63rd International Astronautical Congress, Naples, Italy. Copyright 2012 S. Engelen et al. All rights reserved.

IAC-12-D1.2.3
NanoSAR CASE STUDY OF SYNTHETIC APERTURE RADAR FOR NANO-SATELLITES
Steven Engelen
Delft University of Technology, The Netherlands, s.engelen@tudelft.nl
Maarten van den Oever
Delft University of Technology, The Netherlands, maartenvdoever@gmail.com
Pooja Mahapatra
Delft University of Technology, The Netherlands, p.s.mahapatra@tudelft.nl
Prem Sundaramoorthy
Delft University of Technology, The Netherlands, p.p.sundaramoorthy@tudelft.nl
Eberhard Gill
Delft University of Technology, The Netherlands, e.k.a.gill@tudelft.nl
Robert Meijer
TNO Delft, University of Amsterdam, The Netherlands, robert.meijer@tno.nl
Chris Verhoeven
Delft University of Technology, The Netherlands, c.j.m.verhoeven@tudelft.nl

Nano-satellites have a cost advantage due to their low mass and usage of commercial-off-the-shelf technologies.
However, the low mass also restricts the functionality of a nano-satellites payload. Typically, this would imply
instruments with very low to low resolution and accuracy, essentially ruling out applications such as remote sensing.
However, multiple nano-satellites can cooperate to improve the overall system performance, for example by
increasing the frequency of the observations. The objective of this study is to design a radar system that can be
accommodated in a nano-satellite, and investigate the feasibility of using multiples of these nano-satellites to
perform high temporal resolution remote sensing.
In this paper therefore, the concept of a nano-satellite sized Synthetic Aperture Radar (Nano-SAR) is
investigated. Nano-satellites have very constrained power and volume budgets, and there are limits to how much
surface area they can unfold for use in radar. Given these constraints, a SAR system for use in a nano-satellite in a
350 km orbit was sized, and approaches to tackle the deficits in the radar link budget are proposed. When applying
state-of-the-art technologies, both on the component level, as well as on an architectural level, one arrives at a closed
link budget.
The proposed radar system consists of a patch antenna array with a span of 1.14 m by 0.18 m, operating at a
frequency of 5.8 GHz. Power amplification and phase shifting is performed on the panel, using digital radio
frequency (RF) integrated Complementary Metal Oxide Semiconductor (CMOS) circuits. This results in a swath
width of 60 km, with pixel sizes of 10 m in elevation direction. Given these performance values, coupled with the
increased revisit times, it was obvious this radar, when flown in a larger swarm of nano-satellites, would allow faster
now-casting for weather prediction. With significant investment in technology development, it could be possible to
use this system for SAR interferometry, for near-real-time monitoring of fast ground deformation phenomena such as
earthquakes and volcanoes. Other applications could lie in the field of near-real-time ship motion detection and oil
spill spread detection.
Many technical challenges need to be solved still and platforms need to be designed, capable of supporting this
system, before this payload would be ready for deployment. Preliminary design suggests the cost of such an
instrument is substantially higher than what is common for nano-satellite components. However, the potential of
such a system is extremely promising, and merits further investigation.

I. INTRODUCTION
With the prospect of nano-satellite constellations
and swarms [1], a novel application area can be
envisaged, in which multiple cooperating satellites
orbiting Earth reduce revisit times of arbitrary areas to
several hours, or even less. Interesting as this may seem,
nano-satellites traditionally havent had the best of

IAC-12-D1.2.3

instruments, as they are limited in size, mass and

perhaps most importantly, power. They do have a cost
advantage [2] over traditional satellites, as they
primarily use off-the-shelf components, which allow
using them in larger numbers.
Large numbers of nano-satellites do present a
potential space-debris issue [3], which is why in this

Page 1 of 6

63rd International Astronautical Congress, Naples, Italy. Copyright 2012 S. Engelen et al. All rights reserved.

paper, we assumed them to be orbiting in very low earth

orbits. These orbits would be self-cleaning in case of
a loss of orbit maintenance operations of the satellite in
question, as the average orbital lifetime of a 350 km
orbit is less than 100 days. This strategy, though
expensive in terms of propellant and the requirement of
an active propulsion system effectively eliminates faulty
satellites from the system. It has another bonus, in that
the satellites are much closer to their target, allowing for
higher resolution sampling, with more modest
instruments.
Our paper attempts to make a case for a nanosatellite synthetic aperture radar (Nano-SAR), and
investigates the plausibility of such a system in
combination with a nano-satellite platform.
II. NANO-SATELLITE BUDGETS
Nano-satellites are defined as satellites with a wet
mass of less than 10 kg [4]. The most common
platform however is a so-called CubeSat [4], which
measures the satellites in units of 100x100x100 mm,
generally with a wet mass of around 1 kg per Cube.
Given these mass constraints, and also the ensuing
volume constraints, the incoming power, and the
thermal envelope of such satellites, it should be clear
these satellites are severely restricted in their operations.
To put his into perspective: the orbit average
incoming power for a nano-satellite in low earth orbit
varies between 5 to 15 watts [5], when sufficient surface
area is present, and is ever increasing, especially when
applying deployable solar panels [6]. Where higher
power levels are concerned, thermal issues in removing
the excess heat becomes a major obstacle, as dissipating
over 20W in a small nano-satellite is a challenge indeed.
Attitude control accuracies are generally low, with
common values being reported as 5 pointing
knowledge, and 10 pointing accuracy. However, higher
performance solutions are surfacing on the market, with
advanced attitude determination systems using Earthhorizon scanners [7] and even miniaturised star trackers
combined with reaction wheels for more precise control,
allowing pointing knowledge at levels of hundreds to
tens of arcseconds, and pointing accuracies well below a
degree [8].
Given these constraints, a high power, high
resolution, high accuracy system will not be feasible.
In order to circumvent some of these constraints, our
system assumes the presence of an energy storage
device, such as a battery, allowing for higher peakpower availability. This reduces the demand on the solar
panel size, at the expense of requiring a pulsed and/or
duty-cycled system. Furthermore, we assume a total
available payload power of 15 Watts over the course of
its operations, effectively leaving around 5 W for the
spacecraft bus operations.

IAC-12-D1.2.3

In return, the data downlink is assumed to be

handled by the radar system, relieving the spacecraft
somewhat of its tight power constraints.
Also, the large surface area of the radar panel can be
used in combination with solar cells, to provide more
power to the satellite.
III. THE RADAR SYSTEM
When the radar is used for SAR, the swath width is
limited by multiple factors. The first factor that is
important is the Pulse Repetition Frequency (PRF). The
PRF is determined by the speed of the platform and the
antenna length. To create a synthetic aperture without
ambiguities in the final image, the radar needs to
transmit two pulses per antenna length travelled. Thus a
smaller antenna and a higher speed, i.e. a lower orbit,
require a higher PRF. The drawback of a higher PRF is
that the unambiguous range of the radar decreases.
For an orbit of 350 km and an antenna of 1.14 m in
azimuth the minimum PRF is in the order of 20 kHz.
Such a high PRF leads to an unambiguous range of
about 6 km. A swath of 6 km is common in small
satellite systems [9]. However an ambiguous range of
6 km will cause ambiguous projections that are still in
the main beam to show up in the final image. This is a
reason that such a high PRF is not feasible for our
system.
A solution to overcome these problems is to lower
the PRF, and use measurements from the other satellites
to fill the synthetic aperture. However such a solution
causes problems in the demodulation of the signal,
requires tight synchronisation, and requires accurate
orbit control and pointing.
In [10] a technique is presented that enables the
utilisation of small antennas with Digital Beam Forming
capabilities, which enable lowering of the PRF. With
the utilisation of these techniques the PRF times the
number of sub apertures needs to be higher than the
Doppler bandwidth.
The Doppler bandwidth is given by twice the
satellite speed, divided by antenna length. In our case
the Doppler bandwidth is about 14 kHz, and the
resulting PFR is 2 kHz with 7 sub apertures. The
unambiguous range is 60 km in such a solution.
Since the footprint of the radar is about 600 km in
elevation ambiguities remain. A possible way to solve
this is through utilisation of an encoded pulse. Coding
for conventional radar is presented in [11] and [12] .
This technique can also be used to reduce ambiguities
through cycling through different codes. In doing so,
reflections of each pulse can be distinguished, as they
are encoded with a different code.
In SAR the phase of the signal is important for the
mapping of the received power. If coding is used the
phase still needs obtained for the mapping of the receive
power. In our paper we assume the phase to be known.

Page 2 of 6

63rd International Astronautical Congress, Naples, Italy. Copyright 2012 S. Engelen et al. All rights reserved.

An advantage of using coding is that the code can be

used to increase the signal to noise ratio. In our system
the coding is used for this purpose, however the
processing to perform SAR needs to be adjusted, which
is considered outside of the scope of this work. Also the
gain of coherent addition of the signals is lost since
demodulation is performed before addition of the
measurements.
The resolution in azimuth in a radar system is given
by half the antenna length; and, for non-coded systems,
in elevation by
=

[1]

where B is the bandwidth. The effective ground

range resolution is given by,
=

sin

[2]

with the incidence angle. For a bandwidth of

40 MHz the ground range resolution would be 10 m for
an incidence angle of 24 degrees.
The power required to operate a radar system can be
calculated using the radar formula.
=

(4)3 4

[3]

where Pt represents the transmitted power, Pr the

received power, G the antenna gain, the wavelength,
the radar cross section of the target, L the system losses
and R the range form the platform to the target.
Due to the relatively large distances involved in
satellite communications the transmitted power needs to
be increased. The power needs to be increased with the
range to the power four to receive the same power
compared to a conventional system. If the radar is used
for Synthetic Aperture Radar (SAR) the evaluation of
the system needs to be adapted. We introduce the Signal
to Noise Ratio (SNR), which is given by /().
The Noise Equivalent Sigma Zero (NESZ) is a
measure for the sensitivity of the SAR radar instrument.
Introducing the integration of the SAR system the radar
equation can be rewritten to:
=

2(4)3
2 3

[4]

where V is the orbital velocity and Pt the average

power transmitted. With this formula the sensitivity of
the SAR system can be determined.

IAC-12-D1.2.3

III. CODING GAIN

In the radar system the coding is going to be used to
reject ambiguities. By carefully choosing the codes the
multiple signals can be distinguished. Thus ambiguous
projections that normally would show up in the swath
can be distinguished and removed in the processing.
Another advantage of coding is that the SNR increases,
thus the signals with low power levels can still be
recovered. A disadvantage is that the actual power level
of the signal is not easily obtained. This introduces
severe problems in the processing. The entire processing
of the signals to obtain the final image needs to be
changed in order to work with coding. In this first study
the effect of the coding on the processing is not
considered, coding is only used to increase the SNR.
The actual gain by coding in dB is given by
10 (2), in which L represents the code length. For
a code length of 40000 this gain amounts to 49dB. The
received power, excluding the coding gain, is shown in
Table 1.
Item

Value
[linear]
750

0,06

m2

1.5 1022

m4

Value
[dB]
58,75
24
40
0
-24,437
0
222

dBm

-156,42
dBm

Table 1: Received power parameters, and result,

excluding coding gain
Due to the utilisation of the patch antenna the gain is
not the same for transmit and receive. The received
power of -156 dBm is low and might be a problem.
Without coding the SNR in such a system would be -28
dB which does not lead to feasible applications.
However with the coding used the SNR is over 20 dB.
Now if the radar is going to be used for SAR the
impact of the coding is large, since coherent addition of
the samples is not trivial, and the detecting of the phase
and the power levels is more difficult. However with
sufficient computing power it should be possible to
perform SAR-like operations. In that case indication of
the performance can be obtained from the formula for
the NESZ. The duty cycle during operation is set to
0.02. Without the coding considered a NESZ of 25 dB is
obtained which is not sufficient. However if the coding
could be utilized a NESZ of -24 dB would be obtained.

Page 3 of 6

63rd International Astronautical Congress, Naples, Italy. Copyright 2012 S. Engelen et al. All rights reserved.

IV. DIGITAL RF POWER AMPLIFIERS AND

SIGNAL GENERATORS
Another novelty in this system is in the use of socalled digital RF [13] power amplifiers and signal
generators. In such a system, the signal is generated
fully digitally.
This results in extremely high efficiencies for the
power amplifier, which resembles a switch-mode
amplifier. The signal generation, since it is fully digital,
is also entirely flexible, simplifying the phase matching
in the phased array significantly. Also, any type of
signal can easily be generated, allowing for accurate
beam-forming and pattern generation on the antenna.
As these components can be (mass-)produced in
bulk silicon, each patch antenna could easily be
transformed into an active patch antenna, with its own
signal generation and TX/RX front-end; reducing cable
and connector losses. Simultaneously, the pulse power
is generated on the radar panel, allowing for much
easier thermal management compared to a power
amplifier confined to the satellite main body.
V. RESULTS
V.I. Radar properties
The specifications of the resulting radar system are
summarised in Table 2. It can be used in an altimeter
mode (radar mode) and in SAR mode. When used as
a SAR, the resolution is dramatically increased; at the
expense of a 10 times reduction in swath width.
Value
15
60
5,8
118
18
0,5
0,7
2

Unit
W
%
GHz
cm
cm
2
%

Input power
Power Amplification Efficiency
Frequency
Phased array length
Phased array width
Patch dimensions
Efficiency
Pulse duty cycle
Phased Array Element, max
beam width
60

Phased Array Element gain

6
dB
Code length
40000
chips
Number of antennas
322
Module peak output power
2,33
W
Phased Array Receiver antenna
gain
31
dB
Transmit (pulse) power
750
W
Bandwidth
40000
kHz
Pulse duration
40
s
Energy per pulse
0,042
J
Table 2: Specifications of the proposed radar system

IAC-12-D1.2.3

V.II. Revisit times

Satellites in a low 350 km orbit have a fast ground
swath velocity, and a small field of view, compared to
higher orbiting satellites. In order to assure continuous
global coverage, a constellation of satellites is required.
The number of satellites, and also the number of distinct
inclinations of their orbital planes is defined
predominantly by their instrument field of view, and the
desired revisit time. Also, the overlap between swaths
and the potential gaps in observations are defining the
number of required satellites. Earths rotation at these
low orbits does play a role, but it is minor, compared to
satellites in higher orbits.
Table 3 lists the revisit times for various orbital
planes and numbers of satellites, for a stationary Earth
at 350 km, with a given swath width of 600 km in radar
mode, as well as the effective revisit time when the
system is used in pure SAR mode. This effective revisit
time results from the smaller effective swath width in
case the radar is used purely in SAR mode.
Note these scenarios produce oversampling of
certain areas (e.g. the North- and South-pole). The
constellation or swarm is therefore over-dimensioned.
The question of whether full coverage of all of Earth
is required is something left to the mission designers.
However, when proposing for a mission requiring a
large number of objects in low earth orbit, the unit cost
should be as low as possible.
V.III. Scientific use
With a swath width of 60 km and a pixel size of 10 x
3 m, the Nano-SAR system promises to be of
comparable resolution characteristics with existing SAR
satellites such as Radarsat-2 or even TerraSAR-X. With
the possibility of revisit times going down to 25 days,
Nano-SAR could be utilised for many of the
conventional
SAR
applications
that
utilize
amplitude/intensity information, such as change
detection, oil spill detection, surveillance and so on.
Higher revisit times also mean that faster-changing
phenomena can be observed, i.e., the temporal sampling
frequency of the observed phenomenon can be
improved compared to existing single-satellite systems.
Spatial and temporal resolutions comparable to
constellations such as Cosmo-Skymed maybe be
achieved more economically.
For Nano-SAR, we have assumed the phase of the
radar signal to be known. However, with some further
computation cost, this phase information may also be
extracted from the signal itself, and this could lead to
the applicability of Nano-SAR for SAR interferometry
applications. An additional requirement for this would
be that precise positioning information is available on
each of the nano-satellites.

Page 4 of 6

63rd International Astronautical Congress, Naples, Italy. Copyright 2012 S. Engelen et al. All rights reserved.

Desired
satellite
revisit time
[min]

Required number of
satellites per orbital
plane
[-]

RADAR MODE
Swath width = 600 km,
pixel size = 14000x15 m

SAR MODE
Swath width = 60 km,
Pixel size = 10 x 3 m

Number of orbital planes

[-]

Total number of satellites

[-]

150
0.61
66
40
135
0.68
66
45
120
0.76
66
50
105
0.87
66
57
90
1.02
66
67
75
1.22
66
80
60
1.52
66
101
45
2.03
66
134
30
3.05
66
201
15
6.09
66
402
Table 3: Revisit times as function of orbital planes and number of satellites
With effective revisit times down to 45 minutes over
the area of interest, it would be possible to build up a
useful time series of data within about 15 hours, and
using this, various effects of different atmospheric
conditions during the acquisitions may be cancelled out,
and ground deformation history measured from the
residual phase. For a time series of all of Earth, the
effective revisit time amounts to about 25 days, for the
case of 100 satellites, which renders a useful time series
of data within about 500 days.
In case the radar is used in radar mode, nowcasting can be performed with pixel sizes of 14 km x 15
m, with revisit times of 45 minutes or more, for a
reasonable swarm size.
VI. CONCLUSIONS
Advances in nano-satellite platforms and innovative
mission design enable the realization of novel space
applications. Such enabler could be a Nano-satellite
SAR, offering the possibility of medium resolution
Earth observation at affordable costs. Nano-SAR
combines the limited capabilities of multiple individual
nano-satellites to provide both high spatial resolution
and high revisit times making it suitable for an array of
earth observation missions.
In this paper, the initial specifications for a radar
system, tailored to nano-satellite platforms have been
outlined. This radar would provide a resolution of 14
km x 15 meter, with a swath width of some 600 km;
when flown in an orbit at 350 km altitude. The radar can
also be used in SAR mode, in which it generates 7 subapertures, rendering a total swath width of 60km , with
pixel sizes of 10 x 3 meter.

IAC-12-D1.2.3

Effective
revisit time
[days]
63.13
56.81
50.50
44.19
37.88
31.56
25.25
18.94
12.63
6.31

It is also apparent that quite some technology

development is required in order to allow demonstration
of this system, but the applications could well prove to
be worth the effort.
VI. REFERENCES
[1] C. Verhoeven, B. Monna, S. Engelen, A. Noroozi,
P. Sundaramootrhy, M. Bentum en R. Meijer,
Emerging Eco-system: Nano-satellite Swarms and
Large Satellites, in IAC 2011, IAC-11,
D1,1,6,x11045, Cape Town, 2011.
[2] C. I. Underwood, M. J. Crawford en J. W. Ward,
A Low-Cost Modular Nanosatellite Based On
Commercial Technology, in 12th AIAA/USU
Conference on Small Satellites, SSC98-V-4, 1998.
[3] R. Walker, C. E. Martin, P. H. Stokes en H.
Klinkrad, Sensitivity of long-term orbital debris
environment evolution to the deployment of nanosatellite swarms, Acta Astronautica, vol. 51, nr. 19, pp. 439-449, 2002.
[4] J. R. Wertz, D. F. Everett en J. J. Puschell, Space
Mission Engineering: The New SMAD,
Microcosm Press, 2011.

Page 5 of 6

63rd International Astronautical Congress, Naples, Italy. Copyright 2012 S. Engelen et al. All rights reserved.

[5] T. Shimizu en C. Underwood, SUPERCAPACITOR ENERGY STORAGE FOR

MICRO-SATELLITES:DEVELOPMENT AND
POTENTIAL MISSION APPLICATIONS, in
IAC 2011, IAC-11-C3.3.4, Cape Town, 2011.
[6] C. Clark, Huge Power Demand...Itsy-Bitsy
Satellite: Solving the Cubesat Power Paradox, in
24th Annual AIAA/USU Conference on Small
Satellites, Logan, UT, 2010, SSC10-III-5.

[12] Q. I. WeiKong en Y. U. WeiDong, A novel

operation mode for spaceborne polarimetric SAR,
SCIENCE CHINA, Information Sciences, vol. 54,
nr. 4, pp. 884-897, 2011.
[13] K. Muhammad, R. B. Staszewski en D. Leipold,
Digital RF Processing: Toward Low-Cost
Reconfigurable Radios, IEEE Communications
Magazine, pp. 105-113, August 2005.

[7] The 1/2U MAI-400 A La Carte, [Online].

Available: http://www.miniadacs.com/linked/201204-27%20mai400%20a%20la%20carte%20specification.pdf.
[Visited September 1st 2012].
[8] iADCS-100: Intelligent Attitude Control for
Cubesats, [Online]. Available: http://www.berlinspace-tech.com/fileadmin/media/BST_iACDS100_Flyer.pdf. [Visited September 1st 2012].
[9] P. Hoogeboom, M. Otten, S. Monni, F. Elferink, A.
Meta and P. van Duijn, Formation flying FMCW
SAR sensor for remote sensing applications., in
ARSI 2011, Advanced RF Sensors and Remoste
Sensing Instruments, Noordwijk, 2011.
[10] G. Krieger, N. Gebert en A. Moreira,
Multidimensional Waveform Encoding: A New
Digital Beamforming Technique for Synthetic
Aperture Radar Remote Sensing, IEEE
TRANSACTIONS ON GEOSCIENCE AND
REMOTE SENSING, vol. 46, nr. 1, pp. 31-46,
2008.
[11] J. Chen, Z. Li en C. S. Li, A NOVEL
STRATEGY FOR TOPSIDE IONOSPHERE
SOUNDER BASED ON SPACEBORNE MIMO
RADAR WITH FDCD, Progress In
Electromagnetics Research, vol. 116, pp. 381-393,
2011.

IAC-12-D1.2.3

Page 6 of 6