity — a second gateway hundreds of kilometres away picks up the link when the primary site is rained-out. At V-band the problem doubles, and gateway diversity becomes essentially mandatory.
12.2 ITU regulatory framework
The International Telecommunication Union (ITU), a UN agency based in Geneva, is the global authority for spectrum and orbital slot allocation. Its core instruments:
Treaty-level document, updated every World Radiocommunication Conference (WRC, ~every four years). Defines services, bands, and the priority between primary and secondary allocations. Article 5 contains the master frequency table. Articles 9 and 11 govern the satellite coordination and notification process.
- Region 1 — Europe, Africa, Middle East, former Soviet Union
- Region 2 — Americas, Greenland, eastern Pacific
- Region 3 — Asia-Pacific (incl. Iran, Australia, Japan)
Allocations frequently differ between regions, especially in the UHF, L, and S bands.
12.3 Filing & coordination process
Securing operational rights requires several years of regulatory work, in a defined sequence:
- Advance Publication Information (API) — submitted by an administration (a national regulator) to ITU-BR up to seven years before bringing into use
- Coordination Request (CR/C) — under Article 9.7 of the RR for GEO; Article 9.21 for non-GEO; triggers bilateral coordination with potentially affected administrations
- Notification — when satellites are launched; the network is recorded in the Master International Frequency Register (MIFR)
- Bringing into Use (BIU) — must occur within seven years of the API or the filing lapses (ITU "use it or lose it" rule)
- Milestone-based deployment (Resolution 35, WRC-19) — non-GEO megaconstellations must deploy 10% within 2 years of BIU, 50% within 5 years, 100% within 7 years
12.4 National regulators
Each country licenses its own satellites and earth stations, and represents its operators at ITU. Key regulators that constellation operators interact with:
| Region / Country | Regulator | Scope |
|---|---|---|
| USA | FCC (commercial), NTIA (federal/military) | Licensing, spectrum sharing rules, deployment milestones (47 CFR Part 25) |
| UK | Ofcom + UK Space Agency | Outer Space Act licensing, NGSO sharing |
| EU | National regulators + RSPG, CEPT/ECC | Harmonised European spectrum decisions |
| UAE | TDRA + UAE Space Agency | Satellite licensing, Yahsat/Thuraya filings, ground station authorisation |
| Saudi Arabia | CST + Saudi Space Agency | Filing coordination, gateway licensing, national priorities |
| France | ARCEP + CNES | Eutelsat coordination, French ITU filings |
| Japan | MIC | JAXA + commercial filings (e.g. SkyPerfect, QZSS) |
| China | MIIT + SARA | Guowang, Qianfan, BeiDou filings |
12.5 Polarisation & reuse
To squeeze more capacity into a limited slice of spectrum, satellite systems exploit polarisation diversity — transmitting two independent signals on the same frequency using orthogonal polarisations:
- Linear polarisation (Vertical / Horizontal) — common in fixed VSAT, DTH (Ku band)
- Circular polarisation (RHCP / LHCP) — preferred for mobile and GNSS (insensitive to terminal orientation, robust against Faraday rotation)
- Frequency reuse via spot beams — same frequency reused in non-adjacent beams, separated by guard bands or polarisation, multiplying capacity by a factor of N/k where k is the cluster size (typically 4–7)
12.6 ITU services relevant to constellations
- FSS — Fixed Satellite Service (Ku/Ka VSAT, GEO broadcast, HTS gateways)
- BSS — Broadcasting Satellite Service (DTH TV)
- MSS — Mobile Satellite Service (Inmarsat, Iridium, Globalstar, Thuraya)
- RNSS — Radio Navigation Satellite Service (GPS, Galileo, BeiDou, GLONASS)
- EESS — Earth Exploration Satellite Service (Sentinel, Landsat, ICEYE, Capella)
- SOS — Space Operation Service (TT&C uplinks & downlinks)
- ISS / inter-satellite — links between space stations or satellites
- SRS — Space Research Service (deep space, scientific)
Trade-offs & Performance Metrics
Every constellation is the resolution of a multi-dimensional optimisation problem: service quality, cost, schedule, regulatory feasibility, and resilience all pull in different directions. Understanding the dominant trade-offs is the difference between an architecture that closes and one that does not.
13.1 The five primary trade-offs
Latency is dominated by the slant-range round-trip. GEO offers global coverage with three satellites but ~500–600 ms RTT — unacceptable for real-time gaming, voice-over-IP and remote surgery. LEO at 550 km cuts that to 25–50 ms — but at the cost of a much larger fleet for continuous coverage. Each order-of-magnitude reduction in latency typically costs an order-of-magnitude increase in satellites.
For Earth observation the buyer wants both complete coverage and frequent revisit of any tasked target. A single SSO satellite gives global coverage in ~16 days; tripling the fleet does not triple the revisit because of orbit-plane constraints. Constellations like Planet's Dove flock trade individual sensor performance for a swarm that revisits everywhere daily.
HTS Ka-band gives ~100× the per-satellite capacity of legacy C/Ku at moderate incremental cost, which is why the entire commercial GEO industry has moved that way. But HTS narrows beams, which forces a denser gateway network on the ground — moving capex from the space segment to the ground segment.
A minimum-cost constellation has no on-orbit spares, perfectly phased orbits, and a small optimised gateway network. A resilient constellation has hot spares per plane, gateway diversity, multi-band ground stations, and surge replenishment launch capability. Military architectures (PWSA, AEHF) deliberately accept 30–50% higher lifecycle cost for resilience.
Iridium took 11 years from inception to first service, with a $5B price tag — and went bankrupt before achieving market fit. Starlink began offering service with <1,000 satellites, iterating in flight while the constellation grew. Modern megaconstellations are designed for early partial-service operation rather than waiting for full deployment.
Higher modulation orders (e.g. 32APSK) deliver more bits per Hz, but require higher SNR and so collapse first under rain fade. ACM/VCM systems trade modulation order in real time: nominal sunny-day 32APSK degrades to QPSK during heavy rain, sacrificing throughput to keep the link alive.
13.2 Latency benchmarks
| Architecture | Altitude | One-way Latency (user→sat→gateway) | RTT incl. ground |
|---|---|---|---|
| GEO bent-pipe | 35,786 km | ~120 ms (each leg) | ~600 ms |
| MEO (O3b mPOWER) | ~8,062 km | ~27 ms (each leg) | ~150 ms |
| LEO (Starlink ~550 km) | 550 km | ~2 ms (each leg) | 25–50 ms |
| LEO + ISL backbone | 550 km | 2 ms + ISL hops at c | 30–80 ms intercontinental (often beats fibre) |
| VLEO (~300 km, future) | 300 km | ~1 ms (each leg) | 15–30 ms |
Light moves through fibre at ~67% of vacuum speed (n ≈ 1.47). A LEO constellation with optical ISLs propagates signals at ~99.7% of c in the near-vacuum of LEO. For long-haul links — London↔Tokyo, New York↔Singapore — LEO meshes with optical ISLs already deliver lower latency than terrestrial fibre, which is why the financial trading and CDN industries are moving onto Starlink and future Lightspeed/Kuiper backbones.
13.3 Key performance indicators (KPIs)
The following metrics are how operators specify, contract, and report constellation performance:
| KPI | Definition | Typical Targets |
|---|---|---|
| Service availability | % of time the service meets contracted QoS at a given location | 99.5% (consumer broadband), 99.9% (enterprise), 99.999% (safety-of-life GNSS) |
| Coverage | % of Earth's surface (or service area) seen by ≥1 satellite at min elevation | 100% global (Starlink at full deployment), 100% >55° elev (Iridium polar) |
| Revisit time | Time between successive observation opportunities of a target | Daily (Planet Dove), 5 d (Sentinel-2 single sat), <1 hr (HawkEye-360 RF) |
| GSD (EO) | Ground sample distance — pixel size on the ground | 10–30 m (Sentinel-2), 0.3 m (Pleiades Neo, Maxar WorldView Legion), 0.15 m (BlackSky Gen-3) |
| Throughput per beam | User-side capacity within one spot beam | ~150 Mbps (Inmarsat GX), ~1 Gbps (Viasat-3 spot), 4+ Gbps (Starlink V2) |
| System capacity | Total simultaneous user-side throughput | ~1 Tbps (Viasat-3 single sat), ~30 Tbps (current Starlink fleet) |
| EIRP | Effective Isotropic Radiated Power on downlink | 50–55 dBW (Ku DTH), 60+ dBW (HTS Ka spot), 70+ dBW (military Ka) |
| G/T | Gain-to-noise-temperature ratio of receiving antenna | 20+ dB/K (large gateway), 12 dB/K (Starlink terminal), <10 dB/K (handheld L-band) |
| Link margin | Excess SNR above demodulator threshold for atmospheric/aging loss | 3–10 dB depending on band, climate & QoS class |
| Position accuracy | For GNSS, 95% horizontal positioning error | ~3 m (GPS SPS), 0.2–0.5 m (Galileo HAS/PPP), <0.05 m (RTK with corrections) |
| Time-to-first-fix | Cold-start lock time for navigation receivers | ~30 s (cold), <1 s (warm) |
| Constellation MTBF | Mean time between satellite failures | 7–15 yr design life; >25 yr achieved on some MEO/GEO |
| Replenishment cadence | Launches/year required to maintain fleet | 0–2 per yr (legacy GEO), 30+ per yr (Starlink V2 maintenance) |
13.4 Cost decomposition (orders of magnitude)
Indicative ratios for a complete constellation programme — useful for early-phase trade studies:
- Satellite bus & payload: 40–55%
- Launch services: 15–25%
- Insurance & programme reserve: 3–8%
- TT&C network & MOC: 5–10%
- Gateway/payload network: 8–18%
- NOC, OSS/BSS, data centres: 3–7%
- Staffing & sustainment: 4–8% / yr
- Replenishment launches: variable
- Spectrum & regulatory: 1–3%
Typical programmes allocate ~15–25% of capex to the ground segment but face most of their operational risk there. Cyber breaches, gateway outages, and TT&C blackouts have caused more customer-visible service interruptions in the past decade than any space-segment failure. Modern constellation design treats ground architecture, cyber posture, and ground-station diversity as first-order trade variables, not afterthoughts.
13.5 Choosing an architecture — a decision flow
A simplified architectural decision tree, useful as a sanity check on early-phase trade studies:
- What is the service? Persistent broadcast → GEO. Real-time low-latency → LEO. Navigation → MEO. Niche regional → IGSO/HEO. Imaging → SSO/LEO.
- What latency budget? <50 ms → LEO mandatory. <200 ms → MEO acceptable. >500 ms tolerable → GEO viable.
- Coverage area? Equatorial only → GEO most efficient. Polar inclusive → LEO/MEO/HEO with high inclination. Single country → IGSO or regional GEO arc.
- Capacity per user? Mass-market broadband → HTS Ka or LEO Ku/Ka with thousands of spot beams. Handheld voice/SMS → MSS L/S band. Niche broadcast → C/Ku DTH.
- Resilience requirement? Safety-of-life or military → multi-orbit, multi-band, proliferated. Commercial best-effort → cost-optimised single-orbit.
- Spectrum availability? Crowded Ku/Ka in equatorial GEO arc → look at V/Q-band, LEO, or optical ISLs to relieve feeder pressure.
- Time-to-market? <3 yr to revenue → leverage existing buses, hosted payloads, rideshare. >5 yr available → custom platform with high performance ceiling.
- Regulatory feasibility? ITU coordination feasible in target band, country licensing achievable, BIU/milestone schedule realistic.
13.6 The proliferation paradigm
The 2020s shift from exquisite (few, expensive, long-life) to proliferated (many, cheap, short-life) is the dominant architectural trend across both commercial and military space. Its premises:
- Reduced cost-per-bit and cost-per-pixel through volume manufacturing
- Graceful degradation under attack or random failure
- Continuous tech refresh — every refresh cycle deploys current-gen technology
- Lower per-satellite radiation/reliability standards (shorter design life)
- Enables megabit/pixel-class economics impossible with bespoke buses
- Massively higher launch cadence and ground station load
- Spectrum congestion & ITU coordination complexity
- Orbital debris and conjunction-management burden
- Per-satellite capability ceiling (size, power, aperture)
- Astronomy & dark-sky impact
The synthesis emerging from operational experience: hybrid multi-orbit architectures — exquisite GEO/MEO assets for legacy services and bandwidth-heavy broadcast, proliferated LEO/VLEO for low-latency communications and high-revisit observation, with seamless ground-segment integration so users transparently consume capacity from whichever layer is optimal at each moment. Inmarsat-Viasat, SES O3b mPOWER + GEO, Eutelsat-OneWeb, and the SDA Proliferated Warfighter Space Architecture are all expressions of this convergence.
The most common architectural pitfall is treating constellation design as "how many satellites do we need?". The right framing is: what is the minimum-cost end-to-end system — orbit, payload, ground network, spectrum, software, operations, replenishment plan, regulatory pathway — that meets the contracted service level over the programme lifecycle? The number of satellites is an output of that optimisation, not an input.