WEBVTT

00:00:00.267 --> 00:00:03.437
[MUSIC]

00:00:03.437 --> 00:00:07.160
GRETCHEN HUIZINGA: Welcome to Abstracts, 
a Microsoft Research Podcast that puts the  

00:00:07.160 --> 00:00:13.048
spotlight on world-class research in brief. 
I’m Dr. Gretchen Huizinga.

00:00:14.882 --> 00:00:19.484
 In this series, members of the research community at Microsoft give us a quick snapshot—

00:00:19.484 --> 00:00:23.432
or a podcast abstract—of their 
new and noteworthy papers.

00:00:23.432 --> 00:00:24.276
[MUSIC FADES]

00:00:25.000 --> 00:00:29.400
I’m talking today to Tusher Chakraborty, 
a senior research software engineer at  

00:00:29.400 --> 00:00:35.520
Microsoft Research. Tusher is coauthor of a 
paper called “Spectrumize: Spectrum-efficient  

00:00:35.520 --> 00:00:40.700
Satellite Networks for the Internet of Things.” 
Tusher, thanks for joining us on Abstracts!

00:00:40.700 --> 00:00:44.600
TUSHER CHAKRABORTY: Hi. Thank you for 
having me here, Gretchen, today. Thank you.

00:00:44.600 --> 00:00:49.560
HUIZINGA: So because this show is all 
about abstracts, in just a few sentences,  

00:00:49.560 --> 00:00:54.060
tell us about the problem your paper 
addresses and why we should care about it.

00:00:54.060 --> 00:01:01.080
CHAKRABORTY: Yeah, so think of, I’m a farmer 
living in a remote area and bought a sensor  

00:01:01.080 --> 00:01:07.720
to monitor the soil quality of my farm. The 
big headache for me would be how to connect  

00:01:07.720 --> 00:01:14.120
the sensor so that I can get access to the sensor 
data from anywhere. We all know that connectivity  

00:01:14.120 --> 00:01:20.600
is a major bottleneck in remote areas. Now, 
what if, as a farmer, I could just click the  

00:01:20.600 --> 00:01:27.024
power button of the sensor, and it gets connected 
from anywhere in the world. It’s pretty amazing, right? 

00:01:27.024 --> 00:01:32.760
And that’s what our research is all 
about. Get your sensor devices connected from  

00:01:32.760 --> 00:01:39.106
anywhere in the world with just the click of power 
button. We call it one-click connectivity. 

00:01:39.106 --> 00:01:44.840
Now, you might be wondering, what’s 
the secret sauce? It’s not magic;  

00:01:44.840 --> 00:01:50.360
it’s direct-to-satellite connectivity. So 
these sensors directly get connected to the  

00:01:50.360 --> 00:01:57.440
satellites overhead from anywhere on Earth. The 
satellites, which are orbiting around the earth,  

00:01:57.440 --> 00:02:03.400
collect the data from the sensing devices 
and forward to the ground stations in some  

00:02:03.400 --> 00:02:08.820
other convenient parts of the world where these 
ground stations are connected to the internet.

00:02:08.820 --> 00:02:13.640
HUIZINGA: So, Tusher, tell us what’s 
been tried before to address these  

00:02:13.640 --> 00:02:18.640
issues and how your approach contributes to 
the literature and moves the science forward.

00:02:18.640 --> 00:02:25.760
CHAKRABORTY: So satellite connectivity is 
nothing new and has been there for long. However,  

00:02:25.760 --> 00:02:33.600
what sets us apart is our focus on democratizing 
space connectivity, making it affordable for  

00:02:33.600 --> 00:02:39.720
everyone on the planet. So we are talking 
about the satellites that are at least 10 to 20  

00:02:39.720 --> 00:02:46.280
times cheaper and smaller than state-of-the-art 
satellites. So naturally, this ambitious vision  

00:02:46.280 --> 00:02:53.760
comes with its own set of challenges. So when 
you try to make something cheaper and smaller,  

00:02:53.760 --> 00:03:00.160
you’ll face lots of challenges that all 
these big satellites are not facing. So  

00:03:00.160 --> 00:03:07.800
if I just go a bit technical, think of the 
antenna. So these big satellite antennas,  

00:03:07.800 --> 00:03:13.680
they can actually focus on particular part 
of the world. So this is something called  

00:03:13.680 --> 00:03:20.120
beamforming. On the other hand, when we try 
to make the satellites cheaper and smaller,  

00:03:20.120 --> 00:03:25.840
we can’t have that luxury. We can’t have 
beamforming capability. So what happens, they  

00:03:25.840 --> 00:03:32.040
have omnidirectional antenna. So it seems like … 
you can’t focus on a particular part of the earth  

00:03:32.040 --> 00:03:38.640
rather than you create a huge footprint on all 
over the earth. So this is one of the challenges  

00:03:38.640 --> 00:03:44.360
that you don’t face in the state-of-the-art 
satellites. And we try to solve these  

00:03:44.360 --> 00:03:50.040
challenges because we want to make connectivity 
affordable with cheaper and smaller satellites.

00:03:50.040 --> 00:03:56.840
HUIZINGA: Right. So as you’re describing this, 
it sounds like this is a universal problem,  

00:03:56.840 --> 00:04:04.360
and people have obviously tried to make things 
smaller and more affordable in the past. How  

00:04:04.360 --> 00:04:08.280
is yours different? What methodology 
did you use to resolve the problems,  

00:04:08.280 --> 00:04:10.120
and how did you conduct the research?

00:04:10.120 --> 00:04:16.760
CHAKRABORTY: OK, I’m thrilled that you asked this 
one because the research methodology was the most  

00:04:16.760 --> 00:04:22.640
exciting part for me here. As a part of this 
research, we launched a satellite in a joint  

00:04:22.640 --> 00:04:29.440
effort with a satellite company. Like, this is 
very awesome! So it was a hands-on experience  

00:04:29.440 --> 00:04:35.800
with a real-deal satellite system. It was 
not simulation-based system. The main goal  

00:04:35.800 --> 00:04:42.200
here was to learn the challenge from a real-world 
experience and come up with innovative solutions;  

00:04:42.200 --> 00:04:48.080
at the same time, evaluate the solutions in real 
world. So it was all about learning by doing,  

00:04:48.080 --> 00:04:55.000
and let me tell you, it was quite the ride! 
[LAUGHTER] We didn’t do anything new when we  

00:04:55.000 --> 00:05:01.360
launched the satellites. We just tried to see 
how industry today does this. We want to learn  

00:05:01.360 --> 00:05:07.000
from them, hey, what’s the industry practice? 
We launched a satellite. And then we faced a  

00:05:07.000 --> 00:05:13.960
lot of problems that today’s industry is facing. 
And from there, we learned, hey, like, you know,  

00:05:13.960 --> 00:05:19.720
this problem is industry facing; let’s go after 
this, and let’s solve this. And then we tried  

00:05:19.720 --> 00:05:27.520
to come up with the solutions based on those 
problems. And this was our approach. We didn’t  

00:05:27.520 --> 00:05:35.320
want to assume something beforehand. We want to 
learn from how industry is going today and help  

00:05:35.320 --> 00:05:39.180
them. Like, hey, these are the problems you 
are facing, and we are here to help you out.

00:05:39.180 --> 00:05:42.760
HUIZINGA: All right, so assuming 
you learned something and wanted  

00:05:42.760 --> 00:05:45.040
to pass it along, what were your major findings?

00:05:45.040 --> 00:05:51.040
CHAKRABORTY: OK, that’s a very good question. So 
I was talking about the challenges towards this  

00:05:51.040 --> 00:05:57.360
democratization earlier, right? So one of the 
most pressing challenges: shortage of spectrum.  

00:05:57.360 --> 00:06:03.960
So let me try to explain this from the high 
level. So we need hundreds of these satellites,  

00:06:03.960 --> 00:06:10.240
hundreds of these small satellites, to provide 
24-7 connectivity for millions of devices around  

00:06:10.240 --> 00:06:17.280
the earth. Now, I was talking, the footprint of 
a satellite on Earth can easily cover a massive  

00:06:17.280 --> 00:06:24.520
area, somewhat similar to the size of California. 
So now with this large footprint, a satellite can  

00:06:24.520 --> 00:06:29.880
talk with thousands of devices on Earth. You 
can just imagine, right? And at the same time,  

00:06:29.880 --> 00:06:34.760
a device on Earth can talk with multiple 
satellites because we are talking about hundreds  

00:06:34.760 --> 00:06:44.240
of these satellites. Now, things get tricky here. 
[LAUGHTER] We need to make sure that when a device  

00:06:44.240 --> 00:06:51.080
and a satellite are talking, another nearby device 
or a satellite doesn’t interfere. Otherwise, there  

00:06:51.080 --> 00:06:58.880
will be chaos—no one hearing others properly. 
So when we were talking about this device and  

00:06:58.880 --> 00:07:06.400
satellite chat, right, so what is that all about? 
This, all about in terms of communication, is  

00:07:06.400 --> 00:07:12.240
packet exchange. So the device sends some packet 
to the satellite; satellite sends some packet to  

00:07:12.240 --> 00:07:18.880
the device—it’s all about packet exchange. Now, 
you can think of, if multiple of these devices are  

00:07:18.880 --> 00:07:25.200
talking with a satellite or multiple satellites 
are talking with a device, there will be a  

00:07:25.200 --> 00:07:31.360
collision in this packet exchange if you try to 
send the packets at the same time. And if you do  

00:07:31.360 --> 00:07:37.920
that, then your packet will be collided, and you 
won’t be able to get any packet on the receiver  

00:07:37.920 --> 00:07:46.680
end. So what we do, we try to send this packet on 
different frequencies. It’s like a different sound  

00:07:46.680 --> 00:07:54.560
or different tone so that they don’t collide with 
each other. And, like, now, I said that you need  

00:07:54.560 --> 00:08:00.360
different frequencies, but frequency is naturally 
limited. And the choice of frequency is even  

00:08:00.360 --> 00:08:06.720
limited. This is very expensive. But if you have 
limited frequency and you want to resolve this  

00:08:06.720 --> 00:08:13.400
collision, then you have a problem here. How do 
you do that? So we solve this problem by smartly  

00:08:13.400 --> 00:08:20.360
looking at an artifact of these satellites. So 
these satellites are moving really fast around the  

00:08:20.360 --> 00:08:28.520
earth. So when they are moving very fast around 
the earth, they create a unique signature on the  

00:08:28.520 --> 00:08:35.920
frequency that they are using to talk with 
the devices on Earth. And we use this unique  

00:08:35.920 --> 00:08:42.440
signature, and in physics, this unique signature 
is known as Doppler signature. And now you don’t  

00:08:42.440 --> 00:08:51.240
need a separate frequency to sound them different, 
to have packets on different frequencies. You  

00:08:51.240 --> 00:08:57.360
just need to recognize that unique signature to 
distinguish between satellites and distinguish  

00:08:57.360 --> 00:09:04.280
between their communications and packets. So in 
that sense, there won’t be any packet collision.  

00:09:04.280 --> 00:09:11.200
And this is all about our findings. So with 
this, now multiple devices and satellites can  

00:09:11.200 --> 00:09:17.987
talk with each other at the same time without 
interference but using the same frequency.

00:09:17.987 --> 00:09:23.160
HUIZINGA: It sounds, like, very similar to 
a big room filled with a lot of people. Each  

00:09:23.160 --> 00:09:29.520
person has their own voice, but in the mix, you, 
kind of, lose track of who’s talking and then  

00:09:29.520 --> 00:09:34.200
you want to, kind of, tune in to that specific 
voice and say, that’s the one I’m listening to.

00:09:34.200 --> 00:09:38.320
CHAKRABORTY: Yeah, I think you picked up 
the correct metaphor here! This is the  

00:09:38.320 --> 00:09:44.080
scenario you can try to explain here. So, 
yeah, like what we are essentially doing,  

00:09:44.080 --> 00:09:50.120
like, if you just, in a room full of people 
and they are trying to talk with each other,  

00:09:50.120 --> 00:09:57.726
and then if they’re using the same tone, no one 
will be distinguished one person from another.

00:09:57.726 --> 00:09:57.747
HUIZINGA: Right ...

00:09:57.747 --> 00:10:01.400
CHAKRABORTY: Everyone will sound 
same and that will be colliding.  

00:10:01.400 --> 00:10:05.725
So you need to make sure that, how 
you can differentiate the tones …

00:10:05.725 --> 00:10:05.743
HUIZINGA: Yeah …

00:10:05.743 --> 00:10:12.120
CHAKRABORTY: … and the satellites differentiate 
their tones due to their fast movement. And  

00:10:12.120 --> 00:10:18.420
we use our methodology to recognize that 
tone, which satellite is sending that tone.

00:10:18.420 --> 00:10:24.800
HUIZINGA: So you sent up the experimental 
satellite to figure out what’s happening.  

00:10:24.800 --> 00:10:28.060
Have you since tested it to see if it works?

00:10:28.060 --> 00:10:33.334
CHAKRABORTY: Yeah, yeah, so we have tried it out, 
because this is a software solution, to be honest.

00:10:33.334 --> 00:10:33.347
HUIZINGA: Ah.

00:10:33.347 --> 00:10:37.280
CHAKRABORTY: As I was talking about, there 
is no hardware modification required at this  

00:10:37.280 --> 00:10:44.080
point. So what we did, we just implemented 
this software in the ground stations,  

00:10:44.080 --> 00:10:50.700
and then we tried to recognize which satellite 
is creating which sort of signature. That’s it!

00:10:50.700 --> 00:10:52.840
HUIZINGA: Well, it seems like this research would  

00:10:52.840 --> 00:10:58.420
have some solid real-world impact. So 
who would you say it helps most and how?

00:10:58.420 --> 00:11:04.440
CHAKRABORTY: OK, that’s a very good one. 
So the majority of the earth still doesn’t  

00:11:04.440 --> 00:11:08.920
have affordable connectivity. The lack 
of connectivity throws a big challenge  

00:11:08.920 --> 00:11:15.400
to critical industries such as agriculture—the 
example that I gave—energy, and supply chain,  

00:11:15.400 --> 00:11:21.480
so hindering their ability to thrive 
and innovate. So our vision is clear:  

00:11:21.480 --> 00:11:29.960
to bring 24-7 connectivity for devices anywhere on 
Earth with just a click of power button. Moreover,  

00:11:29.960 --> 00:11:35.440
affordability at the heart of our mission, 
ensuring that this connectivity is accessible  

00:11:35.440 --> 00:11:43.720
to all. So in core, our efforts are geared 
towards empowering individuals and industries  

00:11:43.720 --> 00:11:48.920
to unlock their full potential in 
an increasingly connected world.

00:11:48.920 --> 00:11:51.360
HUIZINGA: If there was one 
thing you want our listeners  

00:11:51.360 --> 00:11:54.040
to take away from this research, what would it be?

00:11:54.040 --> 00:11:59.960
CHAKRABORTY: OK, if there is one thing 
I want you to take away from our work,  

00:11:59.960 --> 00:12:05.080
it’s this: connectivity shouldn’t be 
a luxury; it’s a necessity. Whether  

00:12:05.080 --> 00:12:11.400
you are a farmer in a remote village or a 
business owner in a city, access to reliable,  

00:12:11.400 --> 00:12:18.400
affordable connectivity can transform your life 
and empower your endeavors. So our mission is  

00:12:18.400 --> 00:12:24.760
to bring 24-7 connectivity to every corner 
of the globe with just a click of a button.

00:12:24.760 --> 00:12:28.320
HUIZINGA: I like also how you 
say every corner of the globe,  

00:12:28.320 --> 00:12:37.560
and I’m picturing a square! [LAUGHTER] 
OK, last question. Tusher, what’s next for  

00:12:37.560 --> 00:12:42.440
research on satellite networks and Internet 
of Things? What big unanswered questions  

00:12:42.440 --> 00:12:47.320
or unsolved problems remain in the field, 
and what are you planning to do about it?

00:12:47.320 --> 00:12:52.480
CHAKRABORTY: Uh … where do I even begin? 
[LAUGHTER] Like, there are countless unanswered  

00:12:52.480 --> 00:12:59.360
questions and unsolved problems in this field. 
But let me highlight one that we talked here:  

00:12:59.360 --> 00:13:08.520
limited spectrum. So as our space network 
expands, so does our need for spectrum. But  

00:13:08.520 --> 00:13:13.720
what’s the tricky part here? Just throw 
more and more spectrum. The problem is  

00:13:13.720 --> 00:13:19.480
the chunk of spectrum that’s perfect for 
satellite communication is often already  

00:13:19.480 --> 00:13:26.440
in use by the terrestrial networks. Now, a 
hard research question would be how we can  

00:13:26.440 --> 00:13:34.480
make sure that the terrestrial and the satellite 
networks coexist in the same spectrum without  

00:13:34.480 --> 00:13:40.920
interfering [with] each other. It’s a tough nut 
to crack, but it’s a challenge we are excited  

00:13:40.920 --> 00:13:48.342
to tackle head-on as we continue to push the 
boundaries of research in this exciting field.

00:13:48.342 --> 00:13:48.350
[MUSIC]

00:13:48.350 --> 00:13:52.280
HUIZINGA: Tusher Chakraborty, thanks for 
joining us today, and to our listeners,  

00:13:52.280 --> 00:13:59.040
thanks for tuning in. If you want to read this 
paper, you can find a link at aka.ms/abstracts.  

00:13:59.680 --> 00:14:05.880
You can also read it on the Networked Systems 
Design and Implementation, or NSDI, website, and  

00:14:05.880 --> 00:14:25.360
you can hear more about it at the NSDI conference 
this week. See you next time on Abstracts!

00:14:25.360 --> 00:14:26.285
[MUSIC FADES]