The Payload Data Handling and Telemetry Systems of Gaia
ABSTRACT The Payload Data Handling System (PDHS) of Gaia is a technological challenge, since it will have to process a huge amount of data with limited resources. Its main tasks include the optimal codification of science data, its packetisation and its compression, before being stored on-board ready to be transmitted. Here we describe a set of proposals for its design, as well as some simulators developed to optimise and test these proposals.
arXiv:astro-ph/0504285v1 12 Apr 2005
THE PAYLOAD DATA HANDLING AND TELEMETRY SYSTEMS OF GAIA
J. Portell1,2, X. Luri3,1, E. Garc´ia-Berro1,2, E.M. Geijo2
1Institut d’Estudis Espacials de Catalunya, c/Gran Capit` a 2-4, 08034 Barcelona, Spain
2Departament de F´isica Aplicada, Universitat Polit` ecnica de Catalunya, Avda. Canal Ol´impic s/n, 08860 Castelldefels,
3Departament d’Astronomia i Meteorologia, Universitat de Barcelona, c/ Mart´ ı i Franqu` es 1, 08028 Barcelona, Spain
The Payload Data Handling System (PDHS) of Gaia is
a technological challenge, since it will have to process
a huge amount of data with limited resources. Its main
tasks include the optimal codification of science data, its
packetisation and its compression, before being stored
on-board ready to be transmitted. Here we describe a set
of proposals for its design, as well as some simulators
developed to optimise and test these proposals.
Key words: payload, data handling, telemetry, data com-
pression, simulations, Gaia.
The PDHS of Gaia acquires the data coming from the
CCD focal planes, selects and prioritizes them, encodes
and compresses them and finally generates the corre-
sponding source packets to be fed into the telemetry
stream. We have developed a proposal for its global op-
eration, including the main modules and the data flux be-
tween them. For the moment we have focused on the
Astro instrument, for which we describe a possible im-
plementationof its video processingunits. This proposal,
however, can be extended to the Spectro instrument.
Our work includes not only the system design, but also
the specification of the many operations to be performed
on board. These operations include an optimal codifi-
cation of timing data, as well as an optimal transmission
scheme fulfilling the ESA packet telemetrystandard. The
main guidelines for an optimal data compression system
are also described. These guidelines will be the key for
fitting the huge amount of science data into the limited
downlink. A global view of the overall data path is fi-
nally discussed, from the on board instruments to the on
It is worth noting at this point that besides these designs
andspecificationswe havealsodevelopeda set ofsimula-
eters and for verifying the reliability of the telemetry and
data compression systems. One of these simulators is de-
signed as a large software application, receiving the out-
put of some Gaia simulators, simulating all the telemetry
and data compression system, and returning the science
data to be fed into the data base an processing system.
2.1.Payload Data Handling System
All the science data flux within the spacecraft is managed
by the PDHS, from the instruments to the communica-
tions system. The PDHS must be optimizedand designed
as a pipeline,capableofconcurrentlyprocessingthehuge
amount of data at its several stages. This turns out to be
crucial because on average about 200 stellar objects per
second will be measured (and processed), thus implying
internal data fluxes of some hundreds of Mbps.
Figure 1. Overview of the PDHS of Gaia as proposed in
The overall design of the PDHS is shown in Fig. 1.
We propose to deploy Astro in 10 identical sub-modules
which we name Astro Trail Units (ATUs) operating in
parallel. Fig. 2 shows a possible implementation of the
Astro Trail Units. Each of them will manage the mea-
surement of stars transiting over its associated trail of
Figure 2. Possible implementationof anAstro Trail Unit.
CCDs. Video chains and local sequencers are used for
this, which not only combine the digital data from a sin-
gle star measurement but also operate as interfaces be-
tween high-level commands and CCD-level commands
(Portell et al. 2003). The window acquisition manager
(WAM) is in charge of commanding this, for which an
acquisition protocol has been developed. Unnecessary
delays are avoided and the operation is pipelined. The
detection and selection algorithms indicate to the WAM
which sources must be measured and with which sam-
pling scheme option.
The use of a source priority flag is also proposed (Portell
ily discardedduringdownlinkshortages. Anotherrecom-
mended flag, the Field Density Index (FDI), would ease
the control of field-dependant PDHS operations. All of
these data selection procedures will be executed by the
science data selection module. Afterwards, these pre-
selected raw data shall be coded in an optimised way in
order to avoid unnecessary telemetry occupation. This
will be the task of the Data Handling and Compression
module, which will also include an optimised data com-
pression system. Finally, the compressed and packetised
data will be stored on-board, waiting to be transmitted
during the next contact with the ground station.
Figure 3. Proposed timing scheme for Gaia.
2.2.Optimised Time Data Codification
Some of the critical data generated by the instruments
include timing data, which can produce an important
telemetry occupation and, therefore, their codification
must be optimised. For this, science data are grouped
in data sets of 1 second length (in measurement time),
as already assumed in the baseline. In order to optimise
even more this scheme, we propose to partition every
data set in several time slots, in such a way that less bits
are required to time tag every measurement (Portell et al.
2004a). Figure 3 illustrates this scheme.
We have devised a codification scheme (Portell et al.
2004a) that not only implements our optimised timing
scheme, but also fulfills the Packet Telemetry standard
defined by ESA. Source Packets are generated accord-
ingly to our optimized codification guidelines, dynami-
cally adapting to the current observation conditions. The
core of this adaptive system is the Maximum TSM Offset
(MTO) flag, which indicates the length of every time slot
in which we partition a data set. Also, security systems
have been introduced in order to avoid any decoding er-
ror. Figure 4 illustrates our implementation proposal for
this optimized and adaptive codification system.
As ilustrated in Figure5, ourapproachexecutessome op-
erations in an order different of the usual one, packeting
the data before they are compressed. By using this proce-
dure we ensure that we keep each block of sources iden-
tified, making possible the use of optimized source pack-
ets andmakingeasier thedatapriorisationwhiledumping
the on board storage to the ground station.
We have also proposed an improved channel coding for
Gaia (Geijo et al. 2004), in order to decrease the mini-
mum elevation angle required to establish the commu-
nication. This minimum angle in the baseline is 10◦
above the horizon, while we proposed to reach as low
as 5◦. This could be achieved by adding more correcting
codes within the source packet structures, which would
decrease the codification efficiency but only during the
Figure 4. Implementation guidelines for our optimized
5◦-10◦interval. An adaptive system has been simulated
with excellent results.
3.1. Optimisation of the Time Data Codification
The reliability of our optimised proposal for the tim-
ing scheme depends on several parameters. We have
developed a software to simulate the average telemetry
data rate as a function of these parameters (Portell et al.
2004a). The snapshot shown in Figure 6 shows a static
determination of codification parameters, while Figure 7
shows the adaptive coding optimiser. It led us to a dy-
namic codingof the time dependingon the observedfield
density. The telemetry saving achieved (only with this
codificationsystem, that is, without anycompressionyet)
is about 1.2 Kbps in average, reaching up to some 10
Kbps in crowded fields. Our simulator also offers an esti-
mate of theaveragedata rate, whichis about1.2 Mbpsin-
cluding both Astro fields of view (without compression).
3.2. Telemetry CODEC
In order to obtain more accurate telemetry simulations,
a Telemetry CODEC (coder/decoder) software is being
developed (Portell et al. 2004b). A preliminary imple-
mentation of this software has already been successfully
tested, receiving realistic data generated by GASS (the
Gaia System Simulator) and converting it to raw binary
Figure 5. Communicationlayers includingour datahan-
Figure 6. Optimisation software for the codification pa-
data. Inthis way,wecanobtainrealistictelemetrycurves,
based on realistic star counts.
This software, however, is conceived as a large, dynamic
software application, capable of receiving data from dif-
ferent data simulators and performing complex opera-
tions on them. Furthermore, it will be configured with
XML files, which not only avoids the modification of the
code for adapting to different telemetry models but will
also make possible to fulfill forthcoming XML teleme-
try standards. This software is currently being coded,
but preliminarytests are also offeringsatisfactory results.
Even the data compressionsoftware shall be integratedin
the Telemetry CODEC, as well as statistical studies that
will determinethe telemetry occupationand compression
ratio achieved on every data field type.
Figure 7. Dynamicadaptationof the codificationparam-
eters to the observation conditions.
3.3. Data Compression
One of the callenges in Gaia is to transmit the large
amount of data from the satellite to the ground station.
Preliminary estimates (Lammers 2004) show an average
data generation rate of about 5 Mbps, while sustained
downlink capability will be about 1.5 Mbps. This, in
turn, implies that the data must be compressed a factor of
3 or more, using lossless algorithms whenever possible.
This is, in fact, a true challenge, since tests with standard
compression systems do not offer more than 1.5 in the
best of the cases. Therefore,a tailored and optimisedsys-
tem must be developed for Gaia. We have devised a set
of compression techniques, most of them based on PSF
and Galaxymodels,as well as differentialcoding,predic-
tors and stream partitioning (Portell 2000; Portell et al.
2001, 2004b). These will operate as pre-compressors,af-
ter which the application of standard systems offer much
better results, as shown in Table 1. These preliminary
simulations, obtained with GASS v2.1 data, reveal that
we are in the good direction since a factor of 2.4 is com-
pletely feasible by using our methods. Further detailed
studies and developments should bring us to the desired
target of 3 (or even more) in a near future.
We have proposed a set of designs for the payload data
handling system of Gaia, including an overview of the
system, its modulesand the main data flux betweenthem.
Many of these modules have also been defined, whether
as anotherset of submodules(as the Astro trail unit) or as
operationguidelines(such as the data handlingmodules).
The latter include accurate proposals for the timing and
transmission schemes, as well as for an optimised data
compression. All of these proposals take into account
the latest design of Gaia and the need for an optimised
system, in terms of hardware requirements, processing
speed and reduced telemetry occupation.
Many of these proposals depended on parameters which
Table 1. Data compression ratios obtained with different
Pre-compressorCompressor Best ratio
hadtobedeterminedforarealisticcase. Forthis, wehave
also developeda set of simulationtools which includethe
optimisation of timing parameters, the generationof real-
istic telemetry streams from simulated science data, and
the compression of these telemetry data. Although some
of these software applications are still being developed,
their preliminary results are very encouraging. Our data
compression simulator is specially interesting, since it is
offering the highest ratio currently achieved on realistic
This work has been partially supported by the MCYT
grant AYA2002–4094–C03–01, by the European Union
FEDER funds and by the CIRIT.
Geijo, E.M., Portell, J., Garc´ ıa–Berro, E., Luri, X., Lam-
mers, U., 2004, Improved channel coding for longer
contact times with Gaia, Gaia internal technical report
Lammers, U., 2004, Gaia telemetry rate simulations: a
first look at the complete picture, Gaia internal techni-
cal report (GAIA-UL-008)
Portell, J., 2000, Dise˜ no del enlace de datos para SIXE
(Spanish Italian X-ray Experiment), in Spanish, Mas-
ter Thesis, UPC
Portell, J., Garc´ ıa–Berro, E., Luri, X., 2001, Flux data
Gaia internal technical report (GAIA-BCN-001)
Portell, J., 2001, Some ideas for GIBIS and beyond,Gaia
internal technical report (GAIA-BCN-003)
Portell, J., Garc´ia-Berro, E., Luri, X., 2003, Proposals
of a payload data handling unit and internal data flux,
Gaia internal technical report (GAIA-BCN-004)
Portell, J., Garc´ ıa–Berro, E., Luri, X., 2004, Timing and
transmission schemes for Gaia, Gaia internal technical
Portell, J., Luri, X., Garc´ ıa–Berro, E., 2004, Definition
of a Telemetry CODEC, Gaia internal technical report