Asked me to give a little bit of history about cellular.

Last time I taught last year I gave the past history.

This time I want to give a little, even though I have less time, just a very brief

bit on how we got here and then where, where things are going, at least from my

perspective. So let's see if this works.

Oh, it was working before. If you, how do I do it?

Full page? Is that on this?

Oh there, full screen monitor. I'm sorry.

That, I just got I can dismandle. I can do it manually.

Okay. So my career actually in wireless, I was

were you were back in the late 90's. I just did graduated from my Phd program.

Are you guys undergrads or Masters Phd's? Undergrads must be right.

Okay, yeah. All right.

So when I graduted in the late 90's, the late 90's was a very interesting time.

At that point, there were two technologies that were really at their infancy.

The internet and cellular. Today we take both of those completely for

granted and we can't really imagine any digital technology without both of them,

but it's hard to imagine that if you w, wound back, and this is probably before

many of you actually even had your own. You're probably all too young to know

this, but at that time these were still very early technologies and people still

couldn't appreciate how. Dramatic of a difference over make in

their lives. The that's the page, here e go.

What we saw of the last next decade was really the tremendous growth in both of

those technologies and their merger together.

That is really the unwiring of the internet, that's making that internet

available to you wherever you are and is right, yeah where ever you are.

I think will look back at that decade and really see that decade as a one of the

really revolutionary era's in digital technology.

Now, what I'm going to talk about'er here on the.

It is really the top line which is the development in the cellular standards that

enable that. Those are we the beginning of that decade

began with the emergence of 3G and we later saw 4G and put some of the glue

marks here, they indicate some of the key milestones in that.

But parellel to that there were equally and perhaps more important develops.

Also we saw the introduction o because you can't imagine having a computer or cell

phone without it. But in fact that exhaustive technology

that only comes back to like the late 1990's.

So prior to that people didn't even have wi-fi.

All right so by I, I, I grew up in the stone age or something, and maybe in some

sense I did. But the peril of that also, it really

developed my devices. And this is something that we're going to

see that was incredibly important as well. That it wasn't just the network speeds,

that it really needed to be a compelling device that you could experience the

internet with. And really if you know, actually, in many

ways the. 3G technologies that began in 2000 and

got, significantly improved in the early, within a few years after that.

But it was really the iPhone that really, brought out the Internet and made it such

a tremendously valuable, tool. So that, really, the first one came out in

2007, that's a picture of it. Now, of course, as Mung mentioned, they're

just releasing the iPhone5 right now. So, it's five generations.

Shortly after that were the first Android phones in 2008.

But this was really singular device that. Made people really feel that they could

have mobile data on their fingertips. In fact, when I was I had started Flarion

in 2000. We went to carriers at that point early,

this is like in 2001 - two trying to sell the concept of wireless data.

And the response that we got from a lot of carriers was, well who would want to check

the Internet? On their mobile device.

Now, you can't imagine someone thinking that right now but that was the mindset.

And it was really Steve Job and his ability to have some kind of device where

you could really see the, make that experience valuable that was what made it

M-, made this all worthwhile. So, I'm, let's go ahead and try in this

what you have though. Twelve:20 or something may be, twelve:15.

Twenty past twelve. Twenty past twelve is total time.

Okay. So, so in these fifteen, twenty minutes,

I'm going to try to do two things. First of all, talk about this golden

decade if you like, about how we went from how the internet got unwired and really

that's a story about several technology existed prior to that decade.

It was primarily a voice of this and what we're going to see is how we make internet

go from voice to data and within that we're gonna talk about few topics probably

I think things that somethings that Mangu talk about later in his class.

The difference between circuit switch and packet switch.

The difference between peak rate and average rate, and so on.

And that's all cuz they're connected to what is what makes 3G and 4G valuable.

The second part is I am just going to give a little bit of speculative part on what

is going to happen from here. This is not the end of this revolution and

in fact I think is just the beginning and I think there is a lot of a lot of

exciting technologies to, happen. All right.

So. Just so, and again, if you have any

questions, just really stop and answer, be, because it's better just to go, get

things answer things as they come up. So, the, there are really two challenges

for going from a voice based system, which were the systems in the, prior to the

1990s, going to a system that could actually do data.

And that really is without, that data is fundamentally different than voice.

The first aspect that is different is that by the issue of burstiness.

When you are talking on a voice call, it's relatively constant bit rate.

Alright, now so what happens if you, you transit, you talk into your cell phone.

It get's converted into bits, and then the number of bits per second is constant.

The number of, is constant, and it's also relatively small.

How small? For most cellular voice, it's at around

like eight to twelve kilobits per second. So you, where you want to do, if you wanna

design a system for voice, you want to have a very small pipe, if you like,

allocated to each mobile, and to have it so you can stuff as many pipes in that

for, or, connection at on, once, so you can cover the most number of users per

base station. Alright?

So if you try to graph it, if you graph the data rate versus time, it will just

look like a constant line. It's not exactly constant.

They have something called voice activity rejection when they try to attack when you

don't speak and it shuts it off briefly to save a little power but its otherwise it

is relatively constant, so constant and small.

Data of course is fundamentally different. What happens there is that you have in

there very bursty nature. Imagine that you have taken like a web

browser. I click on the web page, when I click on

that web page, I want to get a very, very high at wide because it will download that

web page very quickly. How high?

That could be in the megabits per second. So it could be orders of magnitude more

than a voice call. So its like when you download a web page.

Oh, you just saw that actually, yeah, two megabits per second.

Compare that to eight, eight or twelve kilobits per second.

Talking about a 100 times. So what he was just showing there, was

like a 100 users all calling at the same time. But then you stop, most things are

not like speed test. You know, you got constantly running them.

You download the web page, and then, as soon as you're done, you basically have an

idle link because you're looking at the. You're looking at the web page and you

don't need it and so what the system has to do of course it means it has to, its

not like serving a 100 user at that a small way continuously.

It has to serve you the quicker at 100 once and then as soon as you've done your

shifted over and sets of an abit. So that was the major change that had to

happen. Now to understand why that was needed to

make a radical change the way systems would have done.

Let's look back at the technology that existed prior to 2000.

So GSM was the first 2G system among mentioned what's 2G is.

Two G is the first evolution to digital voice.

All right. So this was begin the global world wide

standard for voice. So that was GSM.

So in GSM, it was deisnged for voice. It was obvious that very logical designs

of voice, what I'm going to do is I'm going to divide the frequency into these

narrow 200 kilohertz channels and then put time slots and give every one out of eight

of these time slots. All right.

Now this is a vey small amount of data. How is that small.

Consider you've compared that to wi-fi, wi-fi has a bandwidth that's twenty

megahertz, so that's a hundred times wider.

All right. And You can get it in principle for, all

the time. Well, at least while you're using it.

Alright. So you can have a much higher rate.

So this is a very small rate but that's fine because all we need is a small pipe.

12.2 kilobits you can actually transmit pretty efficiently in that little pipe.

Alright? So you transmit the small pipe and you,

the other part about this, that if you want to re-allocate or de-allocate it's

pretty slow. But that's okay because it's a med for

voice. The idea is that you're going to make a

call. It's going to take you time to dial the

call. You're going to stay on it for like A

minute probably or something and then, at least and then hang up.

So if it takes hundreds of milliseconds maybe even seconds to set up this little

pipe and turn it down, that's okay. Because it's voice you are not going to

notice it in the long run of the call. Now.

The beginning, actually, it was even prior to 1990, before the internet.

They actually wanted to introduce data services.

Alright. Now, when the, they wanted to introduce it

for other things just like silly things like voice, Yeah,

Basically, equipments to SMS and stuff like this.

And there was actually, I even remember going to conferences, at that time.

People were saying, well, people maybe don't need the internet.

Maybe SMS is just enough, for example. These were all the idiot things that

engineers like me and myself thought about at that time.

So, they said, well you can still maybe do data.

And I said, well, there's two problems with it.

I said, it's bursty, and there's a p-grade problem with it.

Well, at least the p-grade we could solve. I don't lease the little bit.

When they went to this 2.5G which is but among mentioned, you just may be give the

guy more of these time slots. So you can go from these 9.6 kilobits, you

can go to N times 9.6. You can actually may be squeeze a little

bit more into these time slots. So you're getting the around 40, 50

kilobits per second range just by dumping these in there.

So, that was what was called edge. So how do you know of your phone is on

edge. Sometimes you'll see E on it if you're on

EKNT. If it says that you have notified that E

stands for edge. All right.

Its called hands for edge, doesn't matter what it stands for.

So that's how you do it. But, you can see some problems with it.

First of all, what happens when data is bursty.

So if I allocate you a bigger pipe, you are only gonna, you are only going to

allocate it to your p grade. So when you say download on your phone,

alright, so you say okay, you need a lot of data rate, so let me allocate you lots

of time slot. But then you don't use it, and they go to

waste. And why do they go to waste?

Because it wasn't designed to quickly move those time slots from one user to another

user. All right.

So, this could handle the peak rate a little bit.

Not great like these days. It was may be 50, 60 kilobits per second.

All right. But, so, it wan't a great peak rate.

And it still had this bursting problem. It was, in fact, wasteful.

Probably if you go to allocate 50 kilobits per second, you are probably only using a

fraction of that. And you can't let any other user use them.

Alright, so what happens here is you get the date of birth, but all these blue time

slots here go completely wasted. Does everyone understand that?

So you allocate all these slots, but you use it just for these parts and these go

wasted. So it's not a good peak gradient and it's

pretty wasteful. Now, it's when 3g came out, actually in

2000, or actually, what was called the release 99, it wasn't also really a

solution for data. Alright?

When people think about it today, synonymous with data, but at the time, it

was just to use this CDMA technology, which, remind me, we'll talk about later

To enhance the voice quality, but it was still a fundamentally circuit switch

concept. Alright.

So you basically, you just allocated not time slots, you allocated some other type

of channel, but you allocated it in some fixed way.

So, the really the first key revolution came in a few years later in 2005.

What was called high speed downlink packet access, HSDPA.

How many people have seen H on their phone?

Okay. So that, this won't apply if you are a Verizon Customer.

Verizon have it alternate system. You'll see it called EVDO, or it might

just say EV. Or something like that.

It might just say 3D. Alright.

And if Verizon has a 3D, it's a similar system to this.

So we have H that stands for HSD, where it's also out HSPA HS Packet Access. So

when you connect, when it's connected on that it does this.

And this was a system really the first system, commercial system that was really

designed for data. How does it work?

What I'm going to do, is I'm going to divide, just like the GSM, I'm going to

divide the time slots. But first of all the time slots are a lot

wider in frequency. They're about.

3.84 megahertz. So there's no more than ten, fifteen times

larger than that GSM time slots. But I am not going to pre-assign these any

of these time slots. So I'm just going to leave them open and

going to able to schedule them on the floor.

What does that mean? While, suppose there're are packets

arriving for these mobiles, this is a packet system not a circuit switch system.

So if some green packets arrive say for mobile one, I can schedule those to use

your one end user Y and as soon as I'm done that I can schedule to a completely

different user in the next time slot. And then again, scheduled to another user,

and then the, in the third time slot. So this way, you have a very low delay to

actually. Transfer data to transfer the resources

from one user to another. We're gonna show you about the lot of

these costs about the idea of resource allocation.

In some sense, there's power control performance, there are resource allocation

problems, your allocating power between one user and another.

A lot of what we want for data is about the speed of that allocation, right. How

fast you can allocating resources back and forth.

And that was a key insight into 3G data. Alright?

That was So that was introduced in. Well, how fast?

Well these time slots are two milliseconds.

That means if a packet arrives, in principle, within two seconds I can.

Until you start transmitting out. So module or delay.

In comparison it took like, hundreds of milliseconds to reallocate resources in

edge. So this was a significant lower delay,

must higher peak rate and much better utilization of the channel.

Now that being said, the IPhones someone mentioned actually came out in.

Just using 2G. And it came out in 2007, four years after

this technology was available. Does anyone know why Steve Jobs didn't

pick to go on 3G. It was power consumption, alright?

He was very worried that the phones would burn a lot of battery life, right?

And it's something's that actually, you notice that iPhone five wasn't even the

first phone out with 4G. It's just that when you go to higher

bandwidth, they eat a lot of power and he was worried that it would burn a lot of

battery life. And so it, it took a little while for the battery life to come to do

it at an acceptable level. Yeah.

How did they manage to switch it for the digital users? So, what happens is, your

still transmitting. There's a, there's a, all you need to do

is, you just put a little channel, another channel.

Which tells you who this time slot is assigned to.

Alright. So every time you're, you listen to all

the time so that, so your mobile is still running all the time, that's why it eats a

lot of power. You have to listen to this whole bandwidth

the entire time. And then you just look, you're, say you're

mobile one, you look at the, sort of the A small signal in here which says Is this

just like for me, I say, o it is for me how to code it.

But then you look at this one and you say, everyone have to decode this and they

allow all lookers as just for me and as just not for me.

Then they just don't decode, they ignore it.

But they have to eat the power of wow. They have to eat the power of constant

residual. That's actually how wi-fi works too.

How does the, how does your laptop know the packet for it or the base access point

sends the packets out. It has your address in it.

All right. Its imagine like I send a letter, I write

your name on it, but I. Hold it up and say everybody, look at this

letter. Who is this for?

Alright? And then you all look at it and say, oh

this is for John. And then John will take it, and I'll give

it to you. Alright?

So that's how it works. There's a kind of address on it.

All right. And we'll get back much higher. It went from 50, 60 kilobits up to

fourteen or seven was probably more realistic in the early versions.

And today you probably get 40 megabits and even higher.

Alright. On it.

In principle. The peak rate.

4G expanded this further in two ways, we got even a higher bandwidth in principle,

although, most carriers line up bandwidth as not that much larger than 3G but it got

like twenty Mhz. But it also allow this division in

frequency and it's divided into something they call Orthogonal Frequency Division

Multiple Access. It's basically the same, you have

divisions in time and frequency, and you can assign the people arbitrarily blocks

within this. So what that allowed is, sort of, further

granularity of the, of the channel. The problem in 3G is that the kind of

these minimum block. These are still pretty large.

So once you have a small packets of sending that's kind of wasted.

Alright. In 4G you could send the lots of small

packets as well and those are things like TCP acts and for other if you did want to

send voice, and other things like that to be put into the part.

And allow much more flexible schedule. It will allow introduce other parts here

that you could do coordinate transmission of class cells.

I'm going to transmit on frequency one on neighboring cells, because I'm not going

to transmit on that frequency and will also I changes independent of this.

Things like what Mongo mentioned about going to an all IP network.

Alright. Oh my God!

Okay, I've like, almost blown all my whole time.

Alright, let's, oh! I didn't want to go to end, although

probably I should go to end. Okay.

Let's just talk about this. The second challenge for voice turn down

data was try to improve the reliability of the service.

All right, and so, what has to happen there was you needed some method so to be

If data is much more sensitive to having media channel that's much more, that's

reliable, alright. Reliable means when you transmit voice.

I'll just say this and without going I won't have time in this.

When you talk about voice, voice is, say, tolerable to some level of errors.

If you lose some voice packets once in a while, you just say, you will get, you

just might sense a slight degradation in quality.

Data, on the other hand, you need perfect reliability, or very high levels of

reliability for there to be a reasonable connection.

And so you need other mechanisms to try to improve the reliability, which is

fundamentally hard in a wireless channel because wireless is kind of uniquely

susceptible to, to problems of data, because of a process called fading.

So that's with other problem parts that were introduced in 3G and 4G.

To get around that in a way that's delay sensitive because I'm running out of time

I will just leave you with that. So let's just talk, my last few minutes

here, five minutes, let's talk about what's next.

If I have time, I'll go back. .

Sorry, I should have [laugh]. So.

In some sense cellular technology right now is really a victim of it's own

success. What we've seen is a huge growth in data,

but that growth is really going unabated. The more people experience the more people

may want it. The more people experience the more they

want and how much is it growing. Approximately doubling every year.

Now think about that from the perspective of the carrier.

If they cost structure for delivering data remains the same they would have to

basically charge you double the amount to get the same amount of data.

Of course you're not willing to have your phone bill double every year.

Maybe you're willing to pay a little more because.

Some of it's getting better but not double.

So they have to find radically cheaper ways to try to deliver the same amount of

data. All right.

We'll also talk about ways, I think mull you have to talk all about pricing and

things like this. Right.

About may be that the pricing plans to incentivize people to use their data

better. But let's just talk a little about the

technological solutions that you can do. So, there, a sort of two key solutions of

this skip to this one here. I idea is to place just many, many more

sell powers. Right.

If this ness is been a pro, process that's been happening over the last in these, you

know, twenty years. So we look back at the traditional method

to sell your tally, you see these on the highway.

These are large powers that set up. Or may be in your highway with many

antennas in the top. May be, it could be twenty, 30 meters

high. All right.

They were already you know, what was going towards the mtero cell, metro cell sites.

These are things on just the sides of bulidings and more denserous and now we

think about things like what we call sample cells.

These are very small personal base stations that could be manufactured almost

at the cost of just a wi-fi access point or a commercial gradewi-fi access point.

On top of that they could actually deployed inside of subsribers home or may

be in an enterprise. So for example Pristine university

requires purchase on their campus and then the operator wouldn't be have to run.

Operate that mersel. So you can bade many more south hours much

at much cheaper rate. How much cheaper?

These are around the neighborhood of 100,000 dollars when they first few years

ago. These could be at the neighborhood of

1000. So we're getting all this a magnitude is a

cheaper. The other part I've just mentioned just

one other technology which I think is an area is up to an interest to me.

There are many ideas but one is the idea that is expanding to new frequencies.

So sell your signals just other all types of radio wave communiation operate in a

frequency band. Right.

Just like the tune into your AM and FM radio.

All cellular communication sits in a very narrow band between zero and three

gigahertz, pretty much. Alright, that's all today.

But in fact, there's a huge amount of available spectrum available, in the

regions above that, up to say, 300 gigahertz.

This is sometimes called the millimeter waste space because that's the wavelength

of those transmissions. This part here, if it can be used in a

viable way in cellular technologies could expand up the capacity by orders of maybe

200 times because there's much more spectrum That has other problems in space,

because it has problems with a propagation.

But since the cells are becoming much closer right, together.

This might actually be a viable option to expand it I've got, capacity dramatically.

Alright. Let's.

Let me just say a couple last things. There are many other possibilities, you

see. Some are on the devices.

There are things that you are going to see.

There is actually, with the iPhone5, there was an excellent article that was on the

New York Times about just the evolution. It talked about, maybe I could send you

guys a link to it. About maybe iPhone5 seems kind of

incremental, but in fact if you even just look at the info modes, series of

incremental reductions, and improvements that are coming up, that they're still

pretty dramatic increases and improvements of devices that's happened over the last

five years. I think we're also want to see continue

more other types of improvements. One might be.

Radical things, you know, this was something that came out of Nokia's labs,

right now they're experimenting, with these flexible electronics, that you can,

you can have wearable computers or you could expand them out so that you can get

a display that opens up. You're gonna have a lot more new

interfaces into the devices, or gestural devices, or 3d, more sensors, more

cameras, these other things people talking about, in their terms of augmented

reality. And all the ways that the network itself

can be improved by trying to make these other improvements that I've mentioned.

And also maybe more out of the vertical integrated model that four of the carriers

waited to deploy, that may be ended up changing too.

So let's just wrap this up with a few comments.

So the mobility has really revolutionized the digital era.

It's really hard to imagine any, any digital technology that hasn't been

fundamentally changed by the ability of being mob, mobile.

And that's not just voice communication of course, but it's internet communication,

it's gaming, it's entertainment, it's office, it's navigation.

All these things can benefit from connectivity to the web.

There were many technological advances to make that happen.

I just mentioned two. But actually the biggest change was

actually seeing the value of mobility in the first phase.

Their concept of that mobile data is valuable was actually the really, the key

part. The consumer adoption took needed,

combined progress and needed the enabling technology of networks so, things like

what you're citing here, but also devices and applications that can make it

worthwhile. I think we're going to see, this is really

like I said, just the beginning of the full impact of mobile era.

We'll see back in this last decade as a tremendously important one, but I think

there's many, many more things to come, and maybe from studying this kind of class

may be part of that.

Cellular Data: Past, Present, and Future: Sundeep Rangan

网络：朋友、金钱和比特

Meet the Instructors