SWEET CUBESAT – WATER DETECTION AND WATER QUALITY MONITORING FOR THE 21 ST CENTURY

Abstract

Water scarcity and contamination of clean water have been identified as major challenges of the 21 st century, in particular for developing countries. According to the International Water Management Institute, about 30 percent of the world´s population does not have reliable access to clean water. Consequently, contaminated water contributes to the death of about 3 million people every year, mostly children. Access to potable water has been proven to boost education, equality and health, reduce hunger, as well as help the economy of the developing world. Currently used in-situ water monitoring techniques are sparse, and often difficult to execute. Space-based instruments will help to overcome these challenges by providing means for water level and water quality monitoring of medium-to-large sweet (fresh) water reservoirs. Data from hyperspectral imaging instruments on past and present governmental missions, such as Envisat and Aqua, has been used for this purpose. However, the high cost of large multipurpose space vessels, and the lack of dedicated missions limits the continuous monitoring of inland and coastal water quality. The proposed CubeSat mission SWEET (Sweet Water Earth Education Technologies) will try to fill this gap. The SWEET concept is a joint effort between the Technical University of Munich, the German Space Operations Center and the African Steering Committee of the IAF. By using a novel Fabry-Perot interferometer-based hyperspectral imager, the mission will deliver critical data directly to national water resource centers in Africa with an unmatched cost per pixel ratio and high temporal resolution. Additionally, SWEET will incorporate education of students in CubeSat design and water management. Although the aim of the mission is to deliver local water quality and water level data to African countries, further coverage could be achieved with subsequent satellites. Finally, a constellation of SWEET-like CubeSats would extend the coverage to the whole planet, delivering daily data to ensure reliable access to clean water for millions of people worldwide. Nomenclature A = surface area [m 2 ] B = Earth magnetic field [T] c = speed of light [m/s] Cd = drag coefficient cg = center of gravity [m] cpa = center of aerodynamic pressure [m] cps = center of solar pressure [m] D = residual dipole [Am 2 ] Fs = solar constant [W/m 2 ] G = gravitational constant [Nm 2 /kg 2 ] θ = maximum deviation of the z axes from local vertical [rad] i = Sun angle of incidence [°] Iy = moments of inertia about y axes [kgm 2 ] Iz = moment of inertia about z axes [kgm 2 ] M = Earth magnetic moment [Tm 3 ] µ = Earth gravity constant [m 3 /s 2 ] q = reflectance factor R = orbit radius [m] ρ = atmospheric density [kg/m 3 ] Ta = aerodynamic torque [Nm] Tg = gravity gradient torque [Nm] Tm = magnetic torque [Nm] Tmin = minimum torque [Nm] Tsp = solar radiation pressure [Nm] Ttot = total torque [Nm] V = spacecraft velocity [m/s]

Full text

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 1 of 11
IAC-16,B4,1,12,x34041
SWEET CUBESAT – WATER DETECTION AND
WATER QUALITY MONITORING FOR THE 21ST CENTURY
Kelly Antoninia*, Martin Langera, Ahmed Faridb,c, Ulrich Waltera
a Institute of Astronautics, Technical University of Munich, Boltzmannstraße 15, 85748 Garching, Germany,
kelly.antonini@tum.de
b German Space Operations Center, Telespazio VEGA, Münchener Straße 20, 82234 Weßling, Germany
c IAF-Africa Co-Chairman
* Corresponding Author
Abstract
Water scarcity and contamination of clean water have been identified as major challenges of the 21st century, in
particular for developing countries. According to the International Water Management Institute, about 30 percent of
the world´s population does not have reliable access to clean water. Consequently, contaminated water contributes to
the death of about 3 million people every year, mostly children. Access to potable water has been proven to boost
education, equality and health, reduce hunger, as well as help the economy of the developing world. Currently used
in-situ water monitoring techniques are sparse, and often difficult to execute. Space-based instruments will help to
overcome these challenges by providing means for water level and water quality monitoring of medium-to-large sweet
(fresh) water reservoirs. Data from hyperspectral imaging instruments on past and present governmental missions,
such as Envisat and Aqua, has been used for this purpose. However, the high cost of large multi-purpose space vessels,
and the lack of dedicated missions limits the continuous monitoring of inland and coastal water quality. The proposed
CubeSat mission SWEET (Sweet Water Earth Education Technologies) will try to fill this gap. The SWEET concept
is a joint effort between the Technical University of Munich, the German Space Operations Center and the African
Steering Committee of the IAF. By using a novel Fabry-Perot interferometer-based hyperspectral imager, the mission
will deliver critical data directly to national water resource centers in Africa with an unmatched cost per pixel ratio
and high temporal resolution. Additionally, SWEET will incorporate education of students in CubeSat design and
water management. Although the aim of the mission is to deliver local water quality and water level data to African
countries, further coverage could be achieved with subsequent satellites. Finally, a constellation of SWEET-like
CubeSats would extend the coverage to the whole planet, delivering daily data to ensure reliable access to clean water
for millions of people worldwide.
Keywords: CubeSat, Hyperspectral, Africa, Water Quality, Mission Design, Constellation
Nomenclature
A = surface area [m2]
B = Earth magnetic field [T]
c = speed of light [m/s]
Cd = drag coefficient
cg = center of gravity [m]
cpa = center of aerodynamic pressure [m]
cps = center of solar pressure [m]
D = residual dipole [Am2]
Fs = solar constant [W/m2]
G = gravitational constant [Nm2/kg2]
θ = maximum deviation of the z axes from local vertical
[rad]
i = Sun angle of incidence [°]
Iy = moments of inertia about y axes [kgm2]
Iz = moment of inertia about z axes [kgm2]
M = Earth magnetic moment [Tm3]
µ = Earth gravity constant [m3/s2]
q = reflectance factor
R = orbit radius [m]
ρ = atmospheric density [kg/m3]
Ta = aerodynamic torque [Nm]
Tg = gravity gradient torque [Nm]
Tm = magnetic torque [Nm]
Tmin = minimum torque [Nm]
Tsp = solar radiation pressure [Nm]
Ttot = total torque [Nm]
V = spacecraft velocity [m/s]
Acronyms/Abbreviations
Attitude Determination and Control Subsystem (ADCS)
Cyanobacterial blooms (cHABs)
CubeSat Design Specification (CDS)
Commercial off-the-shelf (COTS)
German Aerospace Center (DLR)
Electrical Power Systems (EPS)
International Space Station (ISS)
Institute of Astronautics (LRT)
Micro-Electro-Mechanical Systems (MEMS)
MEdium Resolution Imaging Spectrometer (MERIS)

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 2 of 11
Multi-layer insulation (MLI)
Moderate Resolution Imaging Spectroradiometer
(MODIS)
Munich Orbital Verification Experiment (MOVE)
Not Applicable (N/A)
Near infrared (NIR)
On Board Computer (OBC)
Operating System (OS)
Printed Circuit Board (PCB)
Right Ascension of the Ascending Node (RAAN)
Random-access memory (RAM)
Real Time Operating System (RTOS)
Secure Digital (SD)
Single Event Effect (SEE)
Space Mission Analysis and Design (SMAD)
Sun-synchronous orbit (SSO)
Semi-analytic Tool for End of Life Analysis (STELA)
Systems Tool Kit (STK)
Sweet Water Earth Education Technologies (SWEET)
Secure World Foundation (SWF)
To Be Determined (TBD)
Total Ionizing Dose (TID)
Technical University of Munich (TUM)
Ultra High Frequency (UHF)
United Nations Educational Scientific and Cultural
Organization (UNESCO)
Very High Frequency (VHF)
Visible (VIS)
Women in Aerospace (WIA)
1. Introduction
The aim of the proposed Sweet Water Earth
Education Technologies (SWEET) CubeSat mission is to
provide water quality and water level data to African
countries. The objective is to image inland sweet water
lakes which are a source of drinking water to millions of
people in Africa. The mission is currently at a Phase-0
[1] stage, the work presented here consists of an initial
mission identification and analysis, and a feasibility
study. This paper addresses the question if water quality
measurements using a hyperspectral camera are possible
on CubeSats. The study is mainly limited to the analysis
of the satellite systems needed to support the mission.
The SWEET project will initially focus on a precursor
mission, demonstrating the usefulness of the generated
data. Subsequently, a constellation of SWEET-like
CubeSats will be built to increase the temporal
resolution, essential in detecting rapid changes in water
quality, sometimes occurring at an hourly rate. SWEET
will also have an educational purpose: students at the
Technical University of Munich (TUM), collaborating
with African universities and students, shall build the
precursor mission. Presently the SWEET team is looking
for financial and development partners.
This paper will initially provide background
information about past hyperspectral instruments and
CubeSats, and motivate the intent of the proposed
mission. Section 2 introduces the chosen instrument and
how it fulfils the mission’s objectives. Section 3 focuses
on the orbit selection process, and section 4 on the
SWEET bus, which is subdivided into 6 subsystems.
Section 5 shows the results obtained from this initial
analysis and discusses the outcomes. Finally, section 6
concludes the paper.
1.1 Background & Motivation
Hyperspectral imaging from space has traditionally
been carried out by large vessels aiming at several
scientific goals. Instruments like Moderate Resolution
Imaging Spectroradiometer (MODIS) on Terra and Aqua
[2] or MEdium Resolution Imaging Spectrometer
(MERIS) on Envisat [3] provided hyperspectral images
in a spatial resolution of several hundreds of meters.
Since their introduction in 1999, CubeSats [4] have
evolved from purely educational missions to spacecraft
with a broad variety of scientific and commercial
applications. Hyperspectral imaging on CubeSats has
been proposed for natural hazard response [5], to
improve weather forecasts [6], for measuring the ozone
concentration in the stratosphere [7], for land use
classification, and for vegetation mapping and algae
detection [8].
According to the United Nations, by 2025, two-thirds
of the world's population could be living under water-
stressed conditions, with Sub-Saharan Africa being the
region with the largest number of water-stressed
countries [9]. Human-made factors such as increasing
water usage and climate change [10] will increase the
scarcity of water, threatening successful achievement of
most of the Millennium Development Goals. Amongst
others, harmful cyanobacterial blooms (cHABs) have
significant socioeconomic and ecological impact
[11-13].
As reported by the United Nations Educational
Scientific and Cultural Organization (UNESCO), cHABs
are present in African sweet water lakes [14]. Although
the information is very scarce, they confirm that the
cyanobacterial blooms have caused animal death and a
few cases of human casualties due to contaminated
drinking water [14]. Many factors influence the time it
takes to form cHABs, ranging from water temperature,
wind speed, solar radiation, and rainfall, making it
difficult to determine how often a lake should be imaged
[15]. According to [16], the lake water quality monitoring
procedures currently in place in Northern Egypt rely on
monthly in situ measurements, though the development
of cHABs can occur in timespans ranging from hours to
weeks. Remote sensing from spacecraft has reduced this
time interval in the past. MERIS used to provide data at
1 to 3 day intervals [15], with several research groups
applying the data for cHABs assessment [17-21].
Similar to what the MERIS instrument achieved in
the past, CubeSats could provide continuous monitoring

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 3 of 11
of inland water reservoirs, saving lives, time, and at a
fraction of the cost of a traditional mission. The proposed
SWEET precursor mission will study the cHABs trend in
the interested areas, and subsequently adapt the
constellation revisit time accordingly.
2. Instrument
The SWEET CubeSat main objectives could only be
achieved by selecting an appropriate payload, a
hyperspectral imager. The requirements of the payload
include: appropriate size and power consumption,
accommodated into a CubeSat; wavelength bands in near
infrared (NIR) to measure inland sweet water height and
quality; good spatial resolution, to image a large variety
of lake sizes; and cloud cover quantification in the
hyperspectral image.
The imagers which met the SWEET requirements
were: HyperScout built by cosine [22], and the CubeSat
imager shown in Figure 1 built by the VTT Technical
Research Centre in Finland [23].
Fig. 1. VTT hyperspectral imager [24]
The imager from VTT was selected for SWEET, as
the combination of NIR and visible (VIS) wavelength
bands best fulfilled the payload requirements. At the time
of this writing, the instrument is to be flown on CubeSats:
Aalto-1 (Aalto university) [8, 25-28] and PICASSO
(ESA) [7, 23, 29], verifying its usefulness on CubeSats.
Table 1 summarizes its characteristics.
Table 1. VTT hyperspectral imager properties
Parameter
Values
Wavelength range [26]
500 - 900 nm
(VIS-NIR)
Spectral resolution [26]
10 to 30 nm
Field of view [30]
10 x 10°
Mass [30]
< 600 g
Cubesat Unit Size [30]
0.5 U
Power consumption [26]
3 W
Image size [30]
512 x 512 pixel
The instrument can be accommodated into a
2U-CubeSat and configured for a selection of bands. The
most appropriate will be selected to address the most
dangerous and common substances affecting the quality
of drinkable water. The instrument’s spatial resolution is
expected to be similar to MODIS on Aqua, 250 m at
700 km altitude, at a fraction of the hardware size [28,
31]. The imager uses a Fabry-Perot-based interferometer
to capture the image, and it is provided with a built-in
optical VIS camera, capable of determining the cloud
fraction [28]. Each hyperspectral image is expected to be
0.5 MB in size.
3. Orbit Selection
The orbit selection process considered various orbit
options, selecting the one that satisfied the largest
number of mission requirements. Out of the many
available orbits, an earth-referenced orbit had to be
selected to analyze the surface of the Earth. The SWEET
orbit shall cover as many African lakes per day as
possible.
SWEET has to comply with space debris mitigation
guidelines [32], which prescribes satellites to de-orbit
from a LEO within 25 years. According to NASA, the
altitude cut-off for a CubeSat to naturally de-orbit within
25 years is between 600 and 700 km [33]. As SWEET
will have no propulsion, an initial altitude of 650 km
would fulfil the rule. This altitude provides relatively
large ground coverage, good ground station
communication times and mission lifetime, and last but
not least, a high payload resolution. Table 2 shows the
four possible circular LEO orbit types analyzed.
Table 2. Properties of the four analyzed SWEET orbits
Altitude
(km)
Inclination
(°)
Period
(min)
SSO
650
97.9
97.7
50° Orbit
650
50
97.7
Polar Orbit
650
90
97.7
ISS Orbit
400
51.6
92.6
The advantage of a Sun-synchronous orbit (SSO) is
a constant Sun illumination on ground throughout the
mission, therefore generating consistently illuminated
images. The 50° orbit promises to best fit the mission’s
objective by providing more accesses to Africa and the
chosen ground segments, however, piggyback launch
opportunities to such an orbit are rather rare. The
advantage of the SSO and polar orbit is that it provides
global coverage, whilst the International Space Station
(ISS) orbit, having a lower initial altitude, provides
images with higher resolution, but has a shorter lifetime.
All considered orbits, except for the 50° orbit, have many
launch opportunities. After considerable analysis,
simulating coverage of African lakes and ground station
communication time, it was decided that the SWEET
CubeSat precursor mission shall fly on an ISS orbit.
Despite the disadvantage of a short life-time, this orbit
has the highest number of launch opportunities and

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 4 of 11
provides higher spatial resolution images (137 m at
400 km altitude) due to its low altitude. Since SWEET is
considered to be a precursor mission for a future
constellation, launching at a lower altitude will suffice to
prove the concept.
4. Bus System
To guide the selection of the bus components, the
mission definition approach from Space Mission
Analysis and Design (SMAD) [34] was used. The
satellite shall take at least 4 hyperspectral images a day,
and be able to download two of them (hypothesizing a
50 percent chance of inappropriate cloud cover). The
SWEET CubeSat precursor mission will download
thumbnails of the optical image to the ground station, to
determine whether the cloud fraction is low enough to
justify downloading the complete hyperspectral image.
The main requirement for the SWEET bus is to
maximize use of the Munich Orbital Verification
Experiment 2 (MOVE-II) design. MOVE-II is a CubeSat
currently being developed at TUM [35], which is in turn
a successor of the first TUM satellite, First-MOVE [36].
Re-using parts of the MOVE-II bus subsystem design
reduces risk, uncertainty in the component behavior, and
lowers the SWEET development costs. In addition, the
SWEET CubeSat shall comply to the CubeSat Design
Specification (CDS) [37].
4.1 Communications Subsystem
On the SWEET CubeSat it is planned to use the same
transceivers as on MOVE-II: ultra high frequency (UHF)
to upload, very high frequency (VHF) and S-band to
download [38, 39]. For the VHF channel, a data rate of
9.6 Kbit/s was estimated, choosing an omnidirectional
antenna pattern on the satellite and a field of view of the
antenna at the ground segment of 160° (due to
obstructions and atmospheric interferences, the last 10°
close to the horizon were neglected). Preliminary link
budget calculations showed that the VHF-channel would
allow to download one 0.5 MB image every 8 to 9
minutes of communication time.
For the S-band communication, a rate of 1 Mbit/s is
taken to be a realistic value [39]. Using a conservative
approach, it was estimated that the download of data
using S-band will only be possible on average every two
days, due to power and pointing restrictions (and limited
field of view of 15° half angle of the S-band patch
antenna on the satellite).
To enhance SWEET’s educational purpose,
university ground stations will be used. In the first
simulation, the following three ground stations were
selected: the TUM university ground station at the
Institute of Astronautics (LRT), in Garching, Munich,
and two more ground stations strategically positioned on
the African continent (the current simulation has one in
Abuja, Nigeria and one in Stellenbosch university, South
Africa). With those ground stations, it was possible to
download at least 5 images per day. Ongoing work is
being carried out to determine an optimum balance
between the number of necessary ground stations and
resource usage on the spacecraft.
4.2 Power Subsystem
The power subsystem consists of solar cells and a
battery to provide power to the payload and all other
subsystems. An initial estimate of power to be generated
was defined based on the payload’s power requirement
of 3 W and the MOVE-II bus power requirement.
Including a safety margin of 25 percent, the power
subsystem shall provide at least 7 W on average. The
solar cells selected for SWEET were the AZUR SPACE
3G30 solar cells [40]. The simulation software Systems
Tool Kit (STK) (and its solar panel tool) [41] was used
to analyze the energy generated per period for different
solar panel configurations. The input files were created
using the 3D modelling tool SketchUp [42], as shown in
Figures 2a, 2b and 2c. Initially, all external panels of the
CubeSat were covered with cells (see Fig. 2a), with the
exception of the bottom panel, accommodating payload
and S-band. This configuration generates an average of
only 2.1 W during one orbit period, not even a third of
the estimated required power.
Fig. 2. SWEET solar cell sketches
To counteract this, two deployable solar panels were
added, and two configurations considered (see Fig. 2b
and Fig. 2c), using the heritage of First-MOVE [43] and
MOVE-II. In the ISS orbit, the configuration shown in
Figure 2c generated slightly more power and was
therefore selected. In total, the SWEET CubeSat’s solar
panels will have 24 cells and generate on average 6.5 W
or 10.1 Wh during one orbit period. A 20 Wh battery was
selected, providing a good trade-off between mass and
capacity. Table 3 summarizes the properties of the 20 Wh
commercial batteries (including electrical power systems
(EPS)) analyzed.

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 5 of 11
Table 3. 20 Wh SWEET battery options
Battery
Clyde Space
20 Wh
CubeSat
Battery [44, 45]
GomSpace
Nano-
Power
BP4 [46]
Nano-
Avionics
Power
Unit 1P0
[47]
Material
Lithium-
Polymer
Lithium-
Ion
Lithium-
Ion
Nominal
Voltage
(V)
6 to 8.4
6 to 8.4
7.4
Current
(mAh)
2600
5200
2600
Energy
(Wh)
15-21
31-43
19.24
Mass (g)
246
270
275
Dimensio
ns (mm)
96x91x27.41
96x90x23
96x90x30
Cost (€)
6725
2450
4700
Due to its flight heritage, the Clyde Space 20 Wh
CubeSat battery was chosen as a footprint for SWEET.
4.3 Structure Subsystem
The structure of the SWEET CubeSat will be built in-
house, similar to MOVE-II. It shall comply with the CDS
[37], and therefore be built in aluminum, provide
mounting for the satellite bus and be the correct size to fit
into the standard deployment mechanics.
4.4 ADCS Subsystem
For the Attitude Determination and Control
Subsystem (ADCS) analysis, external disturbance
torques were calculated at SWEET’s initial altitude of
400 km using equations 1, 2, 3 and 4 below [34]. The
results of these calculations are summarized in Table 4.


(1)
 
 
(2)


(3)
 
(4)
Table 4. SWEET total disturbance torques
Disturbance torque
Value (Nm)
Gravity gradient torque
2.1x10-9
Solar radiation pressure
1x10-8
Magnetic field torque
1x10-6
Aerodynamic torque (drag)
1.1x10-6
Total disturbance torques
2.2x10-6
For attitude determination, the same commercial off-
the-shelf (COTS) components planned to be used on
MOVE-II [48] were selected: five sun sensors and a set
of micro-electro-mechanical systems (MEMS)
composed of a 3-axes gyroscope, a 3-axes accelerometer,
and a 3-axes magnetometer [49]. The selected sun
sensors have a field of view of 120° x 120° and an
accuracy better than 3°.
The option of using star trackers was considered and
discarded, as the added accuracy was not required to
guarantee a success of the mission, thus saving on cost
and mass. The sun sensors will be positioned on all but
the bottom panel, due to virtually complete lack of
sunlight on this panel.
For attitude control two scenarios were considered: a
simple scenario with three (and three redundant)
magnetorquers, as well as a more stable configuration
using three reaction wheels in addition to the
magnetorquers. The commercially available
magnetorquers are standalone boards and too heavy for
SWEET. The magnetorquers will therefore be built in-
house, custom made to be integrated into the side panels
of the satellite.
To estimate the minimum required dipole momentum
to be exerted by the magnetorquers (if no reaction wheels
are present), equations 5, 6 and 7 were used [34].


(5)
  
(6)
 

(7)
The result was a minimum magnetic dipole moment
of 0.04 Am2 per axes. The MOVE-II magnetorquer
currently being designed can achieve a value of 0.1 Am2
per axes, providing a realistic margin. The MOVE-II
ADCS design has magnetorquer coils on every side panel
(except for the bottom panel) and another on the main
ADCS panel inside the PC/104 stack [48]. SWEET could
have the same layout, or alternatively have a
magnetorquer on the bottom panel instead of inside the
PC/104 stack, if allowed by the payload.
The reaction wheels selected for the second scenario
are the miniature wheels HT-RW200.15 built by
Hyperion Technologies [50]. Table 5 summarizes their
properties.
Table 5. Reaction wheel properties [50]
Property
Value
Angular momentum
1.5 mNms
Maximum torque
1x10-4 Nm
Slew rate (3 reaction wheels)
1.5°/s [26]

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 6 of 11
4.5 OBC Subsystem
Due to the early stage in the mission design,
assessments of on board computer (OBC) and thermal
(see Section 4.6) subsystems are preliminary and have to
be evaluated further down the design process. Image
compression will occur on-board, and the compressed
image will measure 0.5 MB [30]. A similar compression
algorithm to the Aalto-1 mission will be used. As secure
digital card (SD-card) based OBCs are more susceptible
to failures, only OBCs with on-board flash memory were
considered. Satellite-focused operating system (OS),
such as free real time operating system (FreeRTOS) [51],
proved difficult to be developed by a purely student team
[52], a Linux-OS OBC being more practical. Learning a
new OS requires time and effort. The steep learning curve
inhibits flexible and rapid satellite development by
students. None of the analyzed commercial OBCs
satisfied the SWEET requirements: they did not have a
Linux-OS, and many of them had an SD-card and not an
on-board flash memory. Consequently, it was not
possible to select an appropriate OBC at such an early
mission phase.
As future work, more SWEET bus components and
requirements will be defined, making it possible to refine
OBC requirements in terms of random-access memory
(RAM), memory and processing speed. Radiation doses
on the components will also have to be taken into
account. Given the selected ISS orbit, it is expected that
the risk posed by single event effects (SEE) and the total
ionizing dose (TID) will not be very high due to the short
duration of the mission and the relatively low initial
altitude.
An alternative OBC solution would be to build an
OBC in-house, although this would prove to be a
challenge. Other CubeSat teams are currently working on
innovative OBC solutions such as Astro Pi, built by
Surrey University which is using two Raspberry Pi [53],
and Phone Sat, built by NASA which uses a smartphone
[54]. The field of OBCs and miniaturization is an
expanding one, consequently, faster and smaller
computers are expected to become available over the
upcoming months and years.
The Nano Avionics OBC [55] is selected as a
footprint for the SWEET CubeSat. It does not fulfil the
requirement of a Linux-OS, but it is used here as a
preliminary subsystem, having sufficient reliable
memory and a successful flight heritage of 5 months. The
OBC footprint is very similar to the other analyzed OBCs
in terms of volume, mass, cost and power and it is
predicted that the footprint for the final selected OBC
will be also in the same range. Table 6 displays the
properties of the Nano Avionics OBC.
Table 6. Properties of the Nano Avionics OBC [55]
Property
Value
OBC name
SatBus 1C0
Company
Nano Avionics
Processor and speed
ARM 4: 8-168 MHz
Max power
287 mW
Dimensions (mm)
PC/104: 96x90x10
Mass
35 g
Flash memory data
storage
2x16 MB
SD card data storage
N/A
Flash memory code
storage
1 MB
RAM volatile memory
192 kB
OS
FreeRTOS
Possible modes
Run, sleep, stop
Base cost
3000 €
Flight heritage
Successfully flown on
CubeSat LituacinaSAT-1
for 5 months
4.6 Thermal Subsystem
The thermal environment of the SWEET CubeSat
will be controlled using a passive thermal design. Passive
thermal control is the science of organizing components
within the cube to facilitate working conditions for all
subsystems. Although active thermal control can deal
with more extreme situations compared to a purely
passive design, the strong CubeSat constraints on
volume, mass, and power budget only allow for a passive
one (except for the EPS system which is provided with a
heater). Table 7 shows the temperature ranges of the
SWEET CubeSat components.
Table 7. Component temperature ranges
Component
Operating
temperature range
Solar cells [derived from MOVE-II]
-100 to +80 °C
Payload [26, 30]
+15 to +45 °C
Battery (20 Wh Clyde
Space) [45]
-10 to +50 °C *
EPS board [derived from MOVE-II]
-40 to +85 °C
Communications (VHF
transmitter, UHF receiver
and S-band) [derived from MOVE-II]
-40 to +100 °C
ADCS [derived from MOVE-II]
-40 to +85 °C
Reaction wheel
TBD
OBC Nano Avionics [55]
-40 to +85 °C
*battery storage temperature: -20 °C to +45 °C (3
months)
Except for the payload, the SWEET components
temperature ranges are very similar to the MOVE-II
ones, therefore the same MOVE-II methods will be used:

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 7 of 11
multi-layer insulation (MLI), appropriate stacking,
placement of hotspots, and battery heaters.
Regarding the payload, VTT suggests to maintain a
constant temperature within the instrument to achieve the
best image quality. The Aalto-1 team showed a
temperature range of +15 to +45 °C to be achievable
within a CubeSat, while effecting the image quality as
little as possible [26]. The First-MOVE mission launched
in 2013, measured the internal temperature of the satellite
to be between +5 and +15 °C in the printed circuit board
(PCB) stack [52], guaranteeing small temperature
variations and good image quality. It is possible to ensure
the instrument’s thermal specification by appropriate
stacking or inserting a small heater. The lens of the
payload will consistently face the surface of the Earth,
receiving Earth’s infrared radiation and albedo. Thus, its
temperature will not vary significantly, assuming that it
does not face the Sun due to an ADCS fault.
Table 8. SWEET subsystems integration
Component
Volume (cm3)
Mass (g)
Energy consumed (Wh)
Cost (€)
Structure [derived from MOVE-II]
N/A
368*
N/A
3,200
Solar cells (24) [40]
11.5
61.9
N/A
12,000
Battery and EPS [44]
236.9
246
0.001
6,725
UHF & VHF antenna [derived from
MOVE-II]
259.2
230*
1 (10 min per orbit)
14,000
S-band [derived from MOVE-II]
129.6
35
0.13 (1 min per orbit)
7,000
Payload [30]
451.6
600
0.3 (2 images per orbit: 3
seconds each)
< 150,000
ADCS [derived from MOVE-II]
42.1
90.4
2.46
10,000
ADCS with reaction wheel [50, 56]
70.2
111.4
3.7
37,000
OBC [55, 57]
86.4
35
0.45
3,000
Total
1,217.3
1,666.3
4.3
< 205,925
Total with reaction wheel
1,245.4
1,687.3
5.6
< 232,925
Budget
1,844.6
2,660
10.1
N/A
Budget with 25 % margin
1,383.5
1,995
7.6
N/A
* including development margin,
5. Results and Discussion
The purpose of this section is to first integrate the
individually analyzed subsystems (see Section 4) into a
standard 2U-CubeSat, in terms of mass, volume and
power. This provides an opportunity to make an initial
estimate of the costs of SWEET. Subsequently, the
SWEET precursor mission orbit lifetime is modelled and
discussed, followed by the simulated operations that can
be achieved by flying the satellite over Africa. Finally,
the capabilities and size of a future SWEET constellation
are briefly described.
5.1 Bus Subsystem Integration
When integrating the subsystems, a 25 percent
margin for volume, mass and power was considered. The
mass budget of 2,660 g was provided by the CDS [37],
and the energy budget of 10.1 Wh by the simulated
power generated using 24 solar cells. The volume budget
of 1,844.6 cm3 was calculated by multiplying the width
and length of a standard PC/104 board (90 mm x 96 mm)
with the CDS allowed height within the cube’s structure
rails (213.5 mm) [37]. The characteristics of the SWEET
subsystems are summarized in Table 8.
The thermal subsystem, the side panels and the
deployable mechanism (hold down and release
mechanisms for the deployable panels), are not present in
the table, as their volume, mass, power and cost are very
low and can be neglected at this point (they will be
included in the 25 percent margin). In Table 8, the
volume of the structure is not considered because the
budgeted volume is calculated to be the internal space
inside the structure.
Despite the stability advantages of installing reaction
wheels on SWEET, they will not be used for the
precursor mission due to higher cost, complexity, mass
and volume. Table 8 shows that it is possible to include
reaction wheels in terms of volume, mass and power
budget, hence it might be considered to add them after
the precursor mission.
5.2 Orbit Lifetime
The SWEET precursor mission will have an ISS orbit
with initial altitude of 400 km. The main disadvantage of
this orbit is the short lifetime of the satellite. The CNES
Semi-analytic Tool for End of Life Analysis (STELA)
[58] was used to determine the approximate lifetime of
SWEET. The input parameters are summarized in
Table 9.

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 8 of 11
Table 9. STELA input parameters [58]
Parameter
Value
Mass (kg)
2.66
Average Reflecting area (m2)
0.03
Reflectivity coefficient
1.5
Average Drag area (m2)
0.03
Constant drag coefficient
2.2
Solar activity
Mean constant
In addition, the orbit parameters generated by STK
were inputted [41], and the initial date was set to 1st of
January 2019, a realistic launch date allowing enough
time for development. The reflectivity and drag areas
were calculated by using the average satellite area
(including the side wings).
The STELA simulation showed that the precursor
mission is expected to deorbit after 0.375 years,
approximately 4.5 months. This does not provide any
information about whether the satellite will still be
functional at such low altitudes.
Equation 8 and 9 were used to estimate the altitude at
which the ADCS will not be able to control the satellite
any longer, as the external moments will have exceeded
the maximum magnetic moment of 0.1 Am2 [34].


(8)


(9)
To calculate the drag and worst-case polar magnetic
field equations 4 and 5 were used, respectively (see
Section 4.4). Table 10 summarizes the results obtained,
and shows that the satellite cannot be controlled below
250 km in altitude due to insufficient control moment.
Table 10. Results of maximum controllable altitude
h (km)
350
300
250
200
ρatm
(kg/m3)
6.66e-12
1.87e-11
5.97e-11
2.41e-10
R (km)
6728.1
6678.1
6618.1
6578.1
V
(m/s)
7701.3
7730
7765
7788.6
Ta
(Nm)
6.38e-07
1.80e-06
5.82e-06
2.36e-05
M
(Am2)
0.014
0.04
0.13
0.52
There is an error margin to be considered, as all
equations are approximations and the atmospheric
density values [59] used were average values. According
to the STELA simulation, the satellite is predicted to be
below 250 km for about 10 days, lowering the satellite’s
predicted mission lifetime from 4.5 months to
4.2 months.
5.3 Operations
The operational requirements for the SWEET
precursor mission are to take a minimum of four
hyperspectral images a day and to download at least two
a day (as well as taking and downloading the optical
images’ thumbnails). A typical power consumption and
battery charge plot over one orbit is shown in Figure 4,
where one image is taken, and subsequently downloaded
using first the VHF antenna over 8 minutes and then the
S-band over 4 minutes. The nominal mode is defined as
the mode in which SWEET is nadir-pointing, 3-axes
stabilized, operating and waiting for commands. Figure 4
assumes that more than three ground stations are
available to be able to download for 4 minutes using the
S-band. When analyzing the scenario with three ground
stations evenly distributed along the satellite path, the
average amount of data that can be downloaded in one
day using only VHF is 3.74 MB, and by using both VHF
and S-band 7.54 MB. As a first estimation and by
introducing appropriate margins, the power budget
allows to download about 75 percent of what is possible:
5 images only using VHF and 11 images using both VHF
and S-band. This is a largely improved result compared
to the initial requirement of two images a day, which
could be improved further with the use of better satellite
pointing.
5.4 Preliminary Mission Analysis
The imager solution is good enough to capture
portions of the biggest lakes in Africa down to small
lakes that fit into a 10 x 10 pixel area (1370 x 1370 m at
400 km altitude) of the image (in the future it will be
evaluated whether an even smaller pixel area can also
generate useful information). 62 African lakes were
identified ranging from 68,800 km2 (Lake Victoria),
down to 2.15 km2 (Lake Oguta in Nigeria) in size. All
lakes are African sweet water lakes. Salty and alkaline
(or soda) lakes are not considered as a source of drinkable
water. Lake Victoria is the largest lake in Africa and
although the satellite’s footprint does not allow capturing
the whole lake in one image, imaging a portion of the lake
will still be beneficial.
Table 11 displays the result obtained by simulating
the amount of flybys over the 62 lakes, using one ISS
SWEET orbit. The simulation software STK [41] was
used and flybys were simulated only when the lakes were
in direct sunlight.

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 9 of 11
Fig. 4. Typical power consumption and battery charge over one orbit [41]
Table 11. Lake imaging (one satellite)
Yearly
average
flybys
Average
revisit time
3 biggest lakes
99
3 to 4 days
30 biggest lakes
41
9 days
Total analyzed lakes (62)
31
11.8 days
Smallest lake
17
21 days
5.5 SWEET Constellation
The SWEET satellite properties described so far are
related to a precursor mission during which the concept
will be proven and the usefulness of the generated data to
the African population validated. The long-term plan is
to build a small constellation of CubeSats, enabling
SWEET to image more lakes with higher temporal
resolution.
The most appropriate orbit for the constellation is a
Sun-synchronous orbit, with launch opportunities and a
good orbit lifetime of 21.63 years with initial altitude of
650 km (calculated using STELA) [58]. The simulated
constellation is composed of four CubeSats with two
different right ascension of the ascending node (RAAN):
100° and 280°, respectively. The satellites will be
launched from two different launchers and be evenly
distributed by using differential drag [60]. Table 12
shows the simulated results of flybys.
Table 12. Lake imaging (four-satellite constellation)
Yearly
average
flybys
Average
revisit time
3 biggest lakes
218
1.5 days
30 biggest lakes
128
2.8 days
Total analyzed lakes (62)
103
3.5 days
Smallest lake
68
5.3 days
It is a challenge to determine how quickly poisonous
substances develop in inland sweet water lakes in Africa.
Development of cyanobacteria bloom can occur in
timespans ranging from hours to weeks. The SWEET
constellation will provide data on average every 3.5 days,
very similar to what was achieved with MERIS [15], but
at a fraction of the cost. Compared to traditional in-situ
water monitoring techniques, re-imaging a lake with such
a high revisit rate will provide continuous monitoring,
saving lives, as well as time and money. The SWEET
precursor mission will study the cyanobacteria
development trend in the interested areas, and
subsequently adapt the constellation revisit time
accordingly.
6. Conclusion: is it possible to measure water
quality with a CubeSat?
SWEET is the CubeSat mission designed to help
Africa tackle its drinking water quality problem.
Research conducted so far, and presented here, shows
that it is possible to integrate a CubeSat hyperspectral
imager into a standard 2U-CubeSat. Using an ISS orbit
for the precursor mission, followed by a sun-synchronous
four CubeSat constellation, SWEET will generate
enough data to fulfil the mission’s objective: monitoring
62 fresh water lakes in Africa with an average revisit time
of 3.5 days. The majority of subsystems have been
defined and integrated into the 2U-CubeSat specification.
OBC and thermal subsystem requirements are roughly
defined, and will be further refined as the project
advances. As for many other CubeSat missions, SWEET
also aims to educate students and provide African
universities with the knowledge of building a CubeSat
constellation. With the successful completion of this
project, millions of people, dependent on Africa’s inland
sweet water, will benefit from the mission.
0
2
4
6
8
10
12
14
16
0
2
4
6
8
10
12
14
16
18
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91
Battery Charge (Wh)
Power (W)
Time (min)
Solar Power Generated
VHF Power Consumption
(Data Mode)
Imaging Power Consumption
(Data Mode)
S-band Power Consumption
(Data Mode)
Power Consumption (Nominal
Mode)
Battery Charge

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 10 of 11
Acknowledgements
The authors would like to thank the Secure World
Foundation (SWF) and Women in Aerospace Europe
(WIA-Europe) for funding Kelly Antonini’s participation
at IAC-2016. In addition, the authors are grateful to STK
for providing a free educational license for the purpose
of this Master’s thesis. The authors acknowledge the
funding of MOVE-II by the Federal Ministry of
Economics and Energy, following a decision of the
German Bundestag, via the German Aerospace Center
(DLR) with funding grant number 50 RM 1509.
References
[1] ESA, How a mission is chosen, 2013,
http://www.esa.int/Our_Activities/Space_Science/How_
a_mission_is_chosen, (accessed 29.08.2016).
[2] D. O. A. Savtchenko, et al., Terra and Aqua
MODIS products available from NASA GES DAAC,
Adv. Space Res. 34 (2004) 710-714.
[3] J-L. Bezy, et al., MERIS - A new generation of
ocean-colour sensor onboard Envisat, ESA bulletin 103
(2000) 48-56.
[4] H. Heidt, et al., CubeSat: A new generation of
picosatellite for education and industry low-cost space
experimentation, in Proceedings of the AIAA/USU
Conference on Small Satellites, 2000.
[5] D. Mandl, et al., Hyperspectral Cubesat
Constellation for Natural Hazard Response, in
Proceedings of the AIAA/USU Conference on Small
Satellites, 2016.
[6] T. S. Pagano, et al., The CubeSat Infrared
Atmospheric Sounder (CIRAS), Pathfinder for the Earth
Observing Nanosatellite-Infrared (EON-IR), in
Proceedings of the AIAA/USU Conference on Small
Satellites, 2016.
[7] D. Fussen, et al., PIC.A.S.S.O.: atmospheric
remote sensing and in situ space science from CubeSats,
in 4S Symposium, Valletta, Malta, 2016.
[8] A. K. Jaan Praks, et al., MULTI PAYLOAD
EARTH OBSERVATION MISSION AALTO-1, in 4S
Symposium, Valletta, Malta, 2016.
[9] UNESCO, United Nations World Water
Development Report 4. Volume 1: Managing Water
under Uncertainty and Risk, UN-Water, 2012.
[10] J. Schewe, et al., Multimodel assessment of
water scarcity under climate change, Proc. Natl. Acad.
Sci. 111 (2014) 3245-3250.
[11] G. A. Codd, et al., Cyanobacterial toxins: risk
management for health protection, Toxicol Appl
Pharmacol, 203 (2005) 264-272.
[12] I. R. Falconer, An overview of problems caused
by toxic blue–green algae (cyanobacteria) in drinking
and recreational water, Environ. Toxicol. 14 (1999) 5-12.
[13] W. W. Carmichael, and G. L. Boyer, Health
impacts from cyanobacteria harmful algae blooms:
Implications for the North American Great Lakes,
Harmful Algae 54 (2016) 194-212.
[14] G. A. Codd, et al., CYANONET A Global
Network for Cyanobacterial Bloom and Toxin Risk
Management, UNESCO Technical documents in
hydrology 76 (2005).
[15] R. S. Lunetta, et al., Evaluation of cyanobacteria
cell count detection derived from MERIS imagery across
the eastern USA, Remote Sens. Environ. 157 (2015) 24-
34.
[16] M. F. Mohamed, Satellite data and real time
stations to improve water quality of Lake Manzalah,
Water Sci. 29 (2015) 68-76.
[17] L. Li, et al., Remote sensing of freshwater
cyanobacteria: An extended IOP Inversion Model of
Inland Waters (IIMIW) for partitioning absorption
coefficient and estimating phycocyanin, Remote Sens.
Environ. 157 (2015) 9-23.
[18] S. C. J. Palmer, et al., Validation of Envisat
MERIS algorithms for chlorophyll retrieval in a large,
turbid and optically-complex shallow lake, Remote Sens.
Environ. 157 (2015) 158-169.
[19] E. T. Harvey, et al., Satellite-based water quality
monitoring for improved spatial and temporal retrieval of
chlorophyll-a in coastal waters, Remote Sens. Environ.
158 (2015) 417-430.
[20] S. C. J. Palmer, et al., Satellite remote sensing of
phytoplankton phenology in Lake Balaton using 10 years
of MERIS observations, Remote Sens. Environ. 158
(2015) 441-452.
[21] K. Kallio, et al., Validation of MERIS spectral
inversion processors using reflectance, IOP and water
quality measurements in boreal lakes, Remote Sens.
Environ. 157 (2015) 147-157.
[22] cosine, HyperScout, 2016,
http://hyperscout.nl/product/, (accessed 17.08.16).
[23] H. Saari, VTT develops a miniaturized spectral
camera for ESA’s Picasso nanosatellite mission, 2015.
[24] VTT, Personal Communication: VTT Spectral
Imager Pictures, 29.06.16.
[25] Aalto-1, Aalto-1 is the 1st Finnish satellite,
2016, http://www.aalto1.fi/aalto1_in_english.html,
(accessed 09.08.16).
[26] A. Näsilä, Validation of Aalto-1 Spectral Imager
Technology to Space Environment, Aalto University -
School of Electrical Engineering, 2013.
[27] ESA, Aalto-1: The Finnish Student
Nanosatellite, 2016,
https://directory.eoportal.org/web/eoportal/satellite-
missions/a/aalto-1 (accessed 09.08.16).
[28] A. Kestilä, et al., Aalto-1 nanosatellite -
technical description and mission objectives, Geosci.
Instrum. Method. Data Syst. 2 (2013) 121–130.
[29] ESA, PICASSO (Pico-Satellite for Atmospheric
and Space Science Observations), 2016,

67th International Astronautical Congress (IAC), Guadalajara, Mexico, 26-30 September 2016.
IAC-16-B4,1,12,x34041 Page 11 of 11
https://directory.eoportal.org/web/eoportal/satellite-
missions/p/picasso (accessed 09.08.16).
[30] VTT, Personal Communication: Instrument
Properties, 17.05.16.
[31] A. Kestilä, et al., Aalto–1 Remote Sensing
Campaign, in XXXIII Finnish URSI Convention on
Radio Science and SMARAD Seminar 2013, Otaniemi,
Finland (2013) 143-146.
[32] H. Klinkrad, et al., The ESA space debris
mitigation handbook 2002, Adv. Space Res. 34 (2004)
1251-1259.
[33] NASA, STD-8719.14A, Process for limiting
orbital debris (change 1), 2012.
[34] W. J. Larson, et al., SMAD III: Space mission
analysis and design, third ed., Microcosm Press, El
Segundo, 2005.
[35] MOVE-II, 2016, http://www.move2space.de
/MOVE-II/ (accessed 09.08.16).
[36] First-MOVE, 2014, http://www.move2space
.de/First-MOVE/index.html%3Flang=en.html (accessed
on 09.08.16).
[37] The CubeSat Program, CubeSat Design
Specification (CDS) Rev. 13, 2014,
http://www.cubesat.org/resources/ (accessed 28.08.16).
[38] M. Langer, et al., MOVE-II – der zweite
Kleinsatellit der Technischen Universität München, in
Deutscher Luft- und Raumfahrtkongress (DLRK)
Rostock, Germany, 2015.
[39] N. M. E. Appel, et al., Nanolink: A robust and
efficient protocol for Small Satellite radio links, in 4S
Symposium, Valletta, Malta, 2016.
[40] AZURSPACE, 30% Triple Junction GaAs
Junction Solar Cell Datasheet, 2016,
http://www.azurspace.com/images/pdfs/HNR_0003429-
01-00.pdf (accessed 04.08.16).
[41] AGI, Systems Tool Kit (STK), 2016,
http://www.agi.com/products/stk/ (accessed 09.08.16).
[42] Trimble Navigation Limited, SketchUp, 2016,
http://www.sketchup.com/ (accessed 17.08.16).
[43] M. Langer, et al., Deployable Structures in the
CubeSat Program MOVE, in Proceedings of the 2nd
International Conference Advanced Lightweight
Structures and Reflector Antennas, Tbilisi, Georgia,
2014, 224-233.
[44] Clyde Space, CS 2U Power Bundle B: EPS +
20Whr Battery, 2016, https://www.clyde.space
/products/38-cs-2u-power-bundle-b-eps-20whr-battery,
(accessed 09.08.16).
[45] Clyde Space, User Manual: 3rd Generation
CubeSat Battery Family, 2016.
[46] GomSpace, NanoPower BP4, 2016,
http://gomspace.com/index.php?p=products-bp4
(accessed 09.08.16).
[47] NanoAvionics, Power Unit 1P0, 2016, http://n-
avionics.com/products/eps/ (accessed 09.08.16).
[48] T. Gruebler, Investigating a Mass-Reproducible
Attitude Determination and Control System on the
nanosatellite MOVE-II, in 8th European CubeSat
Symposium, London, UK, 2016, 7-9 September.
[49] Bosch, Bosch Sensortec BNO055 Intelligent 9-
Axis Absolute Orientation Sensor, 2016,
http://eu.mouser.com/new/bosch-sensortec/bosch-
bno55-sensor/, (accessed 17.08.16).
[50] Hyperion Technologies, Hyperion
Technologies HT-RW200.15, HT-RW210.30, HT-
RW210.60 Data Sheet, 2015.
[51] FreeRTOS, Quality RTOS & Embedded
Software, 2016, http://www.freertos.org/, (accessed
09.08.16).
[52] M. Langer, et al., Results and lessons learned
from the CubeSat mission First-Move, in: R. Sandau, P.
H. Roeser and A. Valenzuela (Eds.), Small Satellite
Missions for Earth Observation, Springer, Berlin, 2015.
[53] D. Honess., Compute Module CubeSats, 2015,
https://www.raspberrypi.org/blog/compute-module-
cubesats/ (accessed 09.08.16).
[54] T. Talbert., PhoneSat Flight Demonstrations,
2013,
http://www.nasa.gov/directorates/spacetech/small_space
craft/phonesat.html (accessed 09.08.16).
[55] NanoAvionics, NanoAvionics Primary Flight
Computer Datasheet - SatBus 1C0, 2015, n-
avionics.com/wp-
content/uploads/2015/04/SatBus1C0_data_sheet.pdf,
(accessed 09.08.16).
[56] Hyperion Technologies, Personal
Communication: HT-RW200.15 Reaction Wheel Cost,
24.06.16.
[57] NanoAvionics, Personal Communication:
SatBus 1C0 Price, 04.07.16.
[58] CNES, STELA,
https://logiciels.cnes.fr/content/stela?language=en
(accessed 09.08.16).
[59] Rocket and Space Technology, Atmosphere
Properties, http://www.braeunig.us/space/atmos.htm,
(accessed on 09.08.16).
[60] C. Foster, et al., Orbit Determination and
Differential-drag Control of Planet Labs Cubesat
Constellations, in AIAA Astrodynamics Specialist
Conference, Vail, US, 2015.