Wireless Communications Networks Transcript

This page contains the transcript from the video clips contained in the Wireless Communication Networks module. The linked words and phrases when clicked on will pull up a definition in the glossary.

Introduction (clip 01)


What I'll try to do is give you an overview in what is happening in wireless communication. I want to discuss several systems and existing standards but that will not be the main focus of this presentation. What I want to explain and discuss is mainly what is happening on the wireless channel, what can we do to build a good system and what are critical issues to be addressed and how are they addressed in real life wireless communication systems.

What I find intriguing is that many years ago Marconi started his first experiments in wireless communication and he managed to send signals wireless over the atlantic and across the Pacific. And now we're talking about communication over only a few meters. What progress have we made [in wireless communication]?

Many cellular systems are built according to this basic structure: a geographic area called a cell is served by a particular base station communicating with the users. The signals from the base station go to the mobile switching center that also keeps track of these signals and in a certain instant of time when the signals becomes weak makes sure that other base stations become involved in the communciation. Several mobile switching centers are connected but -of course- these mobile switching networks are also connected with the public switched telephony network so that one can place a call to other users that are on a public network.

What we plan to do in wireless communication, particularly in cellular communication,- is to accommodate as many users as possible - and the way we do this is by very dense frequency reuse. What we do is we split the area that we want to cover into many small cells and in each of these cells we have to use a particular frequency.

The game that we try to play is that if we use a frequency in this geographical area we want to reuse it in another area; we want to make - relative to the size of the cell - the distance between the cells that use the same frequency as small as possible.

The price that you have to pay is that you get interference from this cell to this cell. So you have to build your system in a way that it's receiver is relatively immune to signals that interfere.

The operator wants to increase the efficiency of the network, he wants to accommodate more and more users he typically makes these cells smaller and smaller so that he can put in more cells in a given area.



Problems (clip 03)



The main transmission problem in wireless communication is: two things:

First of all the receive signal is not a strong line-of-sight signal but are a lot of reflections from obstacles. That's called multipath reception [...]

And that gives a lot of nasty effects, especially for digital transmission. Mainly for transmission at high speed.

And the other effect is that also in many other situations at least one of the antennas is moving and it's because of the motion of the antenna that you continuously see changing and continuously varying channels.

So your system has to adapt to all these changes in the channel and you can choose your modulation technique and your way of packetizing a signal so that you can send it over a wireless channel.

But those things can be resolved if you design a system in an appropriate way--if you know how to use digital processing techniques. What has become very important in the last few years are three issues:


Bandwidth (clip 04)


First of all: bandwidth. The radio system is crammed with transmitters and receivers trying to use the spectrum for communication or non-communication services.

So you have to manage your scarce resource. It's in many cases bandwidth.



Energy (clip 05)


Energy. Here at U.C. Berkeley a large research project called the Infopad focuses on low power design. The heaviest thing in a wireless design is often simply the battery. The main thing that limits the talk time or the time during which you can operate as wireless multimedia system.

Energy is a very important thing. There are a few other issues with energy like the amount of RF signals that are radiated from the antenna. If you use a cell phone with an antenna sticking out of it most of the energy is radiated right into your brain.



Mobility (clip 06)


And mobility. The main difficulty with mobility is the complexity of the network that has to track all the users and has to make sure that the right signals arrive at the right user.

For instance, in the pan European cellular phone system if I place a call to somewhere, I don't have to know whether the person that I'm trying to reach is in Italy or in France or somewhere in Sweden... The system has to figure out where it can reach the user.

David Goodman at Rutgers once did some estimates on what happens if you have to keep track of all these users that are moving around and he found that compared to a typical ISDN telephony network a wireless network has to do about six times as much switching and about 15 times as much signaling.

There's a lot of signaling involved which is different from the switching of the signal. If for instance a Dutch subscriber places a call to France but that Dutch subscriber happens to be in Italy then the link only has to go from Italy to France but all the administration has to go through the Dutch network.


Why wireless? (clip 07)


About 20 years ago many people were predicting that radio was either dead or it was not going to have a very bright future because of optical transmission. There was not really a reason to cover large distances anymore with radio transmission.

But what we see in the past years that it's mainly in short range communication and the link from the subscriber to the fixed network is where radio plays an important role.

And typically in wire line phone systems, 50 to 70 percent investments made by the PTT or the phone operator is in the connection from the subscriber to the network. It's the most expensive part in the network.

So if you can replace this by a wireless communication link than you can provide telephony service in a much cheaper way.

This is a curve where we try to estimate the cost of a connection of the subscriber to the network on a function of traffic density by the amount of telephony traffic per square kilometer. So here we have a rural area like a northern part of Sweden with not many users.

Here we slowly move towards suburbs, villages and inner cities. What you see is for wire line communication (like a twisted telephony line), if you have to lay long cables your expenses will grow significantly if you have to put in longer and longer cables in rural areas.

Whereas for radio communication it doesn't really matter how long your subscriber length is--to some extent of course. This cannot be hundreds of kilometers without raising your transmitter power.

It's mainly in this area that the first cellular systems were successful. That's why in Europe Sweden was one of the first countries where cellular telephony was very popular.

[the fastest way to deploy the phone system is by offering it through a wireless system.]



Further Development (clip 08)

Now we see most of the developments in areas with very dense population where it is also expensive to lay down buried twisted-pair cables simply because you have to dig up the street and everyone will complain that the traffic cannot pass...

We see for flexibility and convenience reason a lot of systems being deployed in this range.

Typical cell sizes for the cellular phone system is one to 30 kilometers with a tendency to go to micro cellular systems which have several advantages. With an increased user capacity, the transmit power that you have to use in a micro cellular system is also much smaller because the distance that you need to cover is only a few hundred meters so the battery size in your hand held is also smaller. You can have a more convenient device.

There are a lot of other issues, like if you make a cell smaller it is more likely that I move from one cell to another, so it's more likely that the system has to perform a handover to another cell. The propagation environment becomes different and the way you plan cells becomes different in micro cellular and "pico" cellular networks.

Here you often consider a circular or hexagonal cell layout and in micro cellular systems typically the grid of streets in a city defines how you build your cellular layout.



Why Does the Cost Not Increase (clip 09)

Q: In this chart you show that the cellular radio cost is fixed but cellular radio has a scaling capability of increasing the ability to support more users by deploying, splitting cells and deploying more radios. So I'm a little surprised that the chart doesn't show some increase in cost as you increase the number of subscribers.

A: The range here is still a range of relatively low density of users. If you make a system denser and denser you get several, possibly a little bit canceling, effects.

One of the effects is if you make cells smaller and smaller your system overhead becomes larger and larger because you have more handover from one cell to another.

People move relatively fast out and about. Your network becomes complexer.

The other effect is if you start with a low density of users and then you increase your density you can do something in the network but that's also something you can scale, if you talk about cell sectorization and rather than on antenna covering a circular cell you do part of the cells. That's something that's also scalable so you can even start with building a sectorized system.

But of course you're right in raising the issue that this line is not exactly flat... You're adding a new dimension here--which is the time dimension, not shown here.



Handover (clip 10)

I've used the word handover or handoff several times without specifically saying what I mean by this. If we define a certain area covered by a particular base station and users are driving around in this area at a certain point in time they will get closer to other base stations or at least drive out of the coverage area of on cell.

The problem there is that at a certain point--and it is important that you choose the right point--you hand the user over to another cell.

A lot of things involved here, like exactly at what point in time do we have to do this [handover].

We will see later on that in wireless channels the received signal power is fluctuating rapidly because of multipath reception because of shadowing by all types of large buildings.

Sometimes it's difficult to decide whether a user is simply behind a certain large building or is really driving into another cell--and different systems handle this in different ways

To give you a flavor of how many handovers occur, we took the example of a cell of 16 kilometers and this table gives the handover probability as the function of the length of one call. Of course if a call gets longer and longer the probability that you have to do a handover also increases.

What is also important: if you decrease the cell size it's more likely that you move out of the cell so you have to handle more and more handoffs.

There are several ways to find out whether you have to do a handover. One way is simply to measure the received signal strength, however that is not always a very reliable measure because the main problem in a cellular system is not the noise level, which is fairly fixed, it is the interference from other cells. And simply receive power on a certain channel is not really an indication of the quality of reception. So it's often better to measure bit error rates or signal to interference ratios. Signal to interference ratios is very difficult to measure in practice but bit error rates may be something that maybe measured easily.



Cordless vs. Wireless (Clip 12)

The main difference between cordless and cellular is that in a cellular system it's an operator that wants to provide you a service that is very similar to a normal telephony service except that you can be anywhere where you want whereas in cordless what the operator provides is simply a wireline telephone link to your house. And there you have a small basestation and you can walk around your house with a direct link to your own basestation. But that's nothing like taking your cordless phone, going to another city and placing another call there. It's just a link to a particular basestation.



Basestation (clip 11)

In most cellular system [the organization of handovers] are centralized. It's the base station that recognizes that the signal drops, that the performance, the bit error rate becomes too bad for reliable communication and it starts to take action. It asks surrounding base stations to measure received signal power of a particular user, to find out where the signal is strong enough.

That's done in most existing cellular systems, including GSM, also the American AMPS System does it this way, but the DECT system--that stands for Digital European Cordless Telephony system-- does it the other way round.

The other way round in terms of who takes the initiative: it's the hand held that measures the best channel--so it's not taking action if the signal drops below a certain threshold but it continuously searches for the best.

It's the hand held, not the system that takes action.



Indoors (Clip 13)

Q: How does the channel characteristics change between dealing with vehicular mobility and pedestrian mobility inside buildings? So the things you've been talking about are very much geared towards the complexities of the channel and its measurements when you're driving around with 60 mph in the city, what gets simpler, what gets harder inside buildings?

A: The main difference is that in a macro cellular network, so the vehicular network, propagation is fairly predictable, in the sense that you can have a topographical database and really draw what will be the shape of a cell if you put a base station somewhere. For many reasons that's no longer feasible if you talk about indoor systems.

First of all: the databases that you need have to be very accurate. And the models that we have now for indoor propagation do not allow us to predict everything. Signals may propagate through an elevator shaft or may or may not propagate through the corner inside a building.

That implies that your cell shapes are no longer nice hexagons--that becomes very tricky and involved in an indoor system.

You have to move to a more dynamic frequency assignment technique, where the system finds out what's the best channel rather than to predefine cell sizes.

And what I find interesting is that the GSM system has been designed as a vehicular system and in terms of handover it's completely centralized, it's the base stations that take action. That whereas most of the operators market GSM as something for the business user with his hand held--and it surprises me that the system still works very well. But if all the operators start to use it as a hand held system they may at least run into some problems with the handover administration. The main answer is: try not to make it too much centralized but make it more of a dynamic channel allocation system.



Business (Clip 14)

If I can give you one megahertz of spectrum for a period of one year and you're allowed to do with it whatever you want--and of course there are technological problems--you cannot put a television channel anywhere or a cell phone channel anywhere without having technology that can support this. But if we forget about it for a moment and you can decide whether you put up a commercial radio station, a commercial television system or a cellular radio system for making telephony calls, what would be the most lucrative thing to do?

It turns out that amount of money that you can earn per year per megahertz is the largest for cellular radio. So even from a business perspective it's interesting to do this.

There's also very much a drive to provide wireless personal telecommunication services in an enormous amount of spectrum that's currently used by television systems. About half the spectrum below one gigahertz is used for television and broadcast. Whereas mobile reception of television is not so much a popular thing where you really need the wireless is for mobile applications; so many of the broadcasters now see pressure from the wireless operators to have their channel made available for wireless services.

On the other hand there are also some developments where broadcast systems, particularly once these broadcast systems are digitized, to do a lot of multimedia-type of applications in networks that are very similar to the broadcast network.



Cash Flow (money) (Clip 15)

If we see the evolution of new systems we can understand why some of the cellular operators are complaining about their cash flow. Even though it's a very lucrative business some of the operators saw that they started with manually operated systems in the 50s and then they introduced the first generation automatic cellular system and later on some digital systems.

Initially you have to do significant investments in buying lots for base stations, setting up base stations, interconnecting all these... So you start with a negative cash flow but you expect that once the system is operational that you gain a lot of money. Now if these life cycles become too short and you introduce a digital system before you earn money from you analog system the your total cash flow may continuously be negative rather than positive.

I don't know whether that's one of the reasons why some of the operators delay the introduction of new systems....

The main difference is that in a macro cellular network, so the vehicular network, propagation is fairly predictable, in the sense that you can have a topographical database and really draw what will be the shape of a cell if you put a base station somewhere. For many reasons that's no longer feasible if you talk about indoor systems.

First of all: the databases that you need have to be very accurate. And the models that we have now for indoor propagation do not allow us to predict everything. Signals may propagate through an elevator shaft or may or may not propagate through the corner inside a building.

That implies that your cell shapes are no longer nice hexagons--that becomes very tricky and involved in an indoor system.

You have to move to a more dynamic frequency assignment technique, where the system finds out what's the best channel rather than to predefine cell sizes.

And what I find interesting is that the GSM system has been designed as a vehicular system and in terms of handover it's completely centralized, it's the base stations that take action. That whereas most of the operators market GSM as something for the business user with his hand held--and it surprises me that the system still works very well. But if all the operators start to use it as a hand held system they may at least run into some problems with the handover administration. The main answer is: try not to make it too much centralized but make it more of a dynamic channel allocation system.



Cost (Clip 16)

You see that I've presented this material before several times and this is exactly a point in time where I invite some interaction from you in trying to figure out what is the cost of a basestation. Many operators regard this as strategic information that they do not want to hand out to their competitors.

Some estimates are: well, a basestation costs approximately $1,000,000. A signifact portion of this is simply the lot or all the things you have to do to be allowed to use a radio transmitter from a certain location. In my own country, in the Netherlands, we saw that the proposals from new operators--it was often regarded the key issue that the operators could promise coverage in certain paths where the local government were very much agains putting up new towers because it would pollute the skyline and the nice villages.



Estimate (Clip 17)

A: Of course the cost for equipment. The operational costs of a basestation are mainly in terms of the connection to the mobile switching center. That could amount to about 25 percent of the operational cost for the operator. Some estimates in holland are quite a bit less than the $1,000,000 that someone mentioned in the previous presentation. $300,000 Dutch guilders--that's about $200,000. So anyone from a cellular operator here who wants to disclose his cost and add something here?

Q: In Japan the installation of a basestation generally costs $100,000.

A: I'm surprised that what you mention is significantly less than the $1,000,000 I heard.

Q: Yes, but they have developed quite a bit--of basestations.

A: It's certainly this scale factor in the amount of basestations that you put down there but there may also be a difference in the functionality of the basestation. It depends on whether the basestation already does some kind of multiplexing and such.



Microcost (Clip 18)

Q: If macrocellular basestation costs $1,000,000, how much does a micro-cellular cost?

A: A lot of development has been put into trying to reduce the costs of these basestations. What you do not need is a lot and a huge tower but on the other hand the problem of locating antennas in an inner city are much more complicated. So one of the interesting developments are to build some type of central basestation controller and have an optical fiber go to the basestation location. And this optical fiber carries the high frequency signals. So all you have to do at the basestation is to convert from optical to radio signal. It depends very much on the technology that you use but there is a tendency to make this cheaper and cheaper and to do not a lot of switching and a lot of organization in the basestation.



Change of Calling Pattern (Clip 19)

The main interesting thing about this curve is: apparently the use of cellular phones have changed over the years. It started with people who really needed their cellular phones and made a lot of calls. And now more and more people have cellular phones. So the calling pattern is more similar to a normal telephone calling pattern. Many operators made mistakes when they first planned their system and they computed how many subscribers they could accomodate in a system. And certainly all these customers made twice as many calls that they were expected. And that has changed now into a more normal pattern of calls.



Surcharges (Clip 20)

In all the advertisements you see that GSM allows you--if you go to another country--to place a call to your colleague in the other hotel room. But if you see in practice what is the tariffing structure it becomes a quite expensive thing to do. But let's assume we have a Belgium subscriber and he calls a friend in Rome, whereas he is in Paris. So we have a signalling flow from Paris to Brussels and a call from Paris to Rome. We have a French operator involved who charges the Belgian operator because the subscriber pays his telephone bill through a Belgian operator. So the French operator has to charge the Belgian operator for a wireless link. This person is making an international call and he has to provide an interconnection to an international public telephony system and there is a surcharge of about 15 percent for providing services to subscribers that belong to a different system.

They're not so keen to do services for people who're not paying directly so they surcharge 15 percent. Well, then Belgicom says: well, you're not making calls on our system so we don't gain anything from you except if we have another surcharge of about 15 percent on top of this.

And then there is this Italian operator who says: you're using my system and you're placing another wireless call over my system, so again charges to the subscriber. And this is one of the reasons why there's so much signalling involved. It's also because of the tariffing and administration.



Connection (Clip 21)

Q: Is the connection to the MSC done through the local telephone carrier. Is that how they typically--they lease the line?

A: That's one possibility. And of course you get the regulatory problem whther the charges for this line are really fair charges. And in European countries your have the situation that there's one cellular operator that's also the national PTT-- and that's the only company that's allowed to provide leased lines. And then a competitor comes in and he has to lease lines from the other cellular operator.

That's of course a very strange situation. So this transition to opening the market for long distance telephony and leased lines usually coincides with new cellular operators entering the market.

Typically the railways, the gas and electricity organizations are now also allowed to lease [their telecom] lines.



Future of Cellular Technology (Clip 22)

In cellular it started with analog systems, like AMPS, the nordic mobile telephone system-- NMT, and now we're moving towards GSM, DECT, Japanese systems, into the mobile data.

There are a few networks in operation like the RAM mobitec system. There's also cellular digital packet data [CDPD] and this will finally be integrated into cellular based personal communications systems [PCS]. In the future we hope to see more advanced systems with higher rates, higher performance and many more users.

What I knew from the past is that although many standardization bodies aim at a single system that can do everything, Donald Cox [at Stanford now] always said: well, the propagation environment, the traffic pattern and the user requirements are so different that even though everyone is trying to migrate everything into a single architecture like UMTS, in practice we see many different systems appear on the market with different properties.



Digital Technology (Clip 23)

One of the problems of introducing the digital system is that operators found it very difficult to have their subscribers move to the digital system. When I have a cellular phone and it works well, why should I buy a digital one if I get the same service for the same price? So [operators] had to come up with a new price structure to get people to go digital or they had to admit that the analog system has no security at all, that anyone can listen to your telephone call. But they weren't too happy with advertising this so broadly.



Wireless E-mail (Clip 24)

I'm in fact not really sure whether if you see the current structure of the cellular phone system (their tariffing structure for circuit switched connections) whether that's really what we need in the future to provide all these services.

I don't want my electronic mail to be sent through a cellular phone system if for every short message I have to set up a connection. Where cellular systems are notorious for sending a lot of signalling. Even 24 messages before you set up a connection and then I send my one hundred bits of an e-mail message-- so there are different options to provide this type of services.



Conclusion (Clip 25)

So to summarize all this in one sheet: even though many people dream about having one system that can do everything in wireless communication, different applications require different solutions and if you want to make an economic solution and have cheap products that leads to different standards and it may take a while before we go to one universal radio access technique. We see the opposite in practice. We see so many systems that have different properties and different possibilities.



Wireless Communication © Berkeley Multimedia Research and Jean-Paul Linnartz, 1999.