LOFAR
- A Multidisciplinary Wide Area Sensor Network
Interfacing the World with a Distributed Supercomputer
Prof.
Dr. Heino Falcke
The world
breathes, changes, and lives and we want to observe, understand and shape this
world. Understanding the dynamic world around and above requires us to obtain,
process, and digest an enormous amount of information. This can only by
achieved through modern information and communication technology: it requires
the construction of Wide Area Sensor Networks (WASN). A WASN consists of an
array of geographically distributed sensors which are connected by a
self-organizing, partly wireless network and an ultra-high speed backbone to a
distributed super computer. All three components are mutually dependent: network
and computing resources are constantly fed by a multitude of sensors providing
a huge data flow that in turn could only be processed by this special
high-throughput networks and computers of the next generation. LOFAR is the first large scale attempt to
implement such a concept in the real world on a large scale, driven by a wide
range of user knowledge processing user groups from a variety of disciplines.
This includes but is not limited to astronomy, geophysics, meteorology, and
agriculture where it will be a major player. LOFAR will open up the next level
of distributed computing and networking and provide significant visibility for
the technology partners involved.
Schematic view of the
LOFAR sensor network.
|
Schematic view of the
LOFAR sensor network. |
Table of Contents:
The Basic Concept for a European
WASN-Testbed
Generalization of LOFAR into a WASN
LOFAR Organization and Structure
LOFAR Organization and Structure
Wide Area Sensor Networks
The
changing world around and above us provides an enormous amount of information
that
can
be measured through a distributed
network of sensors, providing a continuous data stream
needs
to be channeled to central data
bases and monitoring and control units
needs to be processed
for our information society to understand, predict, and shape our environment.
How can one
connect different sets of geographically spread-out sensor arrays to a super
computer?
A wide Area
Sensor Network needs three basic
components:
a
large number of cheap and robust digital sensors of various kinds,
a
dedicated high-speed (10Tb/s) network over several hundreds and thousands of
kilometers,
very high performance central and distributed super
computing capabilities.
Each of
these areas requires new technologies and research and can form the basis of a
next level of information services for science as well as for governments and
citizens.
With the
huge amount of information being generated by a large set of continuously streaming
sensors, a WASN has demands beyond traditional high-bandwidth networks and
applications. Different user groups demand different sensors, bandwidths and
modes of operations. Currently planned examples of sensors required by
different user groups are:
radio dipoles
for radio astronomy creating a virtual telescope, studying: cosmology, space
weather, ionosphere, high-energy particles
particle detectors: measuring cosmic ray induced particle showers
geophones
for monitoring tectonic activity and ground water levels through
micro-earthquakes
weather stations for real-time weather and storm evolution monitoring
microphones
for infrasound applications
soil, crop, and livestock sensors for enabling precision agriculture
A Short History
of LOFAR
The initial
driver for LOFAR came from radio astronomy which has a tradition in digitally
connected element interferometers and high-bandwidth applications. The idea for
LOFAR was to replace a conventional radio telescope with a distributed network
(phased array) of simple digitally sampled dipoles. This allows one to realize a
telescope completely in software which can point and focus electronically
without any moving parts – in fact it can even look in multiple directions and
trough data buffering even back in time. At low radio frequencies this is
possible with modern technology and would provide up to three orders of magnitude
increase in resolution and sensitivity over existing facilities. The idea of
and desire for an astronomical Wide Area Sensor Network was born! Within an
international consortium of scientific institutes the basic concept and requirements
for LOFAR were specified:
2x
13,000 small antennas
clustered
in 100 stations distributed over 350 km in a “log-spiral” configuration
>20
Tbit/sec data network
(if full scientific bandwidth)
>40
Tflop distributed supercomputer
innovative
software systems
data
mining and visualization
full
and exclusive control via the Internet
instantaneous view of the full sky, several simultaneous
users.
Now, the LOFAR project is ready to make the next step…
Artists view of a LOFAR station (center) which can electronically form multiple antenna
beams (left & right) for an astronomical application..
|
Artists view of a LOFAR station (center) which can electronically form multiple antenna
beams (left & right) for an astronomical application.. |
Generalization of LOFAR into a WASN
In the
course of the R&D for LOFAR the project realized that the genuine nature of
their concept went beyond astronomy and fundamental science. As the technical
concepts matured the interest from other scientific communities and from commercial
parties steadily increased. The main aspects that make the LOFAR project
attractive for other fields are:
With
a dedicated high-speed network and 100 extended stations, additional sensors
can be added at only a small additional cost.
Digital
sensors are a main driver for high-speed networking and supercomputing, exceeding
the demands and vision of multi-media applications or computer-to-computer
networking.
A
genuine WASN has never been realized on the scale of LOFAR.
The
additional cost for this expansion of LOFAR is the burden of increased
flexibility and a wide range of user demands in terms of bandwidth, robustness,
interconnectivity, and services. For example, radio astronomy has a high demand
on bandwidth and synchronization. Geophones, microphones, weather station have
low-bandwidth but may require local pre-processing. Agricultural sensors can be
widely distributed and non-stationary (e.g. cows!). High-energy particle
detectors for astroparticle physics applications need
special triggering and synchronization. All applications may also demand
differently spaced network access points. As a result an interdisciplinary
knowledge consortium is formed that clusters around a central ICT research and
infrastructure, producing fundamental and applied research.
Key
Technologies
There are a
number of key technologies that need to be developed by various partners in the
consortium. Some of the key areas of research and development are
low-cost sensors: every user group has defined and design its own set of robust, affordable
sensors that can be coupled to the network,
generic sensor interface: coupling sensors into the network requires the definition of a generic
sensor interface that can accommodate the wide range of sensors envisaged,
broad-band data links: given the huge demand for bandwidth a (possibly routerless)
optical network has to developed using a combination of existing dark fibers
and newly laid connections,
IPv6 based protocols: a number of high- and low-bandwidth applications with guaranteed or
ad-hoc allocated bandwidth have to communicate with sensors and operators and
be integrated in a single network,
high performance cluster-computer: central processing requires not
only fast number-crunching capabilities (10 Tflop)
but also massive data throughput (25 Tbit/s) and
buffering capabilities (up to several minutes).
embedded software is needed to implement the majority of the
WASN functionality, especially the signal processing functions for the various
sensor applications, and the handling of the dynamic and self-configuring
behavior associated with some of these.
adaptive monitoring and autonomous control of the network has to cope with continuously
changing environmental conditions, changing hardware configurations, hardware
degradation, and changing user demands
intelligent web-based user interfaces: with the
enormous amount of data being processed, the results have to be stored and then
made accessible to a widely distributed user base via ‘traditional’ internet
services, making the operation fully transparent, building up and providing
access to a large interdisciplinary knowledge base.
While
sensor development and data reduction algorithm will be the prime task of the
various research communities and user groups, the remaining technology issues
are well posed problems for commercial companies and public-private
partnerships are the preferred model.
Why
should one get involved?
LOFAR is a
unique project that combines some of the most advanced aspects of modern technology
and science. As a major international science project it will provide large
visibility for all partners involved through:
Geographic
distribution: football field sized distribution of stations across the northern
parts of
Scientific
excellence: LOFAR will be at the forefront of astronomical research which
nowadays guarantees media interest and public fascination for the next decade.
LOFAR is probably the only major astronomical facility that can still be realized
in the heart of
Education:
LOFAR will provide a strong link to students and schools across
Next
generation networks: stress-testing IPv6 and security issues
Other
new technology: The WASN concept will have many future applications beyond
basic science that can affect everyday life and hence LOFAR will be a trend setter
in this area.
Testbed:
As a testbed LOFAR – driven by the high demands of basic science – allows one
to develop and test already now techniques that markets will demand in the coming
years.
What
is needed?
There are
numerous specific tasks for potential partners in the project. Some of those
specific to the WASN testbed concept include, but are not limited to:
Lay
out the LOFAR network on scales of 2, 50,
300 km distance with newly installed fibers (inner region) or using unmanaged
dark fibers provided by a partner
Provide
LOFAR nodes and stations with advanced GbE network
equipment, GSM/GPRS links, GPS receivers, etc.
Provide
various categories of sensors (e.g., measuring environmental parameters)
Develop
protocols for simultaneous use of these sensors and secure communication protocols
(IPv6 based - multitude of sensors, high performance secure transport of heterogeneous
data, dynamic behavior through radio links or "ad hoc" behavior of
some sensor types)
Conclusions
A wide area sensor network is a new and
innovative approach to sensing the physical world. It is a technology that
easily drives bandwidth demand to a new level, providing a step beyond GRID.
The reason is that here we do not have computers talking to computers, but the
real world talking to us! Computers and networks provide the key interface. The
real world will always provide more information than we can handle in any
virtual world, hence smart information reduction and processing is the key for
a WASN. A European testbed, realized through LOFAR, driven by demanding users,
will push this technology into a reality.
LOFAR Organization and Structure
The early international
LOFAR consortium that initiated the project consists of ASTRON, MIT (Haystack
The LOFAR
project has a strong regional backing in the northern Dutch provinces and on a
national level. Early 2003 a predominantly Dutch consortium has submitted a
70M€ proposal for funding of a significant part of LOFAR and a large program of
ICT related research that uses it. The consortium unites a number of Dutch
universities, research institutes, industrial enterprises, and some international
partners from
ASTRON has set
up a professional management structure with a project office and system
engineering team to handle the managerial challenges associated with this complex
international situation. As of 2003, the total staff involved in LOFAR at
ASTRON is 45.
Who
is ASTRON?
ASTRON is a
knowledge-institute for astronomical instrumentation funded by the Dutch state.
The mission of ASTRON,
the Netherlands Foundation for Research in Astronomy, is to “make astronomical discoveries
happen by providing innovative observing instruments”. As such ASTRON is a technology
provider for a very demanding customer base. We have a
long standing experience in international projects, through our participation
in international instrumentation development.
ASTRON operates
the Westerbork Radio Synthesis Telescope. Since 1993,
ASTRON has been working towards the next generation radio telescopes: the
Square Kilometre Array and LOFAR. The latter will be the first operational facility
that utilises massive signal processing, high‑speed data networks and
full internet based operations. ASTRON has a Technical Laboratory organised
around the competence areas of system engineering, antenna development, RF
electronics, digital signal processing, and software/image processing. The
Technical Laboratory also carries out major projects for optical, infrared and
millimetre instrumentation. ASTRON disseminates its knowledge and expertise to
regional SMEs through the SKAIHigh
program, sponsored by the
Contacts:
Prof.
Harvey Butcher (butcher@astron.nl), ASTRON
General Director
Dr. Mark Bentum (bentum@astron.nl), FP6 Consortium Coordinator
Prof. Heino Falcke (falcke@astron.nl), Senior Scientist LOFAR &
European Coordinator
Dr. Marco
de Vos (devos@astron.nl),
LOFAR System Engineering Manager
ASTRON,
Tel: (+31) (0)521 595 100
Fax: (+31) (0)521
597 332