Radio Frequency (RF) Survey

Municipal Wireless Broadband and the Digital Community Report Radio Frequency (RF) Survey for the City and County of San Francisco July 21, 2006

Preface This document describes Civitium s findings and recommendations following the performance of a Radio Frequency (RF) Survey during May and June 2006 for the City and County of San Francisco. These results are intended only to provide general insight into the spectrum environment throughout San Francisco, specifically for various unlicensed frequency bands being considered for the TechConnect initiative. Should the City and County reach a definitive agreement for the deployment of a community wide wireless broadband network, it is assumed that the party responsible for deploying the network will perform its own spectrum survey and analysis prior to deployment, and that ongoing monitoring of the environment will be required. Since spectrum environments are inherently dynamic, this survey should be considered only a snapshot from the May June, 2006 timeframe. This survey should not be used to predict what the environment may look like at any point in the future. AirMagnet is a registered trademark of AirMagnet, Inc. All other product names mentioned herein may be trademarks of their respective companies. Copyright Civitium, LLC 2006 Page: 2 of 13

Table of Contents PREFACE... 2 TABLE OF CONTENTS... 3 1 INTRODUCTION... 4 2 DATA COLLECTED... 4 2.1 RF SPECTRUM SCANS... 4 2.2 WIRELESS LAN SCANS... 5 3 FINDINGS... 6 3.1 EXISTING SIGNAL LEVELS... 6 3.2 EXISTING WIRELESS LANS... 7 3.3 TOPOGRAPHIC CONCERNS... 7 4 RECOMMENDATIONS... 8 4.1 SNR VALUES... 8 4.2 BACK HAUL FREQUENCIES... 8 4.3 CLIENT ACCESS FREQUENCIES... 8 4.4 DYNAMIC RF CHANNEL SELECTION... 9 4.5 OPERATOR COORDINATION... 9 4.6 LICENSED OPERATORS... 9 4.7 MOUNTING ASSETS... 10 4.8 INDOOR EQUIPMENT...10 APPENDIX A LEAD ENGINEER BIO... 11 APPENDIX B RF SPECTRUM SCANS... 12 APPENDIX C WIRELESS LAN SCANS... 13 Copyright Civitium, LLC 2006 Page: 3 of 13

1 Introduction This report includes the methodology, findings and recommendations in relation to a pre installation analysis of the radio frequency (RF) environment for the deployment of a citywide Wi Fi network for the City and County of San Francisco. The survey team collected data throughout the City s approximate 50 square miles. Throughout the report, this area is referred to as the survey area. 2 Data Collected Civitium collected RF data within the survey area between May 22, 2006 and June 2, 2006. The data consisted of RF spectrum scans and wireless Local Area Network (LAN) scans. This data offers a basis for determining the impact of existing signals on the operation of a citywide Wi Fi network. 2.1 RF Spectrum Scans Spectrum scans were collected using a spectrum analyzer to measure signal activity in the 2.4 2.5 and 5 6 gigahertz (GHz) frequency bands at various locations throughout the survey area. Numerous spectrum scans were performed at ground level using an AirMagnet Spectrum Analyzer Card (shown below) inserted into a laptop running AirMagnet s Survey software. This spectrum analyzer used an external omni directional antenna. The software automatically captured a spectrum scan at rapid intervals while the survey team drove throughout the survey area. The software plotted the location of each spectrum scan on a map based on coordinates obtained through a Global Positioning System (GPS) device, with external antenna for better position accuracy. The survey team drove throughout the survey area in a manner to capture a minimum of two spectrum images of the 2.4 2.5 and 5.0 6.0 GHz frequency bands per square mile. Hundred of spectrum images, however, were actually recorded. Spectrum scans were also performed atop several rooftops and antennas sites distributed throughout the survey area using a Bantam Instruments 425A spectrum analyzer (shown below.) Copyright Civitium, LLC 2006 Page: 4 of 13

Refer to Appendix B for the identification of chosen spectrum image locations and resulting spectrum images. 2.2 Wireless LAN Scans The wireless LAN scans were done through the use of AirMagnet Survey software interfaced with an 802.11a/b/g radio card and a GPS device. The software automatically recorded the SSID (Service Set Identifier), RF channel setting, signal strength and noise levels of wireless LAN access points while the survey team drove throughout the survey area in half mile increments. This provides a sample of the wireless LANs active within the survey area. Refer to Appendix C for images that depict the routes driven throughout the survey area and maps indicating average noise and access point signal amplitudes. Copyright Civitium, LLC 2006 Page: 5 of 13

3 Findings This section of the report discusses various findings based on the spectrum scans, wireless LAN scans, and notes made while visiting the survey area. 3.1 Existing Signal Levels The existing signal levels shown in the spectrum images and wireless LAN scans in Appendix B and Appendix C, respectively, suggest a limited level of noise impact to a municipal Wi Fi network. These existing signals are emanating from wireless LANs and other radio equipment currently operating throughout the City. In most cases, the average signal levels were below 85dBm at ground level throughout the survey area in the 2.4 2.5GHz and 5.0 6.0GHz bands (refer to Appendix C, Figure C 3). This only provides a rough indication of existing signal activity within the survey area, but it offers an initial basis for assessing the impact on the deployment of a Wi Fi system. This consistent average noise does not pose significant potential problems to the use of these frequencies for supporting mesh nodes at ground level within the City. It s comparable to other cities successfully implementing metro scale Wi Fi networks such as Philadelphia, Miami Beach and Anaheim. More specifically, however, the spectrum scans (refer to Appendix B) indicate consistent radio activity at 5.8GHz observed at ground level throughout the survey area. The peak signal values were often 70dBm (moderately high) at 5.8GHz. This may preclude the use of 5.8GHz for backhaul in some areas, depending on the range of the backhaul and the technologies implemented. In addition, there was existing signal noise indicated in the spectrum scans in the 2.4 2.5GHz band at ground level that will require careful design to avoid the impacts of RF interference and optimize the performance of a city wide Wi Fi network. The maximum signal levels in the 2.4 2.5GHz frequencies are moderately high ( 70dBm) in parts of the band at some locations, such as Area 1, Location 9. These signals are likely emanating from existing Wi Fi access points. In most cases, it will be possible, though, to tune the mesh nodes of the city wide Wi Fi network in the applicable locations to significantly reduce the interference. On the rooftop of One Market Plaza, the spectrum scans indicate considerable signal activity in the 2.4 2.5GHz band. Signals were consistently present at 75dBm and even 60dBm to 50dBm (high) in this band. These signal levels are too high to enable effective backhaul should the 2.4 2.5GHz band be used Copyright Civitium, LLC 2006 Page: 6 of 13

at One Market Plaza. The other rooftop locations, such as the Central Radio Station, Clay and Jones Building, and Bernal Heights radio site, have moderate signals levels in the 2.4 2.5GHz band, but the levels are low enough or within parts of the band that will likely enable the use of 2.4GHz backhaul, depending on the range of the backhaul. Despite the presence of these existing signals, the proper design and deployment of the network should allow effective operation of the system. 3.2 Existing Wireless LANs The wireless LAN scans shown in Appendix C indicate that there are numerous wireless LANs operating throughout the survey area. The AirMagnet Survey software captured approximately 5,500 wireless LAN access points as the survey team drove along the survey route show in Appendix C, Figure C 1. This constitutes a sample of the total number of existing wireless LANs. Based on this data and the range of typical access points, it s likely that there are five times more (27,500) access points active within the entire survey area, which results in an average of 550 access points per square mile. Even though there are many wireless LAN access points currently operating within the City, the signal levels from these existing networks are relatively low when measured outdoors, where the city wide network equipment will be deployed. The vast majority of these existing wireless LANs are installed inside buildings, which attenuate the signals as they propagate outside to levels that will likely have minimal impacts on the city wide network. As shown in Appendix C, the signal strength of the existing 802.11b/g wireless LAN access points (measured outdoors) is mostly between 85dBm and 65dBm. The majority (74 percent) of the existing access points are operating on channel 6, the common default channel. By setting the city wide Wi Fi mesh nodes to non overlapping channels in relation to channel 6 (such as channel 1 or channel 11), the majority of the interference from existing access points can be avoided. Only 1.5 percent of the access points found by the AirMagnet Survey software were 802.11a access points operating in the 5GHz bands. Thus, the impact of existing 802.11a access points is insignificant. 3.3 Topographic Concerns Large portions of San Francisco include hills and curved streets, which can be difficult areas to cover with Wi Fi mesh networks. It may be possible, however, to provide adequate signal coverage with Wi Fi mesh nodes throughout the City by increasing the density of these nodes above other density used for other cities. This assumes that adequate mounting assets are available throughout the City at groundlevel to support the nodes, and from rooftops to support the backhaul. Copyright Civitium, LLC 2006 Page: 7 of 13

4 Recommendations Based on the findings above, the following are Civitium s recommendations for deploying a municipal Wi Fi wireless network throughout the City: 4.1 SNR Values The SNR (signal to noise ratio) in all areas of the city wide Wi Fi network should be at least 15dB SNR to enable stable connections. In addition, signal levels of the Wi Fi network should be high enough in all areas to provide required data rates as indicated in the technical specifications of the client radio cards that the City decides to support. If the wireless network will be supporting voice applications, then the SNR will likely need to be higher based on the specifications of the supported wireless voice handsets. For example, Cisco recommends a minimum of 25dB SNR for the Cisco 7920 Wireless IP Phone. The installation of the network should adhere to these values to ensure that existing noise levels doesn t limit range and performance of the municipal wireless network. In addition, it s important to understand that these minimum SNR values must be present at the location of the wireless user device and the mesh node to ensure effective two way communications. 4.2 Back Haul Frequencies In order to reduce RF interference between backhaul and existing systems operating in the 2.4 2.5GHz band, the backhaul frequencies should be in the 5.0 6.0GHz or 902 928 MHz bands in some areas, such as on the rooftop of One Market Plaza. This avoids excessive retransmissions and potential lower performance that may occur in the 2.4 2.5GHz bands. As mentioned earlier, it may be necessary to avoid 5.8GHz in some areas due to current systems operating at that frequency. 4.3 Client Access Frequencies As mentioned before, most (74 percent) of the existing wireless LANs operating in the 2.4 2.5GHz band are set to channel 6. As a result, the mesh nodes of the city wide Wi Fi network should be set to channels that don t conflict with channel 6 (such as channel 1 or channel 11). Copyright Civitium, LLC 2006 Page: 8 of 13

4.4 Dynamic RF Channel Selection The continual proliferation of wireless LANs and other radio frequency equipment throughout the City will introduce additional sources of potential interference in the future. As a result, the use of access points or mesh nodes that dynamically choose the least congested channel may improve the performance of a city Wi Fi network. Keep in mind that static channels should be chosen, however, if the wireless network will be supporting wireless voice applications. Most Wi Fi phone specifications recommend using static channels to avoid excessive handset roaming and resulting dropped calls. 4.5 Operator Coordination Regardless of the amount of pre planning and engineering that occurs, unintentional interference between operators and/or consumers may still occur after the deployment of a citywide Wi Fi network. For this reason, Civitium recommends that the City require the party deploying the network to define and commit to certain measures for identifying, diagnosing, responding to and mitigating the impact of this unintentional self interference. This may include participating in, or creating, a forum for operators to work together voluntarily to address these issues and/or creating a forum for complaints by affected parties. This approach may address 90% of the issues that surface, since networks owned by both the party causing interference and the party being interfered with can suffer when interference occurs. This creates a natural incentive for coordination and cooperation. One such organization, the Wireless Broadband Access Network Coordination (BANC) group, currently exists, with more than thirty members participating 1. The group is defined as a group of fixed wireless operators that are working together to minimize interference and maximize reliability and spectrum efficiency and it may serve as a model for addressing these inherent challenges. 4.6 Licensed Operators While often referred to as unlicensed, Wi Fi services, specified in the FCC s Part 15 Rules, are technically certified by the FCC, and operators do not require licensing. This is a subtle, but important distinction in that there are other services operating in portions of the same shared spectrum that do require a license. One such service, Amateur (Ham) Radio Service, operates in shared spectrum according to the FCC s Part 97 Rules. Ham Radio is defined as a voluntary noncommercial 1 See http://www.wbanc.com/ Copyright Civitium, LLC 2006 Page: 9 of 13

communication service, used by qualified persons of any age who are interested in radio technique with a personal aim and without pecuniary interest.ʺ These Amateur Radio Services often operate at higher output power (technically EIRP) than Wi Fi services, so the possibility of interference does exist. Civitium proposes that the same mechanisms referenced in Section 4.5 above may be used to address these issues if they occur. The common view in the industry is that Ham Radio users are reasonable and are often happy to cooperate with operators of other services 2. 4.7 Mounting Assets Mesh node installation locations. The City should mount mesh nodes on the arm of light poles and traffic light poles, with the node as close to the center of the road as possible rather than along the side of the road. This generally maximizes signal propagation. The mounting assets should be clear of trees if possible. In addition, mesh nodes should be installed on mounting assets that are located ideally at street intersections. This makes best use of the mesh node for covering larger areas. If only vertical poles (no arms) are available, then it s best to stagger installation of the nodes on poles along both sides of the street. It s important to not use mounting assets for wireless mesh nodes that are higher than the structures surrounding them. This avoids existing wireless LAN signals and RF noise from impacting the operation of the node and the users it supports. Gateway/backhaul installation locations. The location of gateways and backhaul components should be installed on mounting assets offering unobstructed line of sight with the headend or a relay point. The designer of the city wide Wi Fi network should take into consideration all tower sites and rooftops that the City has access to when selecting headend and relay points. 4.8 Indoor Equipment While some users may be able to access the network directly from a desktop, laptop of other mobile device while indoors (e.g. inside a residence or business), it is assumed that in the majority of cases, a dedicated wireless bridge or CPE (customer premise equipment) will be required. These devices, which typically cost between $75 and $100 (and are often subsidized for paying subscribers), incorporate directional, high gain antennas and higher output power (often 200mw or more), allowing transmissions from the user to reach its closest Wi Fi mesh node. 2 See Hams are Your Friends at http://www.isp planet.com/fixed_wireless/business/2001/ham_bol.html Copyright Civitium, LLC 2006 Page: 10 of 13

Appendix A Lead Engineer Bio James T. (Jim) Geier is an Associate Partner with Civitium and the founder of Wireless Nets, Ltd. His 25 years of experience includes the analysis, design, software development, installation, and support of numerous wireless network based systems for municipalities, enterprises, airports, homes, retail stores, manufacturing facilities, warehouses, and hospitals worldwide. Jim is the author of several books, including Deploying Voice over Wireless LANs (Cisco Press), Wireless LANs (SAMS), Wireless Networks First Step (Cisco Press), Wireless Networking Handbook (Macmillan), and Network Reengineering (McGraw Hill). He is the author of numerous tutorials for www.wi FiPlanet.com and other publications. Jim has been active within the Wi Fi Alliance, responsible for certifying interoperability of 802.11 (Wi Fi) wireless LANs. He has also been an active member of the IEEE 802.11 Working Group, responsible for developing international standards for wireless LANs. He served as Chairman of the IEEE Computer Society, Dayton Section, and Chairman of the IEEE International Conference on Wireless LAN Implementation. Jim is an advisory board member of several leading wireless LAN companies. Jim s education includes a bachelor s and master s degree in electrical engineering and a masterʹs degree in business administration. Jim has completed training for various municipal Wi Fi system solutions focused on designing, installing, testing, and troubleshooting municipal wireless networks. Copyright Civitium, LLC 2006 Page: 11 of 13

Appendix B RF Spectrum Scans Provided under separate cover. Copyright Civitium, LLC 2006 Page: 12 of 13

Appendix C Wireless LAN Scans Provided under separate cover. Copyright Civitium, LLC 2006 Page: 13 of 13