Source of Acquisition

NASA Johnson Space Center

Space Shuttle Navigation

In The GPS Era
John L. Goodman

United Space Alliance

NASA Johnson Space Center
Houston, Texas

BIOGRAPHY to the present is given, along with how GPS and GPS/INS
technology will change, or not change, the way Space
John L. Goodman is employed by United Space Alliance Shuttle navigation is performed in the 21 5` century.
at the NASA Johnson Space Center, in support of the
Mission Operations Directorate. Mr. Goodman graduated
from the University of Arizona in 1986 with a B. S. in INTRODUCTION
Aerospace Engineering. His experience includes flight
software verification, rendezvous guidance and In the 1970s, the Space Shuttle was designed with three
navigation analysis; and Global Positioning System TACAN units for use during the entry phase of flight. By
(GPS) applications for the Space Shuttle, Crew Return the late 1970s, with the NAVSTAR GPS program
Vehicle, and metric tracking. promising a revolution in navigation, the use of GPS on
the Shuttle orbiters was studied [1]. The orbiters
Discovery, Atlantis and Endeavour were all manufactured
ABSTRACT in the 1980s with GPS antennae and associated wiring.
However, due to budget concerns and the developmental
The Space Shuttle navigation architecture was originally nature of GPS, a GPS upgrade to the Shuttle system was
designed in the 1970s. A variety of on-board and ground not pursued.
based navigation sensors and computers are used during
the ascent, orbit coast, rendezvous, (including proximity In 1990, the introduction of GPS had led the Department
operations and docking) and entry flight phases. With the of Defense and the Federal Aviation Administration to
advent of GPS navigation and tightly coupled GPS/INS plan the phase out of TACAN ground stations beginning
Units employing strapdown sensors, opportunities to in the year 2000. GPS promised better performance than
improve and streamline the Shuttle navigation process are TACAN and reduced operating costs. In response, then
being pursued. These improvements can potentially Shuttle program Manager (and former astronaut) Robert
result in increased safety, reliability, and cost savings in Crippen initiated an effort to study the possibility of
maintenance through the replacement of older replacing the three TACAN units on each orbiter with
technologies and elimination of ground support systems three GPS receivers.
(such as Tactical Air Control and Navigation (TACAN),
Microwave Landing System (MLS) and ground radar). By 1996, the development of Embedded GPS/INS (or
Selection and missionization of "off the shelf' GPS and EGI) units employing strapdown inertial sensors [2]
GPS/INS units pose a unique challenge since the units in motivated the Shuttle program to look at an eventual
question were not originally designed for the Space replacement of the previously mentioned stand alone GPS
Shuttle application. Various options for integrating GPS (the TACAN replacement) and stable member, spinning
and GPS/INS units with the existing orbiter avionics mass gyro Kearfott High Accuracy Inertial Navigation
system were considered in light of budget constraints, Systems (HAINS) Inertial Measurement Units (IMUs)
software quality concerns, and schedule limitations. An with EGIs. The use of strapdown IMUs employing ring
overview of Shuttle navigation methodology from 1981 laser gyro (RLG) technology was identified as a potential

ION National Technical Meeting, Long Beach, CA, January 22-24, 2001. Copyright 02000 by United Space Alliance, LLC. Published
by the Institute of Navigation with permission. These materials are sponsored by the National Aeronautics and Space Administration
under Contract NAS9-20000. ,The U.S. Government retains a paid-up, nonexclusive, irrevocable worldwide license in such materials to
reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the U.S.
Government. All other rights are reserved by the copyright owner.
source of cost and schedule savin ,, s during orbiter Orbit, Rendezvous & Docking
processing.
• Three HAINS IMUs
Both the TACAN replacement and GPS/IMU replacement • Two Star Trackers
efforts were conducted in parallel. Studies were • One Ku Band Rendezvous Radar
performed to determine the best integration architecture. • One Crew Optical Alignment Sight (COAS)
Stand alone GPS units and EGIs had to be integrated in a • Trajectory Control Sensor JCS, a laser)
manner that did not compromise the integrity of an • Two Hand Held Lasers (HHL)
operational, certified navigation system. Once it was • Two Payload Bay Television Cameras With
decided to tailor the GPS receiver requirements to Ranging Ruler Overlays
TACAN replacement, studies were conducted to • Ground based C Band radar tracking.
determine if there were other applications for GPS during • Tracking And Data Relay Satellite (TDRS) S
Shuttle flights. Band Doppler tracking.

Changes to the Shuttle General Purpose Computers Entry And Landing

(GPCs) to support both projects were identified. GPS and
EGI hardware was selected; interface, firmware and • Three HAINS IMUs
hardware changes were identified and made. Both the • Three TACANs
GPS and EGI units were flown on a series of Shuttle • Four Barometric Altimeters
missions for test and evaluation. Modifications were • Three Ku Band MLS Receivers
made to GPS and EGI firmware based on flight test • Two Radar Altimeters
results. • Ground based C & S Band radar tracking.
• At least two TACAN ground stations.
• MLS ground equipment.
OVERVIEW OF SPACE SHUTTLE NAVIGATION
SYSTEM For HAINS IMUs, TACANs, barometric altimeters and
MLS units, redundant measurements are passed through
Navigation requirements for manned spacecraft are Fault Detection, Identification and Reconfiguration
different than those for terrestrial aircraft, unmanned (FDIR) algorithms [7]. Selection filters then process data
Earth orbiting satellites and interplanetary probes [3, 4]. from units deemed to be "good." Selected measurements
The guidance, navigation and flight control system had to are then passed to the Shuttle navigation software.
be designed to ensure safety of flight and mission success HAINS IMU data is used during all flight phases for
for ascent, orbit, rendezvous and gliding entry to a attitude and sensed velocity determination. During entry,
runway landing in the presence of system failures. For PASS and BFS Kalman filters process selected TACAN,
ascent and entry, a total of five computers, running in IMU derived altitude and barometric altimeter
parallel, can control the vehicle. Four of the GPCs run measurements. Only the PASS Kalman filter processes
what is known as the Primary Flight Software Sub- selected MLS data. Selected radar altimeter data is
System (PASS) software. The fifth runs Backup Flight provided for crew situational awareness, but is not
Software (BFS). PASS and BFS software was coded by incorporated into the navigation solution.
different contractors. The BFS contains a subset of the
PASS functionality to enable the vehicle to finish nominal During rendezvous, a PASS Kalman filter processes on-
ascent, an abort or landing in the event of a generic PASS board radar, star tracker and COAS measurements. TCS
software failure. While on orbit, only PASS software and HHL data are processed in Kalman filters residing in
provides guidance, navigation and flight control a laptop computer for crew situational awareness.
functions. Multiple navigation sensors, command paths Television monitors with ranging ruler overlays provide
and power buses within the avionics system ensure range information to the crew during docking.
redundancy [5, 6].

Navigation sensors used in the "pre GPS era" include: INTEGRATING GPS IN A LOOSELY COUPLED
ARCHITECTURE
Ascent
A loosely coupled architecture using cascaded filters is a
• Three HAINS IMUs common way of upgrading existing navigation systems
• Ground based C & S Band radar tracking. with GPS. This method of integration has been used on
platforms such as the F-16 [8], F-1 17 [9], Conventional THE TIGHTLY COUPLED OPTION
Air Launched Cruise Missile [ 10] and B-2 [ I I].
In a tightly coupled integration, processing actual GPS
A cascaded filter approach involves using the position measurements (pseudorange, delta range) from a GPS
output of the GPS receiver as a measurement for a receiver in a host vehicle computer permits the navigation
Kalman titter in the host vehicle navigation system [8]. system designer to have more control over the quality of
Position and velocity aiding data from the host vehicle the navigation solution, rather than having to rely on a
INS are fed back to the GPS receiver to improve signal receiver vendor's proprietary firmware. Processing high
acquisition and tracking performance. However, the rate, unfiltered GPS observables and inertial measurement
Kalman filter is derived under the assumption that unit data in a central Kalman filter permits higher
measurements are not time correlated. This assumption is accuracy navigation and less vulnerability to GPS outages
violated by processing the GPS position vector in the host and jamming. This architecture enables rapid estimation
vehicle Kalman filter. Time correlated measurements can of GPS receiver clock errors and inertial measurement
lead to filter instability. This problem has been avoided unit errors [8].
in many applications by processing GPS position vectors
at a low rate, such as no higher than every 15 seconds Processing of GPS pseudoranges in the PASS and BFS
[ 10]. flight software was also considered, but not chosen. Like
the cascaded filter options discussed previously, it would
One option considered was to process GPS position have required modifications to baseline entry navigation
vectors as measurements in the PASS and BFS Kalman and a new Kalman filter for the on-orbit phase.
filters, as is done in many military applications. Furthermore, there were security concerns with sending
"corrected" pseudoranges outside the keyed GPS receiver
Another option studied was to convert GPS position into to the Shuttle GPCs. Uncorrected pseudoranges could be
a TACAN measurement (known as "TACAN processed in the GPCs, but the entry and new orbit
transparency"). This was an attempt to minimize or avoid Kalman filters would have to solve for the corruption due
changes to the PASS and BFS flight software and to Selective Availability.
input/output. This would have resulted in a high rate
cascaded filter implementation (TACAN measurements
are processed every 3.84 seconds). A "transparent" STATE REPLACEMENT CHOSEN
approach would not have allowed INS aiding to be
supplied to the receiver from the GPC. Crew displays After considering eight integration options, the Shuttle
required GPS receiver specific controls and quality program directed that GPS be integrated "in parallel" with
assessments, and GPS data needed to be sent to Mission the existing "baseline" navigation on the orbiters. A GPS
Control for the flight controllers. For on-orbit processing, state (position and velocity) selected by the Shuttle flight
a new Kalman filter would have to be created. On-orbit software would "overwrite" the Shuttle flight software
processing of "pseudo TACAN" measurements could navigation state. "State replacement" in the Shuttle
only have been performed within the line-of-sight of the navigation software was chosen over treating GPS as a
TACAN stations already in the flight software to support sensor and filtering the GPS measurements or state
landing. vector. This architecture treats the GPS receiver as a
complete navigation system [12]. Any potential problems
Neither of the loosely coupled, cascaded filter options with cascaded Kalman filters would be avoided. State
were acceptable to the Shuttle program. Processing GPS replacement could be implemented so that the same GPC
as a TACAN measurement or filtering the GPS position flight software could be used for GPS processing during
vector could not have been evaluated in flight without both orbit and entry.
actually incorporating the data into the navigation state.
Both of these options would have required retuning the Another advantage of state replacement is that it perm its
entry navigation Kalman filters in the PASS and BFS flying the GPS receiver in a test mode during flights,
flight software. A new PASS flight software Kalman Without having to use GPS for navigation. This allows
filter for the orbital phase of flight would have to be operation of a "mixed fleet" of GPS and TACAN
created. Any modifications to the existing filters for hardware configurations with the same version of PASS
entry, or a new Kalman filter for orbit, would result in an and BFS flight software. Due to the schedule of orbiter
extensive flight software development and certification overhaul periods, not all orbiters would be equipped with
effort. GPS units at the same time. Supported TACAN/GPS
configurations are:
 • Three TACANs for operational use and one GPS the current navigation state and incorporates (forces) the
receiver for data gathering. selected GPS state into navigation.
• Three GPS units for operational use and no
TACANs.
• Three TACANs and no GPS onboard. RECEIVER SELECTION

It was believed that this approach would also make it For TACAN replacement, the Shuttle program desired an
easier to upgrade the Shuttle with more advanced off the shelf, in production military unit designed for
receivers, whereas the Kalman filtering of GPS aircraft to take advantage of the existing production line
measurements or vectors in the Shuttle GPCs might and logistics base. The Shuttle program also desired to be
require retuning the filters in the GPC. Furthermore, if a an authorized user of GPS, to take advantage of the
receiver upgrade occurred, it was believed that this jamming and spoofing resistance provided by military
approach would eliminate the need for detailed GPS units. In addition, the Shuttle program wanted to
knowledge of the GPS receiver's firmware. benefit from firmware that had been "matured" through
development and use by the Department of Defense. The
The GPS receivers are provided with position, velocity Collins Miniaturized Airborne GPS Receiver (MAGR), a
and attitude aiding from the Shuttle flight software 5 channel unit, was selected.
(PASS, or BFS in the event of a PASS software failure).
One aiding state vector is propagated for all three Unlike earlier GPS receivers, the MAGR used digital,
receivers, using selected IMU data from candidates that rather than analog tracking loops. This permitted space
have been screened by a FDIR algorithm. The single missionization to be performed via firmware, rather than
aiding state is periodically reset with the Shuttle hardware changes. Existing "space rated" GPS receivers
navigation state. were not suitable for the Shuttle (i.e. didn't accept inertial
aiding, too big and heavy for the Shuttle, not capable of
The PASS flight software subjects GPS state vectors to "authorized" operation). At the time, there was not a
three Quality Assurance (QA) checks. The QA tests were military aircraft, all-in-view receiver in production.
designed so that retuning of the receiver Kalman filter, or
changes to receiver residual edit tests would not be
necessary for the Shuttle missionization. INITIAL FLIGHT TESTS AND RECEIVER
MISSIONIZATION
• A check of several receiver parameters, one of
which is the Figure of Merit (FOM). The first flight of a GPS receiver on the Shuttle was on
mission STS-51 in September of 1993 [13]. A Trimble
The current GPS state is compared with the TANS Quadrex was flown. This experiment was not a
receiver's previous state propagated to the part of the TACAN replacement project. The Quadrex
current time. was mounted in an overhead window on the flight deck.
Signal attenuation from the glass and limited field of view
• Comparison of the receiver states with each severely impacted receiver performance.
other.
The initial flight test program supporting TACAN
If criteria for any of the QA tests are violated, that replacement involved the 5 channel, Collins 3M receiver.
receiver's state is not a candidate for selection. State The 3M was a pre-production version of the Collins
vectors from candidate GPS units are then processed in a MAGR. This series of flight tests was intended to prove
selection filter. The BFS flight software uses a simpler if a GPS receiver designed for terrestrial aircraft use could
QA and selection scheme than PASS. There are crew function on the Space Shuttle with a minimum number of
controls that allow these QA tests to be bypassed, if firmware changes.
necessary.
INS aiding was supplied to the 3M by the BFS GPC
Once the QA checks are complete and a GPS state has during ascent and entry, while on-orbit the receiver was
been selected, there are two crew commanded methods unaided and operated by a laptop computer. Receiver
for incorporating the selected GPS state into navigation. state vectors and other 3M data were recorded on the
The first involves automatic incorporation at a flight laptop. The 3M was not keyed. Since the purpose of the
phase dependent rate, if the selected GPS state is within a 3M flight tests was purely data gathering, no navigation
tolerance of the current navigation state. The second data was sent from the 3M receiver to the BFS GPC. The
method, called a "force," Ignores the comparison test with Collins 3M unit flew seven times (December, 1993 to
May, 1996) on Endeavor (flights 61, 59, 68, 67, 69, 72 and BFS flight software on several occasions while on
and 77). Modifications were made to the 3M firmware orbit. On STS-103 (December 1999, before the
between flights based on flight test results. deactivation of Selective Availability), the MAGR/S was
intentionally unkeyed prior to entry. The Shuttle will be
Hardware and firmware missionization of the Collins certified to land with the MAGR/S unkeyed. Another
production MAGR for TACAN replacement began in DTO involved a "late power-on" of the MAGR/S just
1995. The version of the MAGR flown on the shuttle is prior to entry on STS-92 (October 2000). This tested the
known as the MAGR/S (MAGR/Shuttle). Collins based ability of the MAGR/S to collect ephemerides, download
MAGR/S firmware on military MAGR Link 008, with the daily key and establish four satellite navigation during
Link 009 and 010 modifications also included. Lessons entry.
learned from the 3M flights were also incorporated into
the MAGR/S firmware. For most of ascent, the Shuttle is in a "heads down"
configuration. The GPS antenna on top of the crew
compartment is facing the Earth, while the antenna below
"SINGLE STRING" TACAN REPLACEMENT the crew compartment is facing the External Tank. In
FLIGHT TESTS spite of the poor antenna visibility, enough GPS signals
"wrap around" the External Tank and orbiter to permit the
A single MAGR/S unit (the "single string" configuration) MAGR/S to track three to four satellites most of the time.
is being flown bn each orbiter for several years during a However, the Geometric Dilution Of Precision (GDOP) is
test and certification-program. The single string MAGR/S high, which can result in large state errors. On most
flown during the test phase has two antennas, one on the flights, a roll to a heads up attitude is performed at about
top and one on the bottom of the crew compartment (GPS 6 minutes into the flight. This provides excellent GPS
2 in Figure I). Input from the antennas are passed satellite visibility to the antenna on top of the crew
through pre-amplifiers and a signal combiner before compartment. Receiver tracking and performance during
reaching the MAGR/S. On many flights, a laptop this phase of ascent is usually exceptional.
computer is used to record instrumentation port data.
This data proved to be invaluable in resolving MAGR/S Periods of noisy GPS velocity were observed during the
performance issues. orbital phase of flight. Velocity noise was as high as I I
GPS 2 feet/second. Noise periods lasted anywhere from 5 to 20
minutes. Analysis indicated that the MAGR/S was
processing noisy delta range measurements due to
ionospheric scintillation [14]. This does not pose a
GPS I GPS 3 constraint to certification of GPS as a TACAN
replacement.

GPS 3 GPS 1 Of particular interest is MAGR/S performance during the

"plasma" region of re-entry. The term "blackout" is not
used, since communication via the TDRS satellites can
GPS 2 permit near continuous communication with the orbiter
Figure 1: GPS antennae locations. during entry, depending on the orbiter ground track and
TDRS satellite visibility.
The first flight of the single string MAGR/S configuration
was on STS-79 in September of 1996. MAGR/S data was Flight tests demonstrated that the MAGR/S does not
available in "real time" to Mission Control personnel experience a complete "blackout" of GPS signals during
during ascent and entry. More flights followed as each entry. Plasma effects begin to degrade lower antenna
orbiter in the fleet was equipped with a single MAGR/S tracking of GPS satellites as high as 320,000 feet. Upper
receiver. STS-91 in June of 1998 was the first flight antenna tracking begins to degrade about the time of the
during which GPS data was available "in real time" to first roll maneuver (typically around 285,000 feet). In
Mission Control personnel daring the orbital phase of spite of the plasma, the MAGR/S still tracks I to 4
flight. satellites during the plasma region. Frequent loss of lock
occurs, along with a loss of carrier frequency tracking.
Detailed Test Objectives (DTOs) are being carried out Code tracking is maintained.
during the single string flights. These tests involve
astronaut execution of MA,GR/S procedures. Selected Upper antenna plasma effects begin to subside around
MAGR/S state vectors were incorporated into the PASS 217,000 feet, and continuous carrier frequency tracking
on 4 satellites resumes by 200,000 feet on most missions MISSION CONTROL AND ON-BOARD,
Lower antenna tracking can be degraded by plasma until OPERATIONAL USE OF GPS
around 185,000 feet.
The following sections give an overview of the current
The impacts of the plasma region on the MAGR/S are Shuttle navigation methodology. How GPS states will be
extended periods of less than four satellite tracking, failed used by the orbiter navigation system and Mission
reacquisition attempts, incomplete data collection (i.e. Control after MAGR/S certification is covered. Range
ephemerides) and loss of delta range measurements. Less safety is outside the scope of this paper.
than optitnutn number of measurements and poor satellite
geometry could result in an increasingly inaccurate GPS
state vector. However, navigation errors during the PRE GPS GROUND TRACKING
plasma region has so far not been excessive. The orbiter
can fly through this region without an update from GPS. Initially, the Shuttle program required continuous
tracking from 2 radars during ascent and entry, so that
Flight data from the MAGR/S as well as the GPC QA Mission Control could determine the Shuttle state
checks are examined both during and after missions. independently of the on-board navigation system. As
Flight tests and simulation results have driven changes to confidence was established in the on-board system, the
the MAGR/S firmware and PASS and BFS flight software ascent/entry ground tracking requirement was changed
that processes -MAGR/S data from mandatory to highly desirable.

Flight experience also drove a change to the position and Prior to launch, if there are certain failures on the vehicle,
velocity vector aiding scheme used by the PASS flight radar tracking to support navigation is required to meet
software. The original design involved propagation of a Launch Constraint Criteria. C Band (range and angles)
separate aiding state for each receiver, with each and S Band (range, Doppler, angles) radar data is
propagator having an independent source of IMU data processed in a Mission Control based Kalman filter. The
(i.e. an IMU is assigned to each receiver). Instead, one filter can process data froth one S Band and two C Band
aiding state vector is propagated for all three receivers, radars.
using selected IMU data from candidates that have been
screened by a FDIR algorithm. Instead of periodically If the Shuttle is to rendezvous with a spacecraft already in
resetting the aiding states with each receiver's own GPS orbit (a "ground up rendezvous," such as with the
state, the single aiding state is periodically reset with the International Space Station (ISS) or Hubble Space
Shuttle navigation state. The original design did not Telescope), Mission Control checks the cross track
perform a "sanity check" on the GPS states used to reset velocity error after Main Engine Cut Off. A state vector
the aiding states in the PASS flight software. Resetting update may be required prior to the OMS-2 maneuver for
the single aiding state with the Shuttle navigation state cross-track velocity error greater than 6 feet/second. The
takes advantage of protection provided by the GPS QA uplink would be based on ascent ground tracking data
checks, the GPS state selection process and the processing.
comparison of the selected GPS state with the current
navigation state before GPS incorporation. The new Initially, S Band communication sites provided tracking
aiding scheme also permits the ground to have more (range, Doppler, angles) during the orbital phase of flight,
insight into and control over the position and velocity when the orbiter was visible. As TDRS satellites were
aiding states sent to the receivers. launched, ground S Band use for orbit was reduced.
TDRS provides near global communications coverage,
Once the MAGR/S units and Shuttle flight software for but only Doppler measurements due to the type of S Band
GPS are certified (anticipated to occur in the spring of transponder on the orbiter.
2002), the three TACAN units on each orbiter will be
removed and two MAGR/S receivers will be added to On-orbit, both S Band TDRS Doppler measurements; and
each orbiter, for a total of three MAGR/S units per C Band radar data (range and angles from a number of
orbiter. TACAN replacement will occur as each orbiter is tracking stations) are processed to estimate the orbiter's
cycled through a regularly scheduled overhaul. Antennas state vector. The on-board navigation state is usually
for the two additional MAGR/S units are on the top and allowed to grow in down-track position error before a
bottom (Figure 1) of the nose in places formerly occupied new state, based on radar and TDRS tracking, is uplinked.
by TACAN antennas. The first "three string GPS flight" For landing, the maximum allowable downtrack position
(no TACAN) is expected to occur no earlier than 2003. error at Entry Interface (400,000 feet) is 20 nautical
miles.
TDRS S Band Doppler tracking provides a good estimate modify the Kalman filter for each application [15]. Filter
of the orbital semi-major axis (which reflects both tuning assumes a worst case inertial measurement unit
position and velocity). C Band tracking resolves planar and receiver clock. The downside to this approach is that
errors. Radar geometry plays a large role in determining the receiver is not optimized for space navigation. It
the accuracy of ground-based navigation. Oil a effectively provides a blended deterministic solution
weighted least squares algorithm is used by Mission without accurate gravity or drag modeling needed for
Control for orbit determination. accurate orbit determination. As a result, MAGR/S
velocity errors, and the resulting semi-major axis errors,
Ground-up rendezvous flights require radar tracking of are not acceptable for orbit maneuver planning and
the target spacecraft. Radar tracking begins from 18 to 24 rendezvous. Poor orbital navigation performance of GPS
hours prior to Shuttle launch. Ground tracking of the receivers originally designed for terrestrial use and the
target spacecraft continues through docking (ISS) or importance of semi-major axis accuracy is covered by
grapple (Hubble Space Telescope). If the orbiter deploys Carpenter and Schiesser [ 161.
a scientific payload that is to be retrieved later in the
mission, ground tracking of the deployed spacecraft will A GPS Ground Filter, for use in Mission Control, has
be performed to facilitate the rendezvous and grapple. been developed by United Space Alliance to support
Shuttle and ISS maneuver planning and collision
For entry and landing, radar data is processed by Mission avoidance with space debris. The accuracy specifications
Control in the previously mentioned Kalman filter for for the ISS GPS receiver were designed to support
independent assessment of onboard navigation systems. antenna pointing. Like the MAGR/S, the ISS receiver has
Since Shuttle landings are planned for the Kennedy Space not been modified for space dynamics, and is not accurate
Center (KSC), and radars are located there to support the enough to support maneuver planning. Once certified,
Eastern Test Range, C Band and S Band radar tracking is the GPS Ground Filter will filter GPS position and
usually available. C Band tracking is also usually velocity to provide Mission Control with more accurate
available for landings at Edwards Air Force Base, since GPS derived state vectors for the Shuttle and ISS.
NASA has radars at the Dryden facility. However, range
scheduling can prevent entry tracking. If there are certain Near continuous availability of ground filtered GPS
failures in the on-board or ground navigation systems, state vectors, in conjunction with radar and TDRS
radar tracking during entry becomes mandatory. tracking, will allow more responsive mission planning.
Burns will be confirmed more rapidly than with radar and
During entry, radar normally becomes available in time TDRS data alone. If the orbiter is subjected to
for Mission Control to assess the TACAN units and perturbations that are not modeled by Mission Control, it
navigation state health prior to TACAN processing in the call 2 or 3 revolutions before enough ground tracking
GPCs. Mission Control can also perform an emergency data is available to quantify the impact on the orbiter
state uplink to the orbiter if navigation errors are state. The GPS ground filter will permit much more rapid
excessive. perturbation determination.

Once GPS is certified for operational use, ground radar

POST GPS tracking will be used oil initial "no TACAN" flights
as a backup to GPS. After experience is gained with the
Once GPS is certified, the Shuttle program will still certified GPS system on the Shuttle, it is expected that
consider radar tracking during ascent to be desirable, but ground radar tracking activities will be reduced. TDRS is
not required, unless there are navigation equipment also used for telemetry and communications, thus the
failures on the vehicle. Doppler measurements will always be available for
processing. However, radar and TDRS will still be
Radar and TDRS tracking will still be used during the designated as backups to GPS, and Mission Control
orbital phase to support maneuver planning in Mission Center personnel will maintain proficiency in C Band and
Control. The MAGR/S Kalman filter has not been tuned S Band measurement processing.
for orbital dynamics, since emphasis was on a TACAN
replacement and a Commercial Off The Shelf (COTS)
approach that minimized MAGR/S modifications. ASCENT AND POST INSERTION

The MAGR Kalman filter and navigation algorithms were Navigation aids (TACANs, barometric altimeters, MLSs,
designed so that the unit could be integrated into a wide radar altimeters) that would be used for all
variety of platforms, withofl . t havin g to substantially landing due to an ascent abort are powered on and self
tests are run prior to the day of launch. The three RAINS ORBIT COAST NAVIGATION IN THE GPS ERA
IMUs are calibrated and aligned prior to liftoff. Only
I IAINS IMU data (accumulated sensed velocity, gimbal Current plans call for two of the three MAGR/S receivers
angles) are processed by PASS and BFS navigation to be powered off for most of the orbit period. MAGR/S
during powered flight [ 17]. No sensor measurements are state vectors will periodically be taken into the PASS
processed by a Kalman filter. flight software to maintain the on-board navigation state
accuracy at an acceptable level. Ground and TDRS
tracking will remain the primary source of state vectors
ASCENT AND POST INSERTION IN THE GPS ERA for maneuver planning in Mission Control, until the GPS
Ground Filter is certified.
There is no change to the Shuttle baseline navigation (use
of HAINS IMU data) for powered flight. The MAGR/S There is no provision for GPS attitude determination in
units will be turned on approximately 5 hours prior to the MAGR/S. Star sightings will still be used for precise
liftoff. This permits receivers to collect ephemerides of HAINS IMU alignment.
satellites that will be over the Trans Atlantic Landing
(TAL) sites at launch. GPS states are not used by the
Shuttle navigation software during powered flight, but PRE GPS RENDEZVOUS
receiver aiding data is supplied to the MAGR/S units by
the GPCs. Successful rendezvous requires an accurate relative state.
Relative velocity errors, in particular, are critical.
After the MAGR/S and the GPS Ground Filter are However, ground tracking of the orbiter and the target
certified, the GPS Ground Filter could be used as a source spacecraft is not accurate enough to guarantee a safe
of post Main Engine Cut Off state vector uplinks. GPS rendezvous within the orbiter's tight propellant budget.
vectors from the on-board MAGR/S units would not be An on-board navigation system that provides an accurate
incorporated directly into navigation. relative state is required.

Shuttle rendezvous is divided into two phases, ground

PRE GPS ORBIT COAST NAVIGATION targeted and on-board targeted. For the ground-targeted
phase, the orbiter navigation state is determined by
During the orbit phase, the on-board navigation state is ground based radar and TDRS tracking as described
monitored and maintained by Mission Control via state previously. Burns to be executed by the orbiter are
vector uplinks. A vent force may also be uplinked by computed by Mission Control (hence the term "ground
Mission Control for use by the GPCs. This takes into targeted") and uplinked to the vehicle for execution by
account non-propulsive forces acting on the orbiter that the crew. The last ground targeted burn is performed on
cannot be detected by the HAINS IMUs, and reduces the day of rendezvous, 40 nautical miles behind the
error growth in the on-board state. Vent values are based target, at about orbital noon. During the on-board
on flight history for specific orbiter attitudes. The orbiter targeted phase (Figure 2), relative sensor measurements
state (and for a rendezvous, the target spacecraft state) are processed in a Kalman filter in the PASS flight
determined from C Band and S Band tracking are used by software [ 19]. The PASS software maintains an estimate
Mission Control for maneuver planning. of the target spacecraft state (initially provided by
Mission Control) and the filtered orbiter state. These
Alignment of the HAINS IMUs is periodically performed states are used to compute burn solutions using an
using star sightings [18]. Two star trackers with near onboard Lambert targeting algorithm [20].
orthogonal lines of sight are located on the nose of the
orbiters. Data from star sightings can also be used by The star trackers used for IMU alignments are also used
Mission Control to determine IMU gyro biases. The to process angular measurements of the orbiter/target
ground determined biases are then uplinked for use in the relative state. This permits resolution of navigation errors
Shuttle PASS flight software. If the orbiter cannot normal to the star tracker line of sight. The star tracker
maneuver to a star sighting attitude due to excessive IMU pass begins shortly after the last ground-targeted
misalignment, a rough alignment can be executed by a maneuver, and will resolve most altitude and out-of-plane
crew member sighting on a star using the Heads Up position errors. Due to change in relative geometry, some
Display (HUD) or the COAS. This would be followed by down-track position error will be resolved at the end of
a precise star alignment using the star trackers. the star tracker pass. The end of the pass roughly
 I. NC. last around
+Vbarl -50 -100 -150 -200 -250 tar g eted burn.
-4 I . Start star tracker.
3. End star tracker.
4. NCC, first on-board
targeted burn.
5. Start Radar.
6. Ti Burn
7. Start star tracker
(for radar fail).
8. MC-I Burn
/^'
1 14 9. Sunset.
20 10. End star tracker
for radar fail.
I 11.MC-2 Burn.
10= 12. Sunrise
13. MC-3 Burn
I
30 14. MC-4 Burn
15.Manual piloting
I ^.
phase using radar,
TCS & I-111L.
Axes are in kilo-feet.
40t NCC — Corrective Combination
♦ +Rbar Ti —Transition Initiation
MC — Midcourse Correction
Figure 2: Day of rendezvous profile with burns and navigation activities indicated. Relative coordinate frame
centered on the tar get spacecraft. "War" is in the direction of target motion, "Rbar" is in the direction of the
center of the Earth.
coincides with orbital sunset, and is followed by the first After the final on-board targeted burn, at a range of about
on-board targeted maneuver. 2,000 feet, the manual piloting phase begins. Radar data
continues to be incorporated into the orbiter navigation
Incorporation of Ku Band radar measurements begins at a state during the manual phase, but the primary source of
range of 135,000 feet. Range, range rate and angular data relative navigation data for piloting comes from the TCS.
are obtained from this point until a range of about 100 TCS is a laser system mounted in the payload bay that
feet. Five more on-board targeted burns are executed to provides relative range, range rate and angular
control the orbiter's approach to the target spacecraft. information. TCS requires retro-reflectors on the target
vehicle. Two HHL range finders are also carried. Both
At a range of approximately 8 nautical miles, there is TCS and HHL can lock on to the target as far out as 5,000
another opportunity for a daylight star tracker pass, in the feet.
event of a radar failure. Due to the size and brightness of
the ISS, star tracking during orbital daylight may not be A Kalman filter is used in the primary and backup laptop
possible for this pass. A tracking light will be on the ISS computers to process TCS measurements. HHL
to facilitate processing of star tracker data at night. The measurements are processed in the primary laptop, in a
ISS will maneuver to a ni ght star tracker attitude at a different Kalman filter than TCS. Television cameras in
predetermined time durin g the rendezvous. Such a failure the payload bay, with ranging ruler overlays on the
occurred on STS-92 (October 2000), forcing execution of monitors, are used by the crew as an additional range
an "angles only" rendezvous. The rendezvous and information source from 15 feet to docking. Ku Band
docking was successful, but with slightly higher radar, TCS and HHL are also used during undocking and
propellant consumption. fly-around, when the orbiter leaves the ISS.

If both the radar and star trackers are failed, the Crew
Optical Alignment Sight may be used manually by the SHUTTLE RENDEZVOUS IN THE GPS ERA
crew to obtain relative an g ular measurements. COAS
measurements have never been taken in flight due to a Even if both spacecraft have GPS units, simple
sensor failure. However, after the undocking from Mir on subtraction of state vectors does not provide an accurate
STS-71, COAS data was processed as a test. enough relative state for use in burn computation. Use of
such data could easily result in an unsafe trajectory and an
undesirable level of propellant consumption.
Development of the on-board rendezvous navigation Kalman filter. It is also passed oil guidance, flight
system for Apollo [20] (on which the Shuttle system was control and crew display functions.
based) proved that rendezvous could be accomplished
with high inertial state errors on both vehicles, but low The first navigation filter measurement to be processed is
relative state errors [21 ]. High accuracy inertial states on called "drag altitude." Measurement incorporation begins
both spacecraft are not enough to permit accurate when the IMU sensed deceleration is greater than 1 I
rendezvous navigation. ft/sec' (Figure 3). Accumulated sensed delta velocity
data from the IMUs are used to estimate atmospheric
If GPS relative navigation were to be used for spacecraft density. A math model of the atmosphere then provides a
rendezvous, identical GPS receivers on the target and rough estimate of altitude. Drag altitude measurements
chaser vehicles are required. Processing of common are rather inaccurate, but are intended to bound error
satellite measurements from both vehicles in one Kalman growth in the event of other navigation system failures
Filter permits the cancellation of common errors and
biases (such as ionospheric error and Selective Kalman filter processing of selected TACAN range and
Availability) [22]. This in turn drives the need for a radio bearing measurements begins at a range no greater than
data link between the vehicles. 400 nautical miles from the runway and an altitude of
roughly 140,000 feet (for a nominal entry and landing).
The Space Shuttle program must preserve the capability TACAN bearing is not processed when the elevation
to rendezvous with spacecraft that are not equipped with angle of the slant range is 35 degrees or greater (cone of
GPS units. The current suite of on board, relative confusion).
navigation sensors (radar, star trackers, lasers) provide
relative states that are just as good as or better than the Significant navigation errors during entry can exceed the
level of accuracy possible with relative GPS. During the ability of the guidance and flight control system to fly the
manual piloting phase and docking, there are concerns orbiter to the landing site. A set of limits, called guidance
about GPS satellite visibility and multi-path. The Shuttle constraints, defines the maximum allowable state error.
program currently has no plans to use relative GPS for In the event of a navigation error that exceeds the
rendezvous. constraints, a correction to the state vector can be
uplinked directly to the PASS and BFS flight software or
voiced to the crew for manual entry. The "delta state
PRE GPS DEORBIT update." which is based on radar tracking data, has never
been executed in flight.
Currently, a state vector uplink is performed prior to the
deorbit burn to both the PASS and BFS flight software. Barometric altimeter measurements are processed to
This uplink is based on Mission Control processing of control navigation errors in the vertical channel. Two
ground radar C Band and TDRS S Band data. barometric altimeter probes are deployed at Mach 5, and
data becomes available for Mission Control evaluation at
Mach 3.5. Each probe provides two independent
DEORBIT IN THE GPS ERA measurements. Kalman filter processing begins at Mach
2.5 and an altitude of about 85,000 feet. Drag altitude
The current operations concept calls for the three processing ends once barometric altimeter processing
MAGR/S units to be operating 6 hours prior to the deorbit begins. Barometric altimeter processing is inhibited
barn. Selected MAGR/S data will be incorporated into between Mach 1.6 and Mach 1.1 (the Mach jump region).
the Shuttle navigation software prior to the burn,
replacing the uplink from Mission Control. As with PASS Kalman filter processing of MLS range, azimuth
ascent and orbit insertion, MAGR/S data will not be and elevation data usually begins at about 17,000 feet.
incorporated into navigation du ring powered flight. Once MLS is acquired, PASS stops processing TACAN
and barometric altimeter data and shifts from maintaining
three state vectors to one state vector. For a landing site
PRE GPS ENTRY NAVIGATION not equipped with MLS, PASS processing of TACAN
stops at an altitude of 1,500 feet, and baro processing
For most of entry, three independent navigation states are continues until 500 feet. BFS does not process MLS, and
maintained in the PASS [23]. Each uses accumulated, will process TACAN and baro data all the way to landing.
sensed velocity data from a different IMU to protect
against IMU failures. A selection filter selects one
navigation state (position and velocity) for use in the
 tips
4 100 400 25
Data Available
Drag ® For Navigation
1800 240 22 Urac = I I fi.iscr '
Data Used If
1200 220 18 ® No MLS
Not Used
800 200 I4
Ground TACAN
500 180 Tracking 1
U
— 10 , Lock On
350 160`
b
200 ^4 140 z 6 t
„ Baro
^% 150 120 Probes Deploy
c — M=5
4
iw 100-- ¢ 100 Calibrated at M = 3.5
c^ z i
GO 80 2 Alt. = 82.5 kli
=z.s
Or Until Baro A = 16
30 60

20 40

MLS Radar
10 20
Altimeter
0 0 U
Alt. = 500 fit I I.Ali. = 1500 ft.

Figure 3: Availability of navigation sensors and systems during entry. BFS does not process MLS, but processes TACAN
and Baro all the way to landing. Radar altimeter data is for crew situaonal awareness only. Ground tracking data is used
by Mission Control for independent assessment of onboard navigation systems.

Selected radar altimeter data is available to the crew for ENTRY NAVIGATION IN THE GPS ERA
situational awareness from 5,000 feet until landing. It is
not processed in a Kalman filter, due to a lack of accurate For the first several operational flights, selected MAGR/S
terrain models for Shuttle landing sites. data will not be taken into the Shuttle navigation software
between the deorbit burn and the acquisition of ground
radar tracking (around 140,000 feet, the TACAN
EMERGENCY USE OF "UNCERTIFIED," SINGLE acquisition attitude). This will permit a comparison of
STRING GPS DURING ENTRY MAGR/S data with ground tracking before incorporation.
Once operational experience with the three string system
During the flight test phase (prior to certification, only has been obtained, navigation system updates with GPS
one MAGR/S on each orbiter), flight rules were will resume after the deorbit burn and continue through
developed permitting use of selected MAGR/S state MLS acquisition (or landing if MLS is not available).
vectors tinder the following emergency conditions: However, selected MAGR/S states do not have to be
continuously incorporated into Shuttle navigation to
• Use in place of a voice delta state uplink during support entry and landing.
entry. Complexity of the voice delta state uplink
makes it more risky than incorporating Drag altitude measurements will still be processed to
uncertified GPS data. bound error growth in the event of GPS outages or IMU
failures. It is expected that drag measurements will have
• Avoid scenarios (low ceilings at landing, on- little impact on the navigation state when GPS is
board and/or ground station TACAN and MLS incorporated in the automatic mode.
failures) that could result in a high risk crew
bailout and loss of vehicle. Space Shuttle navigation software will continue to
process barometric altimeter and MLS data (in the PASS)
• Enable Mission Control to resolve dilemmas after TACAN has been replaced by GPS. Whenever
between redundant navigation sensors (HAINS selected barometric altimeter measurements are available
IMUs, barometric altimeters, TACANs, MLSs). for processing by the PASS and BFS GPCs, they will also
This does not require incorporating GPS states be processed by the MAGR/S Kalman filter. This allows
into the Shuttle navigation software. the GPS units to determine the bias on the barometric
measurements during four satellite tracking. In the event 1)11 FERENTIAL GPS
of less than four satellite tracking, calibrated barometric
altimeter measurements will help maintain accurate Although the MAGR/S does have a differential GPS
MAGR/S states. capability, there are currently no plans to replace MLS
with differential GPS. Antennas and cables would have
GPS states will be an important source of data for to be added to the Shuttle orbiters to enable reception of
Mission Control during emergency landings at sites where differential GPS corrections. There are technical issues
there is no ground radar tracking capability. with placing omni-directional VHF antennas under the
thermal protection tiles in spots formerly occupied by Ku
During entry, Mission Control in Houston will have an Band MLS antennas. Drilling more holes in the vehicle
open line to the GPS Master Control Station (MCS) at for antenna placement is not desirable. Studies have
Schriever Air Force Base, Colorado. Mission Control shown that differential GPS would provide little accuracy
will be informed of any GPS satellite integrity issues that improvement over the current MLS system.
arise. The crew has the ability to command the MAGR/S
units not to track specified GPS satellites for
measurements. SPACE INTEGRATED GPSANS (SIGI)

As GPS receivers became smaller, navigation system

"SINGLE STRING" GPS USE AFTER developers began envisioning the placement of GPS
CERTIFICATION-_ receivers directly inside INS boxes (an "EGI"). This
provides cost, size and weight savings. Since GPS data
After GPS certification, some orbiters in the fleet will would not have to be sent "outside" a box to be
continue to fly with three TACAN units and one processed, a central Kalman filter could incorporate GPS
MAGR/S, until they receive two more MAGR/S units measurements corrected for S/A effects, as well as other
during regularly scheduled overhauls. Post certification, measurements (barometric altimeter, Doppler radar, radar
single string use of GPS during entry has been proposed altimeter).
as follows:
The "tightly coupled" integration scheme overcomes data
• Those scenarios listed under emergency use of latency issues with previous architectures. Less latency in
uncertified, single string GPS. aiding permits the carrier loops to be aided and speeds up
satellite acquisition. The incorporation of all sensor
• Use as an extra level of redundancy in case of measurements in one filter with high integration rates
navigation sensor (TACAN, Air Data Probe, provides a much more accurate solution than "separate
MLS) and/or ground station failures (TACAN, box" architectures. In addition, ground and flight testing
MLS). has shown that the tightly coupled EGI "blended
solution" has far superior performance under less than 4
• Provide redundancy in the event of ground radar satellite tracking conditions than a stand alone GPS
tracking station failures. receiver [8]. Since an aircraft or missile is subjected to
continuous specific force (due to thrust, lift and drag), the
• Avoid early mission termination due to TACAN Kalman filter can obtain estimates of sensor errors as
failures while on-orbit. maneuvering makes sensor errors visible to the filter
through position and velocity errors. Continuous
• Source of vectors to support emergency deorbit. calibration of the inertial sensors permits an accurate
position, velocity and attitude solution during GPS
• On-board navigation updates on-orbit, when not outages (such as jamming). This also allows
performing translational maneuvers or manufacturers to use lower cost inertial sensors. The
rendezvous. Kalman filter also permits alignment of inertial sensors
much faster than INS systems without GPS [24].
• Pernit a launch in the event of a navigation aid
(TACAN, barometric altimeter, MLS) failure on Three test flights of EGis were conducted to determine if
the pad. such a device could be used in space with a minimum
number of firmware modifications. The first EGI unit to
fly was the Litton LN- IOOG, with a Collins GEM I II five
channel SEM-E form factor GPS receiver. The GEM 111
contained firmware modified for space use, based on

12
GEM III Link 010, produced by Collins to support the A laptop computer was used by the crew to operate the
Army Tactical Missile System program. The Litton EGI SIGI and record data. Changes were made to SIGI
flew on missions STS-81 and STS-86 to provide data firmware based on Flight test results.
gathering for NASA's X-33 and X-34 programs. A
Honeywell H-764G EGI, with the same type of GEM I I I In order to get into a precise orbit the IMUs must be
receiver as was flown in the Litton unit, flew on STS-84. accurately aligned. Stable member HAINS IMUs are
aligned by using the sensed Earth rotation and gravity
In September of 1996, Honeywell was awarded a contract direction. The platform is oriented in various directions so
by NASA for the Space Integrated GPS/INS, or SIGI. that each of the accelerometers can calibrate against the
SIGI is a common NASA navigator, based on the H-764G gravity vector and the gyros can calibrate against Earth
EGI, to be used by both manned and unmanned space rotation. With all of the advantages of a space
vehicles [25]. The Shuttle version of SIGI, intended to missionized EGI, preflight strapdown system alignment
replace the MAGR/S and the HAINS IMU, contained the presents a challenge. There are three opportunities at the
same Collins GEM I I I GPS receiver that was previously Kennedy Space Center to make observations when the
mentioned. vehicle is at different orientations: the Orbiter Processing
Facility, the Vehicle Assembly Building, and the launch
Whereas the MAGR/S was integrated into the Shuttle pad. The difference in orientation at the three locations is
avionics system as a "navigation system," the Shuttle not adequate to get a good separation of the alignment
SIGI was to be'integrated as both a sensor and a variables. During frequent flights of military aircraft the
navigator. The SIGI--blended state vector solution was maneuvering, continuous specific force (engine thrust, lift
processed by the Shuttle PASS and BFS flight software and drag) and processing of GPS measurements permits
using the same software used to process MAGR/S data. the EGI Kalman filter to accurately calibrate the gyros
GPS receiver data and control commands were kept the and accelerometers. For the Shuttle, however, the short
same as MAGR/S. This integration concept was known periods of specific force and maneuvering (ascent and
as "MAGR/S transparency," and minimized the amount entry) coupled with the long time periods between flights
of changes to the Shuttle flight software. (months) makes the reliability of Kalman filter inertial
sensor error estimates questionable.
The Shuttle flight software requires a source of IMU data
(integrated attitude rates, change in accumulated sensed Since the HAINS IMUs are projected to be operational
velocity). These parameters were taken from the inertial through 2010, replacement of the HAINS IMUs and
sensors of the SIGI, which are treated as a sensor. The MAGR/S units by SIGIs has been defer red. Laboratory
original inertial sensor integration concept was "HAINS evaluation of the Shuttle SIGI unit is continuing, along
IMU transparency," an attempt to avoid changes to the with the GPC flight software modified to support it.
Shuttle flight software. The SIGI strapdown inertial
instrument data was to be processed within the SIGI to
look like stable member HAINS IMU data, before being LESSONS LEARNED FROM THE MAGR/S AND
passed to the PASS and BFS GPCs. However, the SIGI TEST FLIGHTS
transparency option would have made it impossible for
Mission Control to identify and take action on suspect Over the last 40 years, the U.S. manned space program
gyros and accelerometers. As a result, the HAINS IMU has had a long and successful history of incorporating
transparency option was abandoned. More changes were "off the shelf' hardware into spacecraft and ground
made to the SIGI firmware to support the Shuttle support systems. Both navigation upgrade projects used
missionization than were expected. COTS products that met the requirements of the original
customers. It was assumed that off the shelf military units
The stable members of the three HAINS IMUs are with proven design and performance would significantly
skewed with respect to each other, so that Mission reduce acquisition costs, and require minimal adaptation
Control personnel can identify suspect gyros and for the Shuttle and minimal testing. The ground and
accelerometers if only two HAINS IMUs are operating. flight test philosophy was that the units "worked" until
With the strapdown configuration of the SIGIs, the axes proven to be broken.
of the inertial instruments in all three SIGI units would be
parallel, making it difficult for Mission Control to However, the time and effort need to test, resolve
identify suspect inertial sensors if only two SIGIs were firmware issues and certify the MAGR/S for TACAN
operating. As a result, plans were made to carry four replacement exceeded initial projections. A number of
SIGIs on the orbiters, rather than three. The SIGI DTO important lessons were learned, which are detailed below:
flew on Shuttle missions 86, 89, 91, 95, 88, 96 and 103.

13
Military and civilian GPS units undergo a less rigorous The trend to use Non-Developmental Item (NDI) avionics
firmware requirements definition, firmware design, containing proprietary software may prevent independent
development and testing process than those used in the validation and verification of firmware. The NASA
Shuttle program [26]. Furthermore, development of Independent Verification And Validation (IV&V)
terrestrial, "non-safety of flight" navigation units entail contractor played a significant and valuable role in the
less detailed analysis of flight data than the Shuttle MAGR/S project, as detailed ill
program. The Shuttle program also processes avionics
units in a more thorough and disciplined manner than Both the MAGR/S and SIGI projects demonstrated the
terrestrial users. This includes strict adherence to work need for a close working- relationship between users and
instructions, maintenance of a parts history, thoroughly vendors. The navigation vendor needs to be involved in
documented investigation of problems and their early decisions on architecture and integration. Frequent
resolution, and more rigorous test requirements. and open communication between technical personnel
should be encouraged. This lesson is best summed up as
It is difficult to leverage off of another program's "communicate early, communicate often." Outside
firmware verification efforts. Firmware development consultants, who do not have a vested interest in the
schedules driven by "time to market" pressures and a choice of a particular unit, should be used. Such
desire to lower overhead costs (a small group of consultants have "hands on experience" with COTS boxes
programmers, short development and test cycles) result in and can be an important information source concerning
a higher probability of code with bugs. Some firmware their design, integration and use.
issues resulted from the use of terrestrial GPS receiver
algorithms at orbital altitude. However, many of the The Shuttle MAGR/S and SIGI projects reaffirmed the
firmware issues that surfaced during the MAGR/S and need for rigorous and thorough flight and ground testing.
SIGI flight tests were due to basic computer science When planning a COTS product missionization and
issues, as opposed to problems arising from use of a integration, adequate time and personnel must be set aside
terrestrial box in an environment for which it was not to analyze flight and ground test data. If data is not
designed (space). Firmware issues that don't manifest in thoroughly analyzed in a timely manner, firmware issues
terrestrial applications due to a flight time of minutes or Will go unnoticed. Performance issues arising late in the
hours can manifest during a much longer space flight. development and certification cycle can negatively impact
Shuttle program ground and flight testing of GPS cost and schedule.
receivers has uncovered many firmware issues that will
aid the maintenance and development of terrestrial GPS COTS projects often do not take into account the
receivers. Deep integration of systems makes them more complexity of software. Most of the focus in costing
vulnerable to software issues. As navigation systems COTS projects has traditionally been on hardware. It is
become more complex and more deeply integrated, easier to "missionize" the hardware of a COTS product
software quality and verification becomes more than the firmware. Experience has shown that firmware
important. The test approach must prove that the box will modifications will often result in more delays and cost
meet requirements, rather than having to prove that the increases than hardware missionization. The one
box is broken. exception to this is radiation hardening of electronics for
the space environment [29].
Some firmware modifications made to the MAGR/S
during the flight test program were the result of issues For a flight critical application (i.e. the box is required to
discovered by military users. The information that was safely conclude the mission), a COTS box will undergo
found to be the most useful was from the vendor, after an more modification than in other applications. The user
investigation of anomalous behavior was complete. will also require more detailed knowledge of navigation
Reports of problems from military users received by the unit design and operation than users of non-flight critical
Shuttle program were difficult to judge in terms of COTS boxes. The Shuttle program considers a box to be
potential impact to the Shuttle application. User reports failed more quickly than a terrestrial user. Engineering
tended to be anecdotal in nature, with little or no and Mission Control personnel must have a thorough
supporting data. Field reports of receiver problems could understanding of receiver operation and data. For
be traced to a variety of causes: user errors, lab set up manned space flight, lack of design insight is a safety
errors, receiver hardware or firmware issues, radio issue. The assumption that state replacement (treating
frequency interference, an antenna problem, or a GPS GPS as a navigation system, rather than a sensor)
satellite problem [27]. eliminated the need for detailed knowledge of the GPS
receiver firmware was found to be false.

14
Lessons learned from using COTS software for space true for critical, "safety of flight" applications such as the
applications can also be found in references [301 (Ariane Space Shuttle. Independent verification and validation of'
5), [31] (Lewis spacecraft) and [32] (Multi-Service firmware is critical, and played an important role in the
Launch System). Shuttle GPS project. The difference in verification and
certification requirements between terrestrial navigation
Perhaps the most important lesson is that modifying a users and the Shuttle program makes it difficult to
terrestrial navigation unit for use on a spacecraft should leverage off of firmware verification efforts of other
be treated as a research and development project. programs. There is a need for frequent and open
Although GPS shows great promise for improving Earth communication between participants, at both the
orbit navigation, the assumption that GPS technology has management and technical levels, throughout a COTS
reached maturity in "all applications" is a common device missionization and test program. Use of COTS
misconception [33, 34]. avionics in mission critical, safety of flight applications
that differ from the original mission for which the unit
was designed require more modification and design
SUMMARY insight than is often anticipated.

Integration of COTS navigation units into the Shuttle

avionics system was performed in a manner that REFERENCES
attempted to minimize cost, schedule, and modifications
to a certified, flight proven navigation system. Proven [1] Van Leeuwen, A., E. Rosen and L. Carrier, "The
terrestrial navigation units that were in mass production Global Positioning System and Its Application In
were selected. The state replacement architecture, which Spacecraft Navigation," Global Positioning System —
treats the GPS receivers as navigation systems, permitted Papers Published In Navigation, Volume 1, The Institute
the flight test of GPS receivers and Shuttle flight software Of Navigation, Washington, D.C., 1980, pages 204 — 221.
without introducing GPS states into baseline entry and [2] Schmidt, Dr. George T., "GPS/INS Technology
orbit navigation. GPS will provide a more accurate Trends For Military Systems," Proceedings of the AIAA
navigation solution than TACAN. Cost and navigation Guidance, Navigation, and Control Conference,
performance considerations have led the Shuttle program AIAA, New Orleans, LA, August 1 1-13, 1997, pages
to retain other entry and rendezvous navigation sensors. 1018-1034.
It is expected that GPS will result in a reduction of Shuttle [3] Kachmar, Peter M., and Lincoln J. Wood, "Space
ground tracking, but it will remain as a backup to GPS. A Navigation Applications," Journal Of The Institute Of
Mission Control based GPS Ground Filter, which filters Navigation, Vol. 42, No. 1, Spring, 1995, pages 187-234.
GPS position and velocity to make up for the lack of [4] Miller, Judy L., "Spacecraft Navigation
space dynamics missionization in the MAGR/S, will Requirements," Aerospace Navigation Systems, AGARD-
provide accurate states for on-orbit maneuver planning. AG-331, NATO, June 1, 1995, pages 390-405.
The Shuttle GPS capability will enhance safety during [5] Hanaway, John F., and Robert W. Moorehead, Space
emergency landings at locations where ground radar Shuttle Avionics System, National Aeronautics and Space
tracking is not available. The first "three string" GPS, no Administration, NASA SP-504, 1989.
TACAN flight will occur no earlier than 2003. Eventual [6]Proceedings of the Space Shuttle Technical
replacement of the HAINS IMU and MAGR/S units with Conference, NASA Conference Publication CP-2342,
an EGI has been deferred. Parts 1 & 2, NASA Johnson Space Center, Houston, TX,
June 28-30, 1983.
It was prudent for the Shuttle program to take advantage [7] Thibodeau, J. R. and S. R. Bauer, "Redundancy
of existing, in production technology for navigation Management Of The Space Shuttle Inertial Measurement
system upgrades. However, changes to both the Units," Proceedings of the AIAA 1982 Guidance And
MAGR/S firmware and Shuttle flight software were more Control Conference, San Diego, CA, August 9-11, 1982.
extensive than anticipated. Modification of aircraft [8] Lewantowicz, Zdzislaw H., Danny W. Keen,
navigation units for space flight should be considered a "Opportunities and Challenges In Avionics Integration,
research and development project, rather than a simple INS/GPS - A Case Study," AGARD Conference
modification of an "off the shelf' box. As digital Proceedings 525, Integrated and Multi-Function
computers have proliferated and become more complex, Navigation, NATO, 1992.
the difficulty of integrating "off the shelf' units has [9] Brown 111, James E., "Ring Laser Gyro/Global
increased. it is not safe to assume that COTS avionics are Positioning System Navigation Improvement Program
simple and inexpensive to integrate just because Plus," Proceedings of the IEEE 1998 Position Location
thousands of units have been built. This is particularly and Navigation Symposium, Palm Springs, CA,

15
April 20-23, 1998, pages 30-37. Vehicle Orbital Rendezvous," Proceedings of the Institute
[ 10] Nielson, John T., "The Untold Story Of The of Navigation GPS-93 Conference, ION, September 22,
CALCM: The Secret GPS Weapon Used In The Gulf 1993, pages 301-312.
War," GPS World, Vol. 6, No. 1, January 1995, p. 26-32. [23] Ewell, James J., "Space Shuttle Entry Through
[I I] Abbott, Anthony S., and Douglas E. Atkinson, "B-2 Landing Navigation," Proceedings of the Position
GPS Aided Munition and Targeting System," Location and Navigation Symposium, IEEE, Atlantic
Proceedings of the ION 1997 National Technical City, NJ, December 6-9, 1982, pages 110-1 17.
Meeting, Long Beach, CA, January 14-16, 1997, pages [24] Tazartes, D. A., and J. G. Mark, "Integration Of GPS
779-789. Receivers Into Existing Inertial Navigation Systems,"
[12] Kachmar, Peter M., William Chu, Peter Neirinckx, Journal Of The Institute Of Navigation, Volume 35,
and Moises Montez, "U.S. Space Shuttle - Integrated GPS Number 1, Spring, 1988, pages 105-119.
Navigation Capability, "Proceedings Of ION GPS-93, [25] Lightsey, E. Glenn, Greg C. Blackburn and James E.
ION, Salt Lake City, UT, Sept. 22-24, 1993, p. 313-326. Simpson, "Going Up: A GPS Receiver Adapts To Space,"
[13] Saunders, P.E., M.C. Robel, M.E. Aerni, D.N. GPS World, Vol. 11, No. 9, Sept., 2000, p. 30 - 34.
Feuerstein, S.B. Lowery and C.E. Cohen, "The First [26] Independent Verification and Validation For Space
Flight Tests Of GPS On The Space Shuttle," Proceedings Shuttle Flight Software, National Academy of Sciences -
of the ION National Technical Meeting, San Diego, National Research Council, Washington, DC, 1992.
Calif., January 24-26, 1994, pp. 657-669. [27] Hardwick, Greg, "False Pull Prevention,"
[ 141 Goodman; John L., and Leonard Kramer, Proceedings of the 26"' Joint Services Data Exchange and
"Scintillation Effects.-On Space Shuttle GPS Data," GPS Users Conference, Oxnard, CA, Oct. 23-26, 2000.
Proceedings of the ION 2001 National Technical [28] Beims, Michael A., and James B. Dabney, "Reliable
Meeting, Long Beach, CA, January 22-24, 2001. Tailored-COTS Via Independent Verification and
[15] Greenspan, R. L., "GPS and inertial Integration," in Validation," NATO Symposium: Commercial Off-The-
Global Positioning System: Theory and Applications]], Shelf Products in Defence Applications, Brussels,
edited by Bradford W. Parkinson and James J. Spilker, Belgium, April 3-7, 2000.
Volume 164 of the Progress in Astronautics and [29] Bates, Jason, "COTS Products Face Technological
Aeronautics Series, AIAA, Washington, D. C., 1996, Hurdles," Space News, Volume 11, Number 35,
pages 187-220. September 18, 2000, pages 12-14.
[16] Carpenter, J. Russell, and Emil R. Schiesser, "The [30] Lions, Professor. J. L., et a], ARIANE 5 Flight 501
Importance Of Semi-Major Axis Knowledge In The Failure, Report by the Inquiry Board, Paris, July 19,
Determination Of Near Circular Orbits," Advances In The 1996.
Astronautical Sciences, Volume 102, Part 2, AAS, San [31 ] Anderson, Christine, et al, Lewis Spacecraft Mission
Diego, CA, 1999, p. 1297-1310. Failure Investigation Board Final Report, Feb. 12, 1998.
[ 17] McHenry, R. L., T. J. Brand, A. D. Long, B. F. [32] Stubbs, Rodney E., Whittak H. Huang and Eric
Cockrell and J. R. Thibodeau, "Space Shuttle Ascent Schmitz, "Integration Of GPS Into A Ballistic Missile
Guidance, Navigation, and Control," The Journal of the Navigation Solution," AIAA-97-3695, Proceedings of the
Astronautical Sciences, Volume XXVII, No. 1, January- AIAA Guidance, Navigation, and Control Conference,
March, 1979, pages 1-38. AIAA, New Orleans, LA, Aug. 1 I-13, 1997, p. 1220-
[18] Smith, F. E., M. E. Campbell, T. J. Blucker, C. E. 1239.
Manry and I. Saulietis, "Shuttle Orbiter Stellar-Inertial [33] Bauer, Frank H., Kate Hartman and E. Glenn
Reference System," Journal Of Guidance And Control, Lightsey, "Spaceborne GPS Current Status and Future
AIAA, Vol. 6, No. 6, Nov.-Dec. 1983, p. 424-431. Visions," Proceedings of ION GPS-98, ION, Nashville,
[191 Wylie, Alan D., and Howard G. deVezin, "Onboard TN, September 15-18, 1998, pages 1493-1508.
Rendezvous Navigation For The Space Shuttle," [34] Rush, John, "Current issues In The Use Of The
Proceedings of the 41" ION Annual Meeting, ION, Global Positioning System Aboard Satellites," ACTA
Annapolis, MD, June 25-27, 1985, pp. 64-72. Astronautica, Volume 47, Nos. 2-9, July-November,
[20] Battin, Richard H., An Introduction To The 2000, pages 377-387.
Mathematics And Methods OfAstrodynamics, Revised
Edition, AIAA, Reston, VA, 1999.
[21] Muller, Eugene S., and Peter M. Kachmar, "A New
Approach To On-Board Navigation," Navigation: Journal
of the Institute of Navigation, Vol. 18, No. 4, Winter
1971-1972, pages 369-385.
[22] Zyla, Dr. Lubomyr V., and Moises N. Montez, "Use
of Two GPS Receivers In Order To Perform Space

16
 --- - - - -

Space Shuttle Navigation In The GPS Era

John L. Goodman
United Space Alliance

Supporting

Flight Design And Dynamics Division

Mission Operations Directorate
NASA Johnson Space Center
Houston, Texas

John L. Goodman Institute Of Navigation

January 22-24, 2001 National Techn ical Meeting
Page 1 of 15
Unite d Spa ce Allia n ce

- -- ----- ----~ ---- .. -- - .- ---- - -- - -

r-- ' - -
--- -

Agenda

• Space Shuttle Navigation

• GPS Integration Architecture

• Ascent

• Orbit and Rendezvous

• Entry

• Lessons Learned

• Navigation Improvements Due To GPS

John L. Goodman Institute Of Navigation

January 22-24, 2001 National Technical Meeting
Page 2 of 15
U nited Space Alli...9rJCf)

~----- --- --~ -- - -- ---- - - --- ------

 _. .,.....-------.. - - - - - _.. - ---- - - - - - - -- ----- - - ----- - - -"--

Redundancy Emphasized In Shuttle Design

• During ascent/entry, 4 primary computers and one backup.

Primaries and backup run different software.

• Pre-GPS Ascent/Entry navigation sensors:

Ascent
- Three IMUs
- Ground C & S Band Radar

Entry And Landing

- Three IMUs - Ground C & S Band Radar
- Three TACANs - At least two TACAN ground
- Four Barometric Altimeters stations.
- Three MLS Units - At least two MLS ground
- Two Radar Altimeters stations.
John L_Goodman Institute Of Navigation
Janua ry 22 -24. 2001 National Technical Meeting
Page 3 of 15
Unit ed Space Alliance

L-~ _ _ _ __ _ _ ~ ___ - - - - - - - - - - - - - - -- ----- -- ---~ ------ --- - - --

 Navigation Sensors For Orbit, Rendezvous & Docking

- Three IMUs

- Two Star Trackers

- One Ku Band Rendezvous Radar

- One Crew Optical Alignment Sight (COAS)

- Trajectory Control Sensor (TCS, a laser)

- Two Hand Held Lasers (HHL)

- Two Payload Bay Television Cameras

With Ranging Ruler Overlays
John L. Goodman Institute Of Navigation
January 22-24, 2001 National Technica l Meeting
Page 4 of 15
Unite d Spa ce Alli..9n ce

~----
----- -- - -- -- -- -
( - - _. - ",...---- - - - - - - - -
I
I .
Redundancy Emphasized In Shuttle Design

• Processing of redundant measurements from IMUs,

TACANs, Baros and MLSs:
- Fault Detection and Exclusion algorithms.
- Selection filters process data from units deemed to be "good. "
- Selected measurements are then passed to the Shuttle
navigation software.

• Ground radar (ascent, orbit, entry) and Tracking and Data

Relay Satellite data (orbit) processed by Mission Control to
independently determine Shuttle position & velocity.

John L. Goodman Institute Of Navigation

January 22-24, 2001 Nationa l Technical Meeting
Page50f 15
United SpLIce Allia n ce

- - -- - - ----------- --- - - ._ . _ - - -- - - - - -
Space Shuttle Program Is Certifying GPS To Replace TACAN

• A keyed, 5 channel Collins Miniaturized

Airborne GPS Receiver was chosen as a
Commercial Off The Shelf (COTS) item.

• "Single string " MAGR test flights began

with STS-79 in September of 1996.

• Once the MAGR is certified for T ACAN

replacement, 3 MAGRs will replace 3
TACANs on each orbiter. First flight
without TACAN expected no earlier
than 2003.

John L. Goodman Institute Of Navigation

January 22-24, 2001 National Technical Meeting
Page 6 of 15
United Spnce Alliance
 - - - - -- - - - -- --- - -- -----
l -- -- - - --~- -

- - - - - --

State Replacement Architecture Chosen

• MAGR integrated as a navigation system, not as a sensor.

• Allows data collection during flight tests from both the MAGR
and Shuttle flight software while still using TACAN.

• Minimized changes to the proven, navigation flight software.

• Three configurations supported by the same version of Shuttle

flight software: 3 TACANs and no GPS units, 3 TACANs
and 1 GPS unit, 3 GPS units and no TACANs.

• GPS vectors subjected to quality assurance checks & a

selection filter.

John L. Goodman Institute Of Navigation

January 22-24, 2001 Nati onal Technical Meeting
Pag e 7 of 15
Unite d S pace Alliance

~ - ------
.. -- - - -- -.- -- ------ -- -- - - - -- ' -
 -- I

GPS Not Used For Ascent

• IMUs aligned and calibrated before liftoff, used as a source

'of accumulated sensed velocity and attitude data.

• Navigation aids (GPS, TACAN, Baro, MLS, radar altimeter)

that would be used for an emergency landing are powered
on prior to launch.

• No on-board sensor measurements are processed by a

Kalman filter during powered flight.

• Ground radar tracking not required by Mission Control,

is but highly desirable.

John L. Goodman Institute Of Navigation

January 22-24, 2001 National Technical Meeting
Page 8 of 15
Unite d Space Allisnce
 Limited Use Of GPS For Orbital Operations

• On-orbit maneuver planning currently supported by ground

--based C -Band radar tracking and S Band Doppler tracking
using the Tracking and Data Relay Satellites (TORS).

• MAGR firmware has not been modified to meet accuracy

requirements to support orbit maneuver planning (TACAN
replacement only).

• Mission control based ground filter in development to

process MAGR states to support maneuver planning.

• Mission Control personnel will maintain ground radar and

TORS data processing skills.

John L_Goodman
January 22-24, 2001
Institute Of Navigation
National Technical Meeting
Pag e 9 of 15
SA
U nit ed Space Allis nce

------ --_ .. --
 GPS Will Not Be Used For Rendezvous

+Vbar -50 - 100 - 150 -200 -250

Star Tracker Last Ground ~

~ Pass If Radar Targeted Burn
Manual Piloting 10 .....................................1 Fail
Phase
• Payload Bay
5 More
Lasers Star
On-Board
• Hand Held Lasers 20 . Tracker
Targeted
• TV Overlays Pass~
Burns

30

40

+Rbar

John L. Goodman Institute Of Navigation

January 22-24,2001 National Technical Meeting
Page 10 of 15
Unite d SpLIc e Allia n ce

--_. -. __. _ - - - -
 - -I

Most Entry Navigation Sensors Will Be Retained

4300r 400 r 25 r
(
+ ---- GPS Era Navigation -----+~ :
1800 240 22 I I
I
I

if): 1200
Q)
220 18 Pre GPS Navigation ------+) :
I

~ 800 200 14
C'Cl 500 I
0 .-... 180 c-

+-'
+-'
Q) '- 10
I

~
Lock
::J Q)
C'Cl 350 Q) 160 On
LL ..0
Z 0
I
E
'--"
200 := 140 ::J 6
>- ~ Z
C'Cl '--"
5c 150 Q) 120 ...c
0
::J ""0
C'Cl 4 Probe Deployment
er:: 100 .2 100 ~
+-' & Calibration
0
f- 60 « 80 2
Q)
0)
c 30 60 Mach Jump Region
C'Cl
er:: 20 40

10 20
" )

-
......;:
(
0- o - ~
0

GPS IMU Baro MLS Radar

Ground TACAN
Derived Altimeter Radar
Drag Tracking
Data Available For Navigation ~ Data Used If No MLS D Not Used
John L. Goodman Insti tute Of Navigation
January 22-24, 2001 National Technical Meeting
Pag e 11 of 15
Unit e d S pace Alliance

- ~ - - .- - ._- - - -
 "Easy COTS Integration" Was An Unrealistic Assu mption
\

• Scope of modifications and testing requir!ed to certify MAGR

as a TACAN replacement significantly exceeded expectations.

• As digital computers have proliferated and become more

complex, the difficulty of integrating COTS units has increased.

• Integrators tend to focus more on hardware issues, rather than

software.

• For a flight critical application, high quality software is essential.

John L. Goodman Institute Of Navigation

January 22-24, 2001 National Technical Meeting
Page 12of15
U nit ed Space Alliance

~' ----- - .. - . -
--- --- -- --- ----- -- -- - --- -1

I
I

"Communicate Early, Communicate Often"

• Close relationship with the vendor is required. Vendor must

be involved in integration architecture decision.

• The NASA Independent Verification And Validation (IV&V)

contractor played a significant and valuable role in the
project.

• In flight critical applications that differ significantly from the

original application that a COTS product was designed for,
design insight is essential.

John L. Goodman
January 22-24,2001
Institute Of Navigation
National Technical Meeting
Page 13 0f 15
USA
U nited Space Allia nce

-~, ---- - ---- -~

-~ - ---- - --- - -----
- ------ ------ -- - - - - - ------- ----- --.. -' - - - - - -

The Most Important Lessons

• Modifying a terrestrial navigation unit for use on a

spacecraft should be treated as a resear~h and
development project.

• Assumption that GPS technology has reached maturity

in "all applications" is a common misconception.

John L. Goodman Institute Of Navigation

January 22-24 , 2001 National Technical Meeting
Page14of15
Unite d SpBce Allia n ce

- -- - ~----.--~ - _.---_.- -- -----_ .. ----~-- - --- - -------

 GPS Will Enhance Shuttle Navigation

• Provides more accurate entry navigation than T ACAN.

• Improves safety for emergency landing sites where

ground radar tracking is not available.

• Near continuous availability of ground filtered GPS

states, in conjunction with ground tracking, allows
more responsive mission planning.

• On-orbit, GPS is a near continuous source of state

vectors for emergency deorbit.

John L. Goodma n Institute Of Naviga tion

January 22-24 , 2001 National Technical Meeting
Pag e 15 0 f15
Unite d Space A llia n ce

~- ... - ------ -- -- -.--