The Evolution of Technology in the Deep Space Network
A History of the Advanced Systems Program

Advanced Systems Program and the Galileo Mission to Jupiter

A NASA mission to Jupiter first was discussed by the scientific community in the middle 1960s, and in early 1976 a scientific working group headed by James Van Allen formulated the Jupiter Orbiter Probe (JOP) Mission. The official start of the project was in October 1977, with the name being changed to Galileo in February 1978.

The Probe was designed to carry seven instruments whose measurements during descent through the upper reaches of the Jovian atmosphere are transmitted by radio to the Orbiter, with the microwave signal itself providing additional science information. The Orbiter was designed to carry some 11 instruments whose measurements, along with those received from the Probe, are transmitted by radio to the DSN on Earth. The Orbiter--Earth two-way microwave signals also provide additional science information. It is evident that the entire success of the mission depends on the radio signals connecting the Orbiter to Earth and Earth to the Orbiter.

After the start of the Galileo Project, there followed what has been described as a twisted tale of politics, technology, and science. Launch was to be in January 1982 with arrival at Jupiter in 1985. The Space Transportation System (STS), called the space shuttle, had been approved for development in 1972, and it was chosen as the (only) launch platform for Galileo. Galileo thus became the first deep-space mission to use the shuttle. The space shuttle fell behind schedule with many problems. The choice of the Galileo upper stage flip-flopped several times between the Inertial (Interim) Upper Stage (IUS) and the more powerful Centaur. The launch date slipped from 1982 to 1984 to 1985 and finally firmed on May 20, 1986, with the use of the Centaur G upper stage providing a flight time to Jupiter of about two and one-half years. In December 1985, damaged memory chips were discovered in the finished spacecraft, and a maximum effort had the spacecraft ready again in early 1986, with delivery to the launch site shortly thereafter.

On January 28, 1986, the Challenger disaster occurred. The space shuttle program was suspended, the Galileo May 20, 1986, launch date was cancelled, and Galileo was trapped in the delay of the shuttle reevaluation process. Also, as a fallout of the Challenger disaster, it was determined that a Centaur upper stage would not be used in future shuttle operations. The Galileo upper stage once more became the less powerful IUS. This resulted in an increase of Galileo flight time to Jupiter from two and one-half years to six years, requiring three gravity assists (one at Venus and two at Earth) with a thermally undesirable close approach to the Sun. The launch date again slipped more than three years to October 18, 1989, with arrival at Jupiter in December 1995, some 18 years after the beginning of the project. As one of the Project leaders has said, "One of the more unique aspects of Galileo has been its very rocky history." James Van Allen has referred to the Galileo Project as "the perils of Pauline."

After the October 1989 launch, Galileo's early cruise phase was nominal, with data retrieved over the shorter distances via S-band signals from the low-gain spacecraft antennas. The high-gain "umbrella" antenna with X-band capability remained furled for about the first year and one-half of cruise while the spacecraft was subject to the Venus and first Earth gravity-assist encounters involving the close solar approach.

On April 11, 1991, the Galileo Mission suffered a grave mishap. The command was sent for the spacecraft high-gain antenna to unfurl. The antenna unfurled part way and the mechanism jammed, leaving the antenna in a useless condition. At first there were expectations that the mechanical problem could be solved, but following attempts extending over more than a year to overcome the problem, it became evident that the high-gain antenna and X-band downlink would be unavailable. Without further remedies, the planned downlink data rate from Jupiter of over 100,000 bits per second (b/s) would be reduced to about 10 b/s using a spacecraft low-gain antenna at S-band. Such a four-orders-of-magnitude decrease in data return would be disastrous for the scientific goals of the mission.

Not long after the initial failure of the high-gain antenna, the DSN Advanced Systems Program, against the possibility that the problem could not be solved, reviewed the reservoir of its advanced technologies to see what might be applied to materially increase the 10 b/s provided by the spacecraft S-band low-gain antenna. This reservoir of advanced technologies included not only devices, but also the outstanding people who had developed the technology over an extended period. Four of the advanced technology areas had evident promise, as follows:

  1. Increase the S-band signal-to-noise ratio of the DSN S-band antennas; that is, increase Ae/Top, where Ae is the effective area of the antenna(s) and Top is the operating noise temperature of the antenna/receiver system. Ae can be increased by antenna arraying, and Top can be decreased with ultralow noise amplifiers and feeds.

  2. Improve the efficiency of the modulation of the radio signal, such as with a suppressed carrier. The new Block V Receiver (BVR) has fully suppressed carrier capability.

  3. Use improved channel codes so that the desired bit-error probability requires less energy per bit; that is, reduce Eb/No. New high-performance concatenated codes and hardware have been designed and tested.

  4. Aggressively apply data compression techniques to the various science, engineering, and optical navigation data from the spacecraft to Earth to provide an increase in the effective data rate. This requires reduced bit-error probabilities for the compressed data, which also can be provided by the improved codes of (3).

These considerations were described in a report prepared by Telecommunications and Data Acquisition (TDA) Technology Development.1 It was estimated that the combination of (1), (2), and (3) above would increase the 10-b/s data rate by an order of magnitude, and the application of (4) would provide at least another order-of-magnitude increase. The resulting equivalent data rate of at least 1000 b/s coupled with careful editing and choice of science and other data could provide a viable mission fulfilling much of the original Galileo science objectives.

However, it is one thing to discuss technology capabilities and quite another to move the technology into a constrained engineering application in a relatively short time. The improvements of (1) and (2) would involve mainly DSN systems with little interaction with the spacecraft. However, the improvements of (3) and (4) would strongly interact with the spacecraft, requiring reprogramming and reallocating of spacecraft computer resources within the constraints of spacecraft operability and safety. Also, the compression of science data would extensively involve science team members in evaluating and choosing data compression algorithms. The November 5, 1991, early report was conducted primarily by ground systems people with small participation from Galileo engineers and science team members, who were busy with the high-gain antenna problem, the Gaspra encounter, etc.

With the positive results of the TDA Technology Development early report and the probability of repairing the high-gain antenna fading, a major Galileo S-band Mission study was jointly chartered by TDA and the Flight Projects Office (FPO), with the report issued on March 2, 1992.2 The study was divided into four subtasks: science/mission design; telecommunication systems; ground systems; and spacecraft systems. The basic conclusion of this more detailed study was that by using the technology from the Advanced Systems Program a viable Galileo S-band Mission would be feasible. A design was provided that would meet a somewhat reduced, but very palatable, set of science objectives.

On January 7, 1993, the Galileo S-band Mission was formally approved and funded. Some details of the technologies used in this rescue follow.

  1. Antenna Arraying and Noise Temperature Reduction. The antenna-arraying capability developed by the Advanced Systems Program was made available for up to six antennas---the 70-m and three 34-m antennas at Canberra, the 64-m antenna at Parkes, and the 70-m antenna at Goldstone. This increases the effective area, AE, for Galileo signal reception to the sum of the effective areas of the antennas, which can exceed that of three 70-m antennas.

    The antenna that could make the greatest contribution from reduction in operating noise temperature was the 70-m antenna at Canberra. Its southern hemisphere location gives it more time at higher elevation (less atmosphere noise contribution) when tracking Galileo, and none of the other antennas has a higher Ae. When arraying antennas, the contribution to the overall array signal-to-noise ratio is greatest by the antenna with the greatest Ae/Top. Accordingly, the Canberra 70-m antenna was equipped with the receive-only "ultracone" feed and low-noise amplifier that previously had been developed by the Advanced Systems Program. It provides a receive-only very low Top of 11.8 K, compared to a receive-only Top of 15.6 K provided by the regular operational system.

  2. Improved Modulation Efficiency. Once the Block V Receiver (BVR) was available with its capability of tracking and processing fully suppressed carrier signals, the increase in modulation efficiency is essentially without cost. It is only necessary to program the spacecraft transmitter for an appropriate increase in phase modulation to obtain the increased efficiency. An increase of modulation index from 43 deg to 90 deg (fully suppressed carrier) approximately doubles the available data power. The BVR is based on the prototype Advanced Receiver (ARX) developed by the Advanced Systems Program over the better part of a decade. This prototype utilizes flexible digital implementation of carrier, subcarrier, and symbol tracking loops, which allows very narrow loop bandwidth. A Costas loop allows recovery and tracking of a fully suppressed carrier.

  3. The built-in channel coding available to the Galileo S-band transmitter was a hardware (7,1/2) convolutional code, primitive by today's standards---particularly for highly compressed data. By programming a software (11,1/2) convolutional code on a Galileo computer (possible because of low bit rates) and concatenating it with the hardware (7,1/2) code, a (14,1/4) code is produced. This is used with a software Viterbi decoder (permitted by low bit rates) as the inner code of a novel concatenated (255, k) variable redundancy Reed--Solomon (RS) coding scheme. The redundancy profile is (94,10,30,10,60,10,30,10), and the interleaving depth is 8. The staggered redundancy profile is designed to facilitate the novel feedback concatenated decoding strategy that allows multiple passes of channel symbols through the Viterbi decoder. During each pass, the decoder uses the decoding information from the RS outer code to facilitate the Viterbi decoding of the inner code in a progressively refined manner. The feedback concatenated decoder is implemented in software and provides a bit-error rate of 10 X 10-7 at an exceptionally low signal-to-noise ratio of Eb/No = 0.65 dB. The original (7,1/2) convolutional code would provide a bit-error rate of 10 X 10-3 at Eb/No = 2.3 dB. This error rate is acceptable for uncompressed image data but much greater than the 10 X 10-7 error rate required for compressed image data.

    In addition to this code based on research in the Advanced Systems Program, the processing of the received signals includes predetection recording and noncausal processing to eliminate acquisition delay and minimize dropout intervals. At each antenna, the S-band signal from the spacecraft is downconverted to 300 MHz and fed simultaneously to the BVR for real-time processing and to the Full Spectrum Recorder (FSR) for subsequent nonreal-time and noncausal processing. Noncausal processing involves using future as well as past values of the signal for estimation at a given time instead of being limited to the use of past and present values, as required by real-time (without delay) processing. For example, noncausal processing can eliminate the usual acquisition delay by phase-lock and symbol loops (and the resulting loss of data) by processing the signal in reverse time, where the beginning of the signal becomes the end of the signal and future signal becomes past signal. Further, once the signal becomes available over an extended interval of time via the FSR, computer programs can make near-optimum estimates of the data by using the signal over the entire interval. Among other things, this permits maintaining synchronization during discontinuous changes in data rate used to take advantage of the predictable varying signal-to-noise ratio resulting from changing tracking antenna elevations and arraying combinations of DSN receiving antennas. (The spacecraft is not able to change data rate in continuous fashion.) This prevents loss of data at each change of rate while tracking loops reestablish lock.

  4. In a mission like Galileo, the majority of the downlink data is assigned to images and, thus, the maximum acceptable compression of image data is most important. The Galileo solid-state imaging camera provides 800 X 800 pixel images with 8-bit digitization of each pixel, thus producing 5.12 megabits per full image. In order to obtain a large factor of data compression for images, it is necessary to use compression algorithms, which introduce error. This error increases with the factor of compression. In order to determine the maximum acceptable error in images, the scientists and the Galileo solid-state imaging team conducted an extensive investigator-in-the-loop evaluation of compression algorithms. It was determined that an integer approximation of the discrete cosine transform, previously studied by the Advanced Systems Program, would provide generally acceptable images. The acceptable compression factor may vary from image to image, but on the average, a compression factor of at least 10 results. The integer approximation does not significantly degrade the compression and makes it possible to carry out the discrete cosine transform on a spacecraft computer of limited capability. The decompression operation includes postprocessing techniques in the frequency and spatial domains to remove compression artifacts without increasing distortion.

With the improved S-band downlink having a maximum data rate in the neighborhood of 100 b/s, the spacecraft tape recorder has an additional critical function. In addition to storing data for later transmission, the recorder must "convert" high-rate data produced by some of the instruments to low-rate data for subsequent playback over the improved S-band downlink, which still is a factor of about 1000 slower than the failed original X-band downlink around which the mission was designed. Some further recovery of Galileo's data volume was accomplished by simply allocating a large amount of DSN antenna time to the mission.

The early accomplishments of the Galileo Mission after arrival at Jupiter indicate that application of the Advanced Systems Program technology for the Galileo S-Band Mission has been entirely successful.

On December 7, 1995, the Galileo Probe entered the Jupiter atmosphere and functioned as planned. The Probe data were received and stored on the Orbiter just prior to its Jupiter orbit insertion. The stored Probe data were transmitted to Earth over an extended period via the S-band downlink. The advanced coding and other improvements of the Galileo S-Band Mission were activated later in 1996. The first satellite encounter after Jupiter orbit insertion was that of Ganymede on June 27, 1996, at a distance of 832 km. All of the enhancements provided by the Advanced Systems Program (except for antenna arraying, scheduled for later) functioned as planned for imaging and other data. On September 6, 1996, there was a second Ganymede encounter at a distance of 262 km.

On November 4, 1996, there was the first encounter with Callisto, at a distance of 1104 km. At this encounter, the spacecraft was at one of its most distant points from Earth, and the DSN antenna arraying capability was used for the first time on the mission. The array included the 70-m and 34-m antennas at, the 64-m antenna at Parkes, and the 70-m antenna at Goldstone. At this maximum distance, the spacecraft downlink data rate was programmed among 120, 80, 40, 32, and 20 b/s, depending on the combination of arrayed antennas and their tracking elevation angles. This arraying capability had been tested successfully in September at spacecraft telemetry bit rates up to 160 b/s.

On December 18, 1996, the Europa encounter occurred at a distance of 692 km. All communication enhancements continued to operate as planned. There is every indication these vital technology contributions from the Advanced Systems Program will continue to contribute to the success of the Galileo Mission during the remainder of its operation.

1L. J. Deutsch et al., Galileo Options Study, internal document, Jet Propulsion Laboratory, Pasadena, California, November 5, 1991.

2L. J. Deutsch et al., Galileo S-Band Mission Study, internal document, Jet Propulsion Laboratory, Pasadena, California, March 2, 1992.


Go to:

--> Next Section
--> Previous Section
--> Home Page