2502 West Colorado Ave., Suite 203 · Colorado Springs, CO 80904 ·
Voice 719.636.2040 · Fax 719.528.5869 ·Wireless Web http://wireless.oldcolo.com
Progress Report 12
March 20th, 1997
As this NSF Wireless Field Test project has become more known, we have received overtures from companies making new generations of radios. They have offered to let us test them.
THE MITCHELL HIGH SCHOOL LINK
One test, however, that had been delayed because of difficulties in getting a satisfactory site survey for running 2Mbps Solectek radios between Mitchell High School in Colorado Springs and School District 11 Administration Building, is now about to be consumated.
We learned something from the several tests run by the radio engineers of Solectek of San Diego, with the assistance of LanTech, the system integrators from Colorado. It would take 4 radios to go from Mitchell, across town to Coronado High School, then, because of the narrow fan of the 2.4Ghz directional antenna aimed back at Mitchell would not quite emcompass the district building, a third radio would have to be placed next to the 2d radio at Coronado. A final 4th radio, with directional antenna would be at School District 11 headquarters, and, to clear the great mature trees surrounding the buildings there, require a 120 foot tower for the final antenna and radio. (the School District Headquarters, relative to most nearby schools in its district is down in a hole) Since the radios are in the $6000 class, 4 radios would be at least $24,000 - and the 120 foot tower in the $10,000 range - excessive for the T-1 local loop bypass total value. It would take too many years for the recurring costs of a US West T-1 link directly from Mitchell to the District, to amortize the cost of the radios. Lesson learned.
So we decided to cut the test to two radios - one at Mitchell that would connect up the workstations of the Computer Lab to a radio at Coronado High School, which then would interconnect to the wired T-1 from Coronado to the district. The wired T-1 directly between Mitchell and the District, would be severed, and tests would run through the end of the school year to see how the radio T-1 compares with the previous wired T-1.
Based on that changed plan, we retained Spectracom of Alamosa to erect two light towers, one 20 the other 30 feet, on the respectives roofs of the two large high schools, on which to mount the directional antennas to make the 6.8 mile length. That work was done in early March.
LanTech then mounted their Solectek radios in weatherproof penthouses on the roofs of each building, and the solid small 3' dish antennas on the towers, and, in a brisk windstorm that caused the antennas to oscillate, they configured the radios and established a link.
During the 3 weeks these radios have been up in a test mode, LanTech, responsible for tweaking them to their highest performance, have measured their performance and monitored their reliability. They have operated excellently, dropping few packets as they send test data to each other. However they are not operating at the peak speed expected - as indicated by the status lights on the radios which track the quality of the link. Since the radios and their antennas are cleanly in line with each other the 6.8 miles, we suspect interfererence. One prime suspect is the radio and radar equipment operating continuously at the heliport on top of Penrose Hospital, which building is in direct line with the data link. The line between the radios passes directly over the Hospital. We will attempt to test that thesis.
The link still operates at above 1Mbps, however, which is faster than the fractional T-1 that it replaces between Mitchell and the School District.
On Monday March 24th, in a coordinated effort, techs from District 11, LanTech and we, are going to cut over the Mitchell T-1 line, and IP assignments from its wired configuration, to pass through the radios, and test the link quality. This requires changes in IP routing tables, and putting IP numbers in each radio and their associated routers. Then, if the link looks good, the radios will remain in place through at least the end of the school year, to feed the T-1 link to the Media Lab workstations at Mitchell. We will both technically measure the performance of the link, and ask Becky Modic, who operates the Media Center computers, to observe and record the performance against her experience with the wired T-1 used over the past year. (which, frankly, has been out numerous times)
The District 11 techs are attempting to learn about the radios since the District, with funds in hand from a large bond issue for Technology for the 55 school system, are interested where the wireless may be a cost saver ovr US West T-1 lines. They are aware of the very large savings ($900,000 up front, and $144,000 a year) experienced by School District 20, to the north of them, after its installed wireless completely between its 22 school sites. District 11 has 55 seperate schools. I have pointed out, however, that our analyses of the problems of location and link with District 11's main headquarters building, which is the location of the District POP for all its schools, shows that the architecture of a radio network in that district will take some innovative planning, and IP routing. Radio may best be applicable to linking elementary and middle schools to their associated nearby high school (the school 'feeder' system the district is organized under), rather than as a district 'backbone' of radios to the high school.
BREEZECOM'S NEW RADIOS
Over the past several months, not reported above yet, we have been trying out a pair of BreezeCom radios between our Old Colorado City Communications office location and the OS2 server location in my, (the Principal Investigator's) house - where I have a home office.
This is the site where I have previously, going back three years, have been variously using for daily work, a pair of 120Mbps Tetherless Access (TAL) units, with their modified, direct sequence, Cylink radios, and then later, a pair of 115Kbps FreeWave, frequency hopping radios. So I have had extensive, daily, subjective experiences with both of these radios. I found that the TAL units, both older ones that they first marketted, and later ones, because of their Cylink direct sequence engines, operated faster (thruput) than the frequency hopping FreeWaves - even though both were rated fairly closely.
Both of these double-56Kbs class radio links were reliable and totally satisfactory for both my reaching our Unix oldcolo.com server on our office LAN, and for going directly out on the net through the router to the Internet itself, for newsgroup, web, telnet, and file transfer (ftp) purposes. In fact, subjectively, the increased speed over my 28.8 modem connection from the same location on a laptop/desktop Windows 486, became quite satisfactory. I don't recall sitting there frustrated in the time it takes for a web site to display. Most accesses passed the '7 second' test, which appears, ergonomically the longest it takes a computer to do something, before the user becomes impatient.
However, as soon as we installed the BreezeCOM radios over the same link, the speed increases dramatically, One does NOT want to go back to the slower speeds.
The BreezeCOM units (and AP-10 - access point, and WB-10 - bridge) are rated at 3.2Mbps up to 7 miles, with 24Dbi-gain antennas. Operating in the 2.4Ghz range. Thats fast. While we will make a seperate report on our experiences with BreezeCom products (we are acquiring two more of their newest Pro series, and upgrading the ones we have to be compatible, so we can set up a relay and multiple radio net), suffice it to say we were pretty impressed with the performance to date. Its nowhere near 3Mbps, but there is a reason.
On top of our 3 story office building (2502 West Colorado Avenue) we have installed an antenna mast, which can handle omni, yagi, and corner-reflector antennas for our tests. The 2.4Ghz corner reflector on the roof is affixed to a mast in such a way that the whole mast must be rotated to change the 'fan' (about 14 degrees) covered by the antenna. It is pointed in the direction of a downtown POP, to which we want eventually to be connected, wirelessly, 2.4 miles away. The 2.4Mhz range is more rigerous in its requirements for radios to be inside that 'fan' to get maximum performance.
In fact, Dewayne Hendricks rigged a portable radio, using a 5 volt battery, the small BreezeCOM radios (only 5 inches long), and a hand held omni antenna to run some tests by walking about in the general direction the antenna was aimed, after the base station 'Bridge' radio was set up and operating through the antenna. The radio only operates at 100 milliwatts, .1 of a watt, not at 1 watt which the FCC rules permit. So it is not expected to have the range of higher power radios. And it didn't. Using the status lights on the radio as an indicator, he was able to determine the radio with the directional antenna was *very* directional, both horizontally and vertically. When, at a distance of about 1/4 mile, he walked toward my house from the center of the directional antenna's line east down Colorado Avenue, he quickly lost link when he went beyond the 14 degree arc of the antenna.
In order to regain link, it was necessary to install a corner reflector directional antenna on my house roof, which is 1/4 of a mile in distance and about 25 degrees off the line of the fixed directional antennas axis, but clear rooftop line of sight to the mast at the base location, and aim it at the base corner reflector.
That arrangement reestablished link, but both by the evidence of the status lights on the Bridge unit (WB-10) which has a low, medium, and high link-quality green light led, and by thruput testing, the unit is, and has been operating right at about 1Mbps ever since. FTP transfers from the OS2 system to several distant systems (oldcolo.com, a Linux Unix server, and wireless.oldcolo.com, an NT server) both on a 10Mbps LAN in the office confirmed that speed. And that speed, from a radio link with units costing in the $1,500 range, is quite satisfactory - and addictive - from an end user standpoint. And quite suprising in speed for incoming web-site access traffic to http://184.108.40.206 (the OS2) web server site.
We now expect to set several BreezeCOM radios up, and see how they perform under varying practical uses of urban link range, with high (24dbi) gain directional, omni at 1 mile range, and directly into the Cisco routers of Colorado Supernet, totally bypassing any US West local loop dedicated lines. But that is for a later report.
Wi-LAN AND HOPPER
Wi-LAN of Calgary, Canada, has a family of radios called 'Hoppers' that have been installed in several educational settings, both in Canada and the US. Seperately I have reported on their use at the Fisher Model School in Richmond, Virginia (the 'Dual-Wireless' report.
However we have a rural school situation in mind, where a district has two buildings, split by a highway even though only about 500 linear feet apart, both running seperate Apple Talk, entirely MacIntosh based LANS (Ellicott School District 20 miles east of Colorado Springs, on the plains). These radios, operating at T-1 speeds (1.544Mbps) may be a solution to that school's problem, now of just linking both buildings together into a single LAN, and making way for a third, closeby building to be added later - since the Hopper radios are designed for point to multi-point.
In preparation, however, we have been exercising a pair of these 'Hopper Plus' direct sequence radios, operating in the 902 to 928Mhz Part 15 ranges, as part of our office 'LAN' network. In fact the wireless.oldcolo.com NT server, which is the NSF Project Web Site system, is now linked to the outside world by a pair of these radios.
Over the past three weeks these Hopper Plus radios have been operating flawlessly, and no one has reported any difference in performance through these radios, reaching the Web site (which is very heavily accessed) over the wired ethernet LAN connections used previously. We will soon deploy them elsewhere, however, and test them under longer range conditions.
It took less than 30 minutes to set up these radios upon the first visit by Trevor Pullishy of Wi-LAN to our premises. But when first turned on, they were interferred with, and interferred to the point of unusability of a very close by (15 feet) FreeWave radio operating in the same frequency band of 902-928Mhz. Since the FreeWave's were not in use, we turned them off and all went well. But it underscored the fact that interference between like-band radios is a reality, and anyone setting up a mix of radios from different manufacturers, has to take that into account, in location, power, and, where possible (as it is with the Hopper Plus's) frequency band selection.
AT&T WAVELAN CARDS
As of this writing we have not been able to fully exercise them, but we purchased - at approximately $750 each - a pair of AT&T (Lucent now) WaveLAN 2Mbps, 2.4Ghz wireless LAN cards. We have only gotton as far as installing them in a Workplace for Windows Pentium, and an OS2 486 machine, and gotton, in our workspace, a complete link between the two machines. These are designed specifically for in-building wireless LAN use, and have antenna connectors so made that it would be difficult to extend their range by connecting them to higher gain, and outside, antennas.
However, at the price, and speed, they obviously can have a place in an educational institutions workstation environments where, for one reason or another, wired LANS are a problem (asbestos in walls of schools, short distance external buildings, such as the common menufactured building-as-classrooms on many a school grounds.)
We have provided a seperate pair of these WaveLan cards to the Monte Vista School District, to try out in a setup where the gateway to the Internet is in the main District Headquarter's Building, while a High School is only 100 feet away on the same lots. The WaveLAN is put into a 'mobile' classroom PC, that can be wheeled into various classrooms, while maintaining a connection to the Internet at, presumably, T-1 speed, back in the fixed gateway location in the nearby building.
The Monte Vista techs have gotton good links between the radios, but they are having difficulty, at this writing, getting the software in the PCs talking to (thruputting) the cards. This is a configuration problem, not a radio problem. When they get less busy they expect to solve this problem, occassioned by unfamiliar cards-as-links experiences.
We have encountered several larger 'integration' companies, such as Edutech of Florida, and DSI of Denver, who have integrated the AT&T developed WaveLAN technology cards into their own server routers for installation in schools. In short, offering their own brand name devices to schools, in turn-key packages, but using the WaveLAN as a wireless engine for their products, and as part of their more comprehensive (wired LANs, servers, routers - and wireless) educational installations.
We will be reporting on these kinds of 'integration' companies later - which will be very important to many a school and university system that desires complete network installations by a company, rather than a 'do it yourself' addition of radios to their networks, while cheaper, requiring a steep learning curve by school or university tech staffs.