The limited lifetime of ISO, the severe sky coverage constraints, the complexity of the scientific instruments, along with the necessity to make many short observations under ground station coverage (no on-board data or command storage for instrument operations was possible) dictated that all operations had to be pre-scheduled in order to maximize the time spent acquiring useful astronomical data. Thus, ISO was operated in a service observing mode with each day's observations being planned in detail up to three weeks in advance.
This operational concept drove the design of the ground segment (Kessler et al. 1996, [95]; 1998, [96]), which consisted of the Spacecraft Control Centre (SCC) and the Science Operations Centre (SOC), both co-located at VILSPA, and two ground stations providing approximately 22 hours/day of real-time support.
The SCC team, within the Directorate of Technical and Operational Support (D/TOS), was responsible for conducting and controlling the flight operations of the spacecraft, including health and safety of the instruments. The SOC team, within the Directorate of Science (D/SCI) was responsible for all aspects of the scientific operations ranging from the issue of the `Calls for Observing Proposals', through the scheduling and use of the scientific instruments, to the pipeline data processing, and distribution of the data products. Additional teams, based mainly at the PI institutes, supported the off-line operations of the instruments.
Two ground stations were used to communicate with ISO, providing visibility of the satellite from the ground for the entire scientifically-useful part of the orbit over 16 hours per day. The primary at ESA's Villafranca Satellite Tracking Station (VILSPA, see Figure 4.3), located near Madrid, Spain, and the secondary at Goldstone, USA, provided by the National Aeronautics and Space Administration (NASA). The ISO Spacecraft Operations Center (SOC), based at VILSPA, controlled the satellite via one of the two stations, carrying out the operations in real-time. The orbit of ISO was arranged so that after perigee passage (when ISO was out of view of both ground stations for approximately 30 minutes) the first ground station to re-acquire the ISO signal was VILSPA. Additional resources, enabling ISO to be operated for a longer period per day, were supplied by the Institute of Space and Astronautical Science (ISAS), Japan.
The SCC was led by the Spacecraft Operations Manager and, throughout the routine operations phase, there were 28.3 staff in post. Its main responsibility, as already mentioned, was conducting and controlling the flight operations of the spacecraft, including health and safety of the instruments.
After the end of the Launch and Early Orbit Phase (LEOP) -- revolutions number 0 to 3 -- the satellite control was transfered to VILSPA from the Operations Control Centre (OCC) at ESOC, Darmstadt; all subsequent operations were successfully supported from the SCC.
Starting from manual use of the Flight Operations Plan and associated procedures, operations were gracefully automated during the Commissioning Phase to use, by the end of this phase, a fully pre-programmed Central Command Schedule (CCS) reflecting the output of the Mission Planning Phase 1 (SOC) and the Mission Planning Phase 2 (SCC). This schedule contained all platform and payload commands. On average, some 10000 commands had to be uplinked to the spacecraft every day. Therefore, only minimum operator intervention was recquired for spacecraft and instrument operations.
The CCS contained `dedicated windows' during which either spacecraft or science operations could be scheduled. Additionally, `event designators' and `keywords' were defined that triggered certain command operations to be inserted in those windows, when required. A skeleton schedule for a revolution (orbit) is shown in Figure 4.4.
|
The baseline approach during routine operations was that all four instruments were activated and de-activated automatically by the schedule, irrespective of whether a particular instrument was scheduled for use in that orbit or not.
To optimise the time available for scientific observations, spacecraft operations and instrument activation/de-activation were placed along an orbit in such a way that they did not use science time (defined as the time the satellite spent outside the main parts of the Van Allen belts). Interleaved manual commanding was, in principle, only required to support ranging, ground station handover, and for a few specific operations of the AOCS. The schedule offered `hold', `resume' and `shift' functions in order to recover from, and to minimise the impact of, spacecraft, instrument or ground segment anomalies. When required, recovery from problems was initiated following the relevant Flight Control Procedures (FCPs) and Contingency Recovery Procedures (CRPs) of the Flight Operations Plan. It is worth noting that approximately 1000 FCPs and 500 CRPs had been written and validated with platform simulator before launch.
During pre-launch testing, it was already realised that the command schedule was highly susceptible to ground segment problems because of the very high scientific instrument command rate. In the event of problems, e.g. when commands could not be verified due to loss of telemetry, the schedule was suspended. In the worst case, a short drop in telemetry could cause the loss of an entire scientific observation of several hours' duration.
Throughout the in-orbit operations, a wide variety of efforts were successfully undertaken by the SCC to prevent or minimise the loss of science. Major improvements included the implementation of an automatic telemetry link re-configuration on the ISO Dedicated Control System, which reduced the impact of telemetry drops considerably. The implementation of the Hipparcos/Tycho Guide Star Catalogue in the Flight Dynamics System (FDS) contributed greatly in solving the guide star acquisition problems encountered early in the mission. In a joint effort between the SOC and the SCC, a new observing mode was implemented for the Long Wavelength Spectrometer, enabling it to gather science data even when not scheduled as the `prime' instrument.
Another improvement, which made a major contribution to the science
output, was the reduction of satellite's absolute pointing error from
4
during the Commissioning Phase to 1
level in the Routine
Phase, especially since the system specification was 
11.7
(see Section 5.4.1).
The ISO Mission Control System (see Figure 4.5) performed all aspects connected with operations and safety of the spacecraft, including safety monitoring of the scientific instruments. The hardware of the control system consisted esentially of two VAX 4600 redundant Spacecraft Monitoring and Control computers (ISORT/ISODV), six associated Sun SPARC-20 workstations, associated spacecraft control software, and the mission planning system software as far as Mission Planning 2 was concerned. The system was designated as the ISO Dedicated Control System (IDCS). The FDS consisted of a set of five Sun workstations and dedicated software. These systems were networked on a partially-redundant OPSLAN to prevent single point failures and isolated the SCC from the outside world.
Two redundant micro-VAX 3100-76 computers formed the Operational Data Server system (ODS-1/2). The ODS constitued the interface between the spacecraft control system of the SCC and that of the SOC as far as science real-time data reception in the form of Telemetry Distribution Formats (TDFs) was concerned. The latter contained not only the telecommand history data, but also specially provided derived telemetry parameters. These parameters were used within the SOC for instrument monitoring and control purposes, using the Real-Time Technical Assessment (RTA) and Quick-Look Analysis (QLA) software, which ran on the four instrument workstations (one dedicated per instrument). The ODS was also the interface between the Mission Planning Phase 1 (MPP1) and the SOC and that of the SCC (MPP2) for interchanging mission planning files.
Furthermore, the ODS provided the short history archive of the science telemetry and archived TDFs onto optical disks for access from the SOC Science Data Processing system. The network interface provided the connectivity of the IDCS with the ground stations through the Integrated Switching System (ISS), as part of the OPSNET. Support functions were provided for: Spacecraft Performance Evaluation (SPEVAL), required to determine all aspects of spacecraft performance which could impact the life of the mission and mission efficiency; and spacecraft on-board software maintenance for the AOCS, STR and the OBDH. Communications Services were provided to support a variety of tasks, such as testing and validating procedures, AOCS and on-board software maintenance and validation, and spacecraft anomaly investigation.
One very significant achievement was the mission extension beyond September/October 1997. During this time, ISO's orbital geometry was such that it underwent eclipses of exceptionally long duration. Additionally, during early September, marginal violations of the Earth constraint on the pointing direction could not be avoided for some minutes each day as ISO went through perigee. Since the spacecraft was required to be operated beyond design specifications with respect to power, Sun and Earth constraints, it was necessary to develop and implement a new operations strategy, which deviated considerably from the well-proven Routine Phase operations concept. In addition to the above, there was a strong requirement from the scientific community to observe the Orion and Taurus regions of the sky, which became visible to ISO during this period for the first time in the mission.
During the period 7 September to 7 October 1997, when the eclipses reached
a maximum of 166.5 minutes, i.e. more than twice as long as the baseline
design of 80 minutes, the power of the two batteries had to be preserved by
switching off non-esential units, by restricting scientific pointings to
one observation during eclipse, and by restricting the use of the instruments
to two out of four during the peak eclipse period. To ensure proper pointing
stability in eclipse, a second `roll star' was used by the Star Tracker. This
star, some 2
away from the guide star, was used to control the
gyro drift with respect to the satellite x-axis and hence the telescope
boresight. At the same time, the Earth warning and forbidden regions had
to be violated, since no constraint-free corridor was left around perigee.
This was crucial for the AOCS and therefore for the telescope pointings
around perigee. In order to reduce the impact of the penetration into the
Earth-constraint region, the Sun constraint had to be relaxed.
All of the above required disabling most of the autonomous fallback functions of the AOCS and OBDH subsystems, i.e. the satellite was safeguarded by relying on ground control only. Both on-board batteries showed excellent performance with less than expected depth of discharge and reached full charge each revolution. The effect of violating the Earth constraints was less than predicted. The telescope upper baffle temperatures increased by just under 4 K, returning to nominal temperatures within 45 minutes thereafter. The AOCS pointing performance was very stable and hence scientific observations performed during eclipses did not suffer any degradation in pointing. The period passed uneventfully and routine operations continued until the helium was depleted on 8 April 1998.
The SOC was organised into two teams: the science team, led by the Project Scientist, which was responsible for community support and for setting the overall policy for the SOC; and the operations team, led by the Science Operations Manager, which was responsible for instrument operations and the SOC infrastructure. On average during the Routine Phase, the SOC had 92 members.
ISO Science Operations were organised almost as a factory `production line' (see Figure 4.6). The starting point was the databases into which observers had entered all the details required to implement their observations in service mode. Each observation was technically validated and then loaded into the Mission Data Base (MDB), which at the end of the mission included more than 40000 observations.
The next step was to generate a long term plan, showing when and how the most scientifically-important observations could be implemented. This was particularly important in the case of a mission like ISO with a short lifetime and with only a limited part of the sky accesible at any given time. A coarse pre-scheduling of the next three months was made. This process was extremely time- and resource-consuming and never worked quite as expected since one was dealing with a `moving target'. In other words, the flexibility offered to the observers to optimise their observing programmes meant that the input changed faster than the plan. This flexibility was necessary and greatly enhanced the scientific return. However, extensive and complex manual work was required to enable to ISO to execute successfully nearly 98% of the highest priority observations. Similar missions in the future should be able to generate a representative long term plan within a few days with minimal human intervention.
Next in the production line came the detailed planning of each day's observations to the level of instrument commands at a granularity of 1 second of time. The goal here was to minimise slews and dead time, and generate efficient schedules while preserving the scientific content (i.e. carrying out the high priority observations). The system worked very successfully and produced schedules with an average efficiency of 92%, where efficiency is defined as the ratio of the time the satellite was accumulating scientific data to the available science time. In fact, the actual efficiency achieved can be considered to be even higher since nearly two-thirds of the time for slewing between targets was used to gather serendipitous data at previously-unsurveyed infrared wavelengths with the photometer, and since the camera and Long Wavelength Spectrometer collected data in parallel modes when the observer has specified use of another instrument. Part of the trick was to do `overbooking'. In other words, the mission database was filled up so that it always contained about twice as many observations as could be accomodated during the remaining ISO lifetime. In essence, short lower-grade observations were used to fill in gaps between high-grade ones.
The SOC monitored the instruments in real-time as the observations were executed automatically, and had the capability to intervene manually if necessary. There were few instrument anomalies; typical interventions were, for example, the `closing' of the camera if a bright target entered its field of view. This was required to avoid saturation and its long-lasting effect on the detectors.
The final steps in the production line involved the off-line processing, quality control, archiving and finally the distribution of the data in CD-ROMs. From an operational point of view, the processing and archiving of the data worked flawlessly. Over 10000 CD-ROMs were shipped to observers. The processing algorithms and calibration were initially far from perfect and, in fact, improvements will still continue for the coming years. However, within one year of launch, an ISO-dedicated issue of Astronomy and Astrophysics containing nearly 100 papers, had been published. Given the complexity of the instruments and in particular of the behaviour of the infrared detectors, this can be considered a significant achievement.
One of the major factors in the successful operation of ISO's sophisticated instruments was the assignment to each of an `Instrument Dedicated Team' (IDT) of experts at VILSPA. The teams' responsibilities included: the overall maintenance of the instruments (including the real-time monitoring software and procedures); the calibration; and the design and much of the coding and testing of the data processing algorithms. Other experts, back at the Principal Investigator institutes, worked in close cooperation with the SOC's Instrument Dedicated Teams. These teams were crucial in making instrument operations run smoothly by rapidly diagnosing and fixing anomalies, by optimising the observing modes and by getting the instruments properly calibrated.
Much of the necessary complexity of science operation was embedded in the
over one million lines of code of the SOC software. More than 1700 Software
Problem Reports (SPRs) were responded to and over 250 System Change Requests
and Extra Wishes (SCREWs) implemented in the course of the mission. This
comes on top of the 
1000 SPRs and 100 SCREWs implemented
pre-launch, during and after the period of integration, tests and simulations.
All of the SOC's software maintenance team had been involved in the
development of the SOC software before launch. Such breadth and depth of
experience turned out to be a major factor in the success of ISO science
operations.
The SOC benefited greatly from having all functions (e.g. from establishing observing programmes to data distribution; from system design to software maintenance) integrated into one centre as this streamlined interfaces and improved communications. For the same reasons, the co-location with the SCC was also very beneficial.
Another key factor was the extensive period of end-to-end tests and simulations through which the entire ground segment software and procedures were exercised prior to launch. Not only was this essential in uncovering bugs not found by lower level tests, but it also ensured that the whole SOC was fully trained and operational at launch. In particular, the full 58 days of the Performance Verification Phase had been scheduled and validated on the software simulator prior to launch. This permitted that, 2.5 months after launch exactly as planned, the Routine Phase could start with two-thirds of the observing modes fully commissioned and ready for use by the scientific community.
