Acquisition and Tracking

NASA/JPL's Block V Receiver

The fifth generation deep-space communication receiver has been in continuous use since the mid-90s, and receives all telemetry/pictures/etc from all NASA space probes beyond the moon. We designed the physical-layer acquistion and tracking algorithms, including:
  • FFT subcarrier and correlation symbol search
  • Initialization of phase and delay locked loops for tracking of the BPSK and QPSK waveforms
  • Optimum acquisition bandwidth pull-in in time of the tracking loops from initial uncertainty (FFT accuracy) to final optimum tracking bandwidth
  • Optimum tracking bandwidth determined from white and phase noise to either
    • minimize tracking error, or
    • maximize MTB cycle slip
  • Phase noise models combine both clock Allan variance, as well as ionospheric and solar plasma fluctuations
  • Lock detector, phase extractors, symbol phase windowing 
As a by-product of this project, a new parametrization of phase-locked loops was invented that allowed impulse response control for both high-order loops as well as slowly-sample signals. These advances were documented in the following IEEE paper:
    Controlled-loop formulation for DPLL

Other papers on this receiver:
    Block V Fast Acquisition Algorithm

An example trade-off diagram for tracking loops, showing tracking error vs loop bandwidth, and depending on white noise levels (C/N0), colored noise levels (spacecraft clocks and solar plasma), and shows mean time between cycle slip, which depend on higher moments of tracking error than the standard deviation.

Use during Galileo's Jupiter Orbit Insertion (JOI)

When Galileo arrived at Jupiter, there were two extremely challenging issues to address from a tracking standpoint:
  • The comm link had to use the low-gain spacecraft antenna, since the high-gain attenna failed to deploy. This meant that signal strengths received at even the 70m DSN antenna were very low (~20dB-Hz)
  • The earth was on the other side of the sun from Jupiter during orbit insertion, meaning the signal had to pass very close to the sun, whose solar wind created strong carrier phase variations.
In addition, since Galileo was tracked from ground sites (3 DSN stations), signal acquisition was not a rare occurrence, but happened at least once a day.

The flexibility of the receiver software was validated by these extreme conditions, where FFT acquisition had to occur over the integration time of apx 1 minute, symbol rates of ~20/s, and tracking bandwidth pull-in lasting several minutes.

NASA/JPL's Experimental Tone Tracker

The Mars Observer spacecraft carried an experimental Ka-band link (KaBLE), that was phase coherent but much weaker than the primary/standard X-band communication link.

Using our experience with JPL's TurboRogue GPS receiver as well as the algorithms developed for the Block V Rcvr, the Experimental Tone Tracker (ETT) was built using the TurboRogue receiver, but with an IF feed coming from one of several of NASA communication antennas at Goldstone and other sites.

Coupled tracking loops, between as strong X-band and weak Ka-band, were used to demostrate the feasibility of operating over solar system scale distances at this higher frequency.

While created originally just for KaBLE, the utility and performance of this tool became known around JPL, and was used for many scientific and emergency tasks.

ETT used for Galileo velocity anomaly experiment

The Galileo spacecraft, to gain energy to reach Jupiter, passed by the earth twice for a gravity assist. During the first passage, the spacecraft passed very low to the earth (altitude of about 300km), and was not tracked by the deep space network for about 20 minutes because of this low altitude. After the spacecraft was reacquired leaving the earth, it was found to have 3mm/s higher velocity than expected. After replaying all telemetry and tracking data, no explanation could be found for this error which was apx 10 times larger than uncertainty. The idea that the JPL orbit models were in error, or that a new relativistic effect was being observed by this fast-moving spacecraft, required more careful observations during the second pass to come.

Since any earth-bound telescope would only be able to see part of the encounter, and then would have to be in special parts of the world (for instance, on a ship in the south Atlantic), we designed an experiment that would track Galileo from one of the TDRSS spacecraft, but the signal would be relayed by bent pipe to the TDRSS ground stations at White Sands, New Mexico.

ETT used to search for lost spacecraft

Because of the nimble/flexible design of the ETT, it was the go-to system for rapid-response searches for lost spacecraft. It was deployed at Goldstone to search for the multiple trajectories that Mars Observer might have taken after its loss of communication, and for Landsat IV lost during launch. For Mars Observer, it was key to be able to integrate for long periods of time looking for a weak signal from the UHF balloon relay, while for Landsat IV the search dynamics were the issue, since the possible set of low orbits were large.

Time/frequency analysis of oscillator noise

Typical sources of non-white noise in communication systems are from RF local oscillators/time references. In addition to spectral components, Allan variance is a useful characterization of oscillators in time. The diagram below captures all of the time standards used by NASA by the early 1990s.

Jump to next section: Channel Mitigation