Chapter 3
WIA-COMM-005 (version 1.0.0, category COMM) is a framework for satellite internet systems, and it starts from orbital geometry. It treats broadband from space as a network whose behaviour follows from altitude, inclination and the number of satellites in each plane. Latency, handover rate, ground-station count and terminal design all follow from those choices. The standard covers LEO constellations of the Starlink, OneWeb and Kuiper kind, MEO systems, and traditional GEO satellites. Its stated aim, 홍익인간 (弘益人間, "benefit all humanity"), is access for remote, rural, maritime and aviation users who terrestrial networks do not reach.
The three orbit classes are defined by altitude, and the choice sets the latency floor:
| Orbit | Altitude | Round-trip latency | Typical role |
|---|---|---|---|
| LEO | 500–2000 km | 20–40 ms | Broadband, real-time applications |
| MEO | 2000–35,786 km (the orbit-parameter section uses 8,000–20,000 km) | 100–150 ms | Navigation, regional fill, LEO backup |
| GEO | 35,786 km | 500–600 ms | Broadcast, legacy maritime and aviation, emergency backup |
Modern systems are hybrid. LEO carries low-latency user access, MEO fills regional coverage, and GEO provides broadcast and backup. A GEO satellite sits at a fixed point in the sky, with a period of 23 hours 56 minutes, so a handful of them cover the globe. A LEO satellite crosses the sky in minutes, so a continuous service needs hundreds to tens of thousands of them.
The standard describes a LEO constellation by a few numbers. The reference ranges are 20–80 orbital planes, 20–80 satellites per plane, and 300 to more than 40,000 satellites in total, with an orbital period of roughly 90–100 minutes. Quoted altitudes are 550 km for Starlink, 600 km for OneWeb and 590 km for Kuiper. Inclination selects the coverage pattern:
Phasing spreads satellites evenly so that coverage has no gaps. The standard gives the spacing as 360° divided by satellites per plane, and the phase angle between planes as 360° scaled by plane number over total planes. Its worked example is a Phase 1 Starlink-style shell: 72 planes of 22 satellites each, 1,584 satellites in all, with 5° between planes. The same figures appear in the TypeScript SDK constructor, which is how a constellation descriptor reaches code:
const network = new SatelliteNetwork({
constellation: 'LEO',
altitude: 550, // km
inclination: 53, // degrees
totalSatellites: 1584,
orbitalPlanes: 72
});Whether a user can use a satellite depends on its minimum elevation angle, which the standard puts at typically 25–40°. The geometry is expressed through Earth radius (6,371 km) and satellite altitude. A higher minimum elevation shrinks the usable footprint and raises the number of satellites needed. The SDK exposes this as getVisibleSatellites, which takes a latitude, a longitude and a minElevation.
Satellite links are limited by power, distance and weather. The standard therefore makes the link budget a first-class object. Its summary form is:
EIRP (dBW) = Tx Power (dBW) + Antenna Gain (dBi) - Losses (dB)
Path Loss (dB) = 20 log10(d) + 20 log10(f) + 92.45
SNR (dB) = EIRP - Path Loss - Noise Figure + G/T
Capacity (Mbps) = Bandwidth x log2(1 + SNR)Here d is the distance in km, f the frequency in GHz, and G/T the gain-to-noise-temperature ratio. The detailed downlink budget, satellite to user, subtracts the losses one by one:
P_rx = EIRP_sat - L_path - L_atm - L_rain + G_rx - L_rxThe reference values show where the margin goes. Satellite EIRP is 50–60 dBW, and free-space loss is 165–175 dB. Atmospheric absorption costs 0.5–2 dB. Rain fade costs 1–10 dB depending on frequency and climate, and receiver losses cost 1–2 dB. The user terminal contributes 35–40 dBi of gain. Received power lands between −90 and −110 dBm. Noise is thermal, N = kTB, with a system noise temperature of 100–300 K. Typical SNR is 15–25 dB in clear sky and 5–15 dB in rain fade. The uplink uses the same structure with a terminal EIRP of 35–45 dBW and a satellite antenna gain of 25–30 dBi.
Rain fade is why the standard includes adaptive coding and modulation (ACM). The terminal modem steps down through the DVB-S2X modulations (QPSK, 8PSK, 16APSK, 32APSK) with LDPC coding. The standard's illustration is:
| Condition | Modulation / code rate | Throughput |
|---|---|---|
| Clear sky | 32APSK, 9/10 | 500 Mbps |
| Light rain | 16APSK, 3/4 | 300 Mbps |
| Heavy rain | QPSK, 1/2 | 100 Mbps |
A link therefore degrades in steps instead of dropping. For validation the standard asks for a margin of 3–6 dB that accounts for rain fade and pointing error.
The frequency bands carry different roles:
| Band | Range | Role in the standard |
|---|---|---|
| Ku | 12–18 GHz | Legacy and GEO systems, backup |
| Ka | 26.5–40 GHz | Primary for LEO user and gateway links; uplink 27.5–30.0 GHz, downlink 17.7–20.2 GHz |
| V | 40–75 GHz | Future high-throughput; uplink 47.2–50.2 GHz, downlink 37.5–42.5 GHz |
| Q/V | 40–50 GHz | Gateway feeder links |
Doppler shift is the other physical effect a LEO terminal has to handle. A satellite moving at about 7.59 km/s at 550 km produces Δf = (v/c) × f₀ × cos θ. At a 28 GHz carrier the maximum is about ±710 kHz. The shift also changes quickly, at roughly 10–50 kHz per second, so the terminal needs continuous tracking and cannot correct once at link setup. The SDK and CLI both expose the calculation. The CLI form is wia-comm-005 doppler --velocity 7500 --frequency 28000.
The budget also has a software form. calculateLinkBudget takes distance, frequency, txPower, antennaGain and bandwidth, and returns snr and capacity. A call for a 600 km Ka-band link at 28 GHz with 250 MHz of bandwidth is the SDK's own example. On the wire, link-budget telemetry carries transmit power, antenna gain, free-space path loss, atmospheric loss, rain-fade margin and noise figure. For active links it is reported about once a minute so operators can see fading events as they develop.
A LEO satellite is visible to a given user for only 4–7 minutes at 550 km, depending on elevation angle. Handover is therefore a routine part of the service, not an exception. The standard bases prediction on visibility duration, t_visible = 2 × arccos(R_e / (R_e + h)) / ω, where ω is the satellite's angular velocity. It distinguishes three handover types:
The handover protocol has three phases. In the measurement phase, about 30 s before the switch, the terminal monitors the current satellite's SNR, scans for the next one and measures Doppler. In the decision phase, about 10 s before, it selects the best candidate, reserves resources and prepares beam steering. In the execution phase it switches the antenna beam, re-tunes for Doppler, and keeps TCP connections alive without packet loss. The integration guidance sets the verification target at an interruption below 100 ms. The SDK wraps the prediction:
const handover = predictHandover({
currentSatelliteId: 'SAT-1234',
userLatitude: 37.7749,
userLongitude: -122.4194,
elevationAngle: 25
});
// handover.timeRemaining, handover.nextSatelliteIdThe HTTP interface exposes the same operations. Integrators code against four endpoints, all discoverable through /.well-known/wia-satellite-internet:
POST /satinternet/v1/terminal/register # register a user terminal
GET /satinternet/v1/coverage/{lat}/{lon} # coverage at a location
POST /satinternet/v1/handover/request # request a handover
GET /satinternet/v1/link/{id}/budget # link-budget telemetryThe coverage endpoint takes a location and returns the visible satellites with predicted elevation, predicted link quality and the next visibility window. Terminals use this to point their antennas, and dashboards use it to show where service is about to improve or degrade.
The latency numbers explain why the standard puts so much effort into the path. At 550 km the propagation delay for the round trip is about 3.7 ms, so most of the 20–40 ms total comes from elsewhere: 5–10 ms of ground processing and 5–15 ms of network routing. The optimisations therefore work on the path and not on the orbit. Inter-satellite links route traffic through space and avoid repeated ground hops, which the standard credits with a 50–70% latency reduction. CDN nodes at gateways, and content cached in space, shorten the round trip further. On the transport side, TCP BBR, QUIC and custom congestion control all cope better with the variable delay of a moving constellation than classic loss-based TCP.
Three physical segments carry the traffic between a user and the internet: the gateways, the user terminal, and the links between satellites.
Gateways and teleports. A gateway uses a 3–13 m antenna on Ka-band (27.5–30 GHz up, 17.7–20.2 GHz down) at 10–100 Gbps, and a global system needs roughly 10–50 of them. They provide internet backbone connectivity, traffic aggregation, network management, and telemetry and control. A teleport has three layers. The RF layer holds high-power amplifiers, low-noise amplifiers and frequency converters. The baseband layer does modulation (DVB-S2X, 5G NR), error correction (LDPC, turbo codes) and traffic shaping. The network layer has a fibre backbone of 100 Gbps or more, CDN peering and internet exchange points. The standard recommends placing gateways near IXPs, with low latency to major data centres, geographically diverse, and in regulatory-friendly jurisdictions. Gateway location is therefore a latency decision and also a regulatory one.
User terminals. The terminal is a phased-array antenna, 30–60 cm across in Starlink-class designs and 50–70 cm for OneWeb-class. It has 1,000–2,000 elements and steers electronically, with no moving parts, over ±60° from zenith, in dual circular or linear polarisation. The steering relation is θ = arcsin(λ × Δφ / 2πd), where Δφ is the phase shift between elements and d is the element spacing. Electronic steering lets the terminal retarget between satellites within a handover window, which a mechanical dish could not do. The modem uses DVB-S2X modulations, LDPC coding with rates from 1/4 to 9/10, and symbol rates of 100–500 Msps. Reference performance is 100–500 Mbps down and 20–50 Mbps up. Power draw is 20–50 W idle, 50–150 W active and 100–200 W at peak, which matters for off-grid and maritime installations. A new terminal enters the network through POST /satinternet/v1/terminal/register.
Inter-satellite links. ISLs turn the constellation into a mesh. Optical ISLs use 1550 nm, carry 1–100 Gbps per link over up to 5,000 km, and have beam divergence under 10 microradians. The standard lists their advantages as 10–100 times the bandwidth of RF, no spectrum licensing, minimal interference, and resistance to interception. Pointing loss enters the budget explicitly: P_rx = P_tx + G_tx + G_rx − L_space − L_point. RF ISLs are the alternative, at Ka-band 23–27 GHz (100 Mbps–1 Gbps), V-band 60 GHz (1–10 Gbps) or E-band 71–76 GHz. A mesh reduces dependence on ground stations, shortens paths through space routing, adds resilience and gives coverage where no ground infrastructure exists. The routing protocols named are OSPF-based routing, BGP between constellations, and DTN (delay-tolerant networking) for deep-space cases.
Two rule sets limit what an operator may build: how spectrum is used, and what happens to a satellite after its mission.
On spectrum, the standard follows ITU filing practice. The steps are advance publication about 7 years before launch, a coordination request about 5 years before launch, notification within 30 days of operation, and recording in the ITU Master Register. The regulatory chapter cites the ITU Radio Regulations by article: coordination (Article 9), notification and recording (Article 11) and operational procedures (Article 22). Interference is managed by frequency reuse (orthogonal polarisations, spot beams, geographic isolation) and by dynamic spectrum management with real-time monitoring and adaptive power control. Coexistence with GEO uses off-axis EIRP limits and exclusion zones of ±2° around the GEO arc. National licensing then runs per jurisdiction: FCC filings through IBFS with milestone requirements, and Ofcom spectrum access licences in the UK. The standard also lists MIC (Japan), KCC (Korea) and EU member-state regulators. An operator declares its licensed jurisdictions in the discovery document, and data envelopes carry a jurisdiction identifier so compliance can be traced per region.
On debris, LEO satellites must deorbit within 25 years of end of mission. Active deorbit uses remaining propellant for a controlled reentry in 1–12 months. Passive decay by atmospheric drag suits satellites below about 600 km, in 1–5 years. A GEO satellite is raised at least 300 km into a graveyard orbit. Collision avoidance relies on daily screening against the TLE catalogue, an alert when collision probability exceeds 10⁻⁴, and a manoeuvre when miss distance falls below 1 km. Design for demise limits surviving components to under 5 kg by choosing aluminium and composites over titanium or inconel, and passivates propellant tanks to prevent on-orbit explosions. The CLI check wia-comm-005 debris-check --altitude 550 --inclination 53 applies these rules to a proposed orbit.
Summary. This chapter has followed one thread, from the shape of the orbit to the shape of the service. Altitude and inclination fix latency, visibility and satellite count. The link budget, Ka-band choice and ACM keep a link running through weather. Doppler of up to ±710 kHz at 28 GHz, and visibility windows of 4–7 minutes, mean terminals must track and hand over continuously. Gateways, phased arrays and ISL meshes make the path physical. Filing and debris rules shape how much of this an operator may deploy. A short implementation checklist from the standard ties these together: define the orbit, size the constellation, design the budgets, file with the ITU, build the ground network and terminal, implement handover, plan deorbit, and obtain approvals.
The next chapter turns from physics to records. It looks at how the standard's canonical envelopes represent constellation descriptors, inter-satellite link records, and ground-station and terminal data. These are the structures through which the quantities introduced here (altitude, EIRP, rain-fade margin, elevation, jurisdiction) cross organisational boundaries.