Experience Developing Wireless Networks
for Interactive Multimedia Instruction

James Griffioen, W. Brent Seales, James E. Lumpp Jr., Tom Kay

Department of Computer Science
and
Department of Electrical Engineering
University of Kentucky

Abstract

This paper presents our experiences developing the network and computer infrastructure needed to support the interactive multimedia wireless classrooms of the future. Despite their advantages, wireless networks have several drawbacks and potential pitfalls. The heavy loads placed on the network by interactive multimedia applications used in class such as collaborative whiteboards, collaborative editors, collaborative web surfing, application sharing, and even video/audio feeds to distance learning sites only aggravate the problems.

In this paper, we highlight the problems that can arise in a wireless setting, and we describe the advantages and disadvantages of the various hardware technologies as well as the software and network protocols that are best suited for interactive multimedia wireless classrooms.

Introduction

At the University of Kentucky, we have been studying ways to apply advances in wireless network technology to enhance education. The Wireless Classroom Project is investigating new classroom environments that make use of laptops and wireless networks to bring interactive multimedia into every class. The project has two primary objectives: (1) to investigate and develop the computing, networking, and software infrastructure needed to support wireless classrooms, and (2) to experiment with and develop new teaching methods that incorporate interactive multimedia to improve the learning process. A third, but less critical goal of the project was to investigate how wireless classrooms can be integrated with distance learning environments to effectively breakdown the distinction between local and remote students and create a truly virtual classroom environment.

In our wireless classroom environments, each student is equipped with a laptop computer and a wireless network connection. The instructor's laptop may additionally connect to a microphone, video camera, art pad, and projection device. The instructor's display is projected on an overhead and video/audio feed of the instructor can be multicast across the Internet to remote students. The entire environment is mobile.

Unlike university programs that put laptops in the hands of students[3,6], the wireless classroom project focuses on the network and how interactivity and collaborative multimedia, facilitated by the wireless network, can revolutionize the way we teach. This paper describes the technological challenges of designing, implementing, and managing wireless classroom environments used for interactive multimedia instruction. Another paper, [5], describes the educational benefits and challenges of wireless classrooms.

Why Wireless?

Wireless classrooms offer a unique combination of unprecedented technologies. Not only does it bring impressive computing power into the classroom in the form of multimedia laptops, but it also brings the network into the classroom[5,8].

Having the network in the classroom obviously means access to the vast information resources of the Internet, but more importantly, it provides the opportunity for the instructor and the students to to interact with each other in radically new ways. For example, collaborative (networked) whiteboards create the opportunity for interactive class notes. Instructors can write notes on the whiteboard, work through examples, or annotate preloaded notes and have all their markings transmitted immediately to student machines. Students can then add their own public or private annotations. A student can ``step to the chalkboard'' to work through a problem without leaving their seat. Moreover, the interactive class session (whiteboard notes) can be recorded and played back at a later date. Collaborative tools such as app sharing are powerful teaching tools, allowing teachers to demonstrate applications or lead students through information with the ability to involve the students by letting them interact with the application. Collaborative web browsing tools are a powerful way for instructors to lead students through the material while allowing students to go off on their own web site explorations. It also serves as an excellent way to illustrate difficult concepts by leading them to images, animations, video, or sound clips. Live, hands-on experience that has historically been deferred to dedicated computer lab session can now occur in any class at any time. Moreover, when the above collaborative tools are combined with multiconferencing software that support live video/audio, wireless classrooms essentially eliminate the difference between students sitting in class and remote students at distant locations.

Wireless classrooms also have the potential to be highly cost-effective. The expense of installing a wired connection for every seat in every classroom is staggering. Moreover, many older classroom and office buildings were not designed for wired networks and require substantial amounts of conduit work just to get the wires to the room. Wireless technology provides an unlimited number of network hookups in any classroom (even old, technologically archaic classrooms) without the enormous expense of installing the wire. It also provides access to campus areas that are not usually networked such as library carrels, study rooms, music practice rooms, coffee shops, or the campus park/mall. Furthermore, the mobility facilitated by wireless networks allows increased flexibility in scheduling and room assignments. Every room on campus can be used as a computer lab during any class.

Finally, wireless networks offer wonderful aesthetic benefits. Entire buildings can be placed on the network without any visible changes to the existing classrooms or physical structure. Thus classrooms can continue to be used in a conventional manner or as a high-tech wireless classroom.

Unfortunately, wireless networks do not currently offer the performance of their wired counterparts and thus must be carefully designed and planned if they are to support the interactive multimedia found in wireless classrooms.

The Wireless Classroom Project

This paper describes our experiences with the Wireless Classroom Project at the University of Kentucky. We designed and installed a wireless network that provides coverage to the College of Engineering's four main buildings where the classrooms, conference rooms, faculty offices, and research labs are located. The network consists of five access points. Simply stated, an access point acts as a bridge between a wireless machine and a wired network. Figure 1 illustrates the role of an access point in a wireless network. Each laptop is equipped with a wireless network card that transmits packets via a radio frequency (rf) signal to other wireless laptops and to the closest access point. The access point, which is connected to the wired network, then forwards the packet to machines and routers on the wired network. Similarly, when an access point sees a packet on the wired network that is destined for a wireless machine, it forwards the packet via a rf signal to the wireless machine. The wired network in our case was the College of Engineering ethernet. The engineering network (including the wireless network) connects to the university backbone and the Internet. The university network supports multicast and has a connection to the Internet multicast backbone (MBONE). In some cases a building could be covered by a single access point, while other buildings required multiple access points. As a result, some buildings have overlapping coverage cells as illustrated by Figure 1. We selected Lucent Technologies WaveLAN wireless network technology for this project (section 2.1 provides the motivating reasons for our choice or hardware).

  

Figure 1: Example Wireless Network Layout

Students who used the wireless classrooms were given a laptop computer (Gateway 2000) and a wireless network card/antenna. The laptops were loaded with both Windows 95 and Linux and all the necessary software, including a variety of collaborative tools and distance learning software packages. Each of the instructors' machines also included a Connectix Color Quickcam for video, audio via the built-in microphone and speakers, a WACOM artpad to write on the whiteboard, and a Boxlight SVGA projector to display the instructor's screen.

Several of our core classes have used the wireless classroom environment for class lectures. All classes occurred in the classroom normally assigned to the class. No modifications were made to any of the classrooms. Class sizes in most cases where around 20 students. Although larger class sizes are not uncommon, we found that 20 students was easily enough to swamp the network, and that new teaching approaches, and more efficient applications, and protocols needed to be used if interactive wireless classrooms are to scale to larger class sizes. We discuss this more in section 3.

During each class period, the instructor loaded the lecture slides into a shared whiteboard application and then used an artpad to write additional comments on the notes, fill in missing pieces, work through example problems, etc. The collaborative whiteboard immediately transmits all markings to the student machines and displays them on their screens. Lecture slides were prepared using standard presentation software or word processors and often included figures, graphs, or color images which meant that the pages being transmitted were of substantial size, in some cases several megabytes! Students were also able to add private or public markings on the whiteboard.

Instructors use a variety of other instructional tools depending on the course and instructor's preferences. A common technique was to use a (collaborative) web browser to guide students to images, video clips, animations, or example programs because of the ease of running or launching these from within a browser. All instructors agreed that app sharing is a powerful instructional model. However, in almost all cases, we found it required enormous amounts of bandwidth and so its performance over the wireless network was unacceptable. Thus app sharing was used sparingly. Being in computer science we often ran applications on the computer science lab Unix machines where the X-display or telnet session appeared across the network on our laptop screen. Being unconstrained, students would also do unexpected things across the network such as using chat sessions with classmates to ask simple questions about the lecture such as ``is that an x⁴ or x⁹'', exploring web links related to the topic being discussed, or unrelated things such a watching the NASA space shuttle mission over the MBONE or reading email..

In situations where there were remote students that were participating in the class, the instructor would also run multiconferencing software to transmit (across the wireless network to the wired network) a live video and audio stream from the instructor's laptop to the remote students. Remote students would also transmit audio (and possibly video) back the instructor and students in the classroom.

In short, instructors made heavy use of image filled notes, animations and video clips and possibly video/audio feeds, all desiring interactive (soft) real-time performance. We quickly learned that supporting this type of network traffic in a wireless setting takes careful planning and the use of the appropriate applications and protocols.

Design Issues

There are several issues that must be addressed when designing interactive multimedia classrooms. Although some problems are similar to those of conventional computer labs, the use of wireless technology combined with an emphasis on heavy use of real-time, interactive multimedia introduces a variety of new problems that must be addressed. The remainder of this paper describes some of the issues that must be considered when designing wireless classrooms. We also describe our experiences and provide recommendations based on what we learned.

Hardware Issues

Laptop and wireless network hardware is improving so fast that it is useless to talk about any specific hardware technology since it will be obsolete next in a few months. Instead, we will try to address the hardware issues that involve aspects of the technology that are unlikely to change. We will also try to provide network design principles that are invariant to the technology.

Network Issues

  Currently there are a lot of wireless network vendors and options available. This is unlikely to change in the near future given the rapid growth and advances in wireless technology. The following outlines some wireless hardware characteristics that should be considered when designing the wireless infrastructure.

Coverage Range:

The area covered by wireless networks will vary, and depends on things such as the antenna used, orientation of the antennas, the frequency used, the physical structure of the building, and interference from other devices. Wireless vendors have charts that show the typical coverage of their equipment, but it is important that you test out the coverage with the equipment yourself or have the vendor do a site survey. As a general rule, lower range radio frequencies have greater coverage higher frequencies. Currently most wireless technology operates at 915 MHz or 2.4 GHz. We found that 915 MHz coverage was a little better than 2.4 GHz but only by about 30 feet in the typical classroom building.

Interference:

Interference from other devices can affect the signal quality and range. Both 915 MHz and 2.4 GHz are susceptible to interference from things such as microwave ovens, copier machines, elevators, etc. Although the 915 MHz range has better distance than 2.4 GHz, it is susceptible to interference from other radio frequency devices operating in roughly the same range (e.g., cordless phones). In our case, we decided the risk of interference at 915 MHz was not worth the few extra feet of coverage. Another factor influencing interference is the signalling method. Two methods are currently used: Direct Sequencing Spread Spectrum (DSSS) and Frequency Hopping Spread Spectrum (FHSS). [1] Both provide reasonable protection from interference by spreading the signal across a range of frequencies. However, FHSS uses a seemingly random series of frequencies that is not as susceptible to interference as DSSS technologies however there are many arguments as to which is better. Unfortunately, FHSS typically operates at substantially lower data rates than DSSS.

Security:

Although one can argue whether it is more difficult to tap into (and snoop) a wired network or a wireless network, our experience indicates that there is more paranoia about wireless network security than wired (particularly among administrators and network support staff). To provide security, many vendors sell wireless equipment with encryption. Clearly encryption provides better security but it typically increases the price. Spread Spectrum also provides limited security with FHSS being more difficult to decipher than DSSS. Security was not a issue for our environment. When we want secure communication we use higher level protocols and applications such as ssh (secure shell).

Interoperability:

Because the laptops will be communicating with one another and with the access points, interoperability among the cards used is crucial. Fortunately, a standard ratified in June of 1997 called the IEEE 802.11 wireless standard, defines a common protocol for vendors to adhere and will allow hardware from different vendors to interoperate. Unfortunately at the time of this writing, many vendors are still in the process of implementing the standard and interoperability is questionable. At this point in time we recommend using a single vendor campus-wide.

Upgrade Path:

Given the rapid changes in wireless technology it is crucial that you consider the future and ensure that your network has a smooth and hopefully inexpensive upgrade path. Avoid network boxes that will not allow you to swap out older components for newer modules or download/install new versions of switch software.

We decided to use WaveLAN equipment from Lucent. We had been using this technology in our research for several years and had experience with it. It supports both 2.4 GHz and 915 MHz frequency operating at 2 Mbps (the fastest at the time) and they announced a 10 Mbps card available within a year. The access points have a clean upgrade path, simply replacing the pcmcia card and antenna with a new card. New versions of access point software can also be downloaded and installed. WaveLAN is also compatible with DEC Roamabout wireless equipment. Finally it is supported by both Windows 95/NT and Linux.

Laptop Issues

From an management/integration standpoint, all laptops should be identical[7]. Since this is unlikely (our single shipment from Gateway had small differences between machines), wireless classroom designers should be ready for heterogeneity. However, some guidelines are useful.

Batteries:

Wireless operation requires battery power and interactive multimedia can be a big drain on a battery. In addition to the CPU/disk consuming power, the wireless card consumes significant power to transmit data. [2] In many cases, multimedia materials are distributed with the text in CDROM format which students use during the class, consuming additional power. External artpads and video/audio devices will consume additional power. Because battery power is so crucial, we recommend the purchasing the best batteries you can get. In our case these were Lithium Ion batteries. A second backup battery is highly recommended.

Screens:

Students will want to view the collaborative whiteboard, use a word processor for taking notes, surf the net, or rerun the example just discussed. In other words, the screen must be large enough to have several windows open side-by-side at the same time. Thus it is important to get the biggest screens possible with the highest resolution possible. Our student laptops had 11.5'' screens which students generally felt were too small.

Drawing Devices:

Since students are accustomed to handwriting and hand drawing figures when taking notes, a good drawing device is important. Pen-based artpads are by far the best but typically require power from an electrical outlet. However, pen-based scratchpads can be used. Ones that support absolute mode are the best (where there is a one-to-one mapping between the screen and the pad area), but those that use relative mode are usable. Built-in pen-based scratchpads are easier to deal with than serial line or other external scratchpads.

Video:

In distance learning situations, video equipment allows remote students to view the instructor and classroom. Video cameras/cards that do compression are important in a wireless setting with limited bandwidth. Software compression steals valuable CPU cycles from the machine and can severely degrade performance of other applications like the whiteboard.

Network Design Issues

  Where the access points are placed and how they tie into the wired network play a critical role in network coverage and performance. In particular, the following points should be considered when designing the wireless network:

Overlapping Cells:

Placement of the access points is important to ensure proper coverage and can also be used to increase bandwidth and alleviate the load on an access point. Each access point provides coverage of a area called a cell. To avoid dead spots between cells, plan for sufficient overlap between adjacent cells. For dead spots at the edge for periphery cells, make sure the outer edge of the cell extends sufficiently far beyond the edge of the classroom, otherwise transient interference may suddenly disconnect students on an edge of the classroom. Laptop-to-access point coverage is obviously important, but interactive wireless classrooms also require laptop-to-laptop coverage. Although less likely, it is possible that two laptops in the same large room use the same access point, but cannot communicate with one another. In this case, adding additional (carefully placed) access points can solve the problem. Overlapping cells can also be used to increase the access point to laptop ratio, thereby reducing loads on an access point and increasing the aggregate network bandwidth supported.

Finally, care must be taken to avoid routing loops caused by overlapping access points communicating with one another. In general, most access points are designed to prevent such loops via a spanning tree algorithm, but it is not difficult to accidentally disable or defeat these features and create loops that bring down both the wired and wireless network.

Wired/Wireless Interaction:

Current wireless LAN technology operate at rates of roughly 2 Mbps while conventional ethernets operate at 10 or 100 Mbps. Although wireless speeds are expected to increase to 10 Mbps before year's end, they are not likely to match leading-edge wired speeds any time soon. This speed mismatch can be a problem. There is no way that all the packets on the wired network can be forwarded to the wireless network. To solve this problem, most access points have the ability to filter traffic based on destination or traffic class. Thus traffic not originating from or destined to wireless machines never appears on the wireless network. However, multicast and broadcast traffic can still be a problem because they is more difficult to filter. Although one can argue that networks with substantial broadcast/multicast traffic are poorly designed or use poorly designed applications, the fact is that such networks are quite common. The College of Engineering network to which we connected is one example (for a variety of reasons) and so broadcast traffic was initially consuming a significant portion of our wireless network's 2 Mbps bandwidth. Fortunately the access point allowed us to filter the majority of the broadcast traffic by tossing packet types which we were not using. Note that disabling multicast/broadcast is an option but probably not something you will want to do if you are using protocols like DHCP to assign IP addresses. Our access point also allowed us to set a maximum number of broadcast/multicast messages allowed per second which is a useful feature.

Mobile vs. Roaming:

When designing the network, it is important to decide what is required: mobility or roaming. We define mobility to be the act of stopping all network communication, moving to a new location serviced by a different access point, and restarting all network connections. Roaming removes the requirement of stopping all network connections. A roaming machine can move from one access point to another while all network connections and data transfers continue on uninterrupted. Wireless classroom environments do not require roaming support since it is usually acceptable for a student to be ``off'' the network while moving between classes.

While mobility is possible with most vendors and operating systems, roaming is not. Roaming requires that access points cooperate to seemlessly ``handoff'' a machine as it moves from one cell to another. This typically requires handoff support in the access points and roaming support in the operating system. For example, we could achieve mobility in the version of Linux we were using, but not roaming, while both were supported under Windows 95. Note that roaming typically assumes continuous coverage from the starting cell to the destination cell.

Mobility Limitations:

Even if mobility and roaming are supported, there are limitations that you should consider. First, if the access points are connected to different IP networks, roaming is probably not supported. Second, if the access points are attached to a bridged network, roaming may or may not be supported. If ``transparent'' bridging is used, roaming will work as expected. However, some learning or smart bridging hardware may become confused when a laptop moves from one access point to another and will incorrectly deliver its messages.

Our experience shows that the aggregate bandwidth capacity of the wireless network is very important for interactive multimedia applications. In order to support large class sizes (or multiple classrooms located close to one another), substantial overlap between access points can substantially improve performance. Roaming is a nice feature, but we have not found it to be particularly important for the wireless classroom environment. We did run into problems when connecting to our bridged network, both because of the large amount of broadcast traffic, and the fact that our bridges were not all operating in a ``transparent'' mode.

Software and Protocol Issues

Not surprisingly, we found that instructors really liked the ability to include multimedia in their lectures and didn't hesitate to use it extensively. As a result, lecture slides had lots of fancy diagrams, figures, charts, and high-resolution images. They also liked using animations and video clips that came with the textbook or were obtained from the WEB. Although this is exactly what we were hoping for, this type of activity quickly brought the wireless network to its knees and in a few cases even hung the network. It was not uncommon for slides with graphics or images to be a megabyte or more in size. Video clips were often several megabytes in size. The live video/audio feeds were not excessively large (between 30 and 150 Kbps), but they required low variance and packet loss as did the other interactive applications. Hopes of app sharing were essentially abandoned early on, and collaborative web browsers resulted in all the students requesting the same document from the same server over the network at the same time. In other words, wireless networks are not yet capable of handling arbitrarily large amounts of multimedia data.

Given the limitations of existing wireless technology, instructors/students must carefully decide when multimedia is necessary (as opposed to just flash) and must also select applications and protocols that limit the load imposed on the network. Collaborative applications that support multicast communication exhibited, by far, the best performance. Unfortunately most applications written for Windows 95 do not support multicast communication. Even collaborative applications like Databeam's Distance Learning Server and Teamwave's Collaborative Workplace use a centralized server and unicast communication and do not scale up to support live multimedia traffic. Preloading slides and other multimedia data at the beginning of class can increase performance during class but often delayed the start of class and did nothing for live multimedia.

We observed the best performance with the MBONE collaborative application suite because it supported IP multicast communication. We typically used a whiteboard (wb), shared text editor (nt), video feed (vic), and audio feed (vat). We were able to run all of these simultaneously with excellent performance. Unfortunately, these tools were not developed for the wireless classroom and often omitted features that would be useful in such a setting. We also wrote our own collaborative web browsing system that used standard web browsers but replaced the unicast communication model with a multicast communication model. The browsers were driven by multicasting the URL to be followed and then by having the web server multicast the desired pages back to the student machines.

In order for us to make effective use of the wireless network, we had to use authoring tools that generated small data files. For example, when including images in web pages, jpeg images were used over gifs. Images were also scaled to the desired size rather than sending the unscaled image and then scaling it while displaying. When creating animations, shockwave or other compact representations were used. When generating postscript, compressed postscript formats were used.

Conclusions

Wireless classrooms and interactive multimedia classes have the potential to revolutionize the way we teach. However, the heavy loads imposed by interactive multimedia lectures and distance learning, combined with the current limitations of wireless networks, means that the network must be carefully designed and used. We have described the features of a network design and software that can make wireless classrooms effective even with today's technology.

References

1

WAVELAN Web Page.
http://www.wavelan.com.

2

AT&T.
WaveLAN Users Guide, 1995.

3

J. J. Burg D. G. Brown and J. L. Dominick.
A Strategick Plan for Ubiquitous Laptop Computing.
CACM, 41(1):26-35, Jan 1998.

4

Col. S. Grier and Lt. Col. L. Bryant.
The Case for Desktops.
CACM, 41(1):70-71, Jan 1998.

5

J. Griffioen, W.B. Seales, and J.E. Lumpp.
Teaching in Realtime Wireless Classrooms.
In Proceedings of the 1998 Frontiers in Education Conference, November 1998.

6

D. Mutchler L. Kiaer and J. Froyd.
Laptop Computers in an Integrated First-Year Curriculum.
CACM, 41(1):45-49, Jan 1998.

7

Jr. R. LeBlanc and S. Teal.
Hardware and Software Choices for Student Computer Initiatives.
CACM, 41(1):64-69, Jan 1998.

8

M. Weiser.
The Future of Ubiquitous Computing on Campus.
CACM, 41(1):41-42, Jan 1998.

Footnotes

...Project

The Wireless Classroom project is supported in part by Databeam Corporation and the Kentucky Information Resource Management Commission and the University of Kentucky.

...sharing

Application sharing (often just called app sharing) is a way in which two or more users can ``share'' (see and interact with) the same application. The application appears on both users screens with changes made by one user occurring on the other user's screen as well.

...screen

The projector is not necessary because the students' machines also receive and display the instructors notes. However, we had some students auditing the class without laptops.

...email.

At first we were concerned about students misusing their newfound freedom, but to our surprise we found that most of them used it in constructive and innovative ways with minimal non-constructive use.

...infrastructure

Although there are a variety of wireless technologies, we will focus our attention on radio frequency systems which are the best suited for the types of wireless classrooms we envision.

...data

A 2.4 GHz card uses about half the power of a 915 MHz card to transmit the same data.

...supported

There is research going on, but you are not likely to find support for this in off-the-shelf hardware.