1. The book that watched its own battery
Neal Stephenson’s The Diamond Age returns as the recalled work because the Young Lady’s Illustrated Primer must manage its own resources while interacting with a complex external world. The desktop’s power manager must do the same: sense, forecast, decide, and act, all while staying within policy.
This entry asks what autonomous power budgeting needs from the desktop.
2. Sensors
The platform needs data to decide. Possible sensors include:
- Array telemetry: voltage, current, and temperature from each solar panel string.
- Battery telemetry: voltage, current, temperature, state of charge, and state of health.
- Load telemetry: current draw from each cell, payload, and protected channel.
- Attitude and orbit data: sun vector, eclipse predictions, and array pointing angles.
- Environment data: predicted and observed solar activity that affects generation.
No single sensor is enough. The battery state of charge tells you what is left; array telemetry tells you what is coming in; load telemetry tells you what is going out; orbit data tells you how long the eclipse will last.
3. Models
Sensors need interpretation. The platform needs models that map data to decisions:
- Generation model: predicted solar power as a function of attitude, orbit, array degradation, and illumination.
- Storage model: battery state of charge, state of health, and safe operating limits.
- Load model: expected power demand by operating mode and scheduled tasks.
- Eclipse model: when the platform will be in shadow and for how long.
- Degradation model: how array output and battery capacity drift over time.
These models do not need to be perfect. They need to be conservative enough to avoid brownouts and discriminating enough to avoid unnecessary shedding.
4. Actuators
The platform must be able to act on its decisions. Actuators include:
- Power switches for shedding loads.
- Voltage and clock regulators for throttling compute.
- Task scheduler for deferring and resuming jobs.
- Communications manager for adjusting data rates and windows.
- Thermal controller for adjusting heater power and setpoints.
- Safe-mode controller for escalating to a minimal operational state.
5. Policy
Autonomy without policy is just panic. The desktop needs rules such as:
- Thresholds: at what state of charge or power deficit does each level of shedding trigger?
- Priorities: which loads are protected, deferrable, or sheddable?
- Time limits: how long can a shedding level persist before ground must be consulted?
- Customer guarantees: minimum power and availability commitments.
- Override: how does ground regain control if the platform’s response seems wrong?
- Recovery: what conditions must be met before normal operations resume?
6. What this changes
- Autonomous power budgeting is a closed-loop system of sensors, models, actuators, and policy.
- Most of these pieces already exist for health management and autonomy; power budgeting is a specialized use of them.
- The next entry will close the autonomous power budgeting wondering arc and record the decision.