Sunday, April 29, 2012

The Radio Science Campaign, Part 1/2 — The Basics

Matt, you've had another one of those drastically long periods of time without a MER Blog post.

I know, I know. But I have an excuse!

If you say "work," we're no longer friends.
Can I say, "Partly work, partly that I just bought a new guitar amp, partly that there are lots of words to write"?

I'll allow it. On one condition...
Don't make it boring.

We Talk With Our Rover

Since the 1st of January this year, priority number one for the Opportunity rover tactical team has been to sequence communication sessions with antennas at the Deep Space Network (DSN). 


I mean, yes, it's a critical bit of the rover surface operational scenario. But, I mean, you know…

Why "yawn"? Because we have done this every roughly sol since the beginning of the mission. 

There are several ways that Opportunity has to communicate with us. Most of them are "canned": We determine them weeks or months ahead of time and load them up to the rover, and then the rover's flight software autonomously does all the hard work for us when the time comes, unless of course we choose to tweak a particular track* tactically.

But, hey, wait a minute, there are a few things that have been different about these communication windows…

More DTEs, Less DFEs ("she's shouting back at us")

First, I must be clear: I am speaking about the UPLINK path only; DOWNLINK is another beast altogether as it also involves relay satellites, occurs at a different time of (Martian) day, and can actually piggyback on uplink sessions. There are subtleties here that I will omit the sake of the reader. (You're welcome!)

Historically speaking, most of the communications passes with Earth are "Direct From Earth" (DFE): meaning, the rover only listens for data — 0's and 1's that map to instructions for the upcoming sol(s) — coming from Earth and sends nothing back except for the infamous "beep" signal to confirm that it has successfully received the data and has "handed over" to the next sol's set of sequence**. 

We have another type of communication session that we still consider to be on the uplink side of the data flow to the rover: Direct To Earth (DTEs). In DTE passes, the rover sends back a limited amount of information — like, almost nothing, but a little — that allows the tactical downlink team to somewhat confidently assess the health and status of the rover. Although there is "downlink" going on here, we colloquially treat DTEs as uplink windows, because DTEs are inclusive of the activities performed in DFEs and they often (but not always) occur during the time of the sol when we typically execute uplink windows with the rover. 

DTEs provide the opportunity for additional, non-critical information, which brings me to the next thing…

There's a Scientific Goal to Achieve — Radio Science ("the vindication of our woes")

The additional information that comes down with a DTE allows engineers to derive timing information to help correct what the rover thinks "now" is. A non-perfect, drifting clock — something that plagues all spacecraft — causes many problems. Some of them are serious, like that the rover could suddenly think's a different sol and try to execute the wrong set of commands. Some of them are mere inconveniences, like that 360-degree panoramas don't show the horizon at the same local-level elevation. This effect is particularly notable in Stu's infamous vertically exaggerated images, even after rotating it to get it so that horizontal is actually "flat" on the Martian surface:

(Notice the strange bias of the horizon, not all of which is explained by the fact that we are peering into the giant Endeavour crater in this image.)

Again speaking historically, DTEs have been used primarily to get this timing information. This past winter, however, we are doing DTEs for another reason — with the added benefit of timing information. In these DTEs, we also get very precise Doppler-shift data (more on how in a bit) that helps us achieve a scientific, rather than engineering, goal. That goal is our famous Radio Science. From these are borne the infamous acronym, "RS-DTEs" (Radio Science DTEs).

Winter DTEs Create Tactical Sequencing Challenges ("sometimes we win, sometimes Mars wins")

We were able to do radio science this Martian winter primarily because we knew we weren't going to move for at least a few months while we passed through the depths of Martian winter. Again, more on why non-movement is important in a bit. The problem is that this is difficult: RS-DTEs require lots of power from the rover when the rover may not have that power available due to low solar insolation and inability to charge up her batteries fully. Balancing the goals of Radio Science with keeping the rover alive and healthy was a day-to-day, sol-to-sol battle. 

Although the end result of the challenge was a simple RS-DTE window, the means of getting there were brain-busting. I have never had to think so hard about uplink windows and how to fit them into the day's plan until these last few months. I'll give some insight into this process later on.

Right, so: Now we know how this winter has been a little different than past winters. We have done this four-month campaign of RS-DTEs, getting a little over of 1 hour's worth of Doppler-shift data per week to meet the requirements of the science. So how does this all work?

Before We Get There: A Brief Overview of Cool Things

To get the basics of the concept behind our beloved RS-DTEs, you cats first need to learn a little. First, some pretty:

Hastily created. Deal with it.
This summarizes all of the communication paths that can occur between the rover and Earth:

1) Relay, Forward and Return Link: We use Return link (rover to orbiter to Earth) most often. Forward link (Earth to orbiter to rover) has not been used regularly in years.
2) DFE: Earth to rover only. 
3) DTE: Everything that a DFE is, plus some small downlink back to Earth.

Now: The DSN antennae and the spacecraft (rover) can be in one of three Doppler states, shown below: 

0-way (not shown): No signals are traveling between the spacecraft and DSN.

1-way: Downlink from the spacecraft to the DSN. Fun fact: Whenever you have a scheduled time slot with a given DSN antenna, that means you at least have 1-way (downlink only) capability. Most of the time you get the uplink, too, but not always.

2-way: Downlink from the spacecraft to the Earth with simultaneous uplink from the DSN to the spacecraft. 

3-Way: The spacecraft simply radiates its downlink signal at Earth and this signal's "footprint" is bigger than the Earth itself due to the spreading of the signal over the vast distances of the solar system. Therefore any DSN antenna that can see the spacecraft will see its downlink signal. The DSN stations are located so that "view periods" of any given point in the sky overlap in time — and therefore DSN antennae on separate ends of the Earth (not to mention those at the same local complex) can see the same signal. When the spacecraft is 2-way with a station and another station starts listening for the same downlink signal, it is said to be 3-way with the spacecraft. (Subtlety: Only a DSN antenna is said to be in a 3-way state.)

Now: More detail. 2-way and 3-way aren't simply 2-way and 3-way. They need another qualifier, something called "coherency state."

When a spacecraft has a track scheduled with a particular antenna, there are several complications regarding the dependence of uplink and downlink paths on one another. Let's say you're a newbie and you want to listen to the spacecraft's downlink signal. You get out some documentation and find out that the spacecraft downlinks data on a radio signal that oscillates with a frequency of, say, 1 bazillion hertz. So you fire up your antenna and start listening at 1 bazillion hertz — but you can't detect the signal. What's going on?

Well, several things. Most importantly, the signal is Doppler-shifted (not to be confused with the Doppler state) because the spacecraft is moving relative to you (or you to it, same thing). That is, not only is the spacecraft hurtling through empty space at well above Earth-escape velocity, but you are on a planet that is moving and rotating through space at a high speed as well. The signal from the spacecraft appears to come down at a different frequency from 1 bazillion hertz due to this relative velocity between observer (Earth) and source (spacecraft). It's just like that canonical example of an ambulance driving by you: the frequency of the sirens increases right as the vehicle moves by you because of the relative speed to your ears.

This Doppler shift data is the bread and butter of not only deep space navigation, but also the science behind our beloved RS-DTEs. Another nice bit about Doppler information is that is gives us ranging data that allows us to determine velocity toward or away from the observer. A little magic math later and you can place this velocity in a reference frame at the center of the solar system, which gives you better geometric context of the spacecraft's position. 

Now, for a dose of reality: the Doppler-shifted signal coming from the spacecraft is noisy and is almost unusable to do accurate ranging measurements. So why have I told you that we do them?

Coherent downlink, is why. The "noise" and "error" in a downlink signal generated by the spacecraft comes from the imperfect reference oscillator source, among other things. There's some hardware in the spacecraft that generates a reference signal about which the spacecraft sends "modulated" signals that represent actual data, actual 0's and 1's. Due to these engineering imperfections and limitations, the frequency of this signal isn't stable and varies in time. Although the changes are small and imperceptible to the layman, it is significant — significant enough to render Doppler ranging data almost useless for high-precision navigation. It is impossible to predict this noise and error, as it is random, thus it is very difficult to extract good data from the Doppler-shifted downlink signal.

The elegant solution devised by the DSN is something called "coherency." What you need in order to have an ultra-stable downlink reference is a whole lot of heavy equipment cooled with… very cold things. Like, a handful of degrees Kelvin kind of cold. You can't put this on a spacecraft because it would be bigger than the spacecraft itself, quickly rendering all deep-space spacecraft un-launchable. Instead, the antennas at the DSN all have this equipment themselves to generate ultra-stable uplink signals. 

The spacecraft turns off its internal oscillator reference signal and instead listens for the uplink signal coming from the station. The spacecraft can measure this uplink signal very accurately and therefore the ultra-stable nature of the uplink signal is not lost. it takes this signal's frequency, multiplies it by a very precise number, and shoots that signal out as its downlink. The downlink signal is then said to be coherent. The communications link between station and spacecraft is also designated as "2-way coherent" (not just "2-way"):

With this ingenious method of uplink-downlink dependence, the antenna on Earth can receive downlink signals that are as stable and accurate as their uplink signal. This means much more precise Doppler shift data — so precise, in fact, that we can use this data to see if a planet is wobbling by placing a stationary radio source on the surface at an arbitrary location. 

How to Measure Wobbliness

If you make the assumption that the planet Mars is spinning perfectly and immutably about its normal spin axis (coming out of its North Pole, like Earth), you can easily*** subtract all so-called "systematic" effects of the Doppler shift in the coherent downlink signal. Most of these are, again, due to the relative velocity of the observer and source. There are other effects that aren't important to us. With all of these systematics removed from the signal, you'll still see variance in the downlink signal. Why? 

Because the planet is wobbling, not just spinning, is why.

The rotation of any object, planet or otherwise, will always be imperfect. It will "precess" and "nutate" over long periods of time:

The exact amount of precession and nutation tells you a lot of things, such as the internal structure of the planet. For instance, if the planet is solid through and through, it won't "slosh" around a lot and will be more stable — and the doppler shift of a stationary thing on the planet will reflect this. If a planet has a liquid core, then it will precess and nutate about its spin axis in a particular way — and the doppler shift of a stationary thing on the planet will reflect this. Measurements of these small movements allow you to lock down, but not fully determine, this structure. Therefore, all that extra downlink signal variance can be derived into the structure of the planet.

Way. Effing. Cool.

That wraps up this part. Up next: The results of Opportunity's winter Radio Science campaign.

The Radio Science Campaign, Part 2/2 — The Results


*Track = pass = communication session. "Track" is the most widely used term at JPL, insofar as my ears can tell. I'll use these terms liberally at the expense of the sanity of people that think there's a difference between them. Also, "track" is not really used as a term for relay sessions with the orbiter, even if the orbiter is a "bent-pipe" and sends Oppy's data immediately back to Earth via a track that it, the orbiter, has scheduled with the DSN. I mention this because we also maintain the capability to change particular parameters with these relay sessions with the orbiter.
**The "beep" actually doesn't always tell you anything about whether or not the uplink was successful. It may, but only by implication, and not in every instance. What it REALLY means is more technical and not worth delving into. Ah, engineering subtleties…
***It's actually really hard. 


Buck said...

I'm an amateur radio operator: a ham. I'm particularly fond of microwave frequencies (10GHz and up). Nice job explaining the noise and jitter problems in oscillators. Not easy to do!

What really caught my eye was the idea that Oppy gets her frequency reference from the incoming signal and then multiplies that up to her transmit frequency. Wicked cool! It's easier (but still hard) to make an ultra-stable multiplier than it is to make an ultra-stable oscillator.

I'm off to troll the recesses of the internet to see if Oppy's multiplier design are published. Thanks for this wonderful post!

p. said...

Thanks for the very detailed post. I always look forward to read your insider view from the MERs' operations.

@Buck: JPL's DESCANSO has a fantastic series of documents about the communication systems of deep space missions called "Design & Performance Summary".
The one about the rovers is located at:
There you'll find the information you're curious about.