4
American Institute of Aeronautics and Astronautics 1 Planning Science Data Return of Mars Express with Support of Artificial Intelligence Erhard Rabenau * NOVA Space Associates Ltd, Bath, BA1 2AB, UK and Thomas Léauté European Space Agency, 64293 Darmstadt, Germany Mars Express (MEX) has been in orbit around Mars since the beginning of 2004. A complex science payload, including a camera, spectrometers, a particle detector, a radar and a transponder for communications with Mars landers, generates a large amount of science data that need to be transferred to the ground. For MEX, a dedicated ESA ground station and a number of DSN stations are available to receive the data. As the mission is operated in a store and forward-type manner, all data are stored on board before it is down-linked. Depending on the season, the downlink data rate varies between 28 kbps and 182 kbps. The generation of science data has to take into account: a) the downlink bit rate b) the available downlink windows and c) the available mass memory space on the spacecraft. The amount of data to be down-linked varies between 800 Mbit and more than 5 Gbit per day. The allocation of downlink windows is driven largely by the science requirements and changes from orbit to orbit, i.e. there are no dedicated data downlink slots allocated as a fixed daily pattern, downlink slots continuously vary in length and time of day. Data have to be dumped from the onboard mass memory before more data are written into it to avoid loss of data. Consequently, the generation of a data downlink plan has been a challenge since the start of the mission. It became obvious that a dedicated tool was needed to help resolve the conflicting requirements. An operational software system was produced, based on experience gained from early MEX operations, and on an ESA study, referred to as MEXAR (Mars Express Scheduling Architecture), that led to a prototype of a software system demonstrating that artificial intelligence techniques for planning and scheduling can be beneficially applied to a real space mission. The tool has been successfully deployed in the MEX mission planning process as part of a suite of tools to support planning. Major objectives of the tool are the automation of the data dumping planning process and the maximization of the amount of dumped data. The paper describes the operational requirements for a data dumping tool, its iterative implementation process up to the final product and the operational assessment in terms of reduction of manpower effort and increase in the volume of dumped data. Nomenclature CCSDS = Consultative Committee for Space Data Systems DSN = Deep Space Network ESA = European Space Agency ESTRACK = ESA Tracking Network Gbit = 1000 million bits kbps = kilobits per second Mbit = 1 million bits MEX = Mars Express MEXAR = Mars Express Architecture MMI = Man Machine Interface * Senior Mission Planning Engineer, NOVA Space Associates Ltd, 11 Kingsmead Square, Bath, BA1 2AB, UK Young Graduate Trainee, Mission Control Technologies (OPS-OSC), Robert-Bosch-Str. 5, 64293 Darmstadt, Germany SpaceOps 2006 Conference AIAA 2006-5931 Copyright © 2006 by NOVA Space Associates Ltd. European Space Agency. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

[American Institute of Aeronautics and Astronautics SpaceOps 2006 Conference - Rome, Italy ()] SpaceOps 2006 Conference - Planning Science Data Return of Mars Express with Support

  • Upload
    erhard

  • View
    212

  • Download
    0

Embed Size (px)

Citation preview

Page 1: [American Institute of Aeronautics and Astronautics SpaceOps 2006 Conference - Rome, Italy ()] SpaceOps 2006 Conference - Planning Science Data Return of Mars Express with Support

American Institute of Aeronautics and Astronautics

1

Planning Science Data Return of Mars Express with Support of Artificial Intelligence

Erhard Rabenau* NOVA Space Associates Ltd, Bath, BA1 2AB, UK

and

Thomas Léauté†

European Space Agency, 64293 Darmstadt, Germany

Mars Express (MEX) has been in orbit around Mars since the beginning of 2004. A complex science payload, including a camera, spectrometers, a particle detector, a radar and a transponder for communications with Mars landers, generates a large amount of science data that need to be transferred to the ground. For MEX, a dedicated ESA ground station and a number of DSN stations are available to receive the data. As the mission is operated in a store and forward-type manner, all data are stored on board before it is down-linked. Depending on the season, the downlink data rate varies between 28 kbps and 182 kbps. The generation of science data has to take into account: a) the downlink bit rate b) the available downlink windows and c) the available mass memory space on the spacecraft. The amount of data to be down-linked varies between 800 Mbit and more than 5 Gbit per day. The allocation of downlink windows is driven largely by the science requirements and changes from orbit to orbit, i.e. there are no dedicated data downlink slots allocated as a fixed daily pattern, downlink slots continuously vary in length and time of day. Data have to be dumped from the onboard mass memory before more data are written into it to avoid loss of data. Consequently, the generation of a data downlink plan has been a challenge since the start of the mission. It became obvious that a dedicated tool was needed to help resolve the conflicting requirements. An operational software system was produced, based on experience gained from early MEX operations, and on an ESA study, referred to as MEXAR (Mars Express Scheduling Architecture), that led to a prototype of a software system demonstrating that artificial intelligence techniques for planning and scheduling can be beneficially applied to a real space mission. The tool has been successfully deployed in the MEX mission planning process as part of a suite of tools to support planning. Major objectives of the tool are the automation of the data dumping planning process and the maximization of the amount of dumped data. The paper describes the operational requirements for a data dumping tool, its iterative implementation process up to the final product and the operational assessment in terms of reduction of manpower effort and increase in the volume of dumped data.

Nomenclature CCSDS = Consultative Committee for Space Data Systems DSN = Deep Space Network ESA = European Space Agency ESTRACK = ESA Tracking Network Gbit = 1000 million bits kbps = kilobits per second Mbit = 1 million bits MEX = Mars Express MEXAR = Mars Express Architecture MMI = Man Machine Interface

* Senior Mission Planning Engineer, NOVA Space Associates Ltd, 11 Kingsmead Square, Bath, BA1 2AB, UK † Young Graduate Trainee, Mission Control Technologies (OPS-OSC), Robert-Bosch-Str. 5, 64293 Darmstadt, Germany

SpaceOps 2006 Conference AIAA 2006-5931

Copyright © 2006 by NOVA Space Associates Ltd. European Space Agency. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Page 2: [American Institute of Aeronautics and Astronautics SpaceOps 2006 Conference - Rome, Italy ()] SpaceOps 2006 Conference - Planning Science Data Return of Mars Express with Support

MPS = Mission Planning System S/C = Spacecraft SSMM = Solid State Mass Memory

I. Introduction ARS Express (MEX) is ESA's first mission to Mars, developed and operated on a small budget with about 2 years from concept to launch, also sometimes referred to as 'flexi' mission in the ESA long term scientific programme. Nevertheless, the mission objectives are manifold and include the exploration of

Mars with high resolution stereo imaging and mineralogical mapping of the surface of the planet, subsurface radar sounding, atmospheric sounding as well as environmental and radio studies. MEX was launched in 2003 and has been in orbit around Mars since early 2004 successfully collecting science data. The primary mission ended in 2005. Currently MEX is in the extended mission phase for another Martian year till the end of 2007. Six instruments and the spacecraft contribute to the volume of data that is generated and needs to be transferred to ground (further on referred to as dumped or down linked). As a first for any ESA mission, an Artificial Intelligence (AI) tool was developed based on operational experience and successfully implemented in the operational environment. The process from dump planning till the data are actually received on ground is described in the following chapters.

M

II. The MEX Downlink Problem The MEX spacecraft operations are performed in a store-and-forward manner, i.e. all science and

housekeeping data are held in a Solid State Mass Memory (SSMM) till they are down linked to earth (see Fig. 1). The SSMM is divided into dedicated packet stores or data areas for each instrument and/or packet type. Different types of packet stores exist: a) cyclic and b) non-cyclic. Cyclic packet stores can be modeled as circular buffers. When more data are written to the packet store than its capacity, the oldest data in the store are overwritten. Data from cyclic packet stores can be dumped by opening the file for dump and closing it after a certain time interval. Non-cyclic packet stores are like a bucket. Once the container is full, any excess data will be lost, i.e. the newest data will be lost. Additionally, non-cyclic packet stores always need to be dumped completely.

S/C housekeeping data are generated at a constant rate and stored in packet stores that are swapped once per day in order to avoid overwrites. On MEX, instruments are switched on and off for each observation. One or more observations may be performed in an orbit. The instrument observation schedule is defined in the pointing timeline. Thus, the data volume generated by an instrument varies from orbit to orbit and from 50 Mbit to more than 1 Gbit between instruments. For some instruments the actual amount of data generated may vary from the predicted values by 10% or more depending on the applied compression algorithm of the processing software. The ASPERA instrument generates data continuously. One instrument uses more than 1 packet store.

SSMM performance constraints need to be considered in downlink planning. The system does not support parallel downlink of data stores. Virtual channels have not been implemented in the CCSDS transfer frame configuration.

The dumped data are received by ground stations of the ESTRACK and DSN networks. The station allocation varies from day to day as the station usage is driven by science requirements and geometric constraints, i.e. interruption of a downlink pass by a science pointing, S/C maintenance, eclipse or occultation. Due to varying distance between Mars and Earth, the downlink bit rates also vary between 182 kbps and 28 kbps. The station allocation includes different antenna sizes which results in different downlink bit rates.

S/C housekeeping needs to be down linked with highest priority after a packet store swap. Science data generation and dumping of the data are part of the activities planned for during the medium

term planning cycle. The data downlink is allocated on agreed instrument data shares at science planning level. From the beginning of the mission, science operations planning has been targeted at maximum science return. In the low data rate season, this very often results in no or very little margin in the usage of the data downlink resource, i.e. the downlink channel is fully utilized, leaving very few or no opportunities to retrieve data in case of data loss during downlink.

Figure 1. Data Flow Configuration on the MEX S/C

III. Dump Planning Constraints In order to be able to generate a feasible dump plan, the following assumptions have been made: • S/C housekeeping telemetry is dumped as a fixed volume each day;

American Institute of Aeronautics and Astronautics

2

Page 3: [American Institute of Aeronautics and Astronautics SpaceOps 2006 Conference - Rome, Italy ()] SpaceOps 2006 Conference - Planning Science Data Return of Mars Express with Support

• The predicted science data volume is specified in dedicated data profiles over time that are part of the operations requests files for each instrument;

• The generation of a second data stream for an instrument is not modelled in data profiles as the mechanism does not support multiple packet stores per instrument;

• All dump activities are grouped in a so-called dump day which contains all downlink windows in a given day starting with the first window after midnight;

• Dumps can only be performed during dump windows. A dump window is a ground station pass during which data can be down linked.

IV. Intermediate Solutions During cruise and Mars in-orbit commissioning, all dump

operations were scheduled manually. This was quickly abandoned because of the enormous amount of work involved as science operations started to be stepped up and of the large potential for error.

The next step was a tool developed in Visual Basic that semi-automated the generation of the dump plan. The tool generated a dump plan based on data share allocations received from the science planning team. The tool allowed a quick generation of a dump plan but was limited due to the fact that the data share allocations were not linked with the data generation time. As the science data generation increased, it became more and more important to dump science data as early as possible after generation. Several iterations of manually changing the data allocations were usually necessary to arrive at a feasible dump plan. The output of the tool were dump operations requests as inputs to the MEX Mission Planning System (MPS). Also, as the dump generation tool resided on a PC, and the MPS on a UNIX platform, input and output files had to be exchanged across platforms through a firewall which added to the time of the process. The work flow is illustrated in Fig.2, dump planning workflow without MEXAR.

Science Planning

MPS++

Dump Planning On PC

Data Share

Pointing TL

Station Allocation

Mission Planning System

Dump Analysis using EXCEL

Dump List

SOR template

SOR

Dump Windows Bit RateEdit

Data Share File

Figure 2. Dump Planning Workflow Without MEXAR.

V. The MEXAR Solution As can be seen, a different solution was required. Before the launch of MEX, the MEX downlink problem

had been addressed in an ESA study in a much more simplified manner. Because of missing operational inputs at that time, the study remained at a theoretical level. After the launch and initial science operations, it was proposed by the team who had performed the study to try and utilise the study result by turning it into an operational system.

MEXAR2 was born. The mission planning team was able to base the requirements on the operational experience and products already used by the MEX mission planning team with the VB-based tool. The requirements were specified such that, for MEXAR2, no change to the input and output products of the software were required in order to maintain the interface with the rest of the mission planning environment. Instead of the data allocation, MEXAR2 reads in the instrument data profiles from the operational request files. The system allows adding further dump requirements based on absolute time, periodic and event-based dumps (dumps that refer to another dump event). The MEXAR2 tool has been developed in JAVA and therefore runs on PC and UNIX platforms.

MEXAR2 provides the following improvements over the previous approach (as illustrated in Fig. 3, dump planning workflow with MEXAR):

Science Planning

MPS++

Mexar

POR / ZD

Pointing TL

Station Allocation

Dump Windows Bit Rate

Mission Planning System

Dump Analysis using Visualisation

tool

Dump List

SOR template

SOR

Rerun Mexar with different conf

settings

Figure 3. Dump Planning Workflow With MEXAR.

• Data are dumped as early as possible after generation because the system now has knowledge of the data generation time (provided in the data profile records);

• The software can be configured to look for more optimal results, e.g. in order to avoid overwriting;

American Institute of Aeronautics and Astronautics

3

Page 4: [American Institute of Aeronautics and Astronautics SpaceOps 2006 Conference - Rome, Italy ()] SpaceOps 2006 Conference - Planning Science Data Return of Mars Express with Support

American Institute of Aeronautics and Astronautics

4

• The software handles priorities; • The software provides extensive lists of statistics and unused dump windows; • The software provides a graphical Man-Machine-Interface (MMI) for ease of use; • With MEXAR2 the mission planner can perform what-if analyses, by quickly generating dump solutions

with different configuration settings. Since MEXAR2 was developed using the experience and existing products from the operational

environment, together with the existing output of the study, a first prototype was quickly available to the mission planning team. Because the input and output data products had not changed in MEXAR2, testing was straight-forward and the tool could be used operationally from the first delivery onward. The development of the tool was performed in stages following a rapid iterative prototyping approach where more and more functionality was added at each stage. Testing and bug fixing went hand in hand with the overall approach usually resulting in a quick turn-around after reporting a problem.

From the first delivery onward, even though some functionality was initially missing, the robustness of the tool was very good, the results exceeded those of the tools used previously and most importantly, the data dump generation could be performed quicker and required less effort. Since MEXAR2 could easily be integrated in the MEX Mission Planning System, no time-consuming cross-platform operations were required any more. Furthermore, the tool allows for easy integration of other JAVA-based applications, e.g. a tool generated by a third party to visualise the generated dump plan.

VI. Conclusion Space Missions are challenging domains for applying Artificial Intelligence (AI) techniques for planning and

scheduling. MEX provides the platform for the research and resolution of an automated problem solving technique. In this respect, as well as in many others, MEX has been a first. With MEXAR2 the MEX Mission Planning Team has a decision support system that integrates human strategic capabilities and automated problem solving algorithms to find high-quality solutions. These quality metrics can be expressed in data availability on ground as early as possible after generation, minimisation of data overwrites, improvement of robustness by means of what-if analyses, optimisation of the downlink channel. Most importantly, the overall time saving for the end-to-end dump planning activities of the mission planning team compared to previous approaches is in the order of 50%.

As a direct result of the success of MEXAR2, a similar application for solving the MEX uplink problem is being developed.

References 1Cesta, A., Oddi, A., Cortellessa, G., Policella, N., "Automating the Generation of Downlink Spacecraft Operation in

Mars Express: Analysis, Algorithms and an Interactive Solution Aid", Technical Report, MEXAR-TR-02-10, ISTC-CNR [PST], Italian National Research Council, 2002

2Oddi, A.,and Policella, N., "A Max-Flow Approach for Improving Robustness in a Spacecraft Downlink Schedule", Fourth International Workshop on Planning and Scheduling for Space - IWPSS '04, Darmstadt, 2004, pp. 151-158