Coverage and Network Traffic of Public Wireless Local Area Networks: WiFi Measurements and Modeling
By Chen Na & Jeremy K. Chen Schlotzsky�s�, founded in Austin in 1971, is a leading upscale sandwich franchise chain with over 600 restaurants worldwide. In 2001, Schlotzsky�s Deli was a pioneer in providing Internet access to its customers and since 2002, has been advancing the service one step further: to provide wireless connections for laptops and handheld devices with wireless network cards. In addition to providing indoor network access, Schlotzsky�s will enlarge its coverage to reach surrounding neighborhoods and businesses. WNCG is collaborating with Schlotzsky�s to solve two challenging issues with its public network initiatives. The first is to achieve the greatest coverage. Building layouts, obstacles, and antenna characteristics can affect signal transmissions, thus change coverage areas. The second is to efficiently use network bandwidth. For example, if only a few people run file sharing but occupy a large portion of the bandwidth, Schlotzsky�s may need to restrict their usage. In order to better manage public network bandwidth, user traffic patterns have to be determined. WNCG is aiming at solving these two research problems: maximizing the coverage and analyzing user traffic patterns. Coverage design of wireless local area networks (WLAN) is highly site-specific in nature. The traditional approach is by field measuring, as we see the industrious engineer from Verizon Wireless every day on TV: �Can you hear me now?� Engineers now can use the computer-aided software called LANPlanner to analyze how environmental factors such as construction material and tree density influence network coverage. LANPlanner was developed by Wireless Valley, a startup company located in Austin. The company, founded in 1998, makes complicated RF engineering very easy for network designers while providing a rapid yet powerful WLAN design environment. The computer-aided design takes four quick steps. The first step is to prepare a building layout. Second, designers assign partitions in the layout to be trees, concrete walls, dry walls, etc. The attenuation models of various partition types are provided in LANPlanner, and taskassigning is as simple as drawing multiplecolor lines in a picture. Third, one places antennas at several possible locations where power outlets and Internet plug-ins are available. Last, the software predicts the coverage according to all the given factors and arranges the locations of antennas to give the maximum overall coverage. One can further optimize prediction results by taking field measurements into account. We followed the procedure mentioned above and predicted the WLAN coverage in and around the restaurant located on Parmer Lane in Austin. The prediction helped Schlotzsky�s determine the optimal place to put an antenna while the restaurant was still under construction, when managers could not find the best antenna location on their own intuitions. For this restaurant, one antenna was sufficient to cover the indoor dining area as well as outdoor patio tables. Figure 1 shows the coverage prediction where a directional antenna is placed at the bottom-left corner and faces the bottom-right of the diagram. LANPlanner found that the optimal antenna location covered all desired sites. The three enclosed contours denote signal reception levels of -55, -60, and -65 dBm, respectively. Typically, a -65dBm signal level allows WLAN to operate at the highest rate, 11 Mbps. The -65dBm contour covers both the dining area and outdoor patio tables. In addition to software predictions, LANPlanner can analyze field measurement data and then adjust the attenuation factors of buildings and trees accordingly. By incorporating field measurements into its computation models, LANPlanner predicts network coverage more precisely. Measurements can only be done after wireless equipment is installed, so we conducted a measurement campaign at the restaurant on South Lamar Blvd. The WLAN there was equipped with a powerful antenna that uses six sub-antennas to enhance its coverage. The manufacturer of the antenna is Bandspeed, a company in Austin that provides a wide range of access-point products for next generation wireless networks. We used LANPlanner to collect signal measurements at 40 locations. Then, LANPlanner plotted a more precise prediction, shown in Figure 2. The three enclosed contours in the bottom of the figure denote signal reception levels of - 85, -75, and -65dBm, respectively. The minimum detectable signal level is typically -85dBm. The figure illustrates that coverage of the antenna can reach 200m, which is considerably greater than the coverage radius of a typical WLAN. Usually, a WLAN can reach at most 110m. The second research problem is analyzing real-world WLAN user traffic in Schlotzsky�s Deli restaurants. We plan to trace network usage in the restaurants for several weeks. The results of the measurements will provide insights into the types of uses of the pubic wireless network, typical user session experience, types of applications used by individuals, and communication bottlenecks. Furthermore, we will investigate how multiple computers share network bandwidth and how signal strengths affect data transmissions. Our observation will reveal how the current wireless protocol (IEEE 802.11b) performs and might lead to improvement schemes. Based on the network measurements and models, Schlotzsky�s Deli can improve efficiency of the network and better understand customer desires for marketing purposes. -- NYCwireless - http://www.nycwireless.net/ Un/Subscribe: http://lists.nycwireless.net/mailman/listinfo/nycwireless/ Archives: http://lists.nycwireless.net/pipermail/nycwireless/
