SSC18-II-06

Making the Invisible Visible: Precision RF-Emitter Geolocation from Space by the
HawkEye 360 Pathfinder Mission
K. Sarda, N. Roth, R.E. Zee
Space Flight Laboratory, University of Toronto, Institute for Aerospace Studies
4925 Dufferin Street, Toronto, Ontario, Canada, M3H5T6; +1-416-667-7923
[email protected]

Dan CaJacob
HawkEye 360 Inc.
196 Van Buren Street, Suite 450, Herndon, VA, USA, 20170; +1-517-203-0360
[email protected]

Nathan G. Orr
Deep Space Industries, Inc.
2188 Bering Drive, San Jose, CA, USA, 95131; +1-855-855-7755
[email protected]

ABSTRACT
To be launched in Q4 2018, the HawkEye 360 (HE360) Pathfinder mission will validate key enabling technologies
and operational methods necessary to provide unprecedented analysis of wireless signals for commercial and
government applications using small satellites. Applications range from logistics monitoring and tracking of aircraft,
ships, and ground transportation, to emergency response and other data analytics and services. The mission will
nominally consist of three Pathfinder satellites, operated in formation, to demonstrate and validate an initial
operational capability. Following the Pathfinder demonstration a constellation with more than 18 satellites will be
deployed.

HE360 has contracted Deep Space Industries (DSI) and major subcontractor Space Flight Laboratory (SFL) to
design and manufacture the spacecraft platform for the Pathfinder demonstration mission. In addition to being a
world leader in low-cost high-performance small spacecraft, SFL is a pioneer in low-cost precision spacecraft
formation flight, a key enabling technology for HE360 mission. DSI, a world leader in state-of-the-art launch safe
propulsion systems, is providing the CometTM water-fueled resisto-jet propulsion system for the mission.

This paper describes the HawkEye 360 Pathfinder mission, with a focus on the core enabling platform and payload
technologies.

INTRODUCTION
HawkEye 360 (HE360) has developed an innovative
Each of the three spacecraft will be identical and their
combination of classical and novel geolocation
primary payload is a Software Defined Radio (SDR)
algorithms that will enable precise space borne
and custom RF front end, along with band-specific
geolocation of terrestrial and aerial radio frequency
antennas. The frequency agile payload will enable
(RF) emitters related to a broad array of business reception of many different types of signals, covering
enterprises. In late 2018, the HE360 Pathfinder
various RF segments spanning VHF through Ku-band,
mission, a formation-flying cluster of three
which will then be geolocated by applying signal
microsatellites, will launch to demonstrate the
processing to the combined received data of all three
commercial capability of HE360’s high-precision RF
spacecraft. The three spacecraft, each with its own
geolocation technology. The spacecraft will be placed propulsion system, will establish a relatively wide-
into a Sun synchronous orbit (SSO) at a 575km altitude baseline, geometrically diverse formation and continue
and a local time of descending node (LTDN) of
to maintain the relative position formation for the
10:30am.
duration of the nominal three year mission.

Sarda 1 32nd Annual AIAA/USU

Conference on Small Satellites
The Pathfinder mission serves to demonstrate the monitor spectrum usage and to identify areas of
practicality of the geolocation mission and paves the interference. In the field of transportation, RF signals
way for a future commercial constellation. Initially, an transmitted from the air, ground or sea could be
eighteen satellite constellation (arranged as six clusters precisely monitored. The system may also be used to
of three) is envisioned for commercial, global service. expedite search and rescue operations by quickly
However, the final constellation size and geometry will pinpointing activated emergency beacons.
depend on market factors including the results of the
Pathfinder mission. RF geolocation as it pertains to this mission means the
identification of a terrestrial signal emitter’s location
HE360 selected Deep Space Industries (DSI) and major through signal processing and analysis of the received
subcontractor Space Flight Laboratory (SFL) to design signal at one or more remote observation platforms. In
the platform for the Pathfinder mission. DSI is the this case, the observation platforms are the three HE360
prime contractor, and the manufacturer of a novel spacecraft in the Pathfinder cluster. Hereafter the
water-fueled electro-thermal propulsion system which spacecraft will be referred to as “Hawks” and
will fly on each spacecraft. SFL is responsible for the individually as Hawk-A through Hawk-C.
design and manufacturing all three spacecraft
platforms. SFL’s versatile flight-proven 15kg Next- As an example of the utility of the technology which
generation Earth Monitoring and Observation (NEMO) will be made available by this mission, consider an AIS
microsatellite bus was selected for the mission. In detection case. There are 21 different types of
addition to being a world leader in providing low-cost Automatic Identification System (AIS) messages, many
high-performance small spacecraft, SFL was selected of which include the maritime vessel’s location
for this mission as it is a pioneer in low-cost precision provided by the vessel’s GPS receiver. Many existing
spacecraft formation flight, a key enabling technology satellites decode or receive this information and use the
for HE360 mission. SFL has developed compact, low- embedded geolocation data for commercial or national
cost formation flying technology at a maturity and cost purposes.
that no other small satellite developer can credibly offer
at present. This precise formation control was Unfortunately, it has been demonstrated that AIS data is
demonstrated on-orbit by SFL in the highly successful not universally reliable. It is fairly easy for individuals,
CanX-4/CanX-5 mission in 20141. With 18 successful such as pirates or illegally operating fishing fleets to
spacecraft missions on-orbit, SFL’s solutions have “spoof” their AIS emissions, effectively changing the
demonstrated high reliability and high availability GPS positions they report to make it look as if they are
products, which can be depended upon for a wide array somewhere other than where they actually are or simply
of commercial applications. By leveraging SFL’s changing their identifier. Furthermore, bad actors with
successful spacecraft platforms and formation flying less technical capability frequently turn off their AIS
technology, along with DSI’s pioneering innovations transceivers - “going dark” and disappearing from port
and next-generation propulsion systems, the mission and satellite AIS data feeds while engaging in criminal
will deliver unparalleled performance in smaller, activities. HE360 will demonstrate that independent
affordable satellites. geolocation of AIS and other signals is possible without
having to trust potentially false data in the
THE MISSION transmissions. In the event that an AIS transmitter is
disabled, other well-known signals commonly
Clearly understanding the world around us is becoming transmitted by ships can be substituted to maintain
more important than ever. Many of the big problems we position knowledge of an emitter when traditional AIS-
face as a society require solutions that contextualize the
receiving satellites would lose contact.
world around us. This applies directly to the RF
domain. HawkEye 360 is capitalizing on the explosive The three Hawks will fly in formation, with co-
growth of RF signals and their application to tracking visibility of a large number of terrestrial emitters at any
assets. Opportunities and applications that arise from one time. Pairs of satellites or the entire trio may
this high-precision radio frequency mapping and intercept the same transmission when the transmission
analytics technology are enormous and appeal to a originates from the common footprint of the
broad array of business enterprises and government intercepting satellites. The satellites will synchronize
users. The mission is filling a void by bringing a level clocks using GPS receivers, and these same GPS
of visualization to a domain that has historically only receivers will stabilize the phase locked loops (PLLs)
been understood by governments. For example, the governing tuning frequency in the satellites’ digitizing
ability to locate and characterize RF signals across RF tuner payload.
many bands from space will allow regulators,
telecommunications companies and broadcasters to

Sarda 2 32nd Annual AIAA/USU

Conference on Small Satellites
 intended to be used for the receive-only Pathfinder
mission. The payload supports three 9361s so that up
to three receive channels can be processed
simultaneously and on separate frequencies. Although
the 9361 has two receive channels, they are tuned via a
common local oscillator (LO), which limits the tuning
range of one channel to within the instantaneous
bandwidth of the other. The embedded processor
system is based on the Xilinx Zynq 7045 SOC, which
combines a dual-core ARM processor with a Kintex
FPGA. The two devices are very tightly integrated on a
single chip, which facilitates easy cross-domain
switching between the processor and FPGA. This is
advantageous for signal processing applications.

Figure 1: Local horizon footprints of the three

spacecraft in formation
Signals will arrive at the three receivers at separate
times corresponding to different slant ranges between
the satellite and the emitter. Signals will arrive at
different apparent center frequencies corresponding to
velocity components in the direction of the signal’s path
of travel from the transmitter to the receiver (Doppler
effects). Comparing time-of-arrival (TOA) and
frequency-of-arrival (FOA) measurements between
pairs of receivers serves as a basis for discovering the
position of the transmitter using multi-lateration. GPS
receivers provide precise estimates for the position and
velocity of the receivers, furnishing the remainder of
the information required for multi-lateration.
Figure 2: Payload simplified block diagram
THE PAYLOAD
Each spacecraft will have an identical payload, The HE360 designed custom-RF front end connects to
consisting of two high-level components: i) A SDR the baseband processors and provides a number of
comprised of an embedded processor and FPGA unique, switchable RF paths and antennas to support a
resource, and a baseband signal processor, and ii) a range of bands and frequencies of interest. Each
custom-RF front-end with antennas, as illustrated in switchable path has custom filters, low noise amplifiers
Figure 2. (LNA) and even attenuators tailored to a specific band.
A low noise block down-converter (LNB) is included to
The SDR flown on the Pathfinder satellites is extend the SDR’s frequency range up to Ku-band (~18
comprised of an embedded processor system and three GHz). A range of antennas, including quarter-wave
baseband processors. The baseband processor is built dipoles, patches, and wide-band button and horn
around the Analog Devices 9361 product. This is a antennas support the full frequency range, from VHF to
highly integrated RF transceiver that combines high- Ku-band.
speed ADCs and DACs, RF amplifiers, filtering,
switching and more on a single chip. The transceiver The processor system takes advantage of open-source
product is capable of tuning from 70 MHz to 6 GHz, signal processing software and firmware to maximally
with an instantaneous bandwidth of up to 56 MHz. The mimic desktop SDR products. This allowed ground
9361 has two receive chains and two transmit chains. development to proceed agnostic of the final space
Although the device has transmit capability, it is not hardware and foster adoption of a “fly as you try”

Sarda 3 32nd Annual AIAA/USU

Conference on Small Satellites
philosophy. For the software side, GNURadio was be associated with the Pathfinder mission was significantly
used. GNURadio is “a free and open-source toolkit for reduced.
software radio.” It is widely used in small space
projects for ground software processing and may have The HE360 Pathfinder platform is essentially a
been used on previous spacecraft in similar embedded 20x20x44 cm form factor with an additional ~7 cm
environments. high ‘mezzanine’, with a launch wet mass of 13.4kg.
Similar to spacecraft designed to the CubeSat standard,
In operation, the payload can be commanded to tune the four launch rails interface with the separation system
baseband processor to a center frequency and stream and guide the spacecraft during ejection from SFL’s
samples at a given sample rate. Nominally, the XPOD separation system. An external view of the
baseband processor will produce complex (quadrature) Pathfinder spacecraft is found in Figure 3. The bus
samples. The RF front end will also be configured structure is predominantly lightweight magnesium, with
based on the signal of interest. Samples will be careful arrangement of structural components to
conditioned to some extent in the FPGA, including provide high mechanical margins. The structural
filtering and balancing associated with the ADCs. concept of the spacecraft is a dual tray based design, as
HawkEye, however, will maximize on-board shown in Figure 4. Most of the platform avionics are
processing wherever doing so contributes to the bottom clustered towards the +Y end of the spacecraft. This
line in terms of the product delivered2. Constraints allows for integration and harness design ease, and
inherent to the mission in terms of downlinking and offers considerable payload accommodation volume.
crosslinking data motivate reducing full-take RF to
meta-data surrounding that RF. To accomplish this
reduction, user-defined signal processing chains
optimized for the embedded platform are implemented.

The payload had gained considerable in-field aerial test

experience in parallel with development, building
confidence prior to the actual launch of the Pathfinder
mission. Indeed, the SDR payloads and receiving
antennas were fitted onto three rented aircraft, flown in
diverse formations over live RF emitters (including
maritime vessels and commercial maritime radar,
amongst other targets), yielding RF signal detection and
geolocation with unprecedented accuracy. Figure 3: Artistic Rendering of the HE360
Pathfinder Platform
THE PLATFORM
The HE360 Pathfinder mission employs SFL’s versatile
flight proven NEMO platform. This state-of-the-art
microsatellite bus has been employed by a wide range
of commercial and government users, and depended
upon in applications and business models which would
only allow for a high-performance high-reliability yet
affordable platform. Indeed, the NEMO bus has been
selected by the Norwegian government for the
NORSAT-1, -2, and -3 satellites (scientific, maritime
AIS, VDES, and radar applications), the Indian
government for NEMO-AM (aerosol monitoring), and
GHGSat Inc. for the GHGSat Constellation
(greenhouse gas emissions monitoring). The platform
supports a full suite of heritage SFL subsystem
hardware. The NEMO platform is configurable, with
many design aspects tailorable, if needed. The NEMO-
platform itself builds upon the extensive heritage Figure 4: HE360 Pathfinder Internal Layout
gained from SFL’s widely used Generic Nanosatellite
Bus (GNB). By leveraging heritage designs and As the spacecraft carries a sensitive RF payload,
experiences gained through many cumulative years of electromagnetic interference (EMI) mitigation was an
on-orbit operation, the cost, schedule, and risk important consideration in design. The spacecraft was

Sarda 4 32nd Annual AIAA/USU

Conference on Small Satellites
segregated into three distinct RF zones: i) the payloads operate between 32kbit/sec and 2048kbit/sec (scaled
isolated within their enclosures, ii) the balance of the on-the-fly), in either BPSK or QPSK modulation and
platform, and iii) the environment external to the 0.5 rate convolutional encoding, is used on the
spacecraft. The zones were setup by creating downlink. The platform is also equipped with
boundaries, essentially Faraday cages, which would dedicated high-data rate payload links: uplink in S-
significantly attenuate noise. This was accomplished band, downlink in X-band and cross-link to other
by: satellites in S-band. The X-Band transmitter is capable
of 3 – 50 Mbps usable data rate. The transmitter uses
• The use of RC-filtered connectors, sized to reject Offset Quadrature Phase Shift Keying (OQPSK) and a
signals above a design cut-off frequency, ½ rate convolutional encoding Forward Error
Correction (FEC) scheme. A high-rate S-band uplink is
• The use of conductive gaskets to ensure DC and RF implemented within the payload SDR itself, with a
seals across all interfaces of the faraday cages, LNA positioned between the radio and the body-
mounted patch antenna. A SFL S-Band inter-satellite
• Strict aperture control, to significantly attenuate RF link, although not required for the mission, is integrated
noise, but yet still comply with spacecraft venting to demonstrate the capability to perform the geolocation
requirements. This is particularly important for the calculations entirely on orbit. In this scenario,
spacecraft exterior, as strict aperture control was information must be exchanged between the satellites
enforced to prevent transmission of noise which so that all measurements reside on a single spacecraft
may otherwise be picked up by the payload receive where the geolocation problem can be solved.
antennas.
The attitude determination and control subsystem
The Pathfinder spacecraft employs a single-string employs six sun sensors, a three-axis magnetometer,
design that results in a compact, low mass spacecraft. and a three-axis rate sensor for attitude determination.
The power architecture is based on SFL’s modular Attitude control is achieved through three vacuum core
power system (MPS), which generates power from the magnetorquers and three reaction wheels. Orbit position
body mounted high-efficiency triple-junction solar and velocity measurements are sampled by a L1/L2
arrays, and uses a 12V lithium ion battery for energy GPS receiver and active antenna. Several modes of
storage. A solar array and battery regulator (SABR) attitude control are available including de-tumble (for
unit within the MPS provides peak power tracking initial stabilization after kick-off from the launch
functionality to optimize power generation. The MPS vehicle), inertial pointing, nadir tracking, align-
also provides power conditioning to generate 3.3V and constrain, and ground target tracking. This system
5V regulated buses in addition to the unregulated 12V allows for 2σ pointing accuracy with only 2.1o and 4.2o
bus, as well as load switching and protection against error in sunlight and eclipse respectively.
off-nominal voltage and current events.
DSI is providing a novel electro-thermal propulsion
The command and data handling architecture is centred system that uses liquid water as the working fluid,
on two SFL-designed on-board computers (OBCs), significantly reducing integration and launch risks
which interface to the uplink and downlink radios and relative to other market options of similar performance.
all other spacecraft hardware. One OBC is nominally The unit has a qualified specific impulse (Isp) of 182
designated as the house keeping computer (HKC), and seconds, giving it exceptional performance with
is responsible for telemetry collection, routing packets comparison to a typical cold-gas system. Conversely,
to and from the radios, payload operations, and while it has a lower Isp than newly available low-power
execution of time tagged commands. The second OBC electric propulsion systems, the higher thrust means that
is designated as the attitude determination and control DSI’s system can be used quasi-impulsively. This
computer (ADCC) and is responsible for polling reduces the time required for maneuvers. Electric
attitude determination sensors, running the estimation propulsion systems also typically utilize high voltage
and control algorithms, and commanding actuators. power supplies or RF-amplifiers that produce wide-
Both computers are cross-connected to all on-board band RF noise, which is detrimental to the RF payload.
hardware, providing a level of redundancy. In this The propulsion system on Pathfinder has a ΔV of 96
configuration, either computer can take on the tasks of m/s, though, the system features an easily expandable
the other if required. propellant tank, allowing for simple propellant volume
tailoring. The water propellant needs to stay liquid at all
Primary telemetry and command is provided in S-band
times. The thermal design of the spacecraft passively
and UHF respectively. A SFL UHF receiver is used to
maintains the propellant in a liquid state, but auxiliary
provide the uplink channel at a fixed 4kbit/sec data rate.
heaters are positioned to augment this in an emergency.
A variable data rate SFL S-band transmitter, which can

Sarda 5 32nd Annual AIAA/USU

Conference on Small Satellites
THE FORMATION ignored and only linear drift in the formation is
SFL has a strong history in the development and controlled. The quasi-nonsingular elements cannot be
implementation of technologies and algorithms aimed used in equatorial orbits, but this is not considered a
detriment since such orbits are not beneficial to HE360
towards operational formation flying missions. The
CanX-4 and CanX-5 spacecraft were the first from a ground-coverage perspective.
nanosatellites to demonstrate autonomous formation
reconfiguration and control with a control error of less
than one metre1. This was enabled by a real-time
relative navigation algorithm based on carrier-phase
differential GPS techniques, which was shown to have
a typical RMS error of better than 10 cm 4. In addition,
the drift recovery and station keeping (DRASTK)
software was developed and used successfully to design
and implement a guidance trajectory for rendezvous
following initial spacecraft separation from the launch
vehicle, and to maintain a coarse along-track separation
in a passively safe relative configuration by
appropriately phasing in-plane and out-of-plane
motions3. It is with this proven track-record of success
in applied formation guidance and navigation that SFL
Figure 5: The CanX-4 and CanX-5 spacecraft,
is uniquely positioned to implement these techniques
which successfully demonstrated in orbit precise,
operationally for the HE360 Pathfinder mission. controlled formation flight at the nanosatellite scale
The baseline orbit for the Pathfinder mission is a in 2014
circular Sun-synchronous orbit with an altitude of 575 The required formation control is 5 km (1σ), which
km and a local time of descending node of 10:30. In the must also be tolerant to 1 week ground station outages.
target formation, the three spacecraft are equally spaced The guidance, navigation, and control strategies
along-track by 125 km. The middle spacecraft has its selected can be implemented on-board the spacecraft,
right ascension of the ascending node (RAAN) adjusted however at present control maneuvers are to be
such that it has a 20 km peak-to-peak out-of-plane computed on the ground and uploaded to each
oscillatory motion, whose maxima are achieved at the spacecraft given the relatively coarse formation-
equator. For a RAAN-offset orbit, the formation keeping requirement. This strategy removes the
becomes co-linear at the maximum and minimum sub- complexity and risk in implementing autonomous
latitudes of the cluster, which occurs near the northern relative navigation and control where it is not
and southern polar regions. The reduced geolocation warranted. The target mission duration is two years,
precision in the polar regions is tolerable since the with a stretch goal of three years. Over this time, only
human population and activity in this region is limited. two of the three spacecraft shall be actively controlled.
Also, the payload data will be downloaded to X-band From a power perspective, the spacecraft are
earth stations in this region frequently. No inclination constrained to applying orbit control maneuvers at least
difference is desired, due to the large cost in 45 minutes apart.
maintaining this formation owing to the required
RAAN corrections. This formation provides a good Conceptually the formation control strategy is broken
balance between ground target viewing geometry for down into two phases: formation initialization, and
geolocation of RF signals, and fuel cost of formation station keeping. Following a two-week commissioning
initialization and maintenance. The quasi-nonsingular period for the spacecraft systems, the initialization
mean orbital element set from [6] is adopted in this phase is expected to last approximately six weeks.
work for several reasons. First, this parameterization During initialization two of the three spacecraft are
results in an intuitive geometric representation of the maneuvered into the target formation – exactly which
formation design variables given its relationship to the two depends on the initial relative configuration upon
solutions of the Hill-Clohessy-Wiltshire equations of separation from the launch vehicle. It is expected that
relative motion. Second, the equations of relative all three spacecraft will be deployed approximately five
motion are significantly simplified, so formation minutes apart from SFL’s XPOD separation system,
guidance and control tasks can be moved onboard more each at a velocity of roughly 1.8 m/s in an uncontrolled
easily. Finally, the use of orbital elements easily lends direction relative to the local orbital frame. Given the
itself to analysis of “mean” or averaged relative motion, GPS telemetry from each spacecraft, a guidance plan
such that short-period and long-period oscillations are can be simulated for each permutation of controlled

Sarda 6 32nd Annual AIAA/USU

Conference on Small Satellites
spacecraft. The spacecraft pair leading to minimum fuel infrequent, owing to the fact that all spacecraft will
consumption will be selected as the controlled mirror their attitudes thus minimizing the impact of
spacecraft going forward. The total initialization phase differential drag on the formation. A side-benefit of this
is broken down into sub-intervals (ΔTinit), during strategy is maximizing the time spent performing
which roughly 85% of orbits are allotted for control, payload observations.
while 15% are reserved as maneuver-free periods for
the purpose of orbit determination used as input for the The formation control simulations are performed with
next initialization window. the aid of Systems Tool Kit (STK). The orbit model
includes an EGM2008 gravity model of degree and
The guidance law during formation initialization is order 30, third-body perturbations due to the Sun and
based on [6], where the fuel-optimal reconfiguration moon, solar radiation pressure, and atmospheric drag
from some initial state to a final desired state is framed with a Jacchia-Roberts atmospheric density model.
as a problem of minimizing the net total change in the Thrusts are modeled as impulsive with a mean error of
differential mean orbital elements. This is possible zero and a standard deviation of 5%. A thrust timing
since incremental changes in the orbital elements can be error with standard deviation 10 seconds is applied as
equated to impulsive thrust maneuvers (i.e., well. Thrust minimum impulsive bit and saturation
instantaneous changes in velocity). The guidance plan effects are also accounted for, as well as attitude control
generates a set of waypoints in differential mean orbital errors with standard deviation of 2 degrees. Guidance
element space from the current time to the desired and control calculations are performed with an “error”
initialization time in ΔTinit intervals. The waypoint at differential orbital element state, whereby the true states
the start of the next sub-interval is used as the target are corrupted with a Gaussian noise whose mean and
during the current control period. standard deviation are provided in Table 1.

The set of control maneuvers during each initialization In the initialization simulation, the spacecraft are
sub-interval is computed using the method of Roscoe et assumed to begin in the aforementioned baseline orbit,
al., which exploits a duality between the continuous and after which they separate from the launch vehicle at a
discrete time optimal formation reconfiguration relative velocity of 1.7 m/s. One is deployed along the
problem in order to iteratively solve for a set of velocity vector, the second along the orbit normal
maneuver locations and magnitudes that result in a vector, and the third towards anti-velocity. The
minimum-fuel maneuver plan to reach the target spacecraft are allowed to drift with no control for 2
waypoint at the target time5. This control strategy is weeks, after which there are 6 weeks allowed for
augmented to enforce a minimum time-spacing between formation initialization. The desired relative orbital
maneuvers, and to prevent maneuvers from being elements are given in Table 2, where the values
planned inside configurable “no thrust” windows, provided are relative to Hawk-B which is assumed to be
which are specified by operators as a set of intervals. at the formation center and is uncontrolled. The results
of the initialization simulation are provided in Figure 6,
The station keeping guidance law is designed to keep which shows the relative trajectory of Hawk-A with
the spacecraft within a designated control window respect to Hawk-B during the initialization period. The
while keeping the spacecraft passively safe using the initial plan and the actual trajectory match fairly well
eccentricity/inclination vector separation concept7. The for the along-track separation and out-of-plane
station keeping phase is conceptualized as a long period oscillation, but there appears to be a large offset in the
of no control (the drift period; approximately 1 week), relative eccentricity vector. This indicates that the
followed by a short window within which the control relative eccentricity vector model does not capture the
maneuvers occur (the control period; approximately 4 true dynamics adequately at the global scale. In a
orbits). The strategy is motivated by [8], whereby practical sense, this is not an issue because the global
during each control window the active spacecraft trajectory is recomputed after each initialization sub-
targets a specific differential semi-major axis which interval effectively closing the loop. The ΔV required
will cause a drift from one side of the control window for this initialization is 5.43 m/s. This value is heavily
to the other. Likewise, the relative eccentricity vector is dependent on the initial separation dynamics, as well as
adjusted such that it will be parallel with the relative the total initialization period – longer periods lead to
inclination vector half-way through the drift period, greater fuel savings, as the natural drift due to J2 can be
which maximizes safety during the drift period. The leveraged to aide in adjusting the in-track offset and
relative inclination vector is simply readjusted to its RAAN.
target value during each control period, since there is no
drift desired here. The long drift period is allowable The results of a station-keeping simulation are shown in
because control maneuvers are expected to be Figure 6. After achieving the target orbit, the spacecraft

Sarda 7 32nd Annual AIAA/USU

Conference on Small Satellites
are in their maintained within the desired control capable of maintaining the spacecraft within the desired
bounds using only four maneuvers every week. The control window. The set of maneuvers over the two-
departures from the reference orbit seen in Figure 7 are year period is given in Figure 8. The fuel consumption
due to mismatches between the true and modeled throughout the mission is regular, showing that a
relative motion in the simulation, as well as relative steady-state of control has been achieved. Overall, the
position determination errors, thrust timing and spacecraft use about 2.2 m/s per year of operations to
magnitude errors, and attitude pointing errors modeled maintain the formation, yielding propellant margins in
in the simulation. In spite of the modeled errors and excess of 80%.
non-idealities, the station-keeping method employed is

Table 1: Navigation error applied to differential mean orbital element states in formation simulations.

Mean 3 -7×10-7 -6.78×10-9 -5.93×10-9 -6.5×10-9 -1.13×10-8

Stdev. 1 1.82×10-8 5.88×10-7 5.934×10-7 1.35×10-8 1.95×10-8

Table 2: Target differential mean orbital elements for initialization relative to Hawk-B.

Hawk-A 0 0.018 0 0 0 1.44×10-3

Hawk-C 0 -0.018 0 0 0 1.44×10-3

Figure 6: Initialization trajectory in differential mean orbital element space (blue is the initial plan, black is
the actual trajectory, and the red X marks the final desired state).

Sarda 8 32nd Annual AIAA/USU

Conference on Small Satellites
 Figure 7: Station-keeping results for 2 years of operation.

Figure 8: Fuel consumption over 2 year station-keeping period.

Sarda 9 32nd Annual AIAA/USU

Conference on Small Satellites
THE FUTURE specific studies, but the actual constellation size and
At the time of writing, all three Hawks are in final flight geometry will depend on requirements that stem from
integration and environmental test. The payloads have the results of the Pathfinder mission. Figure 8 shows
one example constellation. The clusters are in 650 km
been fully flight integrated, tested at the spacecraft
circular orbits and divided into three planes: 97°, 44°,
system-level, and ready for flight. In the coming
and 63.5° (chosen for this example because of their
months, the spacecraft will be subject to
common availability in cluster launches). Two clusters
electromagnetic compatibility (EMC), vibration, and
thermal vacuum (TVAC) test before being shipped to are distributed per plane, with the clusters separated by
the launch integration facility in early August 2018, for 180°. It is evident that even with a simple constellation
design, global revisit rates are quite high, especially in
a launch in October 2018.
those latitudes most commonly populated.

Finally, the Pathfinder mission is a successful example

of well-co-ordinated execution on a multi-organization
project. Often, missions being developed within multi-
part groups tend to face financial and schedule burdens
of bureaucracy, logistics, documentation and
communication restrictions. This partnership was built
for success at the outset as each organization was able
to focus on the core competencies each brought to the
table, allowing for the rapid development and fiscally
responsible approaches synonymous with the
microspace design philosophy.

Figure 10: Example 18 Satellite Constellation Revisit

Figure 9: A HE360 Pathfinder satellite in EMC test ACKNOWLEDGEMENTS

The authors of would like to thank the teams of
The ground segment is currently under development engineers, managers and support personnel at the
and test. The network will involve UHF/S-band offices of HawkEye 360, Deep Space Industries, and
telemetry and command stations located at the Virginia Space Flight Laboratory who devoted their time and
headquarters of HawkEye 360, however, the mission passion in enabling this project to take shape.
will largely be operated out of northern latitude KSAT
S-band/X-band stations. REFERENCES
Following the successful demonstration of the 1. G. Bonin, N. Roth, S. Armitage, J. Newman, B.
Pathfinder mission, HE360 plans to deploy a Risi, and R. E. Zee, “CanX-4 and CanX-5
commercial constellation of similar clusters of Precision Formation Flight: Mission
spacecraft. This constellation would provide similar Accomplished!”, in Proc. of 29th Annual
geolocation services on a global scale with high revisit AIAA/USU Conference on Small Satellites
rates. HE360 has modeled constellations with as many (SmallSat 2015), Logan, UT, USA, Aug. 2015.
as eighteen spacecraft (six clusters of three Hawks) for

Sarda 10 32nd Annual AIAA/USU

Conference on Small Satellites
2. D. CaJacob, N. McCarthy, T. O’Shea, R.
McGwier, “Geolocation of RF Emitters with a
Formation-Flying Cluster of Three
Microsatellites”, in Proc. of 30th Annual
AIAA/USU Conference on Small Satellites
(SmallSat 2016), Logan, UT, USA, Aug. 2016.
3. J. Newman, and R. E. Zee, “Drift Recovery and
Station Keeping Results for the Historic CanX-
4/CanX-5 Formation Flying Mission”, in Proc. of
29th Annual AIAA/USU Conference on Small
Satellites (SmallSat 2015), Logan, UT, USA,
Aug. 2015.
4. E. Kahr, N. Roth, O. Montenbruck, B. Risi, and
R. E. Zee, “GPS Relative Navigation for the
CanX-4 and CanX-5 Formation Flying
Nanosatellites”, Manuscript submitted for
publication, 2017.
5. C. W. T. Roscoe, J. J. Westphal, J. D. Griesbach,
and H. Schaub, “Formation Establishment and
Reconfiguration Using Differential Elements in
J2-Perturbed Orbits”, Jour. of Guidance, Control,
and Dynamics, Vol. 38, No. 9, pp. 1725-1740,
Sept. 2015.
6. G. Gaias, S. D’Amico, and J.S. Ardaens,
“Generalized Multi-Impulsive Maneuvers for
Optimum Spacecraft Rendezvous”, in. Proc. of
5th International Conference on Spacecraft
Formation Flying Missions and Technologies
(SFFMT 2015), Munich, Germany, May 2013.
7. S. D’Amico and O. Montenbruck, “Proximity
Operations of Formation-Flying Spacecraft Using
an Eccentricity/Inclination Vector Separation”,
Jour. of Guidance, Control, and Dynamics, Vol.
29, No. 3, pp. 554-563, May-June 2006.
8. S. D’Amico, Autonomous Formation Flying in
Low Earth Orbit, PhD thesis, TU Delft, 2010.

Sarda 11 32nd Annual AIAA/USU

Conference on Small Satellites