TechTalks Transcript
Directions in Wireless Networking
 Judith Boettcher [JB] |
 Howard Strauss [HS] |
 Alex Hills [AH] |
February 25, 1999
Audio
• Streaming
MP3
• Download
MP3 (Download
Tips)
JB: Welcome to the CREN TechTalk series for Spring of 1999 and to this session on Directions in Wireless Networking. You are here because it's time to discuss the leading core technologies in your future. This is Judith Boettcher, your CREN host for today, and I'm pleased to welcome the TechTalk technology anchor, Howard Strauss of Princeton. Howard is a well-known Web and all-round Information Technology expert. And also I would like to welcome our guest expert for today, Alex Hills from Carnegie Mellon.
Welcome, Howard and Alex.
HS: Thank you, Judith.
I'm Howard Strauss, the technology anchor for the TechTalk series of technology Webcasts. The job of the technology anchor is to engage our guest expert in a lively technical dialogue that will answer the questions you'd like answered -- and ask those very important follow-up questions. You can ask our expert, Alex Hills, your own questions by sending e-mail to expert@cren.net anytime during this Webcast. If we don't get to your question during the Webcast, we'll provide an answer in the Webcast archives.
By now, you've probably wired every building and residence hall on campus, and maybe you've even replaced much of your coax and twisted pair network with fiber -- maybe even fiber to the desktop in some places. Your students, faculty and staff no doubt can access your network remotely, and you certainly have had universal e-mail and Internet access in place for some time now.
Even the laptop revolution has fit in well with your network plans. You are surely putting laptop network connections in all your computer cluster labs and possibly in other places where students might like to connect their laptops.
The laptop was designed to be a mobile computing device, not just a portable device that can be unplugged from one place and reconnected to another. Sure, you can do your spreadsheets and word processing under a tree, in the cafeteria or at an archaeological dig -- but not if you need to be network-connected to do your work, which you really do need to be. Also, maybe in the field, a laptop is more than you want to carry. Perhaps a palm-sized computer is a better choice as a mobile intelligent device.
To get this kind of well-connected mobile computing, you need wireless connections. Wireless connectivity needs to be superimposed over your existing, evolving network infrastructure -- adding new problems of coverage, capacity and wireless network management to your existing network headaches.
But before spending lots of scarce dollars, we need to ask if we really need to do network computing under trees, at every seat in every classroom and at every remote site. Where, if anywhere, does wireless computing make sense, and when, if ever, is it really cost-effective? If wireless does make sense, is the technology ready?
The commercial world seems to think it is. This spring, Discover Brokerage customers who use palm computer or Windows C operating system devices will be able to do online wireless trades, track their accounts and receive real-time quotes. Fidelity Investments will soon offer online trading via wireless pagers and hordes of other commercial wireless services are now in place or will soon appear.
At Carnegie Mellon University, the second incarnation of the Wireless Andrew Project is underway. It will make wireless mobile computing available to 10,000 students, faculty and staff, but it will not replace the hard wired network connections.
What can we expect from this wireless revolution? Will it change the way we do instruction, research and administration? Where is wireless computing going anyway? Which uses are silly and which ones make sense? And how do universities get wireless computing started on campuses and keep it going?
Alex Hills has been in the middle of these issues for some time now. We'll address these questions and others with him on this edition of TechTalk.
Judith?
JB: Thank you very much Howard. And as you say, it's exciting to think about everything that's happening on campus and how connected we are on campus. But then thinking about, gee, how do we move, perhaps, in making the next step in terms of not only being connected but being connected in a wireless manner. So I've been really looking forward to our discussion with Alex today.
And I would like to go on, then, and welcome Alex Hills as our guest expert today. Alex has a prolific background in this area. He is responsible for the operation and development of Carnegie Mellon's computing and telecommunications system. Prior to joining Carnegie Mellon, he was responsible for computing and telecommunications activities across all campuses of the University of Alaska system, and he was also Alaska's Deputy Commissioner of Administration and the state's Chief Telecommunications Official. Shortly after arriving at Carnegie Mellon, he began Carnegie Mellon's wireless initiative in 1992 (which we will be hearing more about during this session). Alex is also interested in using wireless to build telecommunications infrastructure in the developing world. And I think you've probably noticed on our website that we've got links to articles that he's published in journals such as Scientific American, IEEE Personal Communications and Telecommunications Policies.
Welcome, Alex, and thanks for being here on TechTalk today.
AH: Thank you, Judith, it's my pleasure. I've been looking forward to this.
HS: We have, too, Alex -- so we have zillions of questions already for you here.
One thing I thought we could start off with (just to set the picture of this thing) I wonder if you could just talk about why we should bother to do wireless computing at all. What are some of the goals or the things we hope to achieve by doing all this wireless computing on top of all the stuff we're doing already?
AH: Well, I can tell you what it means to us here at CMU in the context of our wireless initiative. We think about all this primarily in the context of mobility, so we can carry our laptops and smaller computers around, and therefore they're mobile. But what our wireless initiative is doing is also providing connectivity in the context of that mobility. So in our wireless initiative, it's really mostly about the combination of mobility and connectivity.
On the research side of our wireless initiative, we're also --
HS: The research side -- that was Wireless Andrew 1?
AH: Well, there's a wireless initiative which comprises our research, and the Wireless Andrew 1 was the infrastructure that we built to support that research. And the research has brought about some quality of seamlessness to this mobile connectivity in the sense that we have software that allows one to move not only between base stations seamlessly, but also between different wireless networks -- say, between the campus network and the cellular CDPD network that serves Pittsburgh metropolitan area. And that hand-off can take place seamlessly where the user doesn't have to do anything to make it happen.
HS: Okay, but you're talking a little bit about how we can make this thing work in a very nice way. But what kind of things can we do, or do you expect that we can do with wireless computing that we couldn't do if we didn't have it?
AH: Well, I mean, one way to answer that would be to say that your mobile computer can be connected to the Internet at all times, where it wasn't before.
Now, how do we see people using that? We have a few hundred people on our campus right now with laptops which are equipped with wireless, and we see them doing, in many cases, the same kinds of things that they do at their desktops -- but they do it from their laptops. So for example, if we're sitting in a meeting and I have my laptop, and I suddenly want to show you something that's on the Web, then I just bring that up on my computer and I show it to you. Or if I want to access a PowerPoint presentation that I have on a server somewhere, I can bring it up for you. Or I can send an e-mail or check my calendar or whatever from the laptop or the mobile computer.
HS: Alex, when we were talking about classrooms (which we talked about a bit in our planning session) you said that classrooms are probably not going to have wireless in them. Could you talk a little bit about that?
AH: Sure. We don't have the classroom use as a high priority right now, and probably the simplest way to answer that would be to say that when students are in classrooms, they're not mobile. So we don't see that it's necessary to use wireless to serve them. If we need to have students using computers in the classrooms, then we can have them connected to a wired network.
Another, maybe somewhat more subtle reason is because there are some capacity or speed issues with wireless. Wireless is more limited in its speed or in its capacity to handle traffic than the wired network is, and so when we get concentrations of users like we do in a classroom, we have more difficulty taking care of the speeds and capacities that are needed using wireless. So since it's not a mobile situation to begin with, there doesn't seem to be a strong reason to meet that need using wireless.
HS: Would that also be true --in fact maybe even truer -- in something like a chemistry lab or physics lab, where again, the data rates might be very high because people are doing very fancy stuff? So would you envision in the physics lab or chemistry lab or electrical engineering kind of lab having hard-wired connections that students would just plug their laptops into, rather than wireless?
AH: Yes, that's our preferred method at the moment, although we may change as we start to lick this capacity situation that I mentioned. Actually, I think the most dense -- you see, when your users are packed most densely, that's when you have the greatest capacity problem, and maybe as we move on I can explain why that is.
HS: So you're thinking like in the big lecture halls, when you have the real high density.
AH: That's right.
HS: And in a lab with 20 students, it's not that big a deal?
AH: Exactly. So the lecture hall would be the highest density situation and therefore the most difficult one for us to handle using wireless because of the capacity issues.
JB: Before we go on into that, down that path a little bit, maybe it would be good, Alex, to come back and differentiate between the kinds of wireless. I know when I first started looking at wireless, I thought it was just kind of one thing. But you said that there were three different categories of wireless computing that we should look at.
AH: Right. I like to categorize wireless networks along a couple of dimensions. One is whether they're voice or data networks, and then also according to the coverage area that's permitted by each kind of technology.
So on the voice side, we've got cordless telephones that provide coverage over a very small area, and then cellular over a metropolitan area, and satellites may be worldwide.
On the data dimension, which is the one that we're most interested in here, the small coverage area technology is wireless LAN. Then the metropolitan area, there are a couple of different metropolitan area networks. Those would include packet radio services like Ardis, RAM and Metricom and also cellular technologies like CDPD.
HS: I think there's lots of folks listening who are not going to know what CDPD is.
AH: Oh, sorry. CDPD stands for Cellular Digital Packet Data.
HS: That makes things perfectly clear!
AH: Right. Well, what it is it's an add-on to a cellular network, so it's an easy way for cellular carriers to easily provide packet data service. So that's kind of the easy way. So you can subscribe to this service through your cellular carrier in most major cities nowadays.
JB: So is that linked to what both Sprint and AT&T are starting to offer nationwide PCS services? Is that the same as the CDPD?
AH: Well, I would say that PCS is similar to or the same as cellular. The only difference between PCS and cellular really is the radio frequencies that they're using. And then some of these are offering packet data as an add-on, so CDPD can be used on either the cellular or the PCS bands in order to provide packet data.
So the CDPD can add onto either cellular or PCS. And to just complete the spectrum of wireless data technologies, then, the widest area one would be low-earth-orbiting satellites like Iridium, which just became operational a few months ago. And that one is worldwide and serves even places like Alaska, where I came from.
JB: That was really helpful, Alex.
HS: Alex, I'd like to just go back to one thing that you said early on. You were talking about using wireless in meetings, and to me that sounded a little bit like using them in classrooms. When we go to a meeting room, why not just have the meeting room wired?
AH: Well, I think in meetings it's more of an impromptu situation. In other words, people bring their laptops to meetings to take notes, to access material from the network -- whatever. In the classroom, at least, I think of a more structured situation where the instructor has planned ahead of time how he or she is going to use the technology. So in classroom use, I think that there is a lot more prior planning. We've got to make sure that everyone's got appropriate hardware, software and so on. That's the distinction I would make.
HS: So you don't see CMU or other universities putting in network connections, laptop connections, in conference rooms, meeting rooms and things like that? Instead, they'll rely on the wireless network there.
AH: Yes, that's the way -- in fact, we're doing that right now, so a few hundred people who are using wireless-equipped laptops are doing that right now, using wireless, and it seems to be working out just fine.
HS: Alex, just a little while ago (and I mean a little while ago in technical time, whatever that is) wireless just was impractical. I mean, it was too expensive, battery life was too short. How are those things changing? Why is Carnegie Mellon deciding that now is the time to do this?
AH: Well, as far as what's happened with the technology, wireless LAN technology has been around for quite a few years. But about five or six years ago, the NIC (the Network Interface Cards) became available in PC card size. So that meant that wireless LAN's were no longer about just sort of eliminating building wiring or avoiding wiring problems. But that was when wireless LAN's started to address mobility, was when the NIC card got down to PC card size.
Then a year or so after that, some of the wireless LAN companies introduced the idea of roaming, which means that you can move from one base station's coverage area to another and not lose your connection, so that helped with the mobility. And then the last one is the reduction in the size of the computers themselves. We're starting to work a lot more with hand-held sized machines.
HS: Like palm computers?
AH: Yes, exactly. And I think -- one other thing, you know, size in all this is constrained to a large degree by battery size, and the wireless LAN companies are getting a lot more clever about using power efficiently by, for example, actually putting the wireless card to sleep for very short periods of time in order to conserve battery. So I think it's a combination of those things that I mentioned.
JB: Moving on to talking about mobile computing and where it's all being used, can you tell us a little bit about your Wireless Andrew 2 and where you're at with that on campus?
AH: Sure. Maybe we should just define terms a little bit more on Wireless Andrew 1 and Wireless Andrew 2.
JB: Probably a good idea.
HS: Why don't you do that?
AH: Wireless Andrew 1 was our research infrastructure that we built beginning in 1994 to support our wireless research (I mentioned that at the beginning) that used a Lucent product called WaveLAN 1 and that works at 900 megahertz radio frequency.
A year or so ago, we decided to sort of expand and enhance that. The Wireless Andrew 1 infrastructure covers seven buildings on our campus, which are a fair number of our academic buildings but certainly not the whole campus. With Wireless Andrew 2, we're building that out to cover the entire campus.
HS: Could you give folks a sense of the size of your campus? I mean, is it a mile across, a couple miles across -- what?
AH: It's a pretty small campus. Let's see.
JB: You're right in downtown Pittsburgh, right?
AH: No, I'd say we're in a suburban location. We've got about 20 to 25 buildings in our core campus, and those are our academic and administrative buildings. And so, you know, it's actually more the number of buildings and the size of those buildings that's important than the land area of the campus or something like that, because it's the indoor coverage that's really the challenge.
HS: But isn't there some issue with the range of a base station or access point?
AH: Yes, but the issues are primarily issues of indoor coverage, so I could give you a couple of numbers.
With outdoor coverage, a base station has a coverage radius of something like 800 to 1,000 feet.
Indoors, it's much harder to put a number on things because it very much depends on the construction of the building -- the layout of the building and the type of construction. What materials are the walls constructed of? If they're 2x4's and sheetrock, then there's not too much degradation in signal; but if it's reinforced concrete and steel beams, then there's a lot of degradation in the signal. So indoors, we get coverage radii on the order of 100 feet up to about 300 feet.
HS: That's short!
AH: But it's very much of a local situation inside the building.
HS: Is that affected by the height? I mean, I know you're not doing cellular type stuff, but when I drive around, I see all these cellular towers and they're obviously trying to get these things way up high. Is that one of the things you're trying to do with these stations, these access points, is to get them at the tops of buildings?
AH: No, it turns out that we actually have to provide coverage floor by floor, so we may put them high on the wall, meaning near the ceiling. One of the things that we've learned about designing wireless LAN's is that, as a practical matter, it's fairly important to provide base stations on each floor of the building.
HS: And that's just for coverage, that's not for capacity?
AH: That's just for coverage, right. Capacity is a whole separate set of issues.
JB: Is it a good time to talk about -- that sounds like it costs a lot of money if you have to do that on every floor, Alex.
AH: Well, I could give you some ideas about how much money it costs. Let's just do this in terms of the price of the currently available product from Lucent, which is the product that we're using.
In our Wireless Andrew 2 project (which is the one I was starting to mention that's actually providing service to the whole campus -- and which will, by the way, be available to everyone in our campus community) all 10,000 faculty, staff and students will potentially be able to use this. It will certainly be available to all of them.
HS: What do you mean by available? Are you going to pay for the cards?
AH: What I mean is, we're building the infrastructure. We'll build the network and then the individuals will need to buy a card, much as they need to buy an EtherNet card now if it's not already part of their computer. To give you an idea of the price of that to the consumer, the list price on the current WaveLAN card is $300.00, so I think one can expect street price would be a bit below that.
Judith, shall I come back and now talk about the cost of building the infrastructure, which I think was what you were asking me?
JB: Sure, let's do that.
AH: Okay. So the list price on the access points, which are the base stations, is $1,000 ($995.00, actually) and to do our entire campus, which is a fairly small campus, that's requiring about 400 access points. So you can multiply that out.
JB: We can do the math on that fairly quickly.
AH: Four hundred thousand dollars, right. And that's for equipment. And then you will also have some other costs -- labor, cost of installing power connections, the wired network connections in order to accommodate the access points, and so on.
HS: You also talked about the fact that folks could get access through the cellular telephone network and from that, get access to your network.
AH: That's correct. That's part of our wireless research, and it's not something that we're rolling out in the near term to our campus community, but we have built the software to make it happen. One practical issue about using the cellular network is when you're using the cellular network, you're no longer using facilities that are owned by the university. You're now using the facilities of a public carrier.
HS: So obviously, you're going to pay them directly.
AH: You have to pay for service, right.
JB: So within the local area network on campus, then, the users really have the hardware purchase costs that they have to think about. But then as they move on, then they've got to contract for service with these other agencies?
AH: If they were to want to use cellular, but we're rolling out in Wireless Andrew 2. The only cost that the user sees is buying that card for a few hundred dollars, and then everything else is free -- just as we don't charge users for our wired network service, and similarly, we won't charge them for the wireless, either.
HS: And that gives them all the kinds of access they would have if they were sitting at some fixed desktop machine?
AH: Exactly.
HS: They get Internet access and they get access to your servers and they get access to printing and all the kind of services that one offers on a campus today?
AH: Exactly.
HS: You also mentioned these low-earth-orbit satellites. Is that something different than the cellular telephone network?
AH: Oh, yes, it's quite different. In the satellites, the base station is in the air -- on the satellite.
And I just mentioned that as a way of kind of putting all this into context and into perspective. It's interesting to notice that the bit rates that one can achieve are sort of inversely related to the coverage area. So with wireless LAN systems, you get very high bit rates in the megabits-per-second. With cellular and CDPD, you measure those in the tens-of-kilobits -- like 19.2 kilobits per second for CDPD. And then the highest speed data that's available from Iridium low-earth-orbiting satellite is 2.4 kilobits-per-second.
HS: That's very slow!
AH: And at fairly high cost also -- like measured in the dollars per minute. So you probably don't want to use that last one unless you're really in a remote spot or you need to send out your SOS or something like that.
HS: Yeah! The access points -- you said they're in the megabit-per-second range, but that's like a couple megabits per second, isn't it?
AH: That's right. The IEEE 802.11 standard (and the wireless LAN companies are all building equipment now that complies with that standard) addresses speeds up to two megabits-per-second. So that's in the low megabit range, I agree. And then there's a new feature to the standard that's just coming out, and some of the companies will be introducing equipment which runs at ten megabits-per-second.
HS: But that's still a lot lower than the 100 megabit stuff that lots of folks are running.
AH: Exactly.
HS: It sounds like that simply means that you need lots more access points, and since they're $1,000 apiece, you need lots of those things.
AH: Well, that's a really interesting point. And it's a good observation that the more access points you use, the more aggregate capacity you have on your campus, because the two megabits-per-second refers to the capacity of the access point. And that capacity is being shared by all the mobile nodes that are using that access point.
But the aggregate capacity -- using our campus as an example, if it's two megabits-per-second per access point and 400 access points, then you can multiply that out to get the aggregate capacity that's available across the whole campus. So it is true that the smaller the cell size -- or the more access points you use -- the more aggregate capacity you have.
HS: Alex, we have a question from someone on e-mail, from Mike Simon. I wonder if I could just read you his question and if you could take a stab at answering it. Mike says, "This question is directed more towards point-to-point wireless connections," and Mike asks, "What type of costs would one expect for the hardware/software needed for bridging networks 10 to 20 miles apart at T1 speeds?"
AH: Okay, well, probably the lowest cost way to do that would be to use the wireless LAN technology that we're talking about. You could build a private microwave system, which would be considerably more expensive than what I'll describe, but the wireless LAN operates at approximately T1 speeds. We've been talking about two megabits, so that's a little bit more than 1.5 megabits.
And the way that you would do it is just by installing an access point connected to your wired network at each end, and then instead of using the antenna that comes with the access point, you would use a special high-gain antenna, what radio people call a YAGI antenna. And you would use two of these high-gain antennas, one at each end pointed at each other, and if you had a high enough gain in those antennas, you would get a good solid connection.
So it would actually be fairly inexpensive -- $1,000 per access point at each end, maybe another $100 for each antenna. I mean, that'll definitely work well over distances of a few miles. Up in the 10 to 20 mile range, you'd have to have extremely high-gain antennas, but you could probably still get it to work. If not, then you'd have to go to a considerably more expensive solution, which would be private microwave.
HS: Sounds like Mike has his hands full here! We have another question, from Ken Alden, and Ken asks, "What is your proposed mechanism for handing off the users from the WaveLAN network to CDPD? Will the user notice the hand-off, all circuits dropped, other than the speed drop in their network connection?"
AH: Well, we have some software that we've developed which uses mobile IP, and the mobile IP is necessary because while we have all of our WaveLAN access points on one subnet, as soon as you switch over to the cellular network you are of necessity on a different IP subnet. So we've got the software developed to make all that happen. It's the software that resides on the laptop, so the hand-off is seamless, as it were.
But as far as whether the user will notice it, I'd say the user will notice it if the user is running an application that is sensitive to the speed of the connection. So if you're using an application that is not going to work very well at CDPD speeds of 19.2 kilobits or less, then you will notice the change.
HS: Alex, when we talked earlier, you said that Carnegie Mellon, like many schools, does not require students to buy computers or to have laptops and things. And you said right now that only about 20% of the students, I believe, who had computers at Carnegie Mellon had laptops. I assume you anticipate that putting all this wireless infrastructure into place is going to change that dramatically.
AH: Yes, we do, and I think that we expect that that will provide an incentive to students to use their laptops on campus because what it means is that when they come to campus, from their dorms or wherever they live, they will not need to wait to use a desktop machine in a lab or a cluster. They'll be able to just use their own laptops wherever they happen to be. Also, a couple of our colleges are requiring their students to buy laptops, so that's kind of moving us in the direction of greater laptop ownership in general, but we at this point have no such requirement at the university level.
HS: I believe you said it was the business school that was requiring laptops, not the engineering school.
AH: The business school at this point, and the next one that's likely to require laptops is the school of computer science.
HS: It's strange that the business school and not the engineering school be the ones that were requiring laptops -- but whatever!
AH: I could introduce you to the deans, if you'd like to discuss it.
HS: Well, we'll do that at another session here. What's happening with respect to the faculty and wireless? Are they buying into this, or how are you getting them to buy into this?
AH: Well, so far, we have sort of a core faculty who are involved in wireless, so they're obviously advocates for all this. We have another group of faculty who have expressed interest in using it in the classroom, but as I've said, we've been putting out some caveats to them on that and we're actually encouraging the use of wired connections there. I think the faculty ownership of laptops is rising just as student ownership is.
One thing that we're doing that I think will encourage faculty to own laptops increasingly is we're equipping all of our classrooms with data projection and network outlets which are oriented to the idea that if the instructor is going to use a lecture, a PowerPoint presentation, it's oriented to the idea that the instructor will walk into the classroom with that presentation on the laptop.
And we're giving them, as faculty, some transition into this new world by providing loaner laptops to them which they can check out for a semester. Hopefully, they'll get hooked on it and then they'll buy one and go from there.
HS: I would think so. Once you have a machine for a semester, how could you ever give it up?
JB: That example, Alex, that's where you're really still encouraging faculty to use the wired classrooms. When we were talking about Wireless Andrew a minute ago, we didn't talk about whether this was going to be your mode of choice in the residence halls. And then we had a question coming in from a person, asking whether the WaveLAN technology works well in the residence halls. That question is coming from Joel Cohen. Do you want to maybe address both of those at the same time?
AH: Sure. I'd be glad to. You know, one thing to say is that some universities are using wireless in the residence halls, because the residence halls were not previously wired and wireless is seen as a way of sort of avoiding wiring costs by using something like Metricom or something like that.
We're not in that category. We've had networking in our dorms for a long time, so that's not a motivation for us. We also don't see mobility as much of an issue in the dorms, as we do on campus as I've described. So with Wireless Andrew 2, we are not doing the residence halls, and the reason for that is a combination of the fact that we don't see mobility as being as much of an issue -- although it certainly is an issue. But the other side of it, frankly, is cost. Resources are always limited and so, given the resources that we have, we've chosen to do the core campus at this point, leaving the residence halls for some time in the future.
HS: Alex, we have another question here about security on wireless LANs from John Schaefer. John Schaefer writes, "How do you prevent an unauthorized user with a wireless PC card, perhaps sitting outside the building, from connecting to your network?"
AH: Well, I can tell you an interesting story there. Let me say that what we do about security is we us authentication, but not encryption at the present time, so that in order to get access to our servers and so on, one has to authenticate. One has to log in using our standard university Andrew ID and password. So if the person that's sitting outside the building is, first of all, wireless-equipped and has an Andrew ID and password, then they can access the network -- but we'd want them to access the network because they're an Andrew user and that's fine.
But there's an interesting little story that happened where one of my staff had a Wireless Andrew 1 base station in his house that he was using and just kind of playing around with and, unbeknownst to him, one of his neighbors in the house next door was also using that and thinking that it was part of the normal campus service, so that when he switched over his base station to Wireless Andrew 2 -- switched to a higher frequency, 2.4 gigahertz -- we got a call at our helpdesk from this person in a house off-campus who wanted to know what was wrong with the wireless network. And of course, the answer was, "When you're stealing service that way, we don't support you." So this person was just sort of innocently coming in through someone's private wireless base station (one of our staff members had one) and thinking that he was getting that service from campus.
So I'm not sure if I answered your question about security. We're authenticating in the way that I described. We're not using encryption at this point. So probably the greatest security threat right now has more to do with snooping than it does with logging on and using the network in that way.
HS: What about some of this technology -- the spread spectrum technology? Does that help secure the connections?
AH: Well, yes and no. All of this wireless LAN technology is spread spectrum, and that was originally developed in World War II as kind of an anti-jamming. It's less susceptible to jamming and interference. But I think in the context of our conversation, we're making the assumption that the bad guy has one of these cards and is therefore using the spread spectrum technique because he's got one of the wireless LAN cards. So in that sense, you know, it doesn't help.
HS: These cards are just commercially available?
AH: Yes, that's right. The Lucent card that we're using -- that's the one I mentioned that lists for $300 -- and it is commercially available.
HS: If your network is set up to use the Lucent card, does that mean that folks can't use any other card?
AH: Well, it did mean that before the adoption of the IEEE 802.11 standard. That was adopted in the summer of 1997, and then the wireless LAN companies within about a year began to produce product that was 802.11 compliant. And so the advantage of a standard, obviously, is interoperability. So one would say, "Well, so these 802.11 products should be able to talk to each other." And the answer is, "Yes, sort of."
The problem is that there are a couple of different options within the standard. At the physical layer, there's a direct-sequence-spread-spectrum option and a frequency-hop-spread-spectrum option, and those two don't talk to each other. So the answer is, if you have 802.11 product and if it's all direct sequence or if it's all frequency hop, then it will talk to each other.
HS: We have a question from Don Sutton that actually talks about some aspect of the IEEE 802.11, and Don asks, "If the IEEE 802.11 standard is two megabits-per-second, and you said you're moving to ten megabits-per-second, isn't there a problem if you go through a T1 at 1.54 megabits?"
AH: I'm not sure I completely understand the question, but I should clarify. We're not moving to 10 megabits-per-second. I was explaining that there's an option in the standard that's coming out that will accommodate 10 megabits-per-second. So we're staying with the two megabits, and the reason we're doing that has to do with the coverage area of the access point, which will be reduced if you go to the higher 10 megabit rate.
I think what the questioner is thinking about is if you have a ten megabit access point that relies on a 1.5 megabit link, well, then yes, there would obviously be a problem. That really doesn't describe the situation that we're in, however. We're actually using two megabit access points and we're servicing them on 10 megabit links. So the wired link does not cause any congestion in our case.
HS: We have another question here from Con Dietz, a friend of mine, actually, at Creighton University. Con says, "Alex, it's great to hear your lucid descriptions again." So I guess you know Con as well. And Con asks, "To what extent is wireless going to be driven by or encourage change in the instructional paradigm, which may be departing from the traditional classroom?"
AH: Right, it's nice to hear from Con again. Con and I were in Alaska together for several years -- so it's not a voice from the past but a message from the past.
JB: Well, then, I'll say "hi" too.
AH: But the question has to do with changing the instructional paradigm, right?
HS: Yes.
JB: He's coming from the idea of how wireless might be a catalyst for change in the instructional paradigm and how might it be used in something other than the traditional classroom, or how it might encourage something other than the traditional classroom.
AH: Well, of course, we've already been through a discussion about inside the classroom and wired vs. wireless, but we do a lot of project here, and at our university, project work is a very important part of our instruction.
And so we have architecture projects, for example, that go on and I can very easily think of our architecture and engineering students out on the site of a project, using their wireless-equipped laptops for drawings, designs and accessing the files that they need. So that's one that quickly comes to mind.
Actually, you know, with our technology and instruction generally here at our university, our philosophy is to make the technology available and then let the faculty and the students figure out how to use it. And they end up coming up with all kinds of things that we never would have thought of in the central organization. And I'm sure that'll happen here, too.
JB: It might be a good time, at this point in our program, Alex, to kind of look at the campus as a whole or other institutions that don't have the Wireless Andrew kind of projects going on. Is wireless something that most campuses should seriously take a look at right now, or is it too early? What recommendations might you give to these campuses?
AH: I receive a fair number of phone calls from other universities asking that question and we usually suggest that it would be a good idea to hold off a little bit longer and leave us (people like us who are right out on the bleeding edge at this moment) to solve some more problems.
The big issues that we are dealing with right now -- technical issues -- they fall into three major categories. One, maybe the most obvious category, would be coverage -- would be the radio engineering things that you have to do in order to locate your access points so that you do, in fact, get complete coverage on your campus.
The second set of issues are the capacity issues. The network also has to be designed in a way that you will have enough capacity to take care of all your users. It will always be less capacity than you have on the wired network, but we're just now starting to get our arms around the capacity issues. And we'll continue to work on those as we get not just hundreds but thousands of users on our wireless network.
And then the last category of technical issues has to do with network management. In the world of wireless, we still don't have the kinds of network management tools that we do with wired networks, so we have a ways to go there, too.
So my thought would be unless you have some compelling reason -- our compelling reason has to do with our wireless research -- but unless you have some sort of compelling reason, it might be a good idea to hold off just a little bit longer.
HS: I think that's unexpected advice. Usually when we talk to technology experts, they tell us that, gee, you're two years behind, you'd better get going! So it's nice to hear somebody say that maybe there's some technology we can actually take a little bit slower.
JB: We might want to come back (I hate to leave that like that, Howard) and just ask the question. Alex, have you heard and seen other campuses have a compelling reason to implement a wireless in a fairly fenced area, or for some particular purpose that they really find is a compelling, driving force for them?
AH: Yes, one is the classic reason that people have used wireless LAN's, which is wire problems. Problems with wire. If you have a building that has an inadequate wiring system, and especially if it's an old building, it might be very difficult to upgrade that.
JB: So they might use it as a temporary solution, rather than tearing the building down and starting over.
AH: Right. There were also other situations where people might be interested in the mobility ideas and might like to implement wireless on a more limited scale than we have.
The comments that I was just making were really oriented to the campus-wide -- really jumping in and doing the whole campus. But some engineering schools or computer science schools may be interested in doing things on a more limited scale.
HS: Alex, we have one question, which will probably be our last e-mail question for this Webcast, from Bob Clay at the University of California at San Diego. Bob says, "How did you gauge the faculty/staff/student interest in having ubiquitous wireless service available? Did you size your infrastructure accordingly?"
AH: We have -- as I said, we got into this because of our research initiative in the first place, so we really didn't need to gauge the interest.
HS: You keep referring to your research initiative. You were doing research on how wireless networks worked?
AH: That's right. So we built our Wireless Andrew 1 infrastructure in order to support that research. And then, as people found out about it, as other people from outside that research project found out about it and expressed interest, that made it very clear to us that there was a lot of interest on this campus. And we anticipate that the interest will become even higher when people become aware how cheap these cards are getting.
With respect to sizing the infrastructure: as we discussed earlier, there are both coverage and capacity issues that you need to deal with when you build the infrastructure, and the approach that we're taking in designing that is we are designing for coverage.
So that's the approach that we're taking, and it's not really one where we have really the final design in mind before we start. It's one where we go in and design for coverage first and then deal with what you might call the "hot spots" later.
JB: You may find, too, you don't know what the capacity is because the number of folks using it will naturally grow over time.
AH: That's right. And actually the capacity that you need is affected by not only the number of users within an access point coverage area, but also by the kind of traffic that they're running. And so that's another thing that we don't know -- is just what kind of traffic that they're running.
We're using Web traffic as kind of our standard paradigm. We're doing some traffic simulation work right now and so, in the traffic simulation work, we're using Web traces. But they could be other kinds of things. They could be other kinds of traffic as well.
So there are a lot of unknowns and we won't really know the answers until we have a sizable group of typical people using it, and then we'll start to be able to answer some of these things.
JB: Thank you so much, Alex. You know, I really hate to bring our session to a close. It seems as if these sessions go so very quickly. Let me just ask, Howard, if you have any other question or comment.
HS: I do. Actually, we do have more mail coming in. We tend to get a lot of mail coming in at the end, and unfortunately, we don't have time to answer it on the air. But we'll get to your question in the archives. I'll have Alex answer it later. But the question I've been wanting to ask and I've been trying to save for the end (even though we're overtime): Alex, there's been Wireless Andrew 1 and Wireless Andrew 2. Can you give us a couple hints and clues as to what Wireless Andrew 3 is going to look like?
AH: Well, I'm not sure if there will be a Wireless Andrew 3, but I think we're likely to maintain and refine our Wireless Andrew 2 network.
The next project that we are getting ready to roll out is one that relies on the Wireless Andrew 2 infrastructure, and we call it Hand-Held Andrew, so the idea is that we'd like to bring about the vision of making hand-held machines full peers in the network, as opposed to second class citizens which have to be synchronized with desktop machines from time to time. So that's why we've mentioned a couple of times in this discussion hand-held platforms, and that's really going to be our next focus.
We're particularly interested in the calendar application. We're interested in three applications. Those are e-mail, notification, and calendar. But the calendar is the one that has some challenges to go with it and so we're hoping in the not-too-distant future to be able to roll out a service to our campus which will allow people to use hand-held machines to do particularly these three applications, as well as Web browsing and other things like that.
HS: So is that what you're going to be doing about a year from now?
AH: I think so, yeah. That's the plan.
HS: It'd be really interesting to talk to you again then.
AH: Call me again, Howard!
HS: We'll do that. We'll do that on the air.
JB: Alex, before we close up, is there any other final comment you'd like to leave with the audience?
AH: Oh, this has really been a lot of fun. I'll do my best to catch up on the answers to these e-mails that I see on my screen here.
JB: All right, sounds great.
Well, with that, then, I'd like to thank all of our Web participants for being with us here today for this time with Alex. Please send follow-up questions to expert@cren.net and Alex will answer them individually and then we'll be posting some of this on our Website.
Be sure to mark your calendars for March 11, which is, by the way, three weeks from today rather than two weeks. This Tech Talk will feature guest expert (actually a colleague of Howard's) Dave Kohler from Princeton. And he'll be talking about perhaps one of the most important and popular topics today, Y2K. I'm not certain we'll find out from Dave what we can do about Y2K since we're so close to it coming up here.
Do plan on joining this session, inviting your friends and colleagues as well. And check the Website for upcoming spring experts, and as always, we welcome suggestions from you as to what you'd like to see and hear on TechTalk.
Thanks to everyone who made this possible today: the board of CREN; guest expert Alex Hills; technology anchor Howard Strauss; Web content producer, Terry Calhoun; to Brad Peck and Lee Perlis of CREN for operational support; and Paul Bennett at UM Web Services for encoding; and all of you, for being here. You were here because it's time. 'Bye, Alex. 'Bye, Howard.
HS: 'Bye, Judith. Thank you, Alex.
JB: Take care.
HS: 'Bye-bye.
