Document 17859585

advertisement
>> Sharad Agarwal: So as you probably saw in the e-mail, Victor
recently got the Outstanding Contribution Award from ACM SIGMOBILE.
This is sort of equivalent to a lifetime achievement award. It's a
pretty big deal. So this was awarded to him at ACM MobiCom which
happened just a few weeks ago in Florida, in Miami. And it's a very,
very nice globe. I recommend you all stop by Victor's office and pick
it up and have a look at it. It's really beautiful.
>>: Or get their own.
>> Sharad Agarwal: Or get their own, yes, yes, absolutely. Here's a
list of the past recipients of the OCA from SIGMOBILE. You will notice
a bunch of people that you know in there.
>>: Very young crowd.
>> Sharad Agarwal: Very young crowd, as Stefan points out. Victor got
the award for pioneering contributions to wireless Internet broadband
technologies and for inspirational leadership of the mobile computing
community. There was some discussion about whether that pioneering
word should be impactful or pioneering. I think in the end pioneering
is what is on his award.
Now, you all know Victor quite well. Here's his background. He got
his Ph.D. from UMASS Amherst and he worked at this company called Deck
which apparently most of our interns don't know what it was.
[laughter] We're hoping most of you do. And then, of course, you know
his history here in MSR. And he's been very active in the SIGMOBILE
community. He, of course, is a founder of SIGMOBILE as well as a bunch
of conferences that are part of SIGMOBILE or affiliated with SIGMOBILE.
And he's been on a variety of steering committee positions as well.
His main contributions have been in broadband wireless. In particular,
in wide space networking and wireless LANs and mesh networking. There
are a lot of details we can find, of course, on his Web page here, a
bunch of highlights on some of the work. And of course Victor's going
to go into some detail about his contributions here and how those came
about. And, of course, he's collected a large number of awards over
the years, and there's a long, long list of them. Here's some of the
main highlights. And I think you know quite a lot of these from all
the press that he's gotten from them.
He's also gotten lots of accolades. These are descriptions of some of
the words that people have used to describe either Victor or his
contributions. So I'd like to present to you Victor, and he will give
his talk now.
>> Victor Bahl: Thanks. Okay. As I was, as you know this talk was
first given at MobiCom. And it was given on October 2nd, and that's
significant, as you will see as I go through this talk.
So I was thinking about what to say to this audience. I thought about
whether to just do sort of a future talk or just do some technical
talk. In the end I decided in many of my sort of travels and when I
talk to people, they generally ask me how do you select your problems,
and how do you move forward and how do you execute on it. So I thought
maybe that's where I'll focus on today.
So the topic is me and my research. And I joined Microsoft in June of
1997. I was newly minted Ph.D.. I had been working in Deck for
several years, and both Rich and Rich recruited me to start up the
networking research group. And they promised me a land of flexibility
and openness and they've delivered.
So when I looked at MSR it was fairly intimidating. There was people
like Gordon Bell. Chuck Packer, [indiscernible] all these award
winners, and they were collecting these amazing researchers. So it was
a little bit intimidating, but also very exciting. So as I was
thinking through what to do here, I met [indiscernible] and recently
left MSR, but he's a great guy. He's actually also sort of god of
graphics. And he gave me this article by Hamming. And many of you may
have read the article and the article said you and your research. And
I see some shaking your heads. People have not read this article or
not read it I would think. Great, this audience -- if you haven't read
it you should read it. This is actually a transcript of a colloquium
that he gave at Bell Labs in which he was asked -- he talked about why
do some scientists make it big and some don't. And what in particular
do they have that causes them to win or do great things? And he lists
out very nicely all the different things. But one thing that struck me
when I read the article was this thing about courage, about having the
courage to go and ask the questions and do things which others may not
do. May be fearful of.
So this talk is then about a young systems researcher at the time
wanting to be courageous. It's about wanting to have impact and then
potentially sharing some of the things that I've learned over the
course of my career, over a certain time.
Now, in order to do this, I do have to leave some modesty aside,
because I will be talking about things that worked and I ask your
forgiveness for that. So let's start with how we started. So when I
joined in '97, the state of wireless networking was not like it is
today, of course. There was no Wi-Fi. There was no standards. The
IEEE Rule 11 standard was just coming about. There were very few
companies, no Aruba or Broadcom or Symbol. Or Symbol was selling
something. But all the companies you know didn't exist at the time.
And then most of the wireless networking were done with these cards.
In fact, if you come to my office I have some of these still in my file
cabinet that I can show you. And the cost of these things were about
$300 apiece. Now in terms of the community itself, there was really no
home. And in fact I had just started SIGMOBILE just as I was entering
MSR.
So it was a very, very brand new community and the home just started,
the home just started. So I was this new guy in town. And so the
question was I wanted to obviously create an impression. So what do I
do? I thought of building a lab. And so then I had this idea that why
not turn Building 31, which was the place where MSR existed, into a
full lab and personal research at scale there.
The question was how do I do that? Who was going to give me resources
for it. So the first lesson is if you have big ideas, you want to do
it, find an individual who is forward looking, who can support you and
who will be willing to help you move your thing. And for me that
individual was Dan Ling, Dan Ling is now retired but I give him a lot
of credit for what MSR is today in terms of the successes of MSR.
Okay. So then with that support I settled on going to this company
called Internet Wireless. And then the question then for me was, okay,
I'll deploy the network but then I needed to program it, too. Without
programming it would just be a network, but I wanted to do that as
well. And it turned out that Windows did not understand wireless at
the time. There was no way for me to program any of the wireless
cards.
So what do you do? You're a researcher, you can do this, you can go to
the device driver, get the device drivers, which I did. But how to do
it at scale, how do you make sure that everyone is using your work. So
then I thought, hmmm let me write something that we always do, we write
a paper. So I wrote a paper. I wrote a paper called wireless is not
ethernet. I gave justifications of why that is the case. I did some
lab experiments. Showed all that stuff and gave it to the product
groups. Well, the product groups put it aside and kind of ignored it.
If you're smiling, you probably have some examples of your own in this
category. So the lesson is if you want people to spend their time on
your idea you must show them the money first. There's more lessons for
that now.
So I came back and I thought what can I do which is compelling that
would get these guys to listen to me and be sort of understand that
this is important? So at that time, if you worked in location
determination field, Roy and Andy Harper had Olivetti Labs and they had
done some work on IR badge and location determination with this active
badges thing, but there was nothing in the RF space which is what I was
looking for.
Now I had done some work before coming to Microsoft Research. I had
done work in the same space and written a paper which was published in
JSAC a while ago.
So I looked at that paper and then after looking at it for a long time
and deciding whether to implement it or not, decided it was just too
complicated. It was so complicated it was one of the best papers I
wrote because it was very mathematically intensive. It had all the
stuff I knew about signal processing and control theory in it, common
filtering state space all the good stuff but it was difficult for me
even to understand it after I had written at it and looking back at it
for a while.
So you must think about simpler solutions. So then I thought, okay,
what should I do next. Now I had seven, eight years of background in
signal processing. I had done pattern recognition work quite a bit
before I got to Microsoft.
And in fact in Deck I had done a lot of image processing. That's where
I was coming from. So now I worked with this great new hire, Vanker,
who we had just hired, in fact he was the first hire for the networking
group, and with him started to think about pattern recognition, the
context of RF signals. And naturally from that the answer was was
there a pattern to the signals that you saw at any particular location.
And so went down that path and then radar was created. And so there's
actually this demo if you click, if you go to the radar Web page click
on it you can see how the original radar system worked, 1988, 1999.
So the lesson here was that some of the best work in my mind is done
when you actually have ideas in a particular field and then you reapply
them in a different field. Because this is a very fresh outlook to
applying the same ideas and they win.
Now, so radar was an enormous success. I don't know if you've gone and
seen. It's one of the highest side papers I have if I look at the
combination, if you go to Google Scholar and look at it there's more
than 6200 citations for that particular paper.
So it's really successful. But that was in 1999. Today it's 2013 and
where are the products today. There's no products. So research
success does not mean commercial success. For commercial success, you
must think really hard about the value proposition. So when I went and
tried to sell radar to our product groups, they said that they didn't
see the value in terms of the amount of effort that had to be placed to
find the location versus what you would do with it. Now, of course,
there were other reasons there, too, at the time we didn't have
smartphones at that time. We only had laptops and things like that.
So there were many reasons. But just to tell you, the guys in product
groups are pragmatic and practical they know what works I didn't have a
very good value proposition didn't have it for many years. I abandoned
it. I didn't write a single paper after that, even though after that
there were 6,000 papers or more written after that.
So one other lesson was to persevere. If you're doing research, you're
probably way ahead. If you're way ahead you're not going to convince
people right away to take your stuff. And then the other thing that I
learned was that the market will always take the simplest solution.
Even though I had done radar which I thought was simpler than the paper
I had done previously but it wasn't as simply as what Skyhook Wireless
did. If you know what Skyhook Wireless did, all they did was look at
the location of the access point and said that's where your location
is. That's it. Created a whole table lookup and whole class, they
sold made lots and lots of money. The iPhone when it first came out it
had Skyhook in it. It was a degenerate case of radar. I had
completely overlooked it. It was sort of an interesting thing for me.
Now, I want to show you a little video that I did in 1999, which was
shown by Rick at MobiCom 1999 and we showed this one in ->>: Turn left.
Conference room No. 2133 is --
>> Victor Bahl: Wait to see how it looked at the time. But this is
[indiscernible] it was done about a year and a half ago doing
radar-like functionality. This was done in 1999. And just say a few
things.
>>: With RADAR, we can build location aware systems and services. Once
radar knows what your physical location is it can provide that
information to the system and the system can do interesting things.
>> Victor Bahl: Sorry. There's a story behind the beard. The thing
was I was pretty young looking and I was heading SIGMOBILE. I needed
to look old. The beard was there for me to look old. All right. So
anyway, this is radar 14 years ago, yeah, 14 years ago.
So now back to my interesting extension story. I had done radar, and I
had now gone in and understood what the product groups wanted. So then
I had an intern Gavin, first interns, sat down, wrote a spec just like
you would see at product groups. It was not a paper, it was not a
spec. And when we went there and we presented this spec and we showed
them the value at the time and then that wireless indeed was different
and you could never do what I had done with ethernet, they got it.
They got it and they accepted it.
Now industry started to have wireless LAN extensions. Not just that
they actually started a huge program called native Wi-Fi after that,
which was to completely abstract out the wireless part from Windows and
take most of it in the stack above. And that was fairly successful as
well.
So that was all good. Now, this is a memo that I wrote to with MSR.
Actually blew it up a little bit. I thought it was sort of funny.
This was about the building that I had talked to you about, creating
Building 31 into a wireless network lab. And of course send the memo
to all of MSR full access encouraging them to use wireless it's
something you don't think about now we just take it for granted. But
at that time people weren't like that. You had to sort of encourage
them. And I talked about Aeranet [phonetic] and the cost of the
adapter.
Now, this took a momentum of its own. So after MSR did this, we did
this at MSR, we used to have this thing called Micro News. Some of you
know about micro news, but many of you don't potentially know. But we
actually had a paper micro news show up every week or so. And in there
Bill Gates announced that we were going to unwire all 45 buildings of
Microsoft in it. And then in there he says it's sort of aimed at other
companies to take steps to use technology to free what Gates calls
knowledge workers. This was a huge deal in the wireless community,
because this was the first company with that kind of spread and deep
pockets that were saying wireless was going to be the future.
And if you remember that company I talked about, Aeranet [phonetic].
Aeranet got bought for $800 million right there. And I believe that a
lot of what happened in Wi-Fi, the burst that happened, was because of
this and they made a whole lot of money here. So the point here is
that it is very difficult to actually I could never have imagined this
would happen. And it's difficult to sort of see when you start your
project what they will, where they will lead. So don't worry so much
about the big things. The little things will get you there as long as
your vision and large enough and big enough.
So good successes in the enterprise. So why not take it outside.
Well, this was a time when 3G networks were very, very popular and
everywhere you went every place, every newspaper article every magazine
looked at 3G networks were going to solve all things. You recall in
panel discussion, Wi-Fi wars versus 3G. You can think about that
that's silly and stupid but that's the time when the telcos were
pushing and they did not want wireless LAN in the public spaces because
they thought it was going to eat into their existing business.
So the point here is what you read in popular media you should take it
with a grain of salt. You need to be independent thinkers. They will
get it wrong because they're just listening to a lot of marketing stuff
that is happening and we are trained not to do that and think
independently.
So I didn't. And I built the Choice Network, which was then the first
Wi-Fi network in the world. And Rich remembers this probably because
he was backing this up quite a lot. There's a poster in my office. We
had this poster. In the Crossroads shopping center we have here we had
the first Wi-Fi hot spot network in the world.
And this page actually still exists although this morning when I tried
to load it, it wasn't loading properly. Now, there was lots of great
innovation in here, by the way. This is not just the fact that we put
wireless cards in there. We had to decide what the architecture would
be. So one of the big architecture design decisions that I made at
that time was to have smart wireless switches and dumb access points.
That was a design decision. But all the other good stuff you hear
about in the conferences and stuff was there. Location aware services
was there. All this other stuff, security was there, and sort of it's
articulated in this paper I had. Now, once I built this, I was out
there evangelizing. The way I would evangelize this, logically, was
here's a graph from hot net conference, the panel I was sitting in 2002
where I shared okay let's look at where I spend my time. Horizontal
access the days of the week and the number of hours spent. And as you
can see, it sort of divides up into hotels flights offices on the move,
home. Now if you map this to say how much is indoor and outdoor this
is what shows up. You're spending the majority of your time indoors.
This to me was a convincing argument to say that Wi-Fi access and
wireless LANs was the future. And I couldn't see why people wouldn't
see that why I was actually making the argument in the thing. But
that's how I did this. Now, the uptake, of course, is that hundreds of
thousands of wireless hotspots exist. In fact it's almost impossible
to determine how many are in the world. We expect them everywhere.
For me it wasn't a successful thing. It wasn't successful because when
I had gone to the product groups and asked the question why didn't
Microsoft pick up and do something with it. I did. I tried to push it
inside. They didn't get it. They were very busy in what they were
doing.
Then in fact the architecture that I had recommended, the fat AP, the
sort of they embraced innovative architecture where they put most of
the logic in the access points not in the wireless switches. As you
can see in retrospect, the world is where I was then. Which is that
the architecture is much more in the wireless switches than the AP's.
When I submitted a paper to MobiCom the reverse came back with small
things here and there and they rejected it. I was rejected and I said
screw it I'm not going to spend my time trying to sort of move forward
with it.
So the lesson here again is that if you really believe in it, you need
to pursue it. And yes you may be wrong, but you may be right, too.
I moved on. I looked back at some of the mistakes,
mistakes I did. Now, seven years after that -- one
course, we patented all the good stuff. You wonder
happens to the patents you heard all the stories in
this is one of the
thing we did, of
sometimes what
terms of Google and
Motorola. But really in this particular case I did get an e-mail in
2007, which was about seven years after I had done this work, saying
that they had made an agreement, a landmark agreement with Nortel and
was sort of based on the patents we had done for those networks. It
goes on and says good things about it.
So Microsoft did get some benefit out of it but not as much as I hoped
they would get. Okay. So this was about some lessons I was learning
but as I was learning I applied them too. In 2002 we interviewed
[indiscernible]. He's not here but he was there when I gave this talk
originally. And he was a Ph.D. student at Cornell. He had come in for
the MSR fellowship. Talking about ad hoc networking. I liked his
drive and enthusiasm, and I loved him. So I took him on as actually my
prodigy at the time.
So then together with my brother who was in the product groups we
worked on this thing called multi net. We built the real thing. It
was all kernel hack and windows, built and worked well but as usual was
ignored by the business group. So in this particular case did not move
on.
So we persevered, right, we persevered and shipped it to the world we
said we so believe in this we're going to make this all available. And
this became one of the fastest downloaded software for Microsoft
Research. 100,000 of the downloads happened in a very short period of
time.
And then in 2009, seven years afterwards, it finally shipped as virtual
Wi-Fi inside Windows. And this was the first virtualization
architecture that resell. Lesson applied here. We built the real
thing. We showed them the money and we persevered.
Okay. Now, my third story starts with a Chinese dinner. So I was -and Craig Mundie, who was the CTO of Microsoft. I got this mail from
Pierre [indiscernible] you may remember him also. The e-mail went like
this. Victor, Craig would like to get your input on what is feasible
Mac protocol. I'm thinking to myself what the heck, a CTO of Microsoft
asking me about a Mac protocol. Didn't seem very smart. He said we
can meet in this Chinese restaurant and wants me to come come now. I
said fine. I showed up on that day, and that's the day when I got my
first exposure to some of these topics.
So from Craig I started to understand and imagine my sort of mindset at
the time. I was a researcher in a research lab doing things and now I
was certainly exposed by the CTO of the company to much, much bigger
problems in society. One of the problems being that there was no
Internet access in rural areas, while we all had it.
In other words what I didn't know there was a duopoly of Internet
access between cable and DSL and that duopoly was a serious impediment
to the progress that we wanted to see. I also learned from him about
network neutrality. I also learned -- there was a very good looking
lobbyist at that dinner table I didn't know what that person did but I
learned she was a lobbyist, and what they do in Washington D.C.. and I
actually started to get it. That was sort of a watershed moment for me
in terms of the work we do at MSR is not just about products, it's not
just about doing papers but it is about, can be about something much
bigger.
So the problem that was placed in front of me, the technical problem
that was placed in front of me is that we needed to become a threat to
the cable and DSL, because the resolution was that if the cable and DSL
guys knew there was an alternative, then they would try to protect the
turf and the way they would protect it is potentially go out and become
even more pervasive. So the idea was how do we become a threat to them
but we can't break the bank. So the decision wasn't that we just go
and dig fiber and do what Google is doing today, but how do we get
there. So in the meantime so I got on the bandwagon and I also sort of
realized this problem is real about the networks. And as you sort of
dug and I gave a lot of talks on this famous quotes by [indiscernible]
general secretary of the UN and Richard Newton, who passed away, dean
of U.C. Berkeley, was member of our tab, and they and me too were going
out and sort of talking about this thirst for connection that needed to
bridge the notion of digital divide.
If you remember for those who were around there there was this whole
digital inclusion, digital divide thing that we were pushing very hard
which was about bringing more people into the mix.
So mesh network emerged. The idea there was this was going to be an
organic system where remember we didn't want to spend that much money
so we wanted the people to spend the money and we wanted them to buy
the equipment and then through the magic of software we were just going
to connect them. And then we would bring fiber or cable or something
in some middle place and then through this multi hub network connect it
to that and boom they will have Internet access. That's what we wanted
to do. That's what mesh networking was all about.
And so the group really, really jumped into this, and Rich was actually
a core part of that group. And we just took it by the throat and
jumped on doing mesh networking.
Now, you would expect -- I expected, being in the field, that there
were 20 years of research that had happened in this area. And so this
would be easy. This would be just an engineering job because people
had written thousands of papers on the subject so why would it be
difficult. Wrong. It was wrong. So I looked at the number of routing
protocols existed and I ran out of steam. I couldn't fill this slide.
But there were so many. So the question in front of you, if you were
engineering our product team which one should you pick. The
interesting point was that every one of them showed improvement over
the other one they looked at. And the only reason was, and you will
understand that in research we cannot tend to do that, not we but
people tend to do that change the assumption here a little bit and
change something here and then make the graph go in the right
direction. And boom you have a paper. And so what do you do? Well,
this was very difficult. As researchers, it was difficult for us let
alone product groups. That's one of the reasons why when product
groups when you say I have a good solution the product groups sort of a
little confused they're hearing 20 different things they don't know
what to trust. And you know why that is there.
So I, too, lost a bunch of faith in the academic research community in
the networking community. And that is the time I went from a phase
very smart people to people who are just writing papers without
knowing.
Now and MobiCom, which is the home audience in the big conference at
the time and still is, was becoming too academic. We needed a systems
conference where we researchers build real systems. And that was the
time when I fathered MobiSystem in 2013. So now I think that one thing
as system researchers and I believe this group is very good at it, it's
world class and top-notch in role models in it but I do believe there
are not that many people like this, which is that we need to take our
work to the point of ir refutability. We need to take it to the point
where there's no question that it actually works in all kinds of
environments and that you're aiming at.
So we were very successful, I thought. We produced these academic
kits. We wrote a lot of software. We gave the software away. We
wrote papers. We did talks. And there were 1200 universities
worldwide who started using our kit. So by all sense and form this was
a success. We organized summits and workshops and looking at the
energy we were putting into this thing, the community also started to
sort of got integrated because multi hub ad hoc networking has been an
intellectually interesting problem space, and there are many, many
hundreds of papers being used. And then the VC dollar started to flow
as well. So just like Internet of things is happening now. VC
dollars, they see money, they say what the heck, a million dollars
here, a million there for somebody who has a hundred million dollars is
not a big deal they started to spend and immediately you started to see
things about New Orleans will have free city wide Wi-Fi or Michigan or
Philadelphia will have. A whole bunch of these things.
So we had succeeded. We had become a incredible threat to the DSL and
cable monopoly. In the process we had actually introduced some really,
really good concepts which have stood the test of time and one of them
was multi radio wireless networks.
We were the first to write papers on that and sort of explain how they
worked well. Now, the reality is that it was not so successful as you
look back. All the good stuff that was coming started to become bad.
San Francisco formally and city wide. AT&T kills. So the point was
that the VCs got it wrong. They did not understand the technology.
They did not understand that the limitations. And so when people who
wanted to make quick money in start-ups gave them the spiel and said
Microsoft is doing it they went and spent a whole lot of money they
realized they weren't getting money back for various technical
problems, which I could go into but I won't now. They didn't do it
very well. So the lesson here is that businesses decisions that are
made on not understanding the technologies that are bad decisions to
make, and this applies to us as Microsoft as a company as well, which
is that we must be very honest with ourselves when we build a
technology and we go talk to somebody else. How real it is. Again,
this goes back to the point of irrefutability, a point that you're
completely satisfied that this will work. Then you put your weight
behind it. Otherwise there's more work to do. They got it wrong.
They didn't really understand this.
Okay. So my full story is around licensed versus unlicensed. Now,
this also goes back to 2003 when I had the first exposure to the
breadth and complexity of the issues involved in all of this stuff. So
everyone wanted more spectrum like now if you're in wireless go over to
your favorite search engine and check everyone says it's crunch and
it's never going to happen or whatever. But the debate is should we
give it to the telecos or do it unlicensed. And there's reason to do
both, right, because the telecos will come and say look at our cell
phone industry, look at how amazingly successful it is, how many people
use it, how much economic value we've created because of that, and we
can do it right. The unlicensed guys will come and say look at Wi-Fi,
look at bluetooth and Zigby, these are successfully, they would not
have happened with unlicensed. If you think about it there's all these
issues. This is the time when I learned about Coase. He was a British
economist. He passed away. He was also a nobel laureate. And he's
called the father of reform in the spectrum allocation policies. Does
anybody know about him. That's okay when I asked the MobiCom audience
did anyone know nobody knew about it. It was sort of sad to me because
I think these are history is important, because you sort of learn. So
what he had done is he had written a very scathing article about the
FCC. What the FCC had done, just so you know, all this broadcast
spectrum was very political at the time. And so they had these
politicians were also very rich a long time back, and so there was some
sort of deal made where the spectrum was given away to these
broadcasters which are now broadcasters, for free. And one of the
reasons was because these broadcasters would broadcast political
speeches and things they would help the political agenda of the people.
So there was an ulterior motive of giving it to them free. When he
looked at the problem, this was in the '50s, he wrote this huge paper
and sort of saying how wrong this was and how badly they had sort of
looked at all this stuff. Anyway, I was invited to Stanford. There
was a workshop that was organized by Larry Lesig and Thomas Haslett.
They are lawyers but they're also great thinkers.
And I would recommend if you're interested to sort of look at their or
become a fan of their tweeting, which I do. But they brought in a
whole lot of forward looking thinkers, economists, legal scholars,
everybody to this workshop and I badly wanted to be there. I badly
wanted to be there. I wanted to learn beyond my engineering. So if
you want to be there, you must bring something to the table. Okay. So
I did. I actually wrote a draft proposal on what is spectrum
etiquette. And I worked pretty hard on this one because I was working
with parameters about what etiquette meant. I went there. They gave
me a stage to present my work and they promptly rejected it. Etiquette
made no sense in spectrum but at least I've started the dialogue.
Anyway, what was happening there was at that time I became aware that
in 1996 the U.S. Congress was going to open up all these TV bands. And
at stake were what happens to these bands. So the idea of dynamic
spectrum access was in the air, but it was just in the air. These
lawyers and these policy thinkers and regulators were not engineers.
They could only talk about things they couldn't do it. They were
making a lot of assumptions. So somebody had to prove that this thing
works. And we did. So we came back and we worked on this problem. We
worked on the dynamic spectrum access problem and we built the first
wide space network in the world.
And so these are some of the pictures of the time. And once we built
it, we knew that people were going to listen to us, and that's exactly
what happened. These are pictures of the chairman of the FCC who came
here on a Saturday and I gave him a two and a half hour talk and
answered every question he had and he came in with his chief of staff
and some engineering and he asked me very pointed questions about
things. We showed him the demo and everything and at the end he sort
of went back and I believe -- and there's really no evidence to back
this up, except the fact that he was there. A month later they
announced that they were going to open these bands up for unlicensed
spectrum with DSA.
Now, this, of course, led to a lot of regulators all over the world who
were interested in this as well. They came here. We showed it to
everybody. We showed it to IEEE standards bodies and the folks from
FCC and Singapore, all these regulators came here.
And also the good news was that beyond doing all this stuff, I also got
to be a part of the Ph.D. thesis of at least two students and I'm sure
there were a lot more. But these were directly the consequence of the
work we did here because they got these two thesises at Harvard and
UMD.
What we had gone from was this world, on the horizontal access is the
range in meters, vertical axis is the speed and Wi-Fi, for example,
sits here and then these other, some of the other different
technologies to a world that looks like this.
So we had literally created a new world where we could start to think
about what kind of scenarios we could enable. And that's what it
looked like. So what I learned there was that again the work that MSR
lets us afford to do actually goes well beyond what we think it is. So
in order to succeed, I had to think about the society and had to make
the case based on society. I had to understand what the business needs
were, right. We had to sort of think about of course research
engineering and the policy and all these things sort of fit together to
solve that particular problem. So research is only one part of the
story. And so when we do some work, if you may have heard Peter talk
about it, if you do some work, we must have a sense of humor about what
we have done as opposed to what actually happens. Building an idea
because we have the luxury to do it is just not enough. You really
have to think about the whole thing and you have to, if you believe in
it, go down. If you don't, and somebody else is doing that, you must
give him or her the same accolade and perhaps more for taking the risk
for taking it to the end which you couldn't. The point is that this is
a lot more to solving a problem than just doing your own papers, et
cetera.
So we've had plenty of impact beyond Microsoft. I wanted to read this.
This came just out of the blue for us. But it was really heart
warming. I'll just take ten seconds off for you to actually read this.
So you can imagine what we sort of did or what this technology did to
this individual sitting somewhere. And I have to believe that there
are a lot more than just this person who is actually benefitting from
it. And so this is great because we are actually doing something
amazing. So let me show you this way to break things up a little bit
and also to make a point.
>>: What if you could better educate children living in a remote
village who lack access to specialized instruction? What if you could
manage traffic so gridlock doesn't stop your city during rush hour?
What if you could provide state of the art high tech medical care to
people living in remote communities? What if governments could use
technology to better deploy city services such as garbage removal and
reduce costs in the process. What if a boy who dreams of playing in
the premier leagues could finally stream his favorite team's games. If
you look for answers you won't find them in what's out there but you
might find them in what's not there. It's called white space. And
it's a renewable natural resource found in unused radio frequencies.
These unused white space frequencies can be used to increase available
bandwidth, improving the quality of broadband connectivity. And lots
of these unused frequencies are located between TV channels. Signals
sent out on these unused TV frequencies travel far and penetrate trees,
walls and other solid objects. Inexpensively extending the reach of
Wi-Fi and other wireless broadband technologies currently limited to
using higher frequencies. What can TV white spaces enable? You can
manage traffic in your city minute to minute. Adapting the
infrastructure to accommodate conditions. You can more efficiently
deliver public services. You can have medical resources be available
for preventive care by reducing the cost of broadband connectivity, a
school can now become part of the global village. The only thing
stopping this from happening is the lack of favorable consistent
regulations allowing white space technology to be deployed on an
unlicensed or licensed exempt basis. People and the devices they use
will finally be better connected by tapping into the TV white space
available in your country.
>> Victor Bahl: Okay, so this is a marketing video, of course, but I
showed it because there was this last point that I wanted to sort of
help me transition into what I'm going to say next, which is the last
thing. So people often ask me, okay, about a vision talk. And I don't
know -- I don't think I'm a visionary or anything like that, but I
always see a whole lot of problems we need to solve. And the question
is which one are we going to solve. So now in this particular case, I
will tell you about this thing called a PCAST report. I don't know how
many of you know it. PCAST stands for presidential commission, which
was put together by the Obama government to look at the future of
United States and the world and think about what is going to happen
with mobile technologies. And they brought in experts from all over
the world and it's a big report. You can read it if you want. But one
of the things they said the way the FCC dealt with spectrum is messed
up and the norm has to be sharing. They leave it at that, that it has
to be sharing. The question for us as engineers, can we build
technology that allows sharing? And so it's that sort of challenge.
So this brings me to the kinds of things that I'm working on. These
are just a subset of things that I work on. But there's a campus wide
spectrum. And the reason I'm working on some of these is that last
part in the video, which is that you have to build it to prove to the
regulators around the world that it works. And so building it means
not building it in your lab, doesn't mean building it in some sort of
rural setting you have to build it mainstream. That's what we're doing
we're building a DSN network mainstream to show that we have
irrefutable evidence that it works.
Now, we also are going to make this available to the research community
as well as pushing very hard on different universities to build it as
well. And then I'll talk to you a little bit about this in the end.
So this is the networking. Many of you may have seen, the guy this is
the [indiscernible] from MSR Asia, some in Building 99, [indiscernible]
right above. This is where the box sits and we have four buildings
equipped with it and it is a joint effort MSR Asia and MSR Redmond.
Looking at the spectrum crisis thing really quickly there are all these
blogs. Start to talk about we are headed for a serious problem, which
things just wouldn't start working. So the question is can it be
managed. The technical question here is can it be managed?
Now, we did some work, and we looked at how the spectrum has been used.
So this is what the FCC has allocated on the band above. This is what
we see. So you can see there's all this allocation but there's no use.
Then how can anybody say that there's not enough spectrum, right? Of
course, this is not a conclusive set of data. We are now taking it
upon ourselves to not again believe the marketing literature but to get
real data. Science is based on real data. We've got to get real data
to determine if there's actually a spectrum crisis or not. We're going
to show after maybe a year or two years whether it exists or not. It
turns out it probably doesn't. So the technology policy group here in
Microsoft has actually built a spectrum observatory, but it costs
25,000. Not scaleable. What we've been doing at MSR is to bring the
price down to less than $5,000, and we've got many universities lined
up that are going to take this and put this observatory. Think of it
as a planet lab but an observatory has a thousand points all over the
U.S., all around the world. The call to arm to a lot of the students
and universities was to help us build the observatory around the U.S.
So there's no more debate whether it exists, doesn't exist. And we
have real data so once we show that there does exist spectrum, then the
next technical question is how do we use it. How do we use that thing
that somebody else has been allocated but you want to use it. Sort of
like white spaces, white spaced becomes the first manifestation of it.
And you have to build systems. It's not easy. You have to think about
hardware, you have to think about broadband receivers. Think about
power and rendezvous strategies. A lot of problems that people have
tried to solve actually are much simpler in the wi-fi space than the
space I'm talking about now. If we solve them we would have gotten rid
of spectrum crisis forever. That's whatever. I have a few parting
thoughts from a lot of the work I've done.
So the first one is that we must, must always question the assumptions.
Now, when the assumptions were made, they were right. So right now we
make some assumptions about the systems we built. We're right. We're
not stupid people. But the generations that come after us, they would
have made a lot of progress, and the question then is were the
assumptions right or not. So we must always do that. I've also made
this point plenty of times, hopefully, which is that research is only
useful if you really take at least systems research is if you take it
upon irrefutability. And then I -- this was sort of a controversial
point that I made, actually, because people asked me about this later
on, which is I believe publish when you're done. The question is when
are you done. We can talk about that. But really it is a disservice
in my mind to publish and in fact I gave this anecdote when my wife was
technical and she was working at Amazon. Doesn't work there anymore.
And she was doing all this stuff with Hadoop and analytics and all the
other good stuff, big data. And she came to me one day and she said
you guys are totally messed up, because you publish so many papers and
90 percent are garbage and you wasted my time, you wasted so much of my
time and my people's time. And that's just not good. That's just bad
for our community, for our service. And so, again, back to MSR. We
can't -- we have to be very critical about what we publish, I think.
Persevere, that comes out hopefully has come out, the main things I've
sort of said. And then the last one is don't be afraid of failing.
You're not going to achieve anything great if you don't even try the
hard stuff. And I have failed many times for the things that I've
shown you, I can talk to you about many cases I failed. I told you how
I didn't do well in wi-fi. The other projects I can talk about. You
didn't hear about them. You only hear about the successes. But that's
exceedingly important. And then you need to have sort of another
thing. So now in ending I would quote Mahatma Gandhi, I told you I
gave this talk on October 2nd. Turns out it was his birthday. He said
something that I very much believe in that be very slow to have
conviction, but when you have the conviction, you go for it. With
that, I will end. We have time. Thank you for listening to me so
quietly.
[applause]
Thanks. So we can stop recording, but we can have --
Download