Onboard Numeric Watchdog for Analysis of Telemetry Channel Heuristics (ON-WATCH)

Project: Research project

Description

Modern spacecraft have well surpassed the capability to generate more data than can be transmitted to the ground. Much work has been done on compressing science data via data science techniques, but this challenge remains relatively unstudied in engineering telemetry data. Telemetry channels are routinely run at slower refresh cadences due to the limitations of downlink bandwidth; however, lower rates make transient discovery difficult and increases
operational risk.

ON-WATCH proposes a new set of onboard tools for large and small missions to intelligently compress high-cadence, high-quality engineering telemetry data using technologies developed for onboard science analytics. Utilizing a fusion of machine learning (ML) and spacecraft model-driven knowledge, the monitoring functions of ON-WATCH provide optimally informative supplemental data products about the health of a spacecraft. This proposal will develop a software architecture and fully implemented data processing algorithms for four key downlinked products:

1.) Quick-Look interpretable, maximally informative statistics about each telemetry channel to support improved operations monitoring. These are small data volume products for mission phases where full downlink is either impossible or unnecessary.

2.) Snapshot products capturing anomalous time periods where predictive models were inconsistent with observed telemetry. Snapshots prioritize high-res downlink of the most unusual data to provide operators with rapid focus-of-attention.

3.) EVRs to be generated when channels fall out of expected ranges (ML may be used to optimize these values as well)

4.) Telemetry rate change requests utilize mission schedule, spacecraft models, and current telemetry context to either increase telemetry rates when events are suspected or decrease rates when nominal states prevail on a per-channel basis, during calm mission phases, or when engineering bandwidth is particularly low.
StatusFinished
Effective start/end date2/13/189/30/18

Funding

  • National Aeronautics Space Administration (NASA): $10,000.00

Fingerprint

Telemetering
Spacecraft
Learning systems
Bandwidth
Monitoring
Software architecture
Fusion reactions
Health
Statistics