Sept 16th,1997 Connecting Up A Little Old Library Wirelessly Part of our National Science Foundation Field Tests for Education Project includes seeing what wireless can do for connecting up libraries. What have libraries got to do with Education? Well that depends on one's definition of 'education.' One can take the narrow view that it is what occurs only as formal activity in places called schools or colleges with registered students and accredited teachers. And most of which learning activity takes place at 'school' or by study at 'home.' Or one can take the broader view that it includes all attempts by citizens to 'educate' themselves, both by formal institutional programs, and by self-directed, or intermittent 'learning' - much of which has historically been done by access to local public libraries. This project recognizes the latter definition. It assumes that, especially in the current and coming age, K-99 - or lifelong - 'education' seems required - all citizens continually having to learn - or teach others - as academic, vocational, knowledge and skills change, and all must 'keep up' in one way or another. American public libraries have always been open to 'all' citizens, not just those in academic programs. And US libraries have been getting connected to the Internet in increasing numbers. Thus, to the extent that a public library can give its patrons access to the Internet, even citizens who do not own computers at home, or cannot, for one reason or another access a commercial Internet provider, or know how to use the net, the goal of continued education by the public is served by access at library facilities. Libraries also can give net access to students in formal educational programs who do not have away-from-school access to the Internet, a chance to keep up with their more fortunate classmates who do have such access from home. Bandwidth Cost However there is a problem faced by smaller, rural, or even larger libraries in urban areas with many 'branch' libraries. That is the local loop cost of sufficient bandwidth connections to the net that can for one or more computers-as-workstations fully use the multi-media capabilities already becoming standard with the World Wide Web - graphics, sound - speech and music, and video, as well as text. While competition has driven down the costs of microcomputers that can display multi-media (from CDs or other programs), so they seem to be within the range of one-time purchase cost budgets of libraries, high bandwidth connections to the Internet, with recurring monthly costs may not be as affordable, especially to branch libraries or rural libraries. One of the problems is technical - telephone companies can usually only offer either dial-up, 56kbs (or in some places 128Kbs with ISDN), or full T-1 (1.544kbs) capabilities. The Pikes Peak Library District of Colorado Springs is a case in point. This progressive library system, which was the first public library in the United States, in 1980, to give patrons dial-up modem access to its data bases (called 'Maggies Place' - a legend among librarians) has grown along with a rapidly growing El Paso County. It now has 10 branch libraries, some in shopping centers, some in older facilities, spread across the metropolitan area of the city, with the main library very modern and well endowed with technology that has evolved with the growth, but the smaller branches two generations behind - full web net access and computers, rather than just older character terminals, capable of accessing and displaying TCP/IP net resources. It has long since had limited access - at the main library - to the Internet, permitting both patrons inside the main library at terminals, and those dialing up by modem, to access - Lynx character graphics only - to internet resources. And the main library supports a graphical web site, which patrons fortunate enough to have commercial accounts with ISPs, and TCP/IP and Web browsers on their system can access. But the most the District felt it could afford has been business-rate plain phone lines connected to each branch library, which support character-level (non graphical) terminals, which can access the central data bases by patrons or by staff. But no web access from the branches. Which requires both microcomputers and, at the minimum, 56Kbps or ISDN connections to each branch, with DSU/CSUs at each branch, and commercial data services, with recurring monthly costs paid to US West. Or, for really good multi-media access by more than one or two web machines per branch, T-1 local loop data line, at sharply greater monthly costs. ENTER WIRELESS While the District is planning to acquire sufficient microcomputers for all the branches, after a Bond issue recently passed, this NSF Project offered to connect up one branch library to the Internet wirelessly, both to demonstrate the capability, measure the throughput and reliability, see what it takes to integrate a computer with access into a traditional library setting, and gain information on patterns of use by regular patrons. The library we selected is one of the smallest and oldest in the city - the Old Colorado City Branch Library. It is a small 50 by 80 foot brick one floor building, built in 1904 by the Carnegie Foundation. It first served seperate Colorado City from 1904, when the collection from the very first tiny 'Woods Library' that had existed since 1896 was moved to the new Carnegie Library. From 1917 when Colorado City became part of greater Colorado Springs, it has served the Westside of Colorado Springs, with a population less than 15,000, ever since. It is a very traditional, neighborhood library with a staff of 2, some volunteers, and patrons largely from the lower income Westside of Colorado Springs. For its size it is quite active, with at least a 1,000 patrons a week coming into the small, friendly place. It represents for many, the only affordable source of books, magazines, newspapers, and reference works. But its size, a collection of only about 20,000 books, limits its holdings. After a meeting with the library staff - both the technical staff headed by Susan Hazlewood, and members of the policy staff - at the Old Colorado City History Center for a demonstration, where a wirelessly connected web site was already operating under a seperate NSF Grant, the offer to make a similar installation in the Old Colorado City Branch Library was forwarded to the Library board, which agreed to host this test. The staff noted that the library district is only now coming to grips with the new set of use policies which full web and net access for patrons will require. So the test may also help them in the development of those policies. For the duration of the NSF Project Test period - ending in 1998, there is no cost obligation to the Library District. And if the system proves useful, then the NSF will transfer the property to the District. Since the District had yet to make purchases of microcomputers for the branches for Internet access, we installed a 100Mhz 486, with Windows 95, computer that we had used the last year at the San Luis Centennial School for our NSF wireless test project. The computer also has multi-media peripherals, such a Sound Board and earphones, to permit library patrons to listen to Real-Audio or other protocol voice and sound sources. This would also give us more data on how the wireless link compares with a 28.8 modem link, for multi-media displays and streaming audio, which are bandwidth intensive. We then chose a Wi-Lan Hopper Plus wireless data radio, that lists for about $2,500, less antenna, to deliver as close to its rated 1Mbps as it could, linked to a similar radio set, with roof top 16dB gain yagi antenna in the multi-user mode, on the premises of Old Colorado City Communications, the research center offices for the NSF Field Tests, a half mile away. This radio, being the third one linked to the base radio at OCCC, (the other two being one at the History Center, and a demonstration radio connected to a web site on an OS2 system on the business office premises of the PI) helped constitute a Wide Area Network (WAN), almost a Municipal Area Network (MAN). All sharing a 1Mbps link to the LAN in OCCC, which in turn has been connected to the Internet via 3 BreezeCom Pro radios. So a wireless link TO an ISP, and FROM the ISP to the upstream provider. But no local telephone loop involved all the way to downtown Colorado Supernet's Cisco servers. One question was how to find the best physical arrangement in a very small library with minimal electrical infrastructure, of (1) the computer, accessible to patrons yet in a place it could be overseen by the small staff and suitable for use (2) the power cords needed to connect to wall power, (3) the LAN wire connection between the computer ethernet to the radio (3) the radio placement (4) the heavy RF cable to the antenna (5) the antenna itself. The goal was to place all the components without drilling holes or making other structural modifications to the historic building. Or to cross the walking spaces with wires. The challenge is that the placement of the computer for patron use may or may not be close enough to the best place for the radio, which itself has to be readily connectable to the antenna by thick, stiff (low loss) RF cable. INSTALLATION The most suitable place to put the Monitor, keyboard, Mouse and Pad, from the librarian's standpoint, was on an old oaken table in the Reference alcove in the front of the building, within sight of the front desk. Still leaving more than half the table for use by other patrons without being disturbed. The placement was, fortunately, near the very tall, screen covered windows which are on the side of the building open to the directional yagi's 40 degree path from the 4th story (roof) of the brick building in which the base Wi-Lan radio resides. From some parts of the front of the library building the yagi could be seen peeping over the brick walls of intervening large buildings. A large thick pine tree planted next to the library front obscured a plain line of sight between the closest window to the computer, and the distant yagi. Schematic of Wide Area Network To Upstream ISP West Colorado Avenue \ _____________ __________ _______ _________ \ _______ | | | || || | | \ | 24| Bancroft |Col| Commercial Brick Bldgs |25| X <--Base ______ St| Park |Crt| ||3 story|| |St / OCCC Radios |X | | | | || || | /| BLDG | |____| |___________| |________||_______||_______| |_______| ^ Pikes Peak Avenue / History ___________________________________________ ________ Center |__ ___ ________ ____ _____ ____ ____| |____ ____ | | | | |X | | | | | | | | || || | | | |_| |__| | | |__| |___| |__| |__|| ||__| |__| |_____ |______| Residential Homes || X| Library ||___| X=Radio/Antenna Locations P.I. Home Office The walls of the Carnegie library building are quite thick, at least a foot from inside wall to outside white brick face, but given the date it was constructed it appeared not to have interior metal - rebar. So it appeared that the 902-928Mhz frequency emissions should be able to pass through the walls without significant loss of strength. Higher frequency radios - such as 2.4Ghz Part 15 radios, or those with the European standard of 100mW (one tenth of a watt) might not work as well. The Wi-Lan radios selected were 500mW (half a watt) and 915Mhz. FALSE PROBLEM When the Wi-Lan radio was placed on a filing cabinet near the east wall, the ethernet and power cable running along the wall below the window sills to the computer CPU sitting under the table against the south wall, it appeared that the only remaining question would be whether or not a 12 inch by 12 inch 'plate' antenna would be sufficient gain to connect reliably with the distant radio. The antenna was placed on top of the radio itself, and oriented toward the yagi source. The Hopper radio indicated by a steady green light that there was a solid radio link to the distant radio. Then a continuous PING was set up on the Windows 95 machine (which had been assigned IP address 192.160.122.7 in its network properties, to reach 192.160.122.1 - a system directly on the LAN in the OCCC office, to which the base station Hopper radio was connected. But the ping returned largely dropped packets, with only a few returned pings. Something was wrong. So we changed the orientation of the plate antenna, first with the short RF cable, then with a much longer one. And we tried a yagi antenna inside, on a tripod. We moved the entire computer with radio that could be no further than 8 feet away with the Ethernet cables we had, to other locations in the library. But the ping results showed a very poor connection - from inside the library. Then we attached a 100 feet of low loss RF cable, and moved, first the yagi and tripod outside with an unobstructed view of the distant yagi, and then put a large (9 inch in diameter, 30 in length) omni 12db gain antenna in the form of a white cylinder into use. But there was no appreciable improvement. So, since the distant yagi was pointed such that the library building might be at the outer, marginal-signal edge of its 40 degree fan, we rotated the base antenna to insure its fan would cover the library building. Still no improvement. DISCOVERY AND SUCCESS So we began to suspect other components of the network, and sure enough we found a rather rare problem - a new ethernet cable that was just flakey enough to pass some packets, but drop all others. As soon as we replaced it the pings came up perfectly, with only 4ms of time delay. And we were able to go back to the 12 by 12 inch plate antenna (whose ergonomically attractive white surface could pass as a picture frame. In fact a suitable small picture could be pasted to its surface, and the antenna could become part of the decor of the library, without effect on its performance. Over the first several hours of operation, some interference from other devices were observed (strong red lights on the 'wire' light of the radio, together with flickering green connect light, and pauses in data being sent from telnet sessions from the keyboard or monitor. So the radio and antenna have yet to be experimented with to get the optimal placement, and observations made during all hours the Library is nominally open - usually from 10AM to 8PM at the maximum. IMMEDIATE USE BY PATRONS As soon as the Netscape web browser was configured to come up on boot up, and some preliminary test run, patrons started asking questions and showing an eagerness to use the system. Several did, showing some familiarity with web use, and all exclaiming how 'fast' it was. (our observation was that it was probably showing an end to end speed of at least 250Kbps through the network) During the time it took to install the computer and try the various antenna tests, which occured on three seperate days observation of the patrons using the Old Colorado City Branch Library confirmed part of our theory of the potential public value of a free, high performance, wireless connection to the Internet and the World Wide Web. The dress, age, demeanor of the bulk of the patrons coming in, including mothers with children, students, older persons, and persons obviously in the work world, indicated that the greatest majority of them were of lesser income than are observed using the Main Library on the east, newer side of town, where growth has been fueled by very well paying technology and defense-contractor firms. i.e. the more affluent, educated, upscale side of town. While the observed patrons of the small, old Carnegie library seemed to be classic 'Westsiders' whom it could also be presumed have less access or ownership of personal computers and commercial access to the Internet. Indeed among the 10 or so patrons who approached us - even volunteering assistance in the installation - only two admitted to having a modern personal computer at home with access to the net. And one of those was from out of town. Thus it appears that our original theory that the free computer and net access offered by wireless will benefit a population with less access to the net than enjoyed either by institutionally affiliated students, or by home-computer owning persons. It also remains to be seen, but it is a topic we have asked the progressive young new branch librarian - Amy Holtzworth - to monitor - whether the 'extension' by computer and network, to research materials , periodicals, and online publications NOT available in print form among the quite small holdings of the small library really makes the Old Colorado City Branch library a much larger resource for patrons than the size and scale of the physical library appears to offer. i.e. is this an incipient model for the 'virtual' library of the future? The 1Mbps speed of wireless connection can easily be used, if patron demand warrants, to extend by internal LAN (or even PCMIA card wireless LAN) to as many as 10 additional workstations, without, in our judgement unacceptable degradation of net access performance. Something only possible by telephone data lines, at the costly - for a small branch library - T-1 fiber recurring cost level. Which, in the City of Colorado Springs through US West, could cost as much as $500 per month per branch. After training the staff how to bring the system up in the morning, give help to patrons, watch for indicators of radio link problems or failures, and how to shut down the system, and the Log we asked them to keep on (1) patterns and numbers of patron use (2) problems with the network - outages, slow access, or other problems and (3) their own use - we left the system to run for the next several months, and learn what we can of it. So the system is up and operating, and under the control of the library staff. VALUE At contemporary market prices in Colorado Springs, a US West provided link with the same capabilities for 1, 5, 10 years, linking the branch library, which, given the distance and topography between the Old Colorado City Branch and the Main Library would require at least one 'relay' radio would compare as follows. US West Wireless DSU/CSU, Router $1,500 0 Radios $7,500 (3) Installation 500 1,000 Antenna 0 1,500 (3) Local Loop T-1 6,000 0 -------- ------- 1st Year $8,000 $10,000 5 Year total $32,000 $10,000 10 Year total $62,000 $10,000 For 10 Branches $620,000 $60,000 Scaled to 10 branch libraries it is rather obvious what the comparative economics would be between T-1 US West and the equivalent in FCC Part 15, spread spectrum wireless no-licence communication between the branch libraries and the Main Library. (there are a number of assumptions in the above table, of course. Actual costs may differ. But it is illustrative of the very different economics of commercial wired and wireless) There will be later progress reports on this Library-educational test project, and what we find, both technically, and adminstratively from its presence, and how useful it is to patrons. These views are of the NSF Wireless Field Test Staff only. When the final reports are filed, the Pikes Peak Library District will be given the opportunity to comment from their own perspective. Dave Hughes Principal Investigator NSF Wireless Field Tests for Education NSF Wireless Tests for 'Local History by Wireless' dave@oldcolo.com