Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
American Institute of Aeronautics and Astronautics
1
Large Reflector Uplink Arraying
G. Patrick Martin1 and Kathy Minear
2
Harris Corporation, Melbourne, Florida, 32902-0037
Barry J. Geldzahler3
NASA HQ, Washington DC, Mail Suite 7K80, 300 E Street SW, Washington, DC, 20546-001
and
Jason Soloff4
NASA Lyndon B. Johnson Space Center, 2101 NASA Parkway, Mail Stop ZF, Houston, TX 77058
Abstract -- The promise of array technology in support of space operations has long been
appreciated, and receiving array technology is now an important operational asset. Notable
examples include the NRAO (National Radio Astronomy Observatory) array of twenty seven
25m reflectors and the DSN (Deep Space Network) ad hoc arraying of various assets,
particularly 34m antennas , to realize significant G/T increases providing needed bandwidth
and range extensions. However, uplink (transmit) arraying has not kept pace due to the
difficulty of ensuring „open loop‟ beam formation under the conditions of wide spacing, due
to 1) transmission line and circuit variability, 2) precise antenna reference point
determination, and 3) tropospheric effects especially at higher frequencies. Notable success
in solving uplink array calibration issues has been achieved by two different groups at JPL,
one arraying five 1.2m reflectors and the other three-34m reflectors. However, there are
operational issues with each of these approaches. We present an approach for mitigating
these difficulties, offering the potential for continual readiness operationally and
extensibility to Ka band. Additionally the approach is suitable for use on a wide range of
antenna sizes, including both 34m and 12m reflectors. Currently a transmit uplink
experiment is underway at Harris Corporation using three-12m reflectors operating at
DSCS X band. This array architecture provides continuous internal self-calibration using
the transmit signal itself, a method to dynamically solve for the antenna reference points,
and mitigation of propagation effects by using received signals from known sources.
Nomenclature
ARP = Antenna Reference Point, usually the intersection of azimuth and elevation axes
A2D = Analog to Digital Conversion
COTS = Consumer Off the Shelf
DISA = Defense Information Systems Agency
DSCS = Defense Satellite Communications System
DSN = Deep Space Network
EIRP = Effective Isotropic Radiated Power
FO = Fiber Optic
FPGA = Field Programmable Gate Array
GRASP = General Reflector Antenna Software Package, from TICRA Corporation
G/T = Gain divided by Temperature, usually expressed in dB
1 Senior Scientist, R&D Technology, Harris Corp., P.O. Box 37, Melbourne, FL 32902 M/S 19-21T, Member.
2 Senior Mathematician, R&D Technology, Harris Corp., P.O. Box 37, Melbourne, FL 32902 M/S 19-21T R&D
Technology, Harris Corp., P.O. Box 37, Melbourne, FL 32902 M/S 19-11B, Member. 3 Chief Scientist, Space Communications and Navigation Division, NASA HQ, Washington DC, Mail Suite 7K80,
300 E. Street SW, Washington, DC, 20546-001, Member. 4 Avionics & Communications Office Lead, Constellation Program, NASA Lyndon B. Johnson Space Center, 2101
NASA Parkway, Mail Stop ZF, Houston, TX 77058, Member (exp.)
SpaceOps 2010 Conference<br> <b><i>Delivering on the Dream</b></i><br> <i>Hosted by NASA Mars25 - 30 April 2010, Huntsville, Alabama
AIAA 2010-2175
Copyright © 2010 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
American Institute of Aeronautics and Astronautics
2
GUI = Graphical User Interface
HW = Hardware
JPL = Jet Propulsion Laboratory, Pasadena, California
LED = Light Emitting Diode
Rx = Receive
SAR = Synthetic Aperture Radar
STK = Satellite Toolkit, a precision orbit prediction application from Analytical Graphics Inc.
TLE = Two Line Element set, a standard method for orbit description
Tx = Transmit
I. Introduction
ASA is moving forward in space communication capabilities on three fronts: upgrades to ground systems to
incorporate arraying technology, operations at Ka band, and optical communication. The International
Telecommunications Union allocation at X-band for both near Earth and deep space missions is a total of 50 MHz.
Hence, each deep space mission is fortunate to get some 5-7 MHz in spectrum allocation and generally 10 MHz for
near Earth missions although 15 MHz is granted on rare occasion. This in turn limits maximum data rates.
Alternatively, Ka band offers higher data rates via greater spectrum availability and substantial link advantages.
Additionally, substantially higher accuracy spacecraft navigation and tracking can be realized via Ka band Doppler
precision increase and medium scattering error reduction.
Downlink arraying technology is employed today, and for the foreseeable future, on the Deep Space Network
(DSN) in a limited way: relatively few antennas are phased together to form a larger synthesized aperture. One of
the questions NASA has wrestled with is uplink capabilities in the array era. There are two main possibilities: use of
single dishes with higher power transmitters or coherent uplink arraying. The aggregate EIRP of an N -element
array of identical antennas is proportional to 2N wherein a few antennas with moderate power transmitters can
achieve a substantially stronger EIRP than a single dish can alone. JPL has demonstrated this potential both in an X
band uplink using three 34m dishes to the EPOXI spacecraft in 2008 and at Ku band with five 1.2m lower power
reflector antennas to a geosynchronous communications satellite. Although impressive accomplishments, each of
those arraying methods require calibration procedures that are not conducive to operational application and they do
not extend readily to Ka band.
This paper is concerned with operationally feasible uplink arraying and extensibility to Ka band. We present an
alternative approach having the potential for immediate readiness operationally, potential for extension to Ka band,
and suitability for use on a wide range of reflector sizes, including both 34m and 12m.
On the basis of our modeling and simulation results which were first presented at the AIAA 2006, an
experimental demonstration of our method is now in progress using three 12m reflectors operating at X band. This
paper begins with a summary of the three principal challenges, describes the experimental set-up and algorithms,
discusses methods of mitigating the challenges, and discusses future work including extensibility of these methods
to a larger array. Our method offers[1]
:
Continuous calibration using the transmitted signal itself
Instant availability without the need for calibration repointing
Potential for direct extension to Ka band
Mitigation of ARP errors and tropospheric propagation variation
Tolerance for COTS hardware and relaxed circuit stability requirements
Applicability to a wide range of antenna and array sizes Important experimentally measured results are presented in Section 4A, where differential closed-loop transmit
circuit phase was maintained within 8 degrees over 83 hours despite circuit phase variation of hundreds of degrees
due to environmental conditions. These results, obtained with error detection assemblies ready for installation on
the antennas, are encouraging since they indicate that similar performance will be obtained during imminent uplink
testing.
Control algorithms discussed include open loop continuous model-based transmit, continuous Blind Signal
Sorting optimum aperture combining for receive, a novel closed-loop continuous transmit algorithm to mitigate
propagation effects (demonstrated on a small prototype array in 2008), and automatic continuous self calibration of
transmit and receive circuit paths using forward and reverse traveling wave properties.
N
American Institute of Aeronautics and Astronautics
3
II. Uplink Arraying Considerations
A. Three Principal Error Contributors
Principal error contributors in arrays of large reflector antennas are:
1. Circuitry variation between the point where a signal is phased for transmission and the location from which
that signal is radiated
2. Imprecise knowledge of the ARP (Antenna Reference Point) and relationship of a particular antenna‟s phase
center to the ARP so that proper phasing can be calculated and generated
3. Uncompensated differential propagation phase variation (tropospheric effects)
Circuitry variation over time and temperature (including transmission lines, frequency converters, filters, and
power amplifiers) is the dominant error in beam formation, and distributing precisely phased signals to the antenna
elements becomes increasingly difficult as separation and frequency increase. For example, suppose a Ka Band
frequency is to be transmitted and that array elements are located within a diameter of hundreds of meters, then the
distribution network alone will span tens of thousands of wavelengths. Effective arraying requires that circuit phase
error contribution to the total error budget be held to no more than about 10° peak, implying tolerable variation less
than a few parts per million. Typical circuitry includes not just the fiber optic transmission lines, but up-converters,
filters, FO transmitters and receivers, and power amplifiers. All of these components are temperature sensitive and
time variable.
While it may be extremely difficult to accurately determine and continually update the positions of widely
spaced antenna elements in general, in the case of large reflector antennas in fixed installations such determination
can usually be achieved with very high precision; consider for instance the NRAO VLA or multiple 34m BWG
reflectors at Goldstone. Basically, the mechanical accuracy and stability required for achieving high gain and
precise pointing in turn ensures a very stable ARP. Consequently, once determined using a variety of means, such
as laser surveying, interferometry of known stellar objects or spacecraft, the ARP tends to be constant barring
significant geological disturbance such as an earthquake.
The last major error, Uncompensated Tropospheric Propagation, is almost negligible at X-Band except at low
elevation where it can be moderate, but, this error can be very large for Ka band at any elevation. Algorithms for
mitigating this error using known emitters are discussed in Section IV.
B. Alternate Uplink Arraying Methods
Uplink arraying methods have been demonstrated by two different groups at JPL, one using two and three 34m
antennas at Goldstone5 and another using five 1.2m antennas
6. Both of these methods were successful in a
succession of experiments.
Vilnrotter, et al5, recently arrayed three 34m BWG antennas at X-Band. These antennas have circuitry that is
compensated locally or environmentally controlled, and great effort has been applied to make these antennas the best
worldwide. Despite this attention, residual amplitude and phase errors remain which must be sensed and calibrated.
Due to far field considerations and lack of a suitable in-orbit transponder, a lunar bounce was used to derive the final
corrections. As a tribute to the experiment concept, quality and equipment stability, the calibration solution obtained
was applied to transmit to a distant spacecraft for hours, even though the reflectors had to track the target and
continuous background Doppler corrections were necessary. This experiment applied intensive SAR (synthetic
aperture radar) processing to focus on a specific small size lunar target in the calibration process to prevent
averaging of range-dependent coefficients seen differently by each antenna.
Principal disadvantages of this approach include the need to point away toward the moon to obtain calibration,
dependence on circuit stability between recalibration events, lack of availability when calibrating, and the 12 hour
lunar visibility. Additionally, this method does not extend readily to Ka Band due to tropospheric propagation
considerations.
5 V. Vilnrotter, D. Lee, R. Mukai, T. Cornish, and P. Tsao, “Three-Antenna Doppler-Delay Imaging of the Crater
Tycho for Uplink Array Calibration Applications,” The Interplanetary Network Progress Report, vol. 42-169, Jet
Propulsion Laboratory, Pasadena, California, pp. 1–17, May 15, 2007.
http://ipnpr.jpl.nasa.gov/progress_report/42-169/169D.pdf 6 Larry D'Addario, Robert Proctor, Joseph Trinh, Elliott Sigman, and Clifford Yamamoto, “Uplink Array
Demonstration With Ground-Based Calibration”, Interplanetary Network Progress Report, vol 42-176, February
15, 2009: http://ipnpr.jpl.nasa.gov.
American Institute of Aeronautics and Astronautics
4
D'Addario, et al.6, explored the use of five relatively small 1.2m apertures, using low cost commercial
transponder components and an advanced time-transfer method to obtain circuitry synchronization. This method
also must address an unknown phase coefficient that includes filters, antenna feed and internal reflector variations.
Calibration was achieved by providing multiple external sensing points, mostly towers, which enabled simultaneous
solution for the unknowns. D'Addario has noted that with good knowledge of the array elements and sensor position,
only one tower is sufficient. Given small elements, a feasibly tall tower in the center of the array is adequate to
satisfy relaxed far-field constraints. This experiment transmitted to commercial Ku Band satellites, and was entirely
successful, maintaining good beams continuously for many days. Such impressive performance is a tribute to the
concept, architecture and circuit design.
Disadvantages of this approach include the need for a high, stable calibration tower, the need to point away, at
least occasionally, to the calibration tower, and potential internal reflector variations between calibrations due to
temperature differentials, especially with larger reflector sizes. Required calibration tower height increases as the
square of element diameter (to satisfy near field constraints), with a 6m diameter element probably close to the
upper practical limit. Array cost increases quickly for aperture diameter less than about 10m, since the number of
elements required to achieve a particular G/T becomes very large (overcoming potentially low cost per element).
Our models predict that cost of an array of 6m reflectors would be about 58% greater than for an array of 12m
reflectors with the same G/T, and an array of 4m reflectors would be almost 300% more expensive. This cost
increase is almost entirely due to low noise amplification and electronics.
The possibility of employing a geosynchronous satellite for calibration has been noted, eliminating the need for a
stable high tower and removing the small reflector size limit. In the case of the DSN, at least three of these
calibration satellites would be required worldwide. Another problem arises for Ka Band frequencies; due to spatially
dependent tropospheric propagation, as such calibration will include two way propagation variations for a particular
spatial direction that is likely not the direction required. Since propagation phase can change rapidly7, even if spatial
directions were similar, two-way latency would likely invalidate the calibration. Ka band tropospheric compensation
is also an issue with lunar calibration methods for the same reasons.
C. Harris‟ Uplink Arraying Approach
Previously discussed alternate methods are all characterized by a residual unknown phase at each element in the
array which must be determined using a known external calibration target or tower. In contrast, Harris‟ approach
applies closed loop circuit phase control to all transmit pathway circuitry, including transmission lines, frequency
converters, filters, power amplifier, feed with polarizer and diplexer, and the antenna itself. By closing a loop on the
entire pathway, there is no unknown phase coefficient. Consequently, the array is always ready for transmit at the
carrier frequency, does not require calibration periods or pointing away to a calibration target.
Time delay adjustment can then be realized at baseband (zero frequency) with no impact on carrier phase shift.
In a customary system, time delay accuracy must satisfy both carrier phase precision and information content
alignment. With this approach, only information content alignment is required. Since carrier phase is unaffected,
coarse control to nanoseconds or hundreds of picoseconds conveniently realized through digital processing is
sufficient versus analog femtosecond precision required for traditional methods at Ka Band.
All Receive pathways are also placed under closed-loop control in order to support precision interferometry,
precise range and range-rate measurements, precision ARP refinement and tropospheric propagation error mitigation
using known location angularly nearby sources (Discussed in Section IV).
7 Nessel, James, A.; Acosta, Roberto, J.; Morabito, David, D.,” Phase Fluctuations at Goldstone Derived From One-
Year Site Testing Interferometer Data”, National Aeronautics and Space Administration, John H. Glenn Research
Center at Lewis Field, Cleveland, Ohio 44135-3191,
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20090020407_2009019738.pdf
American Institute of Aeronautics and Astronautics
5
Figure 1 conceptually illustrates this method.
Both receive and transmit pathways are
controlled, but we will focus on the right hand
side transmit pathway (shown in red). Assume
that a means is available for generating a precise
phase reference at widely separated physical
locations. This can be accomplished customarily
with two-way time transfer or by Harris‟ novel
phase transfer method.
At a point including as much circuitry as
feasible, a sample of the signal being transmitted
is compared with the remote phase reference.
Any phase deviation can be attributed to circuit
variability. A weighting device in the transmit
pathway applies the measured correction, forcing
the entire pathway to a net zero phase shift
between generation point and transmission.
Figure 2 illustrates how the feedback point can
be located on the reflector surface, thus including
all temperature sensitive feed components and
internal reflector variation. This is the last
accessible physical point before the wave leaves
the reflector. As the reflector and its components
expand and contracts due to temperature changes,
the circuit path length change is sensed and
corrected.
In this way, all unknown circuit phase errors
can be sensed and controlled, making the antenna
ready for instantaneous, blindpointing transmit
operation.
In addition to negligible latency, immediate
availability and no point-away requirement, this
method is potentially low in cost since ordinary
COTS hardware can be substituted for precision,
temperature stabilized components. Transmission
lines need not be environmentally controlled.
Since time delay compensation is accomplished at
baseband, only approximate delays are needed to
satisfy time*bandwidth constraints (RF transmit
phase is unaffected). This approach applies to arrays of any size and is particularly well suited for array expansion.
Since each array element is independently ready for service at any time, partitioning the array into subapertures
(such as 34m equivalents) is straightforward.
III. Experimental Setup
Based on modeling and simulation results for the Harris approach in 2006, an experimental effort was
undertaken to validate and demonstrate the concept. The following paragraphs describe the experimental setup.
A. Antennas
Three 12m Cassegrain reflectors manufactured by Patriot Corporation8 were installed on a deliberately scalene
lattice, each side having more than 60m length. Aside from electrical power, all other interconnection is via optical
fiber. (This includes RF transmit, RF receive, transmit circuit control, receive circuit control, Antenna Control Unit
interface, and hardware status.) No effort was made to protect the fiber from temperature or mechanical variation.
A photograph of the array appears in Figure 3.
8 Patriot was acquired by Cobham. http://www.cobham.com/about-cobham/avionics-and-surveillance/about-
us/satcom/albion/products/large-aperture-antennas/120m-earth-station-antenna-system.aspx
Figure 1. Closed Loop Circuit Compensation
F074N-010ppt (0) 051209F074N-010ppt (0) 051209
Figure 2. Circuit Feedback Sensor on Reflector Surface
American Institute of Aeronautics and Astronautics
6
Figure 3. Array of Three 12m Reflector Antennas on a Scalene Lattice
B. Wave Phase Sensors
Figure 4 is a photograph showing a
wave phase sensing probe installed on a
reflector. Transmit signals are sampled
and compared with a precision
reference, enabling closed loop control
of the entire transmission path. This
process ensures that the transmitted
signal has exactly the desired
beamsteering phase applied remotely
by the weighting hardware at the Ops
Center. Error measurement assemblies
fed by the sensor are located
immediately behind the reflector
surface. These assemblies also receive
a precision reference signal via optical
fiber.
C. Signal Processing Facility
All signals and controls are routed
to a nearby Ops Center, containing
signal generators, FPGA based digital weighting and combining at baseband, up and down converters, a DSCS
modem, and the computers required to execute the arraying and control processes as well as reflector pointing and
tracking. All of the algorithms and control reside in Harris‟s ArrayLab, a MATLAB language modeling and
simulation tool with real-time interface and control capability, described in greater detail in Section IV. Figure 5
shows some of the Ops Center equipment.
Field SensorField Sensor
Figure 4. Phase Error Sensor Installed on the Reflector
American Institute of Aeronautics and Astronautics
7
Figure 5. Op Center Panoramic View; hardware racks, control, status, and algorithm monitoring displays.
D. Satellite Targets
Arrangements were made through DISA to transmit and receive through geosynchronous X-Band DSCS
satellites. Several satellites are visible from the array, including one at about 10° elevation. Instant availability of
the arraying method can be tested by rapidly moving among the several satellites. The low elevation satellite
provides an opportunity to experience modest propagation variation, allowing a test of the propagation
compensation algorithms.
Figure 6 is a STK (Satellite Toolkit) depiction of the experiment. Using TLE‟s from DISA, STK generates real-
time pointing directions, range and range rate, for each aperture during an experiment whenever requested by
ArrayLab. (ArrayLab is described shortly in Section II-F.)
Figure 6. Satellite Toolkit Produces Pointing, Range and Range Rate Using DISA TLE‟s
Array geometry must be known to support beamforming calculations. Each reflector comprising the array is
described by a table of parameters beginning with its ARP latitude, longitude and elevation. These coordinates are
determined initially with surveying methods using a laser rangefinder and theodolite establishing location relative to
a nominal surveyed site reference. Relative coordinates are determined nominally to within a fraction of a
millimeter. Using calibrated receive capability, observation of known emitters allows adaptation of the geometry
model through solution for each ARP, resulting in refinement of the initial location. Interferometry (Angle of
Arrival) is also useful in this context as is the “Instant Return” process for mitigation of tropospheric variation,
which also affects ARP refinement.
Reflector geometry table parameters also include values for typical errors, including gravity distortions, azimuth
axis tilt, and RF axis offset from the mechanical elevation axis. It is expected that these array geometry parameters
are highly stable, but they are to be continually refined and tracked.
Array geometry, reflector orientation, imperfections, etc., are parametric values in an overall model of the array.
Collectively, these parameters in association with the model establish a basis for determining all beamforming
weight settings, thus a model based form of calibration [2]
. By sensing and refining these parameters continually, the
model is adapted to present conditions.
American Institute of Aeronautics and Astronautics
8
E. Algorithm-Centric vs. Hardware-Centric System
The transmit adaptive combining experiment (Tx ACE) prototype system is algorithm-centric in contrast to
state-machines where the HW runs the system. Advanced algorithms provide the modeling, signal processing, and
calibration needed to run each phase of the experiments. The algorithms also control the HW accelerators which are
used for quick multiplies and parameter settings. There are 20+ algorithms for this experiment that use combinations
of the four signal paths: Rx, Tx, Receive circuit calibration, and Transmit circuit calibration.
Figure 7 is a flow diagram showing relationship of the algorithms (tan rectangular boxes) with the hardware.
Note that the error detection assemblies (blue boxes) are physically on the antenna mounted near the back side of the
surface. The hardware accelerator cards are in the upper left yellow box. The two square boxes under the
assemblies are physically located in the pedestals of the antenna and are for signal throughput.
Figure 7. Functional Set-up for Model-Based Adaptive Combining for Tx
F. Harris‟ ArrayLab: Comprehensive Array “Simulation to Prototype” Laboratory
ArrayLab (Array Laboratory) is Harris‟ in-house comprehensive array “simulation to prototype” software suite.
It combines specific mission scenarios with custom communication systems and with advanced proprietary
algorithms providing realistic system-level analysis. It was used on a 2-year IR&D to design and model the
performance of phased array systems consisting of widely spaced dishes; and in particular to model and simulate
solutions to specific operational and mission problems. Analysis from these studies was used to quantify system and
component-level capabilities and limitations .in order to create error budgets for system subcomponents. In
hardware prototype mode, ArrayLab‟s advanced signal processing algorithms and hardware control modules now
run the large reflector transmit uplink array. In both simulation and prototype modes, each element in an array is
modeled with 6 degrees of freedom. The element patterns are either inputs from COTS electromagnetic field
analysis software or analytic formulations. For model-based uplink arraying, the directivity pattern is (particularly in
the main beam) adequately estimated using a Bessel function model with input parameter: frequency, diameter, feed
offset, efficiency, edge taper, and angle off boresight. For spillover analysis the directivity was numerically
calculated in GRASP and the output file was ingested by ArrayLab. It provides 6 degrees of freedom element
configuration and element pattern modeling accepting both analytic patterns as well as electromagnetic software
outputs. Live simulations of the Transmit uplink arraying methods presented in this paper using up to 400, 12m
antennas were first presented for NASA and JPL guests at the AIAA Conference in 2006.
Though not intended for operational usage, the 40K+ lines of code in this baseline are archived and versioned-
controlled using Clearcase (Commercial software providing complete development software configuration
management through version control). It is comprised of a modular baseline containing 8 main modules: array
modeling, signal generating, blind signal sorting, advanced proprietary signal processing, error analysis, link budget,
American Institute of Aeronautics and Astronautics
9
custom analysis, and utilities. The modules were designed for ease of transition from simulations to control of
hardware prototype systems.
The top path in Figure 8 depicts ArrayLab used in simulation or design mode while the bottom path depicts
hardware prototype mode. Both paths are driven by the ArrayLab Baseline. Some proprietary algorithms contained
in this baseline and used in this experiment include widely-spaced continuous optimum model-based adaptive
aperture combining on Tx, widely-spaced continuous optimum blind signal sorting-based aperture combining on Rx,
instant return, and advanced calibration methods such as antenna reference point refinement. The algorithms are
written for both a bit-by-bit signal-level as well as a quick-look or expected value analysis. Custom interfaces
provide the user with control over algorithm and array parameters. In prototype mode additional interfaces provide
hardware parameter control and status. During the simulation phase statistics of various errors sources were varied
using ArrayLab‟s Error Tool. These included compensated and uncompensated circuit error, phase reference point
errors and atmospheric effects.
Figure 8. ArrayLab: Simulation to SW Prototype Algorithm & Control Suite.
It can be noted in Figure 8 that for both simulation and hardware prototype modes, ArrayLab controls AGI‟s
COTs software, Satellite Toolkit (STK). In prototype mode, the target line-of-sight, range, range rate are provided
in real-time. STK provides interactive 4D visualization of the planetary and communications platform motion while
ArrayLab provides instantaneous communications performance analysis through plots, graphs, and raw data.
The bottom path in Figure 8 depicts ArrayLab used in real-time hardware prototype mode. The hardware control
for this experiment includes the racks of equipment in both the operations center and the pedestals, the error
detection assemblies located behind each probe sensor, and the three 12m Patriot antennas. Real signals are
substituted for simulated ones via signal generators and antenna elements. While the antennas are substituted for the
simulated array, the array modeling component remains and is used to calculate an initial estimate of the relative
beam steer weights. Statistical error modeling is replaced with circuit phase errors due to the up/down converters,
phase discriminators, line receivers, fiber optic lines, and other hardware as well as tropospheric effects.
Before beginning the prototype study, custom analyses for this experiment were conducted. Object obscuration
analysis (including terrain, buildings, and trees) was performed on 4 potential antenna sites. Element shadowing was
also analyzed. Various other simulations were also performed including: spillover effects on array G/T, G/T
degradation due to planetary thermal emissions on pattern sidelobes, the ability of the Patriot antennas to track
various targets (slew rate analysis) for various array locations on the Earth, near-field radiation effects on over-head
low-flying aircraft, and determination of how the arrayed beam forms in the array near field region.
American Institute of Aeronautics and Astronautics
10
Other signal processing factors considered include Doppler effects due to planetary rotation motion and wide
reflector element separation, especially at Ka-band, and speed-of-light adjustment as a function of atmospheric
pressure and temperature.
G. Graphical User Interfaces & Controls
ArrayLab has 3 main interfaces for this program as shown in Figures 9 and 10 allowing the experimenters to
control the algorithms to be run, input parameters, which targets to communicate with, test, monitor, and control all
the hardware in the system. Real-time output data is captured and displayed during the tests as well as STK target
information. The first keeps track of the algorithms being run and the real-time STK updates from each of the
targets. New targets can be added and orbit parameters can be updated as new TLE‟s are available. Currently this
occurs once a week. The second interface is dedicated to circuit calibration. Calibration of the circuit error
assemblies themselves are also controlled and monitored here.
Figure 9. ArrayLab Graphical User Interfaces for the Tx ACE Experiment. Top Left: Main Interface
controls algorithms and STK real-time connection. Top Right: Circuit calibration and monitoring.
The third GUI (See Figure 10) is for all the HW
interfaces including the three antenna control units, up and
down converters, modem, electro-optical modules,
attenuator, reference signal generator, Transmit signal
generator, remote digitizers residing in the pedestals,
temperature control plates at the error assemblies and in the
pedestal, as well as measurement devices such as spectrum
and network analyzers. Red, green, and yellow simulated
LEDs provide status for each component.
These interfaces are designed for scientific investigation
and neither for operational nor mission control. The
ArrayLab advanced algorithms, array modeling, and STK
connectivity could be used within mission control software
such as Harris‟ OSCOMET.
H. Patriot Antenna Pointing Calibration
The antennas required pointing calibration after
installation. An algorithm was developed to scan in a grid-
like pattern while tracking a target. A low and high
resolution scan is performed. (See Figure 11) The antenna
pointing biases are the difference between the line-of-sight calculated using the orbit parameters and the centroid of
the 3dB down contour on the 3D antenna pattern formed during the scan.
Azimuth and elevation line-of-sight pointing biases are saved for many targets. These are used to create the
nine-parameter (five azimuth and four elevation) corrections incorporated by the antenna control unit. This
procedure is repeated for each dish. All targets are searched for sequentially and autonomously.
Figure 10. ArrayLab Graphical User Interfaces
for the Tx ACE Experiment. Hardware control
and status
American Institute of Aeronautics and Astronautics
11
Figure 11. Results from Antenna Pointing Calibration Algorithm.
IV. Mitigations of Arraying Challenges
A. Continuous Self-Calibration of Tx and Rx Circuitry
Circuit phase control is considered to be the enabling technology in this experiment [3],[4]
. This control in turn
depends upon a precision phase reference which is derived from forward and reverse traveling waves in a reference
distribution network. Theoretically, this reference is insensitive to variation in physical properties of the distribution
network. To verify the control process, a closed-loop circuit control experiment was conducted using the set-up in
Figure 7. The signal goes to a network analyzer instead of the two antennas shown in the figure. In this way, the
network analyzer can measure phase difference that would otherwise be inaccessible. All the hardware depicted
across the top of the figure as well as variable fiber length is a source of phase error and is compensated for
continuously and automatically. The goal was 3 degrees RMS phase error and 10 degrees peak. After running at
many frequencies and power levels, an extended run was conducted with the results given in Figure 12. The RMS
was 1.4 degrees, the peak was 7.9 degrees and there was a bias of -0.42 degrees.
Figure 12. Results for Closed Loop Circuit Control Experiment
Run Time: 83+ Hours Requirement: 3° RMS 10° Peak Achieved: 1.4° RMS 8.0° Peak.
During this 83+ hour run, Figure 13 shows the phase changes in the Transmit circuit for assembly 1 (yellow) and
assembly 2 (cyan). The large changes are primarily due to temperature changes: the air conditioner coming on and
off (faster cycling) and overnight periods when room temperature dropped due to lack of heat (at about times 1 and
2 times 10^5 seconds). Even these gross changes in circuit phase were controlled successfully as shown in Figure 12
where time is on the x-axis and phase differential between the assemblies is on the y-axis. About two degrees peak-
to-peak short term phase variation is attributable to up converter phase noise.
American Institute of Aeronautics and Astronautics
12
Figure 13. Wide range of phase errors in the transmit circuit paths were corrected during Closed Loop
Circuit Control Experiment.
Because of these results, exceeding the design objective, we have increased confidence that this performance
will be realized on the antennas, enabling control of transmit circuit phase errors during transmit uplink arraying.
B. Antenna Reference Point Calibration
Refinement of the phase center reference point location for each of the dishes is the second major source of
differential phase error. It is assumed that the dishes are stable and the initial calibration of the reference points
resulted in less than one wavelength of error. The phase center refers to the point from which the electromagnetic
radiation generated by the antenna element spreads spherically outward, with the phase of the signal being equal at
any point on the sphere. If the phase center location information is inaccurate, incorrect interference patterns will be
generated during beamforming, resulting in reduced signal strength during reception or misalignment during
transmission. These difficulties are further exacerbated as the size of the array is increased and the distance to the
object of interest is increased. A system of equations is formed where each row or observation represents signals
from various sources and angles of arrival. The phase center locations can be calculated based on differential
distances for the antenna elements [5]
. The term “differential distance” refers to the additional distance a wavefront
needs to travel to reach a phase center of the second antenna element after the wavefront has reached a phase center
of the first antenna element.
An algorithm was developed to minimize the error in the spatial positions of any number of antenna elements
using receive signals. This algorithm is run after the initial locations of the elements are measured using the total
station. It can theoretically correct up to one wavelength of error or about 36mm at X-band and 9mm at Ka-band.
Simulations for various antenna phase center errors have been conducted. In the Figure 14 example, 20mm (mean
spherical random) error or about 0.48 wavelengths at X-band was added to each antenna reference point. Also
compensated receive circuit errors of 5 degrees RMS were added. Initial antenna element estimates were modeled
with 6 degrees of freedom plus analytic antenna patterns. Four satellites were used as targets for the signal
downlink. The satellite and planetary motion were modeled using STK. After 20 observations (5 from each target),
the RMS mean error was
reduced to 1.8 mm from
20.0mm. The true
Cartesian positions of the
antennas are shown in
green and the corrected
positions are shown in
black. We will conduct
this calibration before the
adaptive combining expe-
riments using several
DSCS satellites and other
appropriate targets.
Figure 14. Results from Calibration of Antenna Reference Point Simulation.
American Institute of Aeronautics and Astronautics
13
C. Tropospheric Phase Error Correction
The last of the three primary considerations for Tx uplink arraying is uncompensated differential propagation
phase variation due to tropospheric effects. Part of the Transmit Adaptive Combining Experiment includes
evaluating the feasibility of uplink at higher frequencies such as Ka-band. The tropospheric contribution to phase
errors is significant and unpredictable The tropospheric contribution to phase errors is significant and unpredictable
as evidenced by Nessel, et al.9 of Glenn Research Center, “There does not appear to be any apparent relationship
between surface meteorological data and phase fluctuations. This implies surface measurements are not accurate
indicators of what is occurring higher in the atmosphere and developing a model to predict phase stability at a
particular site would be extremely difficult.” Harris‟ Instant Return‟ algorithm[6]
will be used to mitigate these
effects during our final experiment for this study.
During the 2008 Tx Uplink IR&D,
ArrayLab was used to test the Instant Return
algorithm on a hardware prototype
communications system. A mock-up of the
Orion exploration vehicle was constructed in a
compact range. It was fitted with 2-4 element
phased arrays. (See Figure 15). In this study,
tropospheric effects were not the cause of the
unknown phase errors but rather an anomalous
condition such as mechanical or thermal
variation of a circuit (phase detection
assemblies were not an option due to SWAP
constraint). To simulate these unknown phase
errors, random differential beamsteering errors
were added to the steering weights. The system
then switched to receive mode and used the
downlink signal and „instant return‟ to calculate the correct transmit weights. The Instant Return Method is thought
to be a viable solution for Transmit Uplink Arraying at higher frequencies. We will test this on a target with a low
elevation angle (~10 degrees) at X band. The method can use any signal arriving within an elements field of view.
ArrayLab was also used to model and
simulate tropospheric effects and the Instant
Return algorithm for the Norway Facility site
(See Figure 16a). The simulation includes 30
degrees RMS tropospheric phase errors, half
wavelength antenna phase reference error, and 50
degrees phase circuit errors. Sample results for 3
moments in time are shown in Figure 16b. The
elevation angles at the 3 time samples are about
7.1, 19.4, and 27.1, respectively. In this
simulation the target is the AQUA polar orbiting
satellite. Figure 17 shows a similar simulation
using an array of 25 dishes. For each time
sample a 3x3 plot of pattern with uncorrected
random phase errors, the pattern after using a
Receive signal to determine those errors and
adjusting and applying them for correction to the
Transmit signal. The bottom row depicts the
beam pattern in an ideal system for the same time
sample.
9 Nessel, James, A.; Acosta, Roberto, J.; Morabito, David, D.,” Phase Fluctuations at Goldstone Derived From One-
Year Site Testing Interferometer Data”, National Aeronautics and Space Administration, John H. Glenn Research
Center at Lewis Field, Cleveland, Ohio 44135-3191,
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20090020407_2009019738.pdf
Figure 15. Instant Return HW Orion Mock-up
Prototype Comm System (2008).
Figure 16a. Norway Facility Instant Return
Modeling & Simulation
American Institute of Aeronautics and Astronautics
14
Figure 16b. Norway Facility Simulation Results Mitigation of Tropospheric Effects Using Instant Return.
Each column of the graphics in Figure 16b represents transmission at a particular time sample. There are 3 different
line-of-sight elevation angles. The red line denotes the cross-section of the antenna pattern before using a receive
signal to determine the beam steer weights. The black curve is the beam pattern after the beam steer correction. The
black vertical line is where the peak of the beam should be.
Figure 17. Instant Return Simulation Results Applied to an Array of 25 dishes. Each of the 25 elements is
affected by varying phase errors due to the atmosphere. The top 3 plots show the 3D, 2D, and 1D slice of the
antenna pattern with simulated tropospheric errors. The peak is about 20 dB down. The middle 3 plots show the
corrected patterns after using Receive signals to determine the correct beam steer weights for each element. The
bottom 3 plots show the adaptively combined beam form the 25 elements without errors.
American Institute of Aeronautics and Astronautics
15
V. What‟s Next
A. Harris Experiment Plan
Three experiments remain for the array of 12m reflectors:
Calibrated Uplink arraying
Calibrated Receive arraying
Instant Return Uplink with correction of modest propagation variation After these experiments complete, assuming success with propagation correction, a reasonable next step would
extend the X band experiments to Ka band. The three 12m reflectors offer low impact means for extended testing.
B. Future of Array Technology
The DSN is currently upgrading the transmitter on one 34m beam wave guide (BWG) antenna per Tracking
Complex to 80 kW versus the 20 kW available today. This will provide a backup to the 70m DSN antenna‟s uplink
capability should the 70m experience an anomaly. However, on a near term basis, uplink arraying two 34m
antennas each with the currently deployed 20 kW transmitters could alternatively provide the same EIRP but for
only half the transmitter power budget and without the expense of new 80KW amplifiers. Uplink arraying three of
these antennas provides up to a 9.5dB increase.
As NASA spacecraft push farther out into the reaches of the solar system, larger receive apertures at radio
frequencies will be required to capture the data from increasingly weak transmission signals and higher uplink
radiated power will be required to support both routine and emergency communications at great distances from
earth. The need for higher EIRP and the role of uplink arrays in providing this has been succinctly noted by
D‟Addario10
, showing that an array of smaller antennas with modest power amplifiers could readily produce more
than 1.0 TW EIRP and that this would be needed for emergency spacecraft communications to Neptune and beyond.
An important benefit of arrays of smaller reflectors is the field of view they provide for the array. This greatly
increases the usefulness of the array in signal acquisition and navigation and tracking since more known position
quasars or other spacecraft can be seen simultaneously. Such enhanced visibility would also be beneficial for Ka
band propagation error detection and correction.
Another advantage of arrays of smaller reflectors is that lattices can be constructed that greatly minimize grating
lobes. For future array expandability considerations, optimization of antenna element spatial placement with respect
to minimizing sidelobes was also simulated. Figure 18 compares far field array patterns for three different lattices
populated with 12m reflectors. Random and aperiodic arrangements are seen to offer substantial suppression in
comparison with a regular hexagonal lattice. The pattern for a random lattice with 57 12m antennas (center
illustration) compares favorably with that of a single large dish having about the same overall diameter, while arrays
of a few large antennas with the same gain display pronounced grating similar to that of the hexagonal lattice or the
three element array patterns seen in Figure 16b.
Figure 18. Array Configuration Optimization With Respect to Sidelobe Minimization
10
D‟Addario, Larry R., “Large Transmitting Arrays for Deep Space Uplinks, Solar System Radar, and Related
Applications,” Jet Propulsion Laboratory, California Institute of Technology, M/S 11-116, 4800 Oak Grove Drive,
Pasadena, CA 91109, USA. Email: [email protected]
American Institute of Aeronautics and Astronautics
16
VI. Conclusions
Uplink Arraying architecture providing continuous internal self-calibration using the transmit signal itself, a
method to dynamically solve for the antenna reference points, and mitigation of propagation effects by using
received signals from known sources was presented, offering the potential for continual readiness operationally and
extensibility to Ka band. The approach is suitable for use on a wide range of antenna sizes, including both 34m and
12m reflectors. Recently completed experimental testing has confirmed the critical component of the approach,
closed-loop circuit phase control, and transmit experimentation is imminent using three-12m reflectors operating at
DSCS X band.
In conclusion, uplink arraying is a cost effective means of providing very high EIRP that is ready to be deployed
operationally, at least at X band.
Acknowledgments
The authors would like to thank Irene Bibyk for contributions in the early day of this work, the NASA teams at
the Johnson Space Center and the Glenn Research Center, John Rogers of DISA, the Harris teams at Palm Bay, and
Harris for the use of their facilities. In particular, the authors would like to thank Kathryn Morrison, Harris Program
Manager for handling the many details of this experiment and Wade Minear, Electrical Engineer, who has been a
critical part of the modeling and simulation team since 2005, and whose expertise was key in the transition to a
hardware prototype system. A special thanks to Nancy Gazzola Harris R&D Technology support, whose help
getting this document prepared, was crucial to a timely submission.
References
Patents
[1] Martin, G. P, Minear, K. M., Roach, J.(III), Dianic, A., Adams(Jr.), William C., Ralston, Lynda M Harris Corporation.,
Melbourne, FL, U.S. Patent Application for a “Compensation of beamforming errors in a communications system having widely
spaced antenna Elements,” U.S. Patent Application S/N: 12/273,760 filed 29 Nov. 2008.
[2] Martin, G. P, Minear, K. M., Adams (Jr.), William C., Harris Corporation., Melbourne, FL, U.S. Patent Application for a
“Model-based system calibration for control systems,” U.S. Patent Application S/N: 12/273,001 filed 19 Nov. 2008.
[3] Martin, G. P, Roach, J. Adams (Jr.),William C., Minear, K. M., Hash, R. J., Ralston, L. M., Harris Corporation.,
Melbourne, FL, U.S. Patent Application for a “Closed loop phase control between distant points,” U.S. Patent Application S/N:
12/273,839 filed 19 Nov. 2008.
[4] Martin, G. P, Harris Corporation., Melbourne, FL, U.S. Patent Application for a “Systems for determining a reference
signal at any location along a transmission media,” U.S. Patent Application S/N: 12/273,797 filed 19 Nov. 2008.
[5] Minear, K. M., Martin, G. P, Harris Corporation., Melbourne, FL, U.S. Patent Application for a “Systems and methods for
providing corrected antenna element phase center locations in a communications system,” U.S. Patent Application S/N:
12/273,981 filed 19 Nov. 2008.
[6] Martin, G. P, Minear, K. M., Harris Corporation, Melbourne, FL, U.S. Patent Application for a “Systems and methods for
compensating for transmission phasing errors in a communications system using a receive signal” U.S. Patent Application S/N:
12/273,935 filed 19 Nov. 2008.