Return Data Link

Throughout the DSN, the same antennas are used for both the forward link and the return data link signals. Because the strength of a signal decreases as the square of the distance it must travel, these two signals may differ in strength by a factor of 10 ²⁴ in a DSN antenna. Isolating the return signal path from interference by the much stronger forward signal has posed a significant technical challenge. These two signals differ somewhat in frequency, so at least a part of this isolation is accomplished via dichroic or frequency selective reflectors. These reflectors consist of periodic arrays of metallic/dielectric elements tuned for the specific frequencies, which are to be either reflected or passed through.

One of the strongest design constraints is that noise added to the received signal must be minimized. Prototypes of almost all of this type of reflectors currently in use in the DSN were developed under the DSN Advanced Systems Program, as well as microwave analytical design tools that can be used to affix design details for almost any conceivable dichroic reflector applicable to the frequency bands of the DSN.

Low-Noise Amplifiers

The typical return data link signal is incredibly small and must be amplified before it can be processed and the data itself reconstructed. The low-noise amplifiers that reside in the antennas of the DSN provide this amplification while adding the least amount of noise of any such devices in the world.

The quietest of the operational devices are known as traveling-wave masers (TWMs), which amplify signals that are propagating along the length of a tuned ruby crystal. Noise in a TWM depends upon the physical temperature of the crystal, and those in operation in the DSN operate in a liquid helium bath at 4.2 kelvin. Invented by University of Michigan researchers, early development of practical amplifiers for the DSN was carried out under the DSN Advanced Systems Program, as were many improvements throughout the Network's history. The quietest amplifiers in the world today, which operate at a physical temperature of 1.2 kelvin, were developed by the DSN Advanced Systems Program and demonstrated at the Technology Development Field Test Site, Deep Space Station 13 (DSS 13).

Some of the low-noise amplifiers in the DSN today are not TWMs, but are a special kind of transistor amplifier using high-electron mobility transistors (HEMTs) in amplifiers cooled to a physical temperature of about 15 kelvin. Initial development of such amplifiers occurred at the University of California at Berkeley, leading to their adoption by the radio astronomy community. This in turn spawned the JPL development work that was carried out via collaboration involving JPL and the DSN Advanced Systems Program, radio astronomers at the National Radio Astronomy Observatory (NRAO), and device developers at General Electric. This work built upon progress in the commercial sector with uncooled transistor amplifiers. In the 2-GHz DSN band, the cooled HEMT amplifiers are almost as noise-free as the corresponding TWMs, and the refrigeration equipment needed to cool the HEMTs to 15 kelvin is much less troublesome than that for the TWMs. Primarily for this reason, current development efforts in the DSN Advanced Technology area are focused on improving the noise performance of the HEMT amplifiers for the higher DSN frequency bands.

The first DSN application of the cooled HEMT amplifiers came with the outfitting of the NRAO Very Large Array (VLA) in Socorro, New Mexico for collaborative support of the Voyager-Neptune encounter. The VLA was designed for mapping radio emissions from distant stars and galaxies, and consists of 27 antennas, each 25 m in diameter, arranged in a tri-axial configuration. Within the funding constraints, only a small part of the VLA could be outfitted with TWMs, whereas HEMTs for the entire array were affordable and were expected to give an equivalent sensitivity for the combined full array. In actuality, technical progress with the HEMTs under the Advanced Systems Program during the several years taken to build and deploy the needed X-band (8 GHz) amplifiers resulted in better performance for the fully equipped VLA than would have been possible with the VLA partially equipped with the more expensive TWMs. Since that time, many of the DSN operational antennas have had the cooled HEMT amplifiers installed for the 2- and 8-GHz bands.

Phase-Lock Tracking

Once through the first stages of processing in the low-noise amplifiers, there are still many transformations needed to convert the radio signal from a spacecraft into a replica of the data stream originating on that spacecraft. Some of these transformations are by nature analog and linear, and others digital with discrete quantization. All must be performed with virtually no loss in fidelity for the resultant data stream.

The signal typically consists of a narrow-band "residual carrier" sine wave, together with a symmetric pair of modulation sidebands, each of which carries a replica of the spacecraft data. (Specifics of the signal values vary greatly, but are not essential for this general discussion.) If this signal is cross-correlated with a pure identical copy of the residual carrier, the two sidebands will fold together, creating a low-frequency signal that contains a cleaner replica of the spacecraft data than either sideband alone. Of course, such a pure copy of the carrier signal does not exist, but must be created, typically via an adaptive narrow-band filter known as a phase-locked loop. The recreated carrier reference is thus used to extract the sidebands. The strength of the resultant data signal is diminished to the extent that this local carrier reference fails to be an identical copy of the received residual carrier. Noise in the spectral neighborhood of the received residual carrier and dynamic variations in the phase of the carrier itself limit the ability to phase-lock the local reference to it.

These dynamic variations are predominantly the Doppler effect of the relative motion between the distant spacecraft and the DSN antenna on the surface of a spinning Earth; they interfere with the return data link process, but themselves provide for a radio location function. Over the years, the DSN Advanced Systems Program has contributed significantly to the design for the phase-locked loops and to the knowledge of phase-coherent communications, and thus to the performance of the operational DSN.

Synchronization and Detection

Further steps in converting a spacecraft signal into a replica of the spacecraft data stream are accomplished by averaging the signal over brief intervals of time that correspond to each symbol (or bit) transmitted from the spacecraft, and by sampling these averages to create a sequence of numbers, often referred to as a "symbol stream." These averages must be precisely synchronized with the transitions in the signal as sent from the spacecraft, so that each contains as much as possible of the associated symbol and as little as possible of the adjacent ones. In usual cases, a subcarrier, or secondary carrier, is employed to shape the spectrum of the spacecraft signal, and it will be phase-tracked and removed at this stage. There are several different generations of equipment in the current DSN that perform this stage of processing. Designs for all of these have their roots in the products of the DSN Advanced Systems Program. The oldest current equipment is of a design derived from the Multimission Telemetry Demonstration, done in the late 1960s by a partnership of the DSN Advanced Systems Program and the DSN Implementation Programs. This equipment is mostly analog in nature and, while still effective, is subject to component value shifts with time and temperature, and thus requires periodic tending and adjustments to maintain desired performance.

As digital devices became faster and more complex, it became possible to develop digital equipment that could perform this stage of signal processing. Digital demodulation techniques were demonstrated by the Advanced Systems Program in the early 1970s in an all-digital ranging system. Similar techniques were subsequently employed for data detection in the second generation of the Demodulator-Synchronizer Assembly.

A Digital Receiver

More recently, the ongoing evolution of the capabilities of digital devices has made possible the migration of digital techniques into the filtering, detection, and phase-lock processes of the receiving systems. Because of this, the Advanced Receiver (ARX) developed under the Advanced Systems Program has a demonstrated precision of performance, which is almost inconceivable for conventional analog signal handling methods. A copy of the ARX has been constructed and deployed to the DSN Australian site for ad-hoc support of the Pioneer 10 spacecraft on its way out of the solar system, thus delaying the date at which the signal becomes too weak to be reliably received.

Current implementation efforts have been concluded on a new operational receiver (the Block-V) for the DSN, which builds upon the design techniques of the ARX and incorporates all of the functions of the current receivers and the demodulator-synchronizer equipment. By 1997, all of the older generations of this equipment should be out of the DSN, replaced by the new digital Block-V receiver, which offers improved technical signal handling performance, as well as improved maintainability.

Codes and Decoding

The sequence of numbers from the demodulator-synchronizer equipment is still not the replica of the mission data stream on the spacecraft, because, in most circumstances, special codes have been applied to that data stream to improve the reliability of communications. These codes transform the data before they leave the spacecraft, adding carefully designed redundancy and complexity, and the resultant coded stream is reverse transformed by decoding equipment at the DSN site in order to recover the original data stream. The best codes to be used for reliable data transfer have been identified by research performed under the Advanced Systems Program; that work is still making progress. The present accepted standard, flown on Voyager and Galileo, consists of a short convolutional code that is combined with a large block-size Reed-Solomon code. The standard algorithm for the decoding of convolutional codes was devised in consultation with JPL researchers, and demonstrated by simulations performed under the Advanced Systems Program. Prototypes of the decoding equipment have been fabricated and demonstrated at JPL, also with the support of the Advanced Systems Program.

Evolution of the use of codes and decoding equipment has been paced by the evolution of digital processing capability. At the time of the Voyager design, a convolutional code of length k=7 was chosen as a compromise between performance and decoding complexity, which would grow exponentially with code length. Equipment was implemented around the DSN to handle this code from Voyager and subsequently from Magellan, Galileo, and others. Modern digital technology has permitted the construction of much more complex decoders, and a code of length k=15 was devised with the support of the Advanced Systems Program. This code was installed as an experiment on the Galileo spacecraft shortly before its launch. The corresponding prototype decoder was completed soon afterward. Though not needed for Galileo because of its antenna problem, the more complex decoder will be implemented around the DSN for support of Cassini and subsequent missions.

Efforts of the Advanced Systems Program provided the understanding of telemetry performance to be expected with the use of these codes. The figure displays the reliability of the communication (actually, the probability of erroneous data bits), as it depends upon the spacecraft signal energy allocated to each data bit for uncoded communication and three different codes. One of these is the Voyager k=7 code, shown both alone, and in combination with the Reed-Solomon code. Second is the k=15 code, which was to be demonstrated with Galileo's high-rate channel, shown alone and in combination with the Reed-Solomon code, either as constrained by the Galileo's data system (I=2) or in ideal combination. Third, and finally, is the k=14 code devised by the Advanced Systems Program researchers for the Galileo low-rate mission, shown both alone and in combination with the selected unbalanced Reed-Solomon code and a complex four-stage decoder. The added complexity of the codes, which has its greatest effect in the size of the decoder, clearly provides increased reliability of correct communication. Research on new and improved codes continues.

Other types of codes have been used in the past, and continue to be active on some of the older spacecraft. The DSN Advanced Systems Program has played a part in each. The imaging data of Mariner '69 was encoded using short block codes with a self-synchronizing feature devised by the Advanced Systems Program researchers. A decoder for this code was constructed and used experimentally to provide a substantial increase in the data volume returned from the mission. At about the same time, Pioneer 9 was launched with a very complex code of length k=25, which could be decoded by an iterative approximation technique known as "sequential decoding." The code was chosen to satisfy the needs of Pioneer's experimenters who would accept intermittent gaps in their data caused by decoding failure in exchange for knowledge that successfully decoded data would be virtually error-free.

Decoding was planned to be performed by mission operations at Ames Research Center using a recorded symbol stream delivered from the DSN. In conjunction with Pioneer, the DSN Advanced Systems Program explored and demonstrated the potential for decoding this code in real-time via a very high-speed engineering model sequential decoder. With the rapid evolution in capability of small computers, it became apparent that decoding Pioneer's data in such computers was both feasible and economical. Subsequent implementation of sequential decoding in the DSN was done via micro-programming of a small computer, guided by the knowledge gained via the efforts of the Advanced Systems Program. The subsequent Pioneer-10 and -11 spacecraft flew with a related code of length k=32, and are still supported by the DSN in a computer-based decoder.

Data Compression -- A Mathemagical Twin of Coding

Source encoding and data compression are not typically considered a part of the DSN's downlink functions, but the mathemagics that underlie coding and decoding are a counterpart of those that guide the development of data compression. Simply stated, channel encoding is the insertion of structured redundancy into a data stream, while data compression is the finding and removal of intrinsic redundancy. Imaging data are often highly redundant and can be compressed by factors of at least two, and often four or more, without loss in quality. For Voyager, the combined effect of a very simplified image compression process, which was constrained to fit into available onboard memory, and the corresponding changes to the channel coding was about a factor-of-two increase in the number of images returned from Uranus and Neptune.

As with many other items of technology, properly crediting this end result is difficult. There is an active international community with interests in data compression for many purposes. Researchers under the DSN Advanced Systems Program have contributed significantly to the current state of the art. However, it remained for others to actually establish data compression on Voyager in flight, and on subsequent missions. By the time of the design of the Galileo data system, data compression had become a regular option for about a factor of two in the imaging data system. The failure of Galileo's high-gain antenna (HGA) prompted an intense effort to find compression schemes that would recover some of the Galileo imaging data that otherwise could not be returned.