TechTalks Transcript
Implementing a Campus Wireless LAN: What Are the Realities?
 Judith Boettcher [JB] |
 Howard Strauss [HS] |
 Alex Hills [AH] |
October 19, 2000
Audio
• Streaming
MP3
• Download
MP3 (Download
Tips)
JB: Welcome to the CREN Tech Talk series for fall of 2000, and to this session on "Implementing a Campus Wireless LAN: What Are the Realities?" You are here because it's time to discuss the core technologies for your future campus. This is Judith Boettcher, your CREN host for today and I'd like to thank our CREN member institutions and Cisco Systems for their sponsorship of today's Tech Talk.
I'd also like to welcome Howard Strauss from Princeton who is here with us again today as our technology anchor for Tech Talk. Howard is a well-known web technology expert and portal expert and always brings us interesting, unexpected little thought surprises. Welcome, Howard!
HS: I better-�
JB: Surprising you after all these times!
HS: I better do that today! Anyway, thank you, Judith. Let's see. I'm Howard Strauss, the technology anchor for the Tech Talk series of technology webcasts. In this webcast, I invite you to join Judith and me in a lively technical dialogue with our guest expert, Alex Hills, that'll answer the questions you'd like answered and ask those very important follow-up questions. You can join in this dialogue by sending your questions via e-mail to expert@cren.net any time during this webcast. If we don't get to your questions during the webcast, we'll provide an answer in the webcast archive.
What would the world look like if you could see the radio waves and microwaves that are used for wireless communication? In most places, there would be an overwhelming flood of visible energy that passed effortlessly through some solid objects, ricocheted off others and was completely absorbed by anything made of metal. Our wireless local area network computers and base stations are just one more microwave source added to this dense, chaotic m�lange of spherical electromagnetic emanations.
Unlike our physically connected wired LAN's in which a computer is firmly connected to a fixed point in a network, the microwave signal from a wireless computer spreads off into space in all directions. All base stations within this range absorb the signal and need to deal with it. The base stations also absorb the microwave energy from computers that has bounced off ceilings, floors and other objects and that arrives a bit later-and possibly weaker-than the main signal. The base station must also filter out the rest of the radiation clutter that comes from cell phones, satellites, radar, radio and 300 channel cable TV broadcasts.
You may not realize it, but all that stuff - every radio station, AM and FM, TV channel, police broadcast, long distance telephone transmission and much more is zipping right through you, wherever you are. The signal from cordless phones, the leakage from microwave ovens and the transmission from wireless LAN's, however - all of which are the same unlicensed FCC band - don't go through you. The frequency in this band is strongly absorbed by water, which is why it is used in microwave ovens. When the water in a hot dog absorbs enough of this energy, for example, it gets very hot. To a microwave oven, cordless phone or wireless LAN, you are very close to being just a five or six foot mobile column of water, absorbing radiation that was intended for cooking or communication.
A wireless LAN attaches to a wired network. You need a well-designed, reliable wired network before you can even think about building a wireless LAN. To build a wireless LAN, you'll need to solve many additional tricky problems that don't exist with wired networks. Those microwave ovens, for example, send out lots of noise that may make it difficult for base stations to communicate. Metal floors in buildings will definitely shield wireless signals from passing through them. Every different material - books, file cabinets, furniture and even people - can distort, block and reflect wireless signals. And since the wireless signals spread spherically off into space, there are security problems that transcend those of wired networks. Someone outside the building, or even outside your campus, for example, might be able to pick up wireless transmissions or transmit into your network.
As complex as all this sounds, many universities have successfully implemented wireless LAN's, including my alma mater, Carnegie Mellon University, where Alex Hills prowls the halls with his wireless computer and his wireless expertise. While he can probably not make wireless LAN's as simple as A-B-C, he'll definitely cut through the electromagnetic clutter. As Albert Einstein once said, "Everything should be made as simple as possible, but not simpler." That's just what we'll do for you now on today's webcast of Tech Talk. Judith?
JB: Thanks very much, Howard. You know, I love your image, I think, of walking through all these radio waves, although you know, it reminds me of what we often see on Star Trek as those invisible hard doors. I see them coming down the pike fairly shortly, I think! But with that, let me say I'm very pleased to welcome back Alex Hills as a returning Tech Talk expert. He almost doesn't need an introduction, and we always have more about Alex on the website. But of course, here's an introduction anyway!
Alex is a Distinguished Service Professor of Engineering and Public Policy in Computer Science at Carnegie Mellon University. There, he teaches telecommunications policy and does research on wireless communications and the development of telecommunications systems in rural and remote areas. He served until recently as Vice Provost and Chief Information Officer at CMU and in this capacity was responsible for the development and operation of Carnegie Mellon's computing and telecommunications systems, including the Wireless Andrew project of which there's a lot of links on our website.
Alex in another life is also the Vice President and Chief Technology Officer of New Horizons Telecommunications where he consults on strategic use of communications and information technology. Welcome back to CREN Tech Talks, Alex, and I just have to ask-does the fact of your interest in remote telecommunications have anything to do with your spending the winter in Alaska?
AH: Thanks, Judith. It's great to be back. And yes, you've made the connection. Actually, my roots are here in Alaska building telecommunications systems in the far reaches and that's why I'm back here.
HS: Okay! Alex, why don't we start trying to pin down what a wireless LAN is, anyway, and I also wonder. We're talking about wireless LAN's. Are there wireless wide area networks?
AH: Sure. I like to talk about wireless WAN's, MAN's and LAN's.
HS: What are MAN's?
AH: The wide area network, the widest of the wireless networks, would probably be the low-earth-orbiting satellites like Yridium and GlobalStar that literally cover the earth and operate, by the way, at very low data rates, 2.4 kilobits per second, maybe up to 9.6. Then in the metropolitan area, you could talk about cellular and some of the packet radio systems that also provide data service. And then - oh, and those currently operate at speeds, well, 19.2 kilobits per second is typical for cellular and packet radio. That's a peak rate and not a throughput, which would be lower. And then in the local area network, the wireless LAN's which we're talking about today, there the coverage area-as we'll discuss-is maybe as much as a campus and the data rates are much higher, currently about 11 megabits per second.
HS: Okay, now, on campus, we're not going to be using - we'll only talk about these wireless LAN's, we're not talking about cellular stuff or low-earth-orbiting satellites, right? We're - what are you talking about?
AH: I'm talking about a very specific technology that consists of a unit called an access point, which is actually a combination radio transceiver and data bridge. And what we do is we mount those on the walls in a building, strategically located in order to provide complete coverage.
HS: They could be mounted outside a building, too, right? For outside coverage?
AH: They could be, that's right, um-hum. And so the access point is effectively the interface between the wired network and the wireless link. The mobile computer, the laptop or even a desktop computer then is equipped with a network interface card which talks by radio to the access point.
HS: And when we talk about these global things that are transmitting, do you include PDA's?
AH: Most PDA's nowadays don't have PC card slots, so although there are ways to adapt PDA's for use with various wireless networks, the wireless LAN products that we're talking about today use network interface cards which come in a PC card type II form factor and so you can use those with any platform that has such a slot. And that generally means laptops or hand-held computers, Win CE machines, for example.
HS: Okay, I know that Carnegie Mellon and a lot of universities have done a great deal, including even Princeton has done a great deal with getting wireless stuff all over campus. Why are we doing this? What does this buy us? Or are we just having laptops and desktop machines hooked into the network?
AH: Well, with us it started back in 1993 when we had - when we wanted to create an infrastructure for our wireless researchers, but we had even at that time the vision that we could have people using their mobile computers, laptops or hand-held machines, while moving around the campus. At first, it was our researchers and then later on, when we built the production network, it was all of our campus community. So we're really talking about mobility. We're talking about the ability to freely move about while remaining continuously connected to the campus network or to the Internet.
HS: But I mean, are people really going to walk from one building to another while they're computing? I'm trying to get a better idea of what "mobility" means.
AH: Well, it's an interesting thing to visualize, isn't it?
HS: I think I'd stumble down steps and fall over and things like that.
JB: Well, you know, think about how people are using their cell phones.
HS: It's scary!
AH: Well-�
HS: If I see somebody in a car with a wireless computer, I think I'm going to take another road!
AH: Um-hum. So maybe, you know, it's - I mean, I think most people use the computers while stationery, but the idea is the convenience to use it wherever they happen to be. I do see quite a few people walking around using their computers while they're walking, though. It's sort of like reading a book while you walk - equally dangerous, I guess!
JB: That's interesting, isn't it?
HS: Just cheaper to drop a book.
AH: It is interesting, though, that the roaming capability of the wireless LAN products does provide a continuous connection as you move along.
HS: So you can get out of range of one station and in range of another, and you're just passed off like you would be in a cell phone network?
AH: Exactly, you're handed off from one to the other.
JB: Alex, what kinds of things did your researchers find that they really, you know, used that mobility capability for? Was it a kind of a clear direction, clear vision from the beginning?
AH: The vision was clear from the beginning. In fact, I have PowerPoint slides from 1993 era-�
JB: Yeah?
AH: --that articulate that vision. We originally built it for the researchers to do their work, but we very quickly found them using the capability for their personal use, too. For example, bringing laptops to conference rooms and, you know, maybe tapping away on e-mail or maybe even sending messages back and forth across the room, making comments about the speaker of the moment, something like that. So we found our researchers using their mobile computers for personal use and the word got out and pretty soon we had a demand from the campus community at large. Everyone wanted to use the wireless capability.
JB: Well-�
HS: Okay.
JB: Go ahead.
HS: Okay, you mentioned that the advantage is mobility. What about other advantages like either cost savings or replacing your wired network or doing better teaching or all the other possible things that one might think about?
AH: Well, a couple of other reasons that people are sometimes interested in wireless LAN's are, one is because of wiring problems. Maybe you have an old building that has inadequate wiring or maybe you have a building which is a historic building-and I know you have some of those at Princeton and we do at Carnegie Mellon, too, where we would prefer not to tear up the walls just to put wires in. So in those situations, you can use wireless LAN technology as a wire substitute, and you can use it as a cable substitute, too, if you have to cover some distance, maybe a few hundred yards or even a mile between buildings. There's a way to use the equipment to cover those kind of distances on sort of a point-to-point basis.
HS: But what about the point about does your wired network get replaced? I mean, I know you have to connect the access points together, but do wired connections disappear when you do this?
AH: Well, I think generally, the answer is no. The approach that we've taken at Carnegie Mellon is to supplement our wired network with the wireless, so we continue to operate our wired network. We continue to provide service to desktops using the standard wired network, CAT 5 wiring, and so on. But then we add to that the wireless network which allows the mobility that we talked about. Another point is that there's a lot of wire in wireless, and by that, I mean that the access points need to be tied to the wired network so there's wire that has to actually run, both communications wire and power wire has to run out to each access point.
HS: Okay, we have a question here from Ena Haines at Columbia University and it's a question that I'm sure a lot of people are thinking about, even if they don't ask it. And what Ena says, I'll just read her question to you. She says, "We'd be interested in the current knowledge about health risks of wireless LAN's. Can you give us any good links or references?" What about that? I think a lot of people have expressed some concerns about cell phones and this seems to be right in the middle of the microwave band and everybody knows microwaves are dangerous to people with, say, pacemakers and whatever. And I mean, you know, you're not supposed to put your cat in the microwave, that kind of thing.
AH: There was a good article in IEEE Spectrum within the past few months on this very subject. My understanding of the research on health effects is that no link between radio energy and - radio energy from the kind of devices that we're talking about here - and negative health effects has been demonstrated.
But on the other hand, it also hasn't been proven that there are no health effects because it's very hard to prove something like that. One thing that we can talk about is we can talk about exposure. In fact, that's what the IEEE Spectrum article focuses on. And if you're concerned about health effects, you might want to think about how much exposure you get to radio energy. And I don't think there's any doubt that the most commonly used device that gives the most radio exposure is the cell phone. That's because the antenna is so close to your body, so close to your head. The power drops off as the square of the distance. The power that comes out of a cell phone is about half a watt, but it's very close to your body, so-�
HS: Is that the square or the cube? Because it's a spherical wave.
AH: It's the square, yeah. I think it's the square.
HS: Okay.
AH: So anyway, that's the one, if people want to worry, that would be the one to worry about. The power output on the wireless LAN cards that we're talking about on your laptops, it's about a tenth of a watt and it's arguably farther away from your body than the cell phone is. And then the ones that you want to worry about the least are the access points or the cellular base stations which are not very close to you at all.
JB: That sounds good. You know, another question that comes up often with wireless that a lot of folks worry about is the area of security, Alex. And we had a question from Shana Leonard from the Arizona Health Sciences Library and they have an implementation there where they have a wireless LAN in a classroom and they're thinking about expanding it to cover the whole library. What - do you have any advice as far as security, and is security a problem on the wireless networks?
AH: Well, security is certainly an issue that people worry about because of the obvious fact that these signals are traveling through the air and it's at least theoretically possible for anyone to pick them up. And that's why the wireless LAN manufacturers have two different kinds of capabilities that are helpful. One is encryption and 128-bit encryption is available on wireless LAN products. It is typically shared key, however, so there are going to be some management problems associated with that.
And then the other capability that's starting to come out is authentication. And the way that works is that an access point can be upgraded to look like an authentication client and so it can prohibit access to the wireless network until a user properly authenticates on. And then once the username and password are properly entered, then the authentication server is consulted and once that authentication process is complete, then the person is allowed onto the wireless network. So those two capabilities are there for people who are concerned about security.
HS: Are most people using encryption when they have wireless LAN's, or is that just something you could do but nobody's doing it?
AH: I - we're not doing it at Carnegie Mellon right now.
HS: Why not?
AH: Well, because it's a university and security isn't probably as great a concern to us as it might be in some sort of a commercial enterprise, for example. We also provide other kinds of security in the sense that it's necessary to authenticate to get access to certain servers, for example, so there are some kinds of security that are inherent in our system without encryption.
HS: Sure, but anything else? I mean, anything that comes through this thing that's not encrypted, it seems easier to intercept than it would be if it's wire.
AH: That's correct, and it's also possible for signals to be intercepted on our wired network as well.
HS: Sure, it just seems easier to intercept them from the air.
AH: Um-hum, that's right. And it is possible that the signals can be sniffed. That's why the encryption capability is provided, but we don't happen to be using it at CMU.
HS: Okay, and I mean, with the signals, you're not even sure where they are in terms of - I mean, you don't know how far outside buildings they extend. You know, with a wire, you know where the wire is, kind of.
HS: Okay, at the current technology, the bandwidth and the speeds of wireless devices are lower, right, than network devices? Does that mean there's some applications that just don't work on wireless, or at least don't work today?
AH: Well, I guess streaming video would be the obvious one to be concerned about. Let me just mention what's happened recently with respect to speeds on wireless LAN networks. The 802.11 standard was adopted in 1997, about three years ago, and that's the standard that really has helped this whole industry start to expand because it allows interoperability between different products. The 802.11 standard adopted in '97 is a two megabit per second standard, two megabits maximum, and then soon after that, the 802.11B standard was adopted. That's an 11 megabit per second standard, and both of those, by the way, use a radio frequency of 2.4 gigahertz. It's the 802.11B product that's becoming very popular on the market, so it's the 11 megabit per second product. Now, what's in the works is the 802.11A standard and that one's [inaudible].
HS: We're going back. Or [inaudible] C that follows [inaudible].
AH: I won't be able to defend the numbering sequence to you!
HS: And I also saw something about 802.11F.
AH: I can't talk to you about that, but I can about 802.11A which is probably going to be the next popular wireless LAN standard, and that's a 54-megabit standard, but it also operates at a different radio frequency. It operates at 5.7 gigahertz, so we won't have backward compatibility on that one. This'll be more of a forklift upgrade.
HS: What's the effect of going to higher frequency? Is that going to give you more range or what? What does that do?
AH: If you hold all other things equal, higher frequency will give you less range.
HS: So why is this a good thing?
AH: Why is it good to go to higher frequencies?
HS: Yeah.
AH: Probably two reasons. One is because there's more spectrum available at higher frequencies and so that facilitates ramping up to higher speeds. Generally, there's more spectrum availability, more-and there definitely is at 5.7 than at 2.4. The other one is because, as I think you alluded to in your introduction, is your unlicensed bands and interference does get to be an issue, although I would argue it's not fatal by a long shot. But as interference levels rise, then one response is to go to higher frequencies. In fact, the very first wireless LAN system we had at Carnegie Mellon, which was at 900 megahertz, and then we moved up to 2.4 when we went with the 802.11 product. So as you move to higher frequencies, you get more spectrum availability which lets you run at higher speeds, and you also move away from the interference. But then the interference sort of chases you and, you know.
JB: Like building highways, huh? The more highways you build, the more-�
AH: Yeah, and I think interference is another issue that we might want to talk about, but none of this interference-either at the 900 megahertz band or the 2.4 gigahertz band-it hasn't proved to be fatal to us by a long shot. It's actually just proved to be something that we need to consider when we do our design.
HS: Could you talk about your design now? How do you go about designing this wireless LAN?
AH: Well, that's probably the most interesting question of all.
HS: Good we got it in!
AH: That's right. When we talk about design, we are thinking about two things. We're thinking about where we place the access points and we're also thinking about the frequencies that we assign to the access points. In the 802.11B standard, there are three different channels that you can use without having any overlap or interference between channels. So when we do a design, we select our access point locations. There are almost an infinite number of possibilities [inaudible].
HS: So how do you decide even where to start? I mean, do you want them as high as possible? What are your sort of heuristics for this?
AH: I was just going to say that the results of these two decisions that we make actually affect two things. They affect coverage, which is sort of the obvious one, that you need to place the access points in the way that you do get continuous coverage of the target space. But they also, both of these decisions also affect capacity in a very direct way. That is, how much aggregate throughput you're going to get through your whole system.
Now, when we place the access points, we do generally try to place them high, fairly near the ceiling. That helps to get over-a lot of times you have walls that are not ceiling height walls, or other obstructions tend not to go all the way to the ceiling. So that helps, but given that we're going to place them maybe a foot or two below ceiling height, we still need to decide which parts of which corridors or which rooms we're going to use for those locations. And that turns out to be a very challenging problem.
We developed-several years ago, we developed a design procedure that consists of taking lots of measurements and lots of cut-and-try. And that procedure allows us to select the access point locations, or finally arrive at a set of access point locations, and then also assign the frequencies that are going to give us sort of the optimal solution.
HS: So if I understand this, you pick likely places for the access points, then you wander around the building taking some measurements. And if things don't look the way you want, you move the access points. And you sort of do that iteratively. Is that what's happening?
AH: That's exactly right. It's a cut-and-try process and I think we can say the first try virtually never is the right answer, and so then you go through this iterative process.
HS: And what are you trying to achieve? Are you trying to make it so that every spot in the building sees only one access point, sees two, what? What are you trying to get? How do you know where it is?
AH: You're getting good! We're trying to get complete coverage, that's the first thing. So that every spot in the building sees at least one access point. And then we say, "sees the access point." We also build in some margin to account for this interference problem that we talked about. So there's a little bit of radio engineering in here in that you build in some extra signal strength to allow for that sort of thing. So that's the first thing. You want to be sure that every point is covered by at least one access point.
But then in order to help with the capacity concern, you would like to not have any point covered by two access points that are on the same frequency. The reason is because that will actually slow down the network, if it happens. So what we do is we select the access point locations in a way that minimizes coverage overlap as much as possible, but it'll never be possible to completely eliminate coverage overlap. That's because those radio signals don't know where we'd like them to stop.
HS: [inaudible] no [inaudible].
JB: Okay!
AH: So given that we know our access point design will result in some coverage overlap, then we select the frequencies in a way that minimizes the co-[inaudible] overlap and that is what constitutes the optimal design.
HS: And do you have some - I mean, if somebody wanted to do this at their university, do you have some description or some document that describes how to do this, or some software that helps support this [inaudible]?
AH: Well, we do have some software. As a matter of fact, I mentioned the manual process. We've since developed a piece of software that reduces a lot of this to-well, to something that an untrained person can do and it also speeds up the process dramatically and it also allows us to collect much more data than we would using the more manual process. So we have written a piece of software to do that.
HS: And is that software available?
AH: That software, well, Carnegie-Mellon has the idea that they want to exploit this commercially, so it's possible that that software is going to become available through a commercial outlet.
JB: I think there is a design guide document that we do have linked off the site. Maybe that's part of the earlier manual process that's been documented.
AH: Yes, I think so.
HS: Is that just to whet your appetite for the software when you start to sell it?
AH: Right.
JB: And for-�
AH: I guess while we're on this subject, I could probably also put in a small plug which is, I think, on your website and in your introduction, you indicated that I'm spending my sabbatical year with New Horizons Telecom and that's one of the things that New Horizons Telecom does is to help clients to do these kinds of designs.
HS: Okay, is that company based in Alaska?
AH: That company is based in Seattle and they have an office in Alaska, one in Spokane, Washington, and one in Atlanta, Georgia.
HS: Okay. Can we talk a little bit about the cost of doing a wireless LAN? What's involved besides buying the cards and buying the base station? And in fact, what do those things cost?
AH: Well, the access points list for around a thousand dollars apiece and the street price is more like seven or eight hundred. I'm thinking right now primarily about the Lucent product, which is the one that we use, but others are pretty much in the same ballpark. The network interface card lists for about 175 and street price on that is probably about 150 or maybe a little bit less. So that's equipment cost. But the question you asked me, I think, was what's the total cost likely to be of building the whole system?
HS: Sure. Yeah, that was the question.
AH: Right. And there's kind of a range. It generally ends up costing a few thousand dollars per access point to do the installation. That's after you consider the cost of design, of running the wiring and actually installing the access points.
JB: What about if one-if an institution wanted to start with perhaps a, you know, section of a library that was kind of open up. You know, would they need ten, 15, 20 thousand to get started with a small project?
AH: Oh, sure, that - sure.
JB: I mean, would that cover?
AH: That'd cover it real easily. I mean, if it's an open area, you might be able to just cover it with one access point.
HS: How many square feet - this is probably a question that has no answer - but how many square feet does an access point cover?
AH: It's a good question. The coverage of an access point is very dependent on local conditions. By that, I mean the layout of the building, the floor plan of the building, the type of construction material used in the building and so on. But for indoor installations, we usually use a rule of thumb of about five to ten thousand square feet per access point. So in other words, in New Horizons, if we're talking to a client and we're just trying to develop a rough idea of what it's going to take to do a job, then we use that sort of rule of thumb.
JB: Okay.
HS: Is that going to change when you go to the higher frequency? You said the higher frequency standard has less range. Are we going to be-�
AH: Well, what I said was that if you hold everything else equal, if you hold everything constant, [inaudible]-�
HS: So you're saying if you do things like you don't increase the power.
AH: Right.
HS: [inaudible] there.
AH: That's exactly right. So since we don't have any five gigahertz product yet, we don't really know for sure what the range will be, but we do know that if everything was held constant, that the range would be smaller. Just to illustrate what I'm talking about, when we moved from the Lucent 900 megahertz product up to their 2.4 gigahertz product, we actually ended up with cell sizes or coverage areas which were roughly the same, and that was because Lucent designed their new product to make that happen. And so in other words, they didn't hold everything else equal and so it turned out pretty well.
HS: I mean, is what they did just increase the power? Or were there other things they did?
AH: That was - power is one issue. Another one is what radio engineers called receiver sensitivity. So it's sort of the other end of the link, right? You can increase power or you can reduce receiver sensitivity. Either one of those will have the same effect.
HS: Okay, you talked a little bit about 802.11 and we have a question here from Ryan Cunningham at Murray State University in Kentucky who asks about Blue Tooth. Ryan says, "I've read about the forthcoming Blue Tooth wireless standard. Will this standard be compatible with 802.11?"
AH: Well, the simple answer is no. But let's talk a little bit about what Blue Tooth is. We talked at the top about wide area networks, metropolitan area networks, local area networks, and then we could go down one step more in terms of coverage area and talk about personal area networks. And that's what Blue Tooth is.
Blue Tooth is a new standard, new wireless standard, which is aimed at a range of up to ten meters and it's intended largely as a replacement for cables and connectors so that to sync up your PDA with your desktop machine, for example, you don't need a cable. You know, they just talk to each other by radio. Or another example would be in maybe more of a futuristic scenario, your every appliance and everything in your house has a Blue Tooth chip. It has a Blue Tooth radio on it, and so for example, when my refrigerator notices that I'm out of beer, it tells my PDA to put beer on my shopping list. So there's this idea-�
HS: That's something your company's working on, I assume?
AH: Oh, of course!
JB: And then you can call the Pizza Hut at the same time, right?
HS: Wireless.
AH: Right.
AH: Another serious comment about Blue Tooth, though, is Blue Tooth operates in the 1.4 gigahertz unlicensed band and it does not talk to 802.11. It's a different protocol.
JB: So will that create interference in those channels?
AH: It will, right. It will. And using all of these - these are mostly all spread spectrum techniques. Interference from a non-compatible device actually looks like noise. And-�
HS: I mean, at CMU, do you tell your faculty and staff, "Don't use cordless phones, don't turn your microwave ovens on"? I mean, do you have to go out there and tell them, "Don't use things that generate noise?"
AH: Well, let me - sure - talk just a little bit about that. The most serious interferers at 1.4 gigahertz, as you just implied, are microwave ovens and 2.4 gigahertz cordless phones and actually, microwave ovens aren't that much of a problem, mainly because they're well-shielded. So that's not the one we worry about so much as the 2.4 gigahertz cordless phones which are intended to emit, that's how they-�
HS: Oh, sure, if you shielded them, they wouldn't work!
AH: Yeah. And so we actually do have a policy in place that asks people not to use those kinds of devices. As much as you can control this kind of thing inside a university environment, we do. But let me also say that what happens if you do experience this kind of interference? It's not that everything stops, normally. What actually the user normally experiences is a slowdown in the speed of the connection. But not usually a total disruption.
HS: Okay, and that's just because of the robustness of TCP/IP?
AH: It's, no, well, actually it's-there's robustness on several levels. The spread spectrum technique that's used in 802.11 is quite robust in itself, and that's sort of the whole point of it is to be able to overcome interference and noise. And then you also have, at the higher layers, you've got the ability to retransmit when there are errors and so on.
HS: Okay, if I can get one more building question in, and that is, if someone's building a new building, because new building's going up all over campuses and things, are there things that you should consider or that you should do if you know that you're going to put wireless in that building?
AH: Well, one thing that you might consider doing is looking into a wireless design tool from a company called Wireless Valley. My colleague, Ted Rappoport at Virginia Tech has a company that creates a piece of software which does radio propagation modeling and it's one that I think would be very helpful in this kind of situation where the building isn't yet created. You could use that tool to actually place your access points. In other words, a kind of design methodology before the fact, where the one I talked about before was after the fact, after you're actually working with the building. That might help you to locate those access points and then on the strength of that, you could design your wiring right into the building. You could make it part of your building plan, the wiring that's going to serve the access points.
JB: So getting some of that power up towards the ceiling, then?
HS: Well, I was thinking mostly - well, I guess it would help with that as well, but I was mostly thinking with deciding where to put the access points, which room, which corridors and that kind of thing. And by the way, speaking of wiring for access points, at the installation we did at CMU, we ran communication wire and power to every access point location. The equipment that we're using needs 110 volt AC power. But there are some newer products that are now coming out on the market that allow those units to be powered over the communication wire, right over the count five wire.
HS: So that saves you the cost of dragging the power outside?
AH: Exactly, and that can be a pretty significant cost.
JB: We have a number of questions coming in. Howard, are you going to pull up that one perhaps about the number of folks that can be served by an access point?
HS: Yeah, there's actually a whole bunch of questions that have come in here. Sure. There's a letter from Jeff Nucciarone, I think.
JB: Yes, right, from Penn State.
HS: Yeah, okay. Jeff says, "What's the total effect of bandwidth on the number of people that you can have on an access point?" He wants to know when do you start getting congestion? I guess this goes back to one of the questions of design. I mean, if you were designing this thing and you knew there was a classroom in the building and everybody was going to use one, how do you figure how many of these access points do you need?
AH: Well, let's take the question-one way you can take the question is if you just have a system with a single access point and how many users will that single access point take care of? That's a different question than the one where you're building the whole system and you're worried about the interaction between access points, which is what I was alluding to earlier when I was talking about coverage overlap and things. But if we just isolate the access point, it depends - to a great extent, it depends on what kind of traffic you're running - and I guess that sort of goes without saying, but it's true. Given that, with the 11 megabit product, we think that the range is somewhere between 30 and 50 mobiles per access point. But now-�
HS: And that's, you're assuming some typical user just looking at a web page here and there, that kind of thing.
JB: Or e-mail.
AH: Or running e-mail or something like that, yeah. Now, another - if you're in a situation where you have a really high density of users, then there are things that you can do with your design. For example, let's say you have a lecture hall with hundreds of students in it. One thing that you could do there would be, you could take - and I mentioned that there are three independent channels in the 802.11B standard.
HS: Yeah.
AH: So you could come and you could overlap the coverage areas of those access points. Earlier, I sort of implied that overlap is bad - and it is if it's co-channel overlap - but you can use non-co-channel overlap to help the problem by coming in and overlaying coverage areas on top of each other or nearly on top of each other and that will actually increase the capacity that you have within that particular physical space.
HS: And we have a question about those channels from David Flagg at Wayne State University. I think that's at Wayne State University. And he wonders why only three channels? I mean, even my little cordless phone at home has 11 channels on the thing. Wouldn't it give you more versatility if there were more channels?
AH: Well, let me talk about that a little bit. First of all, there's a certain spectrum that's allocated in this unlicensed band. Now, the spread spectrum signal that's emitted by a wireless LAN unit occupies a fairly broad bandwidth, so the question is, how many of those fairly broadband wireless LAN direct sequence spread spectrum signals can you fit within the amount of space that's allocated by the FCC for this band? In fact, there are 11 channels. There are 11 channels that are possible within 802.11B direct sequence that will fit into the unlicensed band, but the problem is that those 11 tend to overlap with each other. So if you can visualize this, channel 2 is offset from channel 1 by just a little bit and they mostly overlap with each other in frequency. And then 3 mostly overlaps with two. It's just a little bit, offset just a little bit more.
So after you work all that out, it turns out that if you want channels that don't overlap with each other in spectrum, if you don't want them to have any frequency overlap, then you need to use channels 1, 6 and 11, and that's how you get your three channels.
JB: Okay.
AH: Was that clear?
HS: Yeah.
JB: Yeah - no, actually-�
HS: That's very clear. It's like when I used to live in Philadelphia and we had a TV set that had all these channels on it but we only got three.
AH: Okay, I'll accept that!
JB: So to go back to the lecture hall example, though, and those three channels, what you would do then - if I understand this correctly - is you might have multiple access points, some of which would maybe talk the language of the frequency 6 channel and a couple others that might be on the 11, channel 11. Is that correct?
AH: That's right. They all talk the same language, but if they're on different channels, then they won't hear each other. That's for sure.
JB: All right, thank you for clarifying that.
HS: And the little transmitters in these laptops, I mean, they just hunt around for all the channels? I mean, your access point is tuned to a channel. What about the little laptop transmitting? Does it transmit on all channels?
AH: Yes, they are what's called frequency-agile and so they can hunt and they can find the best signal. But it does actually raise an interesting point, which is, which signal will the mobile unit settle on? And the way a lot of the products work right now is, they'll select the strongest signal strength, which might make sense sometimes, to select the strongest signal strength.
HS: Yeah, but not if there's a whole classroom full of people.
AH: If you're in this lecture hall situation that we just talked about, what you'd really like is you'd like more of a load-balancing.
HS: Yeah.
JB: Um-hum.
AH: You'd like them to select the access point that's going to balance out the load, and so that's an improvement that would probably help with some of the products.
HS: You're saying "would probably help." Does that mean it doesn't exist, that kind of facility, now? I mean, that there's no load balancing.
AH: In the products I'm familiar with, I think they may have it on the drawing boards.
HS: So you could have really - I mean, you could have extra capacity in these access points, but one could saturate, right?
AH: That's possible, yes. That's probably a pretty good reason to, if you're going to overlap your coverage with those access points as I described for the lecture hall, it'd probably be good to place them all in fairly close proximity, or you know, maybe at different corners of the lecture hall or something like that, so that people in one part of the room would tend to get one access point and another part of the room get another access point.
HS: Yeah, is there any special software or anything that you need for this wireless stuff to do network management? Are there any special network management tools?
AH: Yes, but in most cases, not very well-developed. Most of - I think maybe even all of the access point products have an MIB that you can use with SNMP.
HS: Tell our folks who don't know what MIB's are what an MIB is.
AH: Oh, an MIB is - what is it? I'm not even sure if I know what the acronym stands for! Do you know, Howard? It's a management information base, I think.
HS: Or something like that. I don't know. But functionally-�
AH: So functionally, it's something that your SNMP workstation can talk to and enable it to monitor that particular unit. But the network management tools are not nearly as well-developed as they are for wired networks.
HS: Okay, we're getting toward the end of our time. Actually, come to think of it, we are past!
JB: We are very close, and actually, I was - you know, we've got a number of other questions, and I think we answered them. So I guess I would ask those folks who did send questions in, if you don't feel as if your question got answered, send it in real quickly here and we'll try and get it answered. Howard, do you want to take another final question?
HS: Yeah. Normally, at the end of this thing, we try to address the question of suppose you're at some university and you haven't done this thing yet. You're at a university, you don't have any wireless stuff in, but you're thinking about it. Alex, how do you get started?
AH: Well, if you're - if by getting started, you mean sort of dipping your toe in the water-�
HS: Yeah.
AH: That's pretty easy to answer. It's probably buy an access point or two and a few cards and hook them up and play around with it a little bit.
HS: Okay, but where's - if you were going to put them in some place - where's a good place? I mean, where'd you put them? Would you start by putting one of these things in a classroom or would you put them in your student center? Where's a good place to - I mean, after you've done the experimentation in your office and down the hallway and things, where's the first application that you would-�
AH: Well, I can tell you what we're seeing at some universities. Some universities just make a decision to just do this on a campus-wide basis. In other cases, we have seen the business school as the first one to wholly embrace this technology, and we've seen that at several different universities, where the business school will make a decision, they want all their MBA students to be equipped with a certain laptop computer and they want them all to be equipped for wireless and they want to cover the business school building or the business school space with wireless. That's one kind of typical situation, if a university doesn't go for doing the whole campus.
HS: And-�
AH: And I suppose another sort of obvious answer to your question is that if they need help, if they need technical help, then companies like the one that I'm working with this year are out there to provide the kind of design help, if that's needed.
HS: And I mean, when you first do this, nobody has the little PC cards, right? What do universities do to encourage people to get those things? I mean, because in the beginning, the thing's available only in some little small area of the campus. I mean, your business school example, would you just go out and tell everybody in the business school, "Do it"? Is that what it takes in a university just to-�
AH: That's what we did is we just built the wireless local area network everywhere and it's such a popular thing that everyone wants to buy the network interface cards, which we sell at our computer store. CompUSA sells them, too. Was that the question? Or-�
HS: Yeah, I was - okay, but you're just saying that because it's such a neat thing, really you don't have to push people too hard to get the cards.
AH: That's been our experience, that's for sure.
HS: And do you offer cards at a discount or-�
AH: [inaudible]�
HS: Go ahead.
AH: At our computer store, we try to do the best we can on pricing, but we never sell anything at a loss. We operate our computer store for the convenience of our university community so we try to price everything as low as we can.
HS: Okay, any other last-minute advice that we might have missed here that you could give to folks who are about to do this?
AH: I think my main bit of advice is don't underestimate this design problem. You can slap a few access points in and you'll get it to work, but you won't necessarily have the capacity that you would get if you did the design properly.
JB: Okay. And there is a little bit of at least preliminary design guidance in the document that's on the-that did come out of the first stages, first couple of stages at CMU.
AH: Right.
JB: Okay. And then there's other resources that we've mentioned throughout the talk. Okay, anything else before we go for the closing comments? This has been a really great session again, Alex, and as always, we could go on for quite some time here and go into more detail.
AH: I've enjoyed it. Thanks a lot!
JB: All right, but we have-�
HS: Next time, we'll have to do this from Alaska.
JB: Well, that's where you are today, Alex, right? I don't know if we shared that with folks.
HS: I don't know if we shared that with folks, but I was expecting CREN to fly me up there, Judith.
JB: Oh, all right, well, next time, Howard. And in the summertime, I suspect. But thanks to everyone for participating with us here today and just a quick reminder to be sure to continue to set aside Thursdays at 4:00 Eastern Time for the series of Tech Talks. Our next session-we're back to a two-week interval-also features a returning Tech Talk expert, Mark Bruhn from Indiana University. He's the Security Policy Officer there and he's got an interesting topic called the Ten Top Mistakes that Users Do that Make It Really Tough on IT Professionals for Security Purposes. So be sure to put that in your calendar.
Many thanks to all the institutions who support these Tech Talks, and thanks to everyone else who made the event possible today. And a reminder, a special thanks to Cisco Systems and their Wireless Aeronet products for supporting this program. Special thanks also to Alex Hills, our expert today; to technology anchor, Howard Strauss; to Terry Calhoun, our Tech Talk web guru; to Jason Russell, Gayle Terkeurst and the whole support team at Merit Network; to Susie Berneis, the audio file transcriber-continue to watch for those-and finally, a thanks to all of you for being here. You were here because it's time. Bye, Alex. Bye, Howard.
HS: Bye, all. Bye-bye.
AH: Bye.
JB: Bye now.
END OF WEBCAST
