PIs: Kaveh Pahlavan, Sridhar Pingali, Hemant G. Rotithor, John A. Orr
Personnel: Ali Zahedi, Andy Messier, Jared Robinson
Sponsored by NSF

CWINS is significantly enhancing its research activities in wireless networks for multi-media applications with the support of an experimental wireless campus testbed that will be used to construct a wireless classroom environment. In this classroom, the instructor presents his/her material that is prepared on a personal computer through an electronic projector, and uses an electronic white board for handwriting. Multimedia services such as voice, text and bit-mapped images will be provided to students dispersed over a geographical area on the campus. In the past, CWINS has been recognized as a leader in wireless data communication research.

The testbed will further the current research in the following areas:

• Development of analytical and simulation models and benchmarks for performance evaluation of multi-media wireless networks using various wireless LAN technologies.
• Development of efficient mobility management methods that can scale to large wireless networks.
• Development and analysis of the scaleable protocols to provide reliable and prioritized transfer of multi-media traffic.
• Several faculty members, graduate students and senior undergraduates who are involved in research programs at the CWINS will share the use of the testbed with others involved in the Computer Communication Networks (CCN) program. The testbed will have a multifold impact that includes: Demonstration of feasibility of access to the information super highway using portable wireless terminals for multi-media traffic that will initiate collaborative exploration of new applications with WLTG industrial partners.
• Development of benchmarks specific to multi-media wireless network applications that will be used for further research in comparative performance evaluation of different wireless LAN technologies.
• Development of multi-media wireless networking protocols and mobility management algorithms that will be used by other research organizations in their prototype design and industrial organizations in the development of next generation products.

The results of research performed in this project will have a significant impact in all scenarios of group interaction including conferencing and meetings in industrial or corporate environments and are not confined to an academic setting.

Network Architecture for the Testbed

In order to support the services described above, we intend to use the network architecture shown in Figure 1. The center of the network is the Atwater Kent Laboratories in which the wireless classroom exists. Three wireless access points cover this building and they are wired to the ATM access switch that bridges the network to the campus backbone. One of the three wireless access points covers the classroom and the other two are designated to the rest of the building. Four more wireless access points, one in each building, cover the other four buildings that are included in the testbed. The wireless access points in the other four buildings are onnected to the central network in the Atwater Kent Laboratories through four sets of point-to-point wireless bridges. The point-to-point wireless bridges in the Atwater Kent Laboratories are wired to the central ATM access switch to provide a communication link to the rest of the wireless network as well as the backbone wired network. The traffic from the wireless network to the backbone is monitored with the HP Open View Network Node Manager of the HP workstation at the ATM switch. Notebook computers with PCMCIA cards can operate in the buildings with the coverage. The portable HP protocol analyzer is used at different locations in the network coverage to test the designed protocols.

Design and Performance Monitoring of a Wireless Campus Area Network

This project is being conducted by two fourth year undergraduate students, Andrew Messier and Jared Robinson.

Abstract: With the 802.11 Wireless LAN standard near finalization, there will soon be a sizable market for wireless local area network technologies. Wireless LANs will likely find many useful and interesting applications in today's high tech society. One of the most probable places that Wireless LANs will start to be implemented is in the college campus environment. Students will benefit by accessing databases and running applications over the school's network without the need to sit down at a designated terminal. The use of lap-tops in the classroom has the potential to add another whole dimension to a modern college education.

Here at WPI we are in the process of setting up such a network. Using Digital's PCMCIA Roamabout cards and radios, wirless Laptops are able to communicate with wireless access points located around campus, and connect to the campus's network backbone. Our project is to setup and monitor the performance of this network.

Objectives

• Installation of the Wireless Campus Area Network. Our first step will be the installation of the Wireless C.A.N.. The router and switch will be based in the CWINS laboratory on the third floor of Atwater Kent. This router is connected to the campus backbone through the EE department's LAN. An access point connected directly to the switch will provide coverage for much of Atwater Kent. Access points in the the buildings will be connected to the switch via point to point wireless bridges.
• Application Planning. Before we conduct meaning application level performance monitoring of the network we must first determine the likely applications that will be used. FTP and Telnet are perhaps the most widely used, however we also have multi-media applications and a whiteboard application that can convert the professor's class notes to a cumputer image that can be broadcasted to all of the students laptops. In addition, a laptop with Novell connectivity will be able to run the school's network software.
• Performance Monitoring of the Wirless LAN. Once we have installed the LAN on Campus we will conduct a series of benchmark tests that can measure delay times and throughputs for FTP, Telnet, and other possible non-campus related Wirless LAN applications such as data-base updating and credit card verification. Our goal is to obtain clear, application lavel data that will help future Wireless LAN planners. The Benchmarks tests we will perform are as follows:
• Effect of Walls
• Effect of Floors
• Effect of Interference
• Effect of Wireless Bridges
• Relation of RX power to throughput
• Comparison of actual power measurements to those obtained through software simulation
• BONeS simulation of Wirless LAN. Due to the lack of 'Real World' experience when it comes to Wirless LANs, it is hard to predict how the LAN will perform in certain situations without actually going through the installation process. However, software such as the Block Oriented Network Simulator ( BONeS ) can help in the planning of a Wirless LAN before much time or money is invested in an installation that will not work well. One interesting situation that we will investigate using BONeS is the case where multiple wireless LAN's are being used in close vacinity to one another. To do this we will use the wireless CSMA modules in Bones and create a layout of three independant Wirless LANs using the same frequencies. The outcome of this type of simulation will be useful for planning Wirless LANs in cities or crowded office buildings where conflicting Wireless LANs are a real possibilty.
• Comparing real power measurements to ray tracing models. The CWINS lab at WPI has always been a leader in indoor propagatin modeling. Ray-tracing software can predict the recieved power inside a building given its floor plan. This type of software will undoubtedly be used to plan the optimal placement for access points in a Wireless LAN. Once we have our actual Wirless LAN installed, we will able to compare real world power measurements to those obtained through the software.