Chapter 4
A satellite-internet operator has to describe a moving network in a form that other parties can read without phoning the operator. That network includes thousands of spacecraft, laser links that open and close with orbital geometry, gateways on several continents, and terminals on rooftops, ships and aircraft. Phase 1 of WIA-COMM-005 handles this by fixing the data format. It defines the canonical envelopes for constellation descriptors, inter-satellite link (ISL) records, ground-station records and user-terminal records. Later phases carry those envelopes over interfaces and protocols.
The standard covers three orbit classes, and each places different demands on a descriptor. LEO sits roughly between 500 and 2000 km, MEO between 2000 and 35,786 km, and GEO at 35,786 km. The specification uses the same short vocabulary throughout, so the terms are worth fixing first.
The wider scope of the standard includes link budgets and frequency planning, latency optimisation, handover and mobility management, regulatory compliance and debris mitigation. Phase 1 turns the facts behind each of these into records. An altitude and an inclination become a constellation descriptor. A laser carrier, a transmit power and a loss figure become an ISL budget. A terminal's last-known link metrics become something a monitoring system can compare across a fleet.
The stated aim is to reach remote, rural, maritime and aviation users who terrestrial networks do not serve. Those are the users whose terminals move, whose gateways are far away, and whose service depends most on orbital geometry. That is why the format records orbital facts and not only commercial ones.
The central Phase 1 record is the constellation descriptor. It states the orbit tier (leo, meo or geo), the satellite count, the altitude, the inclination, the minimum elevation angle that users are expected to operate at, and the date of first launch. The reference example is one orbital shell of a LEO system:
{
"wia_satinternet_version": "1.0.0",
"type": "constellation_descriptor",
"constellation_id": "starlink-shell-1",
"tier": "leo",
"satellite_count": 1584,
"altitude_km": 550,
"inclination_deg": 53.0,
"expected_minimum_elevation_deg": 25.0,
"isl_capable": true,
"first_launch_at": "2019-11-11T12:56:00Z",
"operator_id": "did:wia:satinternet:operator-A",
"signature": { "alg": "Ed25519", "value": "..." }
}A real constellation usually publishes one descriptor per orbital shell, and a consumer sums across shells to work out end-to-end coverage at a given location. A single constellation-wide record would hide the differences between shells that matter at the ground. Shells can differ in altitude, in inclination and in whether they support ISLs.
The numbers in a descriptor are connected by simple relationships that a reader can check. For uniform distribution across planes, the specification gives:
Phase Angle = 360 deg x (plane_number / total_planes)
Satellite Spacing = 360 deg / satellites_per_planeThe Phase 1 example for Starlink has 72 planes with 22 satellites each. That gives 72 × 22 = 1,584 satellites, a plane-to-plane step of 5°, and an in-plane spacing of about 16.4°. A descriptor whose satellite_count contradicts this kind of arithmetic is a data-quality problem that a validator can catch without any orbital simulation.
Inclination shapes where the capacity goes. At 87-90° the constellation covers the whole globe including the poles, which suits scientific, military and emergency use. At 53-70° coverage is concentrated on the latitudes where most people live, which suits commercial broadband. The typical LEO figures in the specification are altitudes of 550 km, 600 km and 590 km, orbital periods of about 90-100 minutes, and constellation sizes from a few hundred to tens of thousands of satellites.
The expected_minimum_elevation_deg field links the descriptor to what a user actually sees. A satellite is usable only when it is above a minimum elevation angle, typically 25-40°. The specification gives the relation cos(θ) = R_e / (R_e + h) with Re = 6,371 km, and a coverage-radius expression built on arccos[Re·cos(θ) / (Re + h)]. The relation as written corresponds to the zero-elevation horizon. Implementers should be aware that in the usual spherical-Earth derivation, the Earth-central angle for a given elevation is that arccos term minus the elevation angle itself, and the ground radius is Re times that angle. A higher minimum elevation shrinks each satellite's footprint, so the satellite count needed for continuous coverage grows. That trade-off is why this angle sits in the descriptor and not in a separate configuration file.
Orbit data is carried in two forms. The Two-Line Element set is kept for legacy compatibility. The CCSDS Orbital Ephemeris Message (502.0-B) is used for high-precision propagation. A consumer that only needs a rough pass prediction can use the TLE. One that has to plan handovers or ISL pointing should use the OEM.
ISLs let traffic cross the constellation in space instead of descending to a gateway after each hop. The specification describes two families. The optical family uses a 1550 nm laser, 1-100 Gbps per link, ranges up to 5,000 km and a beam divergence under 10 microradians. Its advantages are high bandwidth (stated as 10-100 times RF), no spectrum licensing, minimal interference, and resistance to interception. The RF family uses Ka-band (23-27 GHz) at 100 Mbps to 1 Gbps, V-band (60 GHz) at 1-10 Gbps, and E-band (71-76 GHz).
Whether either link works is a link-budget question. The specification writes the received power as:
P_rx = P_tx + G_tx + G_rx - L_space - L_pointHere the gains are in dBi, the powers are in dBm, Lspace is free-space loss and Lpoint is pointing loss. With divergence of only a few microradians, pointing loss is what dominates on a laser link. Two spacecraft moving at orbital velocity must keep a very narrow beam aligned over thousands of kilometres. That is also why a link record has a short lifetime and not a permanent one.
The isl_link_budget envelope carries the outcome of this calculation:
{
"type": "isl_link_budget",
"link_id": "isl_01HX...",
"satellites": ["sat-A", "sat-B"],
"carrier": "1550nm-laser",
"transmit_power_dbm": 35,
"receiver_sensitivity_dbm": -40,
"free_space_loss_db": 75,
"atmospheric_loss_db": 0,
"fade_margin_db": 8,
"link_quality_db": 12,
"rate_gbps": 100,
"valid_until": "RFC 3339"
}Some points about this record are worth stating. atmospheric_loss_db is 0 in the example because a link between spacecraft does not pass through the atmosphere, which is a difference from a gateway feeder link. The envelope reports resulting values and does not itemise every term of the equation. Antenna gains and pointing loss do not appear as separate fields, so the record is a summary for monitoring and not a full design calculation. Active links refresh on a one-minute cadence, and operations dashboards subscribe to the stream to detect fading events. The envelope is signed by the satellite operator, so a downstream consumer can tell which constellation produced the link.
Once ISLs exist, the constellation can be routed as a mesh. The specification lists the benefits as reduced ground-station dependency, lower latency through space routing, greater resilience, and global coverage without ground infrastructure. The routing protocols it names are OSPF-based routing inside a constellation, BGP between constellations, and Delay-Tolerant Networking for deep-space cases.
The envelope types in this phase, and what each carries, are:
| Envelope | What it carries | Typical update pattern |
|---|---|---|
| Constellation descriptor | Tier, satellite count, altitude, inclination, minimum elevation, first launch | One per orbital shell, changes rarely |
| ISL link record | Endpoints, optical wavelength, link rate, link-quality estimate | Refreshed every minute for active links |
| Ground-station record | Station identity, location, supported bands, operational state | Changes with station state |
| User-terminal record | Terminal identity, owner, supported bands, last-known link metrics | Updated for performance monitoring |
The orbit tiers compete on a trade-off between delay, footprint and fleet size. The specification's figures show why operators combine them.
| Tier | Altitude | Period | Round-trip latency | Typical role |
|---|---|---|---|---|
| LEO | 500-2000 km (550-600 km typical) | ~90-100 min | Lowest of the three | Low-latency user access |
| MEO | 8,000-20,000 km typical | 6-12 h | 100-150 ms | Navigation, regional broadband, LEO backup |
| GEO | 35,786 km | 23 h 56 min | 500-600 ms | TV broadcast and DTH, legacy maritime and aviation, weather, emergency backup |
The MEO row shows two altitude ranges, because the terminology section defines the MEO class broadly while the architecture section quotes the 8,000-20,000 km band where systems actually operate.
The GEO latency follows from physics and not from engineering quality. A request and its response cross the 35,786 km gap four times, which is about 143,000 km. At the speed of light that is roughly 480 ms before any processing, in line with the specification's 500-600 ms. LEO's altitude of a few hundred kilometres brings that propagation term down by about two orders of magnitude. That is the reason LEO dominates new broadband builds. The price is fleet size. A GEO system covers the globe with three to five satellites, and a LEO system needs hundreds to tens of thousands.
Real systems are hybrids: LEO for user access, MEO to fill regional coverage, and GEO for broadcast and backup. A single descriptor with a tier field lets a consumer treat all three as one catalogue and choose between them by delay and footprint.
Production operators repeat three deployment patterns, and each one makes a different part of the data format matter most.
Pattern A, high-density urban. Terminals sit high on rooftops and the ground-station network is dense. The usual job is backup connectivity when terrestrial fibre fails. Per-satellite capacity is rarely the limit, so latency optimisation dominates. Route selection draws on the standard's latency-optimisation endpoint, and ISL link-quality records feed that choice.
Pattern B, rural broadband. Terminals are spread thinly across homes and the gateway network is sparse. The service is the primary connection in areas with no fibre. Capacity per satellite is the binding constraint, and constellation sizing translates directly into per-subscriber bandwidth. For this pattern the satellite count and altitude in the descriptor are the numbers that matter commercially, as are the ISL rates that let traffic reach a distant gateway.
Pattern C, mobility. Ships, aircraft and recreational vehicles move through the constellation. Handover frequency becomes the main operational concern. The handover-management protocol must keep pace with terminal motion, and the standard documents a handover budget per operator. Terminal records with last-known link metrics support this pattern, because they let an operator see degradation before the user does.
Conformant operators also publish benchmarks every quarter. These include median RTT per region, 95th-percentile RTT, sustained throughput per terminal class, handover success rate, and fade-event MTBF. The median hides the long tail that real-time applications notice, so the 95th percentile is the more telling figure, and fade-event MTBF links the benchmark back to the ISL fade records described above. The benchmarks serve customers choosing a provider and also feed regulatory reports.
Satellite internet crosses every sovereign boundary, so the format is built around external references and not around proprietary ones. The Phase 1 reference list includes:
Lawful intercept is where national law enters the data model. Jurisdictions that require it, such as the US under CALEA, the EU under ETSI ES 201 671, and Korea under its Communications Secrecy Protection Act, declare the requirement in the discovery document. A session in such a jurisdiction emits a notice envelope when it starts. The intercept itself runs over a separate signed channel, and every access to that channel is audit-logged. Satellite service is delivered across borders in real time, so the compliance obligation attaches to the session and not only to the operator's home country.
Debris mitigation reaches the data format at the other end of a satellite's life. The standard's scope includes debris mitigation as a regulatory topic, and ISO 24113 is the reference. Constellations in the tens of thousands make an accurate, signed, machine-readable record of what is in orbit part of the compliance picture.
A reference container ships mock data so that integrators can test against realistic envelopes before they connect real systems. The suggested path is to stand up the container and run the conformance suite against it. Then replace the mock backend one endpoint at a time, wire up audit-log replication, onboard a single trusted peer, and expand to more peers before going to production. The material estimates that a first deployment following this path reaches production in about 90 days. Most of that time goes to audit and compliance work, which is justified by the number of jurisdictions the service touches. Later jurisdictions reuse the same machinery with a policy overlay for each.
The envelope conventions are shared across the WIA family: UTF-8 JSON, RFC 8785 canonicalisation, signatures, and ULID identifiers. Within the 1.x line, hosts may add optional fields but may not remove existing ones, and breaking changes need a major version.
The domain-specific cautions are more important than that template.
valid_until before using a link in routing.The direction visible in the material is towards more orbit tiers cooperating and more traffic staying in space. As optical ISLs mature and the mesh carries more of the load, the ISL record becomes a routing input and not only a monitoring item, and the one-minute refresh cadence will be tested by faster-changing topologies. Delay-Tolerant Networking, named for deep-space routing, extends the same record types beyond Earth orbit. The 3GPP NTN reference points to tighter integration with terrestrial mobile networks, which is also where handover management will matter most. Phased-array terminals, which steer electronically with no moving parts, are what make fast handover practical in consumer hardware. The 1.x compatibility promise means records written today stay valid as these capabilities arrive.
Phase 1 gives satellite internet a shared vocabulary in the form of records. The constellation descriptor states tier, count, altitude, inclination and minimum elevation, and these relate to each other through checkable geometry. The ISL budget record condenses a laser link's power, loss, margin and rate into a short-lived signed statement. Ground-station and user-terminal records complete the picture at the edge. Behind these records lie the orbit trade-offs between LEO, MEO and GEO. A GEO round trip near 480 ms is a physical floor, whereas a LEO shell at a few hundred kilometres can be much faster. Three deployment patterns, urban, rural and mobile, determine which fields matter most to a given operator.
These shapes are only descriptions, however. They say what a constellation or a link looks like but not how anyone asks for it, subscribes to fading events, or selects a route. The next chapter takes up that question in Phase 2, where the same envelopes are exposed through HTTPS interfaces, including the latency-optimisation and handover-management functions that Patterns A and C depend on.