Truly File-Based Operations at Mars - Lessons Learned and Ideas for Future Missions

摘要

Since 2011 Mars Express has been flying a truly "file-based" concept for commanding the spacecraft and science operations, after a hardware anomaly forced a change from the previous "on-board schedule of commands" approach. As a consequence of the anomaly, new concepts had to be developed and applied to not just the technologies of file transfer and file management, which have received much attention over the years, but also to answer the questions: What are file-based operations? What goes in the files? How are they practically used both on ground and on board? ESA's first generation of deep space missions implement a packet-based large file transfer protocol to enable "files" to be transmitted from ground and guarantee the completeness on-board. However this is a transport-layer protocol and does not address the usage of the files operationally. Similar proposed standards (such as CCSDS File Delivery Protocol) also do not address the functionality of the files, and so by themselves do not necessarily lead to file-based operations. The Mars Express approach relies on seeing spacecraft operations not as a stream of commands, relayed from a mission planning system to the spacecraft via files, but as a collection of discrete activities, with one commanding file per activity. The contents of the file is the responsibility of the science planners, combining elementary sequences as per the rules defined by the spacecraft/payload operations engineers to construct self-contained, fail-safe sets of commands. For example, one science observation activity includes power, thermal, data link and instrument configuration changes that can be combined into one file (of telecommands, as an On-board Control Procedure, or a combination). Rather than individually scheduling the low-level commands, the activity file is simply executed at the correct time. If the observation needs to be modified or cancelled, the file is replaced or deleted. Spacecraft operations can conveniently be abstracted to the simpler paradigm: "which file to execute and when?" - the low-level constraints and resources checking having already been performed at mission planning level when that activity was planned. The ratio of commands-in-files to scheduled "execute" commands (the amount by which operations have been "compressed" by considering the higher-level "activity" file rather than low-level commands) is about 30:1 but is limited by available memory resources. Future missions with larger memories should achieve better than this, with fewer files containing more complex operations, resulting in even more abstracted operations overall. Mars Express has demonstrated a use case and implementation of file-based operations with regards to spacecraft commanding, but this is only half of the picture. The return of science data and housekeeping telemetry remains based on "packet stores". Packet stores are akin to looped magnetic tapes, in that the data is stored in the order in which it was written. For Mars Express this is not usually a problem - all data will eventually be downlinked from each packet store and any data selection is made on ground. On-board data selection is however an effort-intensive process involving copies from one packet store to another in an attempt to isolate the required data. More sophisticated on-board file management capabilities would simplify such operations. As spacecraft capabilities increase in terms of both data collection and data storage, whilst downlink bandwidth remains relatively static, the issue of "selective downlink" of only interesting science data will become ever more prevalent. A file-based operations concept would make implementing selective downlink easier compared to a packet based implementation, as only the required files would be downlinked without having to manage low-level streams of packets. We also consider that autonomous on-board data selection must play a part in future missions. Instruments can use heuristics or tunable criteria to quantify the quality of the science data per file upon which such a data selection will then be made. Our conclusions are that the case for file-based operations is clear for commanding, and the solution implemented by Mars Express demonstrates a successful route for future missions to consider. File-based operations for data downlink are not as straightforward: not all instruments produce clearly defined science data sets which can be easily placed in a file, and advanced file management functions are not as-yet available. The use of files for storage of science data would enable an efficient selective downlink mechanism, which will improve the quality of science data, in a context where data return bandwidth is more constraining than on-board file storage capability.