Native Earth-Fixed-Cell Grids

Astraeus can represent service demand on a latitude/longitude grid or on native Earth-fixed cells (EFCs). EFC mode is useful when the service denominator and beam-target identity should be a stable, beam-scale terrestrial cell rather than a coarse angular bin.

Cells are created before scheduling. The same identity is retained through population, targeting, scheduling, delivery, and reporting artifacts; EFC metrics are therefore tied to the cells actually evaluated by the run.

Native H3 cells provide stable Earth-fixed footprints and centers over the selected service region.

Choose The Denominator Deliberately

Use EFC mode for questions such as:

• how many beam-scale cells were eligible, feasible, targeted, or served?
• how broadly did beam targeting rotate through a service region?
• how does the result compare with an external Earth-fixed beam-cell model?
• how sensitive is service to cell size or boundary treatment?

Use a latitude/longitude grid for broad geographic coverage, coarse regional reporting, or continuity with an existing angular-grid study.

EFC-cell, latitude/longitude-cell, area-weighted, and demand-weighted percentages have different denominators. Compare percentages as equivalent physical claims only when grids, weights, demand, and normalization are aligned.

Configuration

Select EFC mode under the service population grid:

mission:
  kind: service
  service:
    population:
      grid:
        kind: efc
        provider: h3
        diameter_km: 500
        evaluation:
          mode: sampled

diameter_km is required. For H3, Astraeus selects the resolution whose average cell diameter is closest to the request and records both the requested and resolved sizing metadata.

Providers

Provider	Use
h3	preferred discrete global hexagonal grid for regional, wide-area, and global studies
hex	direct longitude/latitude hex-like lattice for compact fixtures or local sensitivities
beam_footprint	longitude/latitude lattice parameterized by a beam-footprint diameter
external	reserved configuration value for a future supplied-cell catalog path

hex and beam_footprint use longitude/latitude construction whose physical geometry varies with latitude. Use H3 for the normal global representation.

Boundary Evaluation

Mode	Behavior
centroid	includes a cell when its center lies in the eligible area; fastest and least detailed at boundaries
sampled	evaluates a compact set of footprint samples; default balance for most studies
exact_polygon	evaluates polygon overlap; best boundary fidelity with higher geometry cost

Use H3 with sampled evaluation for a normal engineering baseline. Use exact polygon evaluation when boundary inclusion materially affects the conclusion.

Stable Cell Identity

Native cells use stable PEFC service-point identifiers. H3 identifiers include the selected resolution and underlying H3 cell reference. Key fields include:

Field	Meaning
cell_id	native service-cell identifier
target_cell_id	cell considered or selected by beam targeting
matched_area_id	matching service region
lat_deg, lon_deg	representative cell center
area_km2	cell area available to reporting and weighting

That identity allows a result to distinguish the cell existing, being eligible, appearing in a target plan, passing feasibility, receiving an assignment, and receiving delivered traffic.

Artifacts

Mission-side EFC artifacts include:

Artifact	Contents
mission/service_population_summary.json	denominator, provider, diameter, selected H3 sizing, evaluation mode, and counts
mission/service_population_cells.*	one row per native service cell
mission/service_population_cell_footprints.*	footprint vertices for maps and spatial review
mission/service_population_beam_target_assignments.*	target/feasibility/application rows keyed by EFC cell
mission/service_population_beam_target_assignments_summary.json	eligible, feasible, targeted, and applied cell counts

When beam targeting is active, transport also writes:

Artifact	Contents
transport/beam_targeting_summary.json	mode, objective, windows, selection concentration, and warnings
transport/beam_target_plan.*	resolved target windows by satellite and beam
transport/beam_target_scheduler_rollup.*	scheduler selections and rejections by target cell
transport/beam_target_link_outcome_rollup.*	feasibility, application, rejection, and steering-angle outcomes

Metric Interpretation

Report the denominator beside every percentage. A useful minimum set is:

• eligible native-cell count;
• targeted, feasible, and applied cell counts;
• delivered traffic or served-demand KPI;
• area- or demand-weighting method where used;
• provider, resolved cell size, and boundary-evaluation mode.

selected_cell_fraction and selection_concentration describe scheduler behavior. Read them alongside demand-served and data-delivery KPIs. A run can deliver most requested traffic to concentrated demand while touching only a small fraction of all eligible cells.

Beam targeting states also remain distinct:

State	Meaning
Targeted	cell appears in a generated target plan
Feasible	at least one candidate passed the relevant physical checks
Applied	beam used the cell as an active target
Rejected	cell was considered but not applied, with a reason where available

Long dwell, sparse demand, strict priorities, or hotspot objectives may intentionally concentrate selections. Coverage-rotation objectives broaden the target set but remain bounded by visibility, link closure, steering, beam count, and dwell policy.

Packaged Example

The beam-hopping example uses a native H3 grid over Europe and West Africa:

astraeus run astraeus-resources/examples/user/beam-hopping-capacity.yaml

Start with a dry run after changing service area or diameter. A smaller diameter can sharply increase cell count, candidate work, and retained artifact volume.

Scale And Fidelity Limits

• Runtime and memory grow with eligible cells, visible candidates, target windows, satellites, beams, and requested artifact detail.
• EFC mode supplies geography and stable identity; interference, beam hopping, forecasting, handover, and optimization require their own explicit model configuration.
• exact_polygon has higher geometry cost than sampled or centroid evaluation.
• external cell catalogs are reserved but not yet executable.
• H3 cell area varies with location and resolution. Use emitted area_km2 for area-weighted summaries rather than assuming equal physical area.

See Protocol-Informed Beam Hopping for ordered allocation and continuity, and Geometry And Coverage for spatial aggregation modes.