1. The array that had to know when to listen

Carl Sagan’s Contact returns as the recalled work because the radio telescopes in that novel cannot listen everywhere at once. They must know where to point and for how long. A LEO desktop has the same constraint in reverse: it cannot transmit everywhere at once. It must know when each ground station or relay satellite will be in view, how long the pass will last, and what the link can carry.

This entry wonders how the desktop predicts communication opportunities and decides what data to send first.

2. Pass prediction as a planning input

The desktop knows its own orbit and the positions of ground stations and relay constellations. From this it can compute future passes: rise time, set time, maximum elevation, and duration. Each pass has a predicted link budget that changes with elevation. A pass that peaks at 10° elevation is different from one that peaks at 85°.

The autonomy needs a rolling pass schedule. The schedule feeds the data management system, which decides which files to queue for which pass. It also feeds the power and thermal managers, which must prepare for transmit operations, and the ADCS manager, which may need to point a high-gain antenna.

3. Data prioritization

Not all data is equally valuable. A reasonable priority hierarchy might be:

  • Critical telemetry and alarms: health, safety, fault reports. These get the first opportunity.
  • Command responses: confirmation that ground commands were executed.
  • Time-sensitive customer data: observations with a delivery deadline.
  • Routine payload data: science or imagery without an immediate deadline.
  • Logs and diagnostics: useful for analysis but not urgent.

But a fixed hierarchy can starve lower-priority data. A better approach is value-aware scheduling. Each data item has a size, a deadline, and a value. The autonomy solves a knapsack problem for each pass: fill the predicted link capacity with the highest-value set of items whose deadlines are met.

4. Preemption and deadlines

Sometimes new high-priority data arrives after a pass has been planned. The autonomy must be able to preempt lower-priority queued data. This requires the queue to be dynamic and the plan to be recomputable. It also requires the platform to know how much of a partially transmitted file has been acknowledged, so it does not restart from the beginning.

5. What this changes

  • Pass prediction is a core input to autonomous operations.
  • Data prioritization should be value-aware and deadline-aware, not just a fixed hierarchy.
  • The communications plan must be recomputable as new data and new passes arrive.
  • The next entry will wonder about adaptive link rates and relay routing.