Thursday, November 3, 2011

This is why we can't have nice things

Well, official title: "A Lesson In DSN View Periods")

(Secondary Title: "Where Has All My Time Gone Because I Have Like Seven Posts In the Queue and I Haven't Posted A Single One Yet")

My mother told me that I could always have nice things. In fact, she still gives me nice things, like giant jars of homemade salsa straight from Colorado. It keeps up the illusion that reality isn't cruel and that it will give you salsa if you call her every once in a while.

And, it turns out, we actually can have nice things. Take Opportunity, for example. She's a nice thing*. She is the proverbial cake. But we can't always eat the cake. This is for several reasons. First, because I'm always one to remind people that the sugar in the cake — proverbial or otherwise — is bad for your blood sugar and makes your insulin response weaker over time. This warning usually disappears into the conversational ether, because, hey, I'd ignore me, too. The second reason is that operating a spacecraft is hard. Very hard.**

It's all about resources, resources, resources. I like to ask people, "What do you think the most precious commodity is for a spacecraft?" The answers are the usual: "Power" (not incidentally, from the power engineer); "Heaters" (also not incidentally, from the thermal engineer); "attitude control thrusters" (also not incidentally, from the guidance and navigation engineer). I've even asked lay people the same questions. The responses are roughly the same.

I disagree with these "average" answers, and many of the folks on both the uplink and downlink sides of operations at JPL would, too. Nay, the most precious resource, we would say, is this:

Time with the Deep Space Network
As we'll come to see, the DSN is an over-subscribed resource, constantly untangling the web of requests of several dozen spacecraft, each with different requirement and desires and idiosyncrasies. I'm going to show some neat-o fun things about the DSN being able to "see" each piece of floating metal (and more) out there.

Learning hats, engaged.


Reader, DSN. DSN, reader.

Notice the nice placement of these complexes: They are each separated by about 120 degrees of longitude, giving us petty humans the ability to continuously watch one part of the sky. Just as an object is approaching the horizon of, say, an antenna at Goldstone, it's coming into view at Canberra. These so-called "view periods" overlap by a considerable amount of time. Each complex consists of several individual and independent antennas. Because each antenna at each complex is located in a different spot, these overlap periods are slightly different at every location. However, for our concerns, it's all the same for a given complex.

Right. So. Remember SPICE? Open-source geometric awesomeness? Well, SPICE is still our friend. I've got a slew of knowledge about the DSN that I can't share, but all the SPICE stuff? Yeah, fair game! It's great. So let's use it. You know, get the nasty icky mathy stuff out of the way. We'll stick to basic view parameters, since that's all that matters to first order. 

Now, who remembers the right-hand rule?

This provides a "coordinate frame" for the antenna. Once you've established a coordinate frame and where its center is in space (literally, in space!), you can use a number of different ways to express the position of other stuff. One way is a three-dimensional vector: "Stuff is this much along the A axis, that much along the B axis, and a little along the C axis". Another way, most common for an antenna fixed on the ground, is the azimuth-elevation-range parameter set. Say you're standing straight up and you see a star somewhere in the sky. You can know everything about its position relative to you if you know its azimuth, elevation, and range in your coordinate frame. 

Usually, azimuth is measured from the North; it's also known as the "right ascension". Elevation is how high the out of the "plane" made by the horizon the stuff is. Range is simply distance to the stuff. Easy. We'll be using these terms quite a bit here.

The reader can imagine that if there are a lots of things in the same place in the sky, things can get complicated for the DSN even though there are multiple antennas at each complex. So, just how crowded is it? Let's use azimuth and elevation to find out. 

Let's stick with just azimuth. We can get a rough idea of how the view angles of some certain objects vary over the course of the year. Remember that the DSN antennas are attached to the Earth, which is tilted relative to the plane created by the orbits of the planets. Also remember that this plane is only notional and ill-defined because the planets are all slightly out of this plane. Pluto's orbit (well, I guess I'll include small bodies!), not necessarily the tilt of the body itself, has a very significant tilt with respect to the orbits of the other planets. Mercury's is also very pronounced. However, those of the 8 primary planets are roughly the same, so it goes to say that if a particular antenna sees the same azimuth for several different bodies, the antenna sees close to the same elevation for these bodies, as well.

Note: I'm going to stick with planets and not be spacecraft specific. I do this mostly because we can get a good sense of the complexity by only knowing where planets are. That's where most of the spacecraft are anyways — and, really, if you want to know where any Mars rover or orbiter is, all you need to know is where Mars is. Anything else is in the details and you don't need that precision anyways. Cassini is basically at Saturn, MESSENGER is basically Mercury, etc. Planets are proxies for many of our spacecraft. Lastly, I'm keeping the list of objects to to watch for short because there are just too many deep space spacecraft to make sense of them all in one plot. Not enough colors or plot symbols.

Let's get an initial sense of things with a plot of azimuth from each DSN complex of all the inner planets through Saturn, but ignoring Jupiter and instead looking for the Juno spacecraft. Let's grab a point in time every hour, and only plot the azimuth if the elevation is greater than 0 — i.e., the object is visible in the first place.

Ew. Busy plot. Let's dissect it. By virtue of the Goldstone and Madrid stations being in the Northern Hemisphere, we would expect that their azimuthal plots of each major body to be roughly the same, offset only by the some "phasing" or spacing in time of a given body. For Canberra, we would expect something different. indeed, most of the bodies cross the 0-azimuth (or 360, same thing) point because they are in the northern sky for Canberra antennas. The azimuth of a body, if it's visible, "flips" over when it crosses 360/0, and that's the behavior we see there.

The major thing to glean is this: Things are going on, and they're going on all the time. When one thing falls out of visibility, another one jumps in. The antenna has to constantly jump back and forth between objects, depending on who wants what and when they want it and for how long they want it.

Well, really, that didn't tell us much. I can think of two instructive ways to tell the story a little better. We'll start with a time-lapse video of the position of the planets with respect to Earth. A while back, I made just a time-lapse: I plotted the positions of all the planets relative to Earth, but not fixed to a particular point on the earth (we call this an inertial frame), every few days for a few years. Our predictions of planetary orbits are quite good, so the positions are reliable to an extraordinary degree. (*rimshot*)

I'm only capturing direction here, and making the size of the arrows the same for convenience. See how the outer planets move much slower? Look at Mercury and Venus: they move so fast that they actually have a significant "going the other way" deal as they hit the other side of their orbit and move in the other direction relative to Earth. What we are looking for here are overlaps: When Mars and Saturn overlap, the DSN has to deal with talking to rovers and orbits of all different kinds at imperceptibly different parts of the sky for several hours. Sometimes, the overlap isn't so bad. Sometimes it is. Sometimes it's in between. The DSN has to deal with this constantly — and for other things in the sky not close to the planets, as well. 

From this video, the major idea to glean is this: The planets' relationship to each other is cyclic, sometimes working to your advantage and sometimes not. Mostly not. Especially mostly almost always not. Long-term planning folks like to use representations like this just to get a good idea of what their DSN "coverage" is going to look like when they, say, want to land a spacecraft on another planet. Because if you can't talk to your spacecraft (when it's most critical to do so) because of conflicts with 17 other missions, why try to? You plan around major conflicts like this. Everything can be just perfect about a mission, and a lack of DSN coverage can totally hose you.

The second instructive way to represent this is with a combined azimuth-elevation polar plot. If you've got one of those old-school GPS units, or even a new one that isn't meant for giving you driving instructions, you might notice a similar kind of plot: Azimuth going around in a circle with elevation represented as concentric circles, increase in value as you approach a view straight overhead. An object directly above you in your local-level coordinate frame will be in the middle of this kind of plot. Take this example: it's the azimuth-elevation plot for the aforementioned bodies from the view of DSS-63, a 70-meter dish at the Madrid complex. (Combined azimuth-elevation plots for antennas at Goldstone and Canberra tell us the same story.)

The red "+" is the first point at which the object is visible over the timeframe specified in the plot title. I have also appended the time of day (UTC, of course) at which this occurs, as this plot has no way of effectively conveying the direction or start of time.

What do we see? Mercury, Venus, and Saturn are all very close in the sky on this particular day (and they will continue to be so for quite some time…). Although their so-called "rise times" (when they first appear on the horizon) are separated by a few minutes, they are close enough to be a pain. 

When accounting for handovers between stations or complexes, having multi-antenna assignments for a single spacecraft, or any other number of a dozen things I can think of, we can easily see how crowded it gets. (And remember: I'm only counting planets as proxies for spacecraft, not even mentioning the ones not close to planets.) And, just because something is visible, doesn't mean it was assigned (or "allocated," as we say) the time with that antenna. 

Let that sink in. Let that soak into your brain. Think about all the things you don't know, all the things that make that DSN engine run. 


And that, folks, is why we can't have nice things. At least, not all the time. The DSN is a monumental achievement of humanity, providing not only communication but also precision navigation services to several dozen spacecraft whose voices are but faint whispers in a background of nothingness. To place Opportunity in this picture is easy, really: She's just another player, just another bidder in a vast landscape of missions with equally crucial requirements to check in on their birds, asking them a simple question:

Hey, are you ok? Let's chat.


*A billion-dollar nice thing, but I fail to see how that makes a difference.

**Once upon a time, my roommates and I decided to play a board game. I didn't know the game, so I asked if it was hard. One of them said, "You fly spaceships, and you're asking if the board game is hard?!"


Astro0 said...

That was one of the best written pieces I've ever read about the role and complexity of the work undertaken by the DSN. Believe me, it is incredible that we can send commands to these spacecraft so incredibly far away and an absolute miracle that we manage to get these signals back at all, through a Network that is at times, stretched to its limits and sometimes beyond. Thank you for this article.

Ed Davies said...

It might startle a few astronomers to hear that right ascension is another name for azimuth.

Matt Lenda said...

@Astro0: My pleasure.

@Ed: The source of endless debate, even within the engineering crowd!