If you’re fascinated by the idea of humans traveling through space and curious about how that all works, you’ve come to the right place.

“Houston We Have a Podcast” is the official podcast of the NASA Johnson Space Center from Houston, Texas, home for NASA’s astronauts and Mission Control Center. Listen to the brightest minds of America’s space agency – astronauts, engineers, scientists and program leaders – discuss exciting topics in engineering, science and technology, sharing their personal stories and expertise on every aspect of human spaceflight. Learn more about how the work being done will help send humans forward to the Moon and on to Mars in the Artemis program.

On Episode 183, Mark Fernandez, principal investigator for the Spaceborne Computer-2, details the experiment on the International Space Station that will further test the capabilities of in-space computing which may help humans explore deeper into space. This episode was recorded on January 29, 2021.

Transcript

Gary Jordan (Host): Houston, we have a podcast. Welcome to the official podcast of the NASA Johnson Space Center, Episode 183, “Spaceborne Computer.” I’m Gary Jordan and I’ll be your host today. On this podcast we bring in the experts, scientists, engineers and astronauts all to let you know what’s going on in the world of human spaceflight. Computing capabilities in space are quite reduced compared to what we have here on the ground. This creates challenges when transmitting data to and from space. Although relying on ground-based computers works for space exploration to the Moon or in low-Earth orbit. The solution will not work for exploration farther into the solar system, like to Mars. Launched in 2017, the Spaceborne Computer from Hewitt-Packard Enterprise serves as a proof of concept to test if commercial off the shelf computer systems can withstand the harsh environment of space and operate for a one-year mission. It successfully performed more than a trillion calculations or a teraflop per second for 207 days without requiring a reset. Now, launching on Northrop Grumman CRS-15 mission, Spaceborne Computer-2 builds upon the successes of this first study to bring a significantly advanced computing system to the space station. It dives deeper into exploring ways in which space exploration can be accelerated through the use of the commercial off the shelf computer systems we use here on Earth, without expensive, time consuming or bulky radiation shielding. The experiment tests additional techniques for recovering or mitigating errors in the extreme environment of space. Hewlett-Packard Enterprise is working with the space community and the ISS U.S. National Laboratory to demonstrate that current Earth-based data processing of space station experimental data can be performed aboard, during the mission of Spaceborne Computer-2. So, going into all the details of computing in space and this latest experiment is its Principal investigator and Solutions Architect for Converged Edge Systems at Hewlett-Packard Enterprise Mark Fernandez, Ph.D. So, let’s get right into it. Enjoy.

[Music]

Host: Mark Fernandez, thanks for coming on Houston We Have a Podcast today.

Mark Fernandez: Great to be here. Thanks for having me.

Host: Hey, your — we’re coming up on the launch here, soon. Northrop Grumman is going to be taking your experiment and I’m sure that the experiment that you and your team have worked on for a while to space. That’s coming up here real soon. How are you feeling?

Mark Fernandez: I am super, super excited. We’ve achieved many milestones to get there. And the first culmination of our efforts is going to be that launch scheduled in February.

Host: That’s right, and what we’re talking about today is something called the Spaceborne Computer. Mark, I’m going to be honest with you. This is a little outside of my expertise, computing. I’m going to rely on you to sort of navigate us through this topic today. And so, with that in mind I wanted to start kind of at a high level, just what we’re talking about today. What are some of the considerations, and some of the aspects of computing in space, that we should be thinking about before we even start talking about the Spaceborne Computer itself. Let’s — let’s start there. When you’re talking about computing in space what exactly is being computed in space? You know what — are those processes look like, give us the bird’s eye view of what we’re going to be talking about.

Mark Fernandez: Sure. So, for the most part, there’s not a lot of computing taking place in space. We really rely on our Earth-based computers in their research labs and their national labs etc., to do that. But Spaceborne got started by a conversation with one of our largest customers, NASA Ames, who has that as their mission to support space exploration with their super computers and such. And they indicated that when we got to the Moon, or when we got to Mars that the time delay between here and there would not allow them to do their mission. So, they ask us if we could take a small part of their computer and see if it would run in space. So, that’s what got us kicked off. If you seen the movie, “The Martian,” you’ll see in there that there’s a 24-minute delay between Earth and Mars, and another 24 minutes to return. So, you can see Matt Damon get very, very excited when he’s been sitting around at Mars waiting for the answer for almost an hour. Well, you’re going to need better than that if you’re going to live on Mars or the Moon. So, all of the things that we do here on Earth. If you dig deep enough, you’re going to find the computer in the background doing some work. You’re going to want to take that computer with you.

Host: So, Mark, when a little closer to home. Let’s — to the example of the International Space Station. You talk about the computing is not taking place in space. It’s taking place on Earth. So, what, what exactly is being transmitted down to Earth for computing?

Mark Fernandez: So, there are thousands of experiments have been run on International Space Station and they primarily collect that data and send it back down to Earth. For example, high resolution imagery of the polar ice caps. All sorts of medical data collected for the crew. All sorts of “telemetry” about the vehicle or about the space station are sent down to Earth for analysis etc. So, we’re talking with several payload developers about using Spaceborne Computer-2 and they include digital processing, signal processing, a lot of biological analysis in space. So, “hey is that mold that’s growing on my potato going to hurt me if I eat it?” Right. You wouldn’t want to wait an hour or more before you have dinner if you’re on Mars and be worried about what you’re eating. So, you may want to do that analysis in space.

Host: So, that’s the idea there, right? You have the — you have quicker communications when you’re closer to Earth. You can rely on those ground-based systems, but the real trick here is when you’re farther away you just don’t — some — in a lot of instances you just don’t want to wait for that answer. That answer is going to take some time to transmit to Earth and back. And you need that information quicker. So, you need the processing capabilities a little closer to where you are.

Mark Fernandez: Absolutely. Another critical one is for Spaceborne -1, we did a simulation of the Mars Lander and by all estimates the Mars Lander is going to be too large and complex for a human to fly it. So, it’s going to be under computer control. So, you’re going to want the computers to be able to control those thrusters as you make that first landing into Mars. And we help test that software and we were pretty successful with that and got some people excited about the possibility of bringing computers with you on these missions.

Host: Now, what are some of the challenges when it comes to taking this computing ability and putting it in the environment of space, what are those challenges that are presented to you when you’re designing these systems?

Mark Fernandez: Oh, that’s a very good question. We want to use the most modern readily available, state of the art, enterprise class systems. And so, one of the challenges, one of the mission requirements for Spaceborne-1 was can you package up what is meant to be mounted in a data center rack? And can you get that into a rocket? Will it fit? Will it survive a shake, rattle and roll of launch? Number two, can non-IT folks install this state-of-the-art IT gear into a non-standard rack onboard a spaceship? And number three, will it function? We had a lot of concerns about the power, the cooling, the networking and others had concerns about the radiation. So, we focused on the power, the cooling and the networking. And a little bit about the software and we were pretty successful. We were challenged with — we were challenged with addressing the concerns of radiation. And we said, there’s not going to be a way to do that easily. So, this is an unshielded, right off the factory floor equipment that we are doing our experiments with.

Host: See, that’s interesting. And what — I’m trying to think about some of the hurdles that come at you when you’re in space. Maybe radiation is one of them. So, the idea here is you really don’t need to do any kind of special modification for the space environment. You don’t need to design a system from the ground up. The — idea here is you can have this computing ability in space by just taking something right off the shelf, it doesn’t need to take those extra steps or maybe that’s exactly what you — the question that you want to ask as part of these experiments.

Mark Fernandez: Bingo. That’s one of the primary questions that we wanted to ask. So, we took modern state of the art hardware and explored all of the fully supported redundancy that’s already built into it. Redundant power supplies, redundant fans, check/recheck and other things that’s in technology today. We added a layer on top of that in the software to closely monitor the system and protect it when things went out of parameters. And we call that hardening with software. So, there’s no traditional — hardening in these systems, except for the hardening with software, which was brand new to HPE. And so, then we ran this in a redundant fashion, and we had a successful mission.

Host: So, what you’re talking about here when you’re talking about a mission is Spaceborne Computer-1, you’ve mentioned it a few times. What — were you around when that idea came up like, “hey, let’s take a computer off the shelf and put it in space.”

Mark Fernandez: Yes. It’d have – it started in 2014 with a casual conversation with NASA Ames and in 2017 we got going seriously and we launched, and we returned in 2019. We returned in 2019 in June is when our splashdown was. And there was so much success and so much enthusiasm, we were asked to mainly start working on Spaceborne-2. So, we splashed down in June, we got our user agreement signed in October. So that was a quick turnaround and we did the turnover to NASA in November. I sort of had a personal goal that we would achieve turnover within a year, which would be outstanding. But with all the COVID issues slowing things down, etc. I’m pretty happy that we’re able to do it in 13 months.

Host: Yeah, that’s an excellent accomplishment. It’s funny that you say, you know, so much enthusiasm, you know, I’m imagining like a lot of cheers and say, “Hey, let’s do it again.” But I’m wondering, you know, you talk about after splashdown. I wonder if there was a period you mentioned that this thing returned, right? It did splashdown back on Earth. It didn’t burn up like in the atmosphere like some cargo ships do. So, what were you doing on the ground? Any further analysis after the mission was completed in space? Anything else you were doing to — extract any last or squeeze any more data out of this experiment?

Mark Fernandez: Oh yeah, a couple of good points there. You mentioned burn up in the atmosphere. We had part of our mission planned. If Spaceborne Computer did not function at all, if it was a total failure. We had permission to go ahead and put it in the trash can and let it go down and burn up. But that was not the case. We ran successfully for 658 days. And we splashed down. Around that time, we were awarded the exceptional technology achievement award by NASA. So, there was a lot of excitement about that and we got to go to that presentation; that was a lot of fun. We had thousands of ideas of what we want to do in Spaceborne Computer-2 and the engineers worked really hard to narrow that down to something that we could do, and what we thought would be under a year. So, I’m pretty excited with what we ended up with and I’m looking forward to Spaceborne-2.

Host: Yeah, Mark, I think you’re explaining it so well, right. Spaceborne Computer-1 and you explained this again very well. You mentioned — you mentioned, you know, the experiment was let’s see if this works. You know, let’s see if it survives launch. Let’s see if it works in space. Let’s see if we can return it. And it sounds like mission success and that’s thanks to a lot of the folks that went into creating this experiment and making it such a success. Now, leading that into Spaceborne-2, right. That’s what we’re going to be talking about today. This is the experiment that’s going up here very soon. Tell me about this experiment. What else is being added on to an already successful mission on one?

Mark Fernandez:So, I’m super, super excited about it. In retrospect, Spaceborne-1 is, was a proof of concept, just as you explained. Can we do it? And were successful? Yes, we were. In our user agreement that was signed just a few months after return to Earth, it actually states that we are to support the space community with onboard processing, outreach and education. So, Spaceborne-2 is not for us. It’s not for Hewlett-Packard Enterprise. It is for the space community. To test out it’s computing, and edge computing is taken over the world and important. Along those lines we are shipping up with Spaceborne-2, a member of the HPE Edge-line computer servers, it is designed for the edge. It has a little bit more ruggedization already built into it for hazardous environments here on Earth. We’re continuing to layer on our hardening with software on top of that. But what’s exciting is we’ve also included a [graphics processing unit] GPU for artificial intelligence and machine learning capabilities. We’ve already got one payload developer that has performed his pre-experiment work before we turned over, taking advantage of that GPU and seeing incredible turnaround times. So, I like to — I like to tell people that the purpose of exploration, space exploration and other, but in general, the purpose of exploration is not data collection. It’s insight. And with these processing capabilities at the edge, where the data is being generated or collected, you can turn that sample to insight time around as fast as possible. And so, we can get that insight back down to humanity much faster than we can get that data back down.

Host: Mark, I think you explained it very well, but I just want to make sure that I understand this. Edge computing, when you talk about this is taking over the world. It sounds like it’s something, you know, computing within the local environment. Maybe I’m oversimplifying, but can you give us sort of an explanation on what this thing is that’s taken over the world, edge computing?

Mark Fernandez: Yes, so it’s computing the far outreaches of your warehouse. It’s computing on your oil rig, on your high-speed train, on your aircraft, in the depths of your mind. So, instead of having just censors out there at the edge, and having to worry about a reliable way to get that data back to the home office, back to the core, if you would, then you process at the edge and you get the go, no go, safe, unsafe, up, down answers that you’re interested in. And those are returned to the core, to the home office for further action if needed. And we can do the similar thing with Spaceborne-2. We plan on, for example, one of the examples that’s been brought up to us that may be of use is a lightning study analysis. And those scientists working on that are really focused on the camera. And capturing those lightning images as best they can. And in the back of their mind they’re not thinking about how to get that data back down to Earth to process. “Oh, we’ll figure that out later.” Well, it’s a lot of data and it takes a lot of time to get it back down to Earth. Whereas they told us, I really just need to see that the 10 seconds of the video before the lightning strike, and the 10 seconds after it. And I’ve got this software that runs, and it analyzes film, and it can pick out via artificial intelligence machine learning, it can pick out a lightning strike. And then we snip it plus and minus 10 seconds here on Earth and that’s what we look at. And I said, well I can do that onboard with you, and then A, I will send you that little snippet down and I’ll send you a little email say, “time to go to work.” A similar one had to do with observing the polar ice caps. And it turns out that the instrument can’t give you accurate data if there’s cloud cover, but they have processing routines that know how to look at the video and determine if it’s good data or not. So, rather than sending all the data back down to Earth, we might be able to help them send down only the data where they can actually see the polar ice caps and make significant efforts towards their research.

Host: Yeah, what you’re saying their Mark is the data that the scientists are getting for analysis is more efficient. They don’t have to sift through the stuff that they don’t really care about. The artificial intelligence, the software within you know, that — I guess closer to home, right. The local environment is doing that for them. So, they don’t have to put that effort in.

Mark Fernandez: You got it. That’s correct.

Host: All right, I think I’m — on the edge of computing myself, Mark, I think. I’m getting more knowledgeable about this.

Mark Fernandez: So, the low-hanging fruit, if you would, is OK, Mr. Scientist what are you going to do with this raw data when you get it? And they say, “well we’ll run the so and so routine on our Linux servers.” And I say, “well, I’ve got a Linux server in space, why don’t I run your code in space and then A, you’ll have less data to look at when it arrives to you on Earth. It will already be pre-qualified, so you don’t spend or waste any of your time looking at data that’s of no interest. And you’re ready to go to the next step.” And we were actually having a conversation with a payload developer and we had just completed that step. We’d done what he referred to as the pre-processing of the data. And got it down from hundreds of gigabytes, down to tens of megabytes, which is a significant reduction in data that he could get to Earth in a much faster time. And I said to him, half-jokingly, “OK, what do you do with this next?” He said, “we’ve got an industry standard routine that we run so that scientists throughout the world end up with a plot that they all recognize, and all understand.” And I said, “OK, so you take this pre-processed data, and you generate a plot?” “Yes.” I said, “well you want me to do that for you?” And that’s when the light bulb went off. So, we can now generate the plot in space and get it to him in seconds, instead of the hours or days or weeks that it took for him to get the raw data that he would then have to run through his routines to end up with the plot ready for publication and analysis.

Host: And so, to clarify, is all of this — what you’re doing is you’re working with different scientists and who are trying to analyze data and are you saying to them, I can work with — take your payload and run it through the Spaceborne Computer-2 and that’s a proof of concept that we process this data, is that’s what’s happening or is this something else?

Mark Fernandez: You got it, absolutely.

Host: Very cool.

Mark Fernandez: We want to have thousands of proofs of concepts so, that onboard data processing can be shown to seriously benefit the scientists and engineers back here on Earth.

Host: See, I’m imagining myself as scientist and engineer and how much time it takes to do this kind of research. Say, I’m doing this lightning research and I have to spend a lot of my time sifting through the right data to find the right parts of a lightning strike 10 seconds before, 10 seconds after. Now, instead of spending that time sifting through the data to find the important parts, I’m just sifting through the important parts and I can spend more time doing that, which likely means I can do my research a little bit better and more efficiently. This is quite a capability that you’re presenting here, Mark.

Mark Fernandez: Absolutely. You’ve summarized it wonderfully. So again, the purpose of exploration is not data collection. It’s insight and I want to get our brilliant minds throughout the world working on the insight rather than the number crunching.

Host: So, it seems like a lot of part — a lot of this experiment, Spaceborne Computer-2. A little bit of it is a proof of concept of the computing capability, right. But it sounds like you are reaching out to different payload developers, different scientists, principal investigators and you are trying to say, “hey, we can work together.” You can do your experiment and I can do my experiment and you’re trying to gather up, you know like you said, 1,000 different examples or use cases.

Mark Fernandez: Mm-hmm, correct. And I — we’ve got a presentation. We’ve got a standard operating procedure. I’m not going to change anything that you don’t already do Mr. Scientist. I’m just going to do it near the data where it’s generated. So, you give me your code, I will run your code, and I will give you the answers. Just as you handed to your intern or your co-worker to run that same code to get you the answers, we’ll do it for you on your behalf.

Host: So, aside from processing the data from multiple different principal investigators and other research that’s onboard station, what are some of your other hopes for Spaceborne Computer-2? Any other technology demonstrations, radiation shielding, any other thing that’s onboard?

Mark Fernandez: The depth of the potential is pretty deep. We’ve got image processing, signal processing, healthcare aspects, DNA analysis. I’m just seriously excited about the potential that we can get and I’m hoping, there’s a – this is a bad expression to use, a ground swell of success. I need a space swell of success across many of these sciences, realizing what we can do with your instrument on location. So —

Host: So, take us through the process here. It’s going to launch on Northrop Grumman, what’s the — what’s the next couple of steps once it gets onboard to get installed and hooked up and ready to go?

Mark Fernandez: Yes, we’ve already got our procedures squared away for it to be installed and powered up and cooled, and we’ll run through our health checks, and then we’ll be ready to begin proof of concept. We’ve got a website, HPE.com/info/spaceborne, if you’d like some information about Spaceborne. And on that website, there will be the place where you can request to use Spaceborne Computer-2 as one of your research opportunities. So, we’re opening this up and it’s not officially a solicitation, it’s an opportunity. And we will evaluate these opportunities based on the benefits that they provide to you Mr. Scientist, your organization, NASA, space exploration as a whole and most importantly, humanity as a whole. And we will rank those and then bring you under wing and get you running on Spaceborne Computer-2.

Host: So, it sounds like the crew is going to hook this up and then really, it’s in your hands to do that, to filter through some of those requests and get as many examples, so that you can be successful with your experiment. How long is the runway here Mark, that — you’re looking to have Spaceborne Computer-2 onboard?

Mark Fernandez: That’s the other part of the Spaceborne Computer-1 success, with Spaceborne Computer-2 they asked us to send up twice the compute capabilities and extend the mission 2X. So, our mission is nominally two to three years, which lines up with the first link of the first mission to Mars link is going to be two to three years. So, we will determine if our hardware can last two to three years. It successfully lasted a year and a half the first time around. And we’ll have plenty of opportunity, plenty of cycles to share with the space community to show that this edge processing can advance their science and engineering.

Host: How about that. Now, Mark you have a lot of excitement about this experiment and it’s definitely coming through and it’s coming up here soon, I wonder if you have further ambitions. We’re talking low-Earth orbit right now. But even in the beginning of our chat today you were discussing looking towards exploration and thinking about a Spaceborne Computer on a mission to Mars to do that edge computing millions of miles away. What are your ambitions going forward?

Mark Fernandez: That’s it. We’re supporting the NASA and ISS National lab missions, space exploration. We have a natural desire to have a lunar presence and continue on to Mars. We want to be there to support that.

Host: That’s awesome, Mark. You know, I feel like I have a good understanding of your goals and ambitions here, Spaceborne Computer-2. There’s going to be a lot of — a lot to learn and a lot of data that’s going to flow through this computer. I wonder if I missed anything before, we call it a day.

Mark Fernandez: Spaceborne Computer-2 is a collaboration between Hewlett-Packard Enterprise and the ISS National Lab. We will jointly evaluate the research opportunities and determine which ones we will function on Spaceborne Computer-2. We’re looking to have commercial entities, as well as the traditional NASA and NASA agencies going beyond that to other government agencies as well. We have an aspirational desire to include an educational, outreach opportunity. Again, that’s aspirational and we hope to squeeze that in.

Host: And that’s one of the great things about the International Space Station as a platform to do stuff exactly like this, to have further development, technology development, to allow commercial industry to come onboard and test some of these things out, so that we can all accomplish this mission going to the Moon and Mars together. Very — interesting stuff and a cool world to be a part of. Mark Fernandez really appreciate your time. This has been fascinating to learn about edge computing. I think I’m a little bit smarter on the subject. Still not quite to your level, but I think I have a good grasp and I’m excited for you and all the teams that put together the Spaceborne Computer-2. Thanks for coming on Houston We Have a Podcast today, I really appreciate it.

Mark Fernandez: Thanks for having me. It’s been a great conversation.

[ Music]

Host: Hey, thanks for sticking around. Really fascinating conversation we had with Mark Fernandez today about the Spaceborne Computer-2 launching soon on the Northrop Grumman CRS-15 mission. I hope you learned something today. You can check out more of our episodes at Houston We Have a Podcast at NASA.gov/podcasts as well as the many other shows we have across all of the space agency. You can check us out at the Johnson Space Center pages of Facebook, Twitter, Instagram and talk to us using the hashtag #AskNASA on your favorite platform. You can submit an idea for the show or ask a question. Just make sure to mention it’s for us at Houston We Have a Podcast. This episode was recorded on January 29, 2021. Thanks to Alex Perryman, Pat Ryan, Norah Moran, Belinda Pulido, Jennifer Hernandez, Nicole Rose and the International Space Station Program Research Office. And the support of Hewlett-Packard Enterprise team. Thanks again to Mark Fernandez for taking the time to come on the show. Give us a rating and feedback on whatever platform you’re listening to us on and tell us what you think of our podcast. We’ll be back next week.