Connectivity Providers Database

N.B. This paper was written by Dewayne Hendricks in June 1997 for publication in MicroTimes. Mongolian Wireless Field Test Expedition Last year we launched an expedition to Mongolia with the purpose to integrate a series of data radios into a wireless network in Ulaan Baatar, Mongolia and to field test them where severe weather conditions prevail, where there is a very poor wired telecommunications infrastructure, where advanced telecommunications technology expertise is limited (although there is considerable local computer expertise) and where US access to Mongolian scientific and research facilities is highly constrained by lack of normal Internet connections. The purpose was to build on and apply knowledge being gained from the National Science Foundation (NSF) funded "Wireless Field Test (WFT) Project for Education" run by David Hughes of Old Colorado City Communications, in Colorado Springs, Colorado. Why we needed to do this The basic problem which this expedition was attempting to solve is the lack of a suitable wired telecommunications infrastructure in Mongolia, including across its capital city. Typical of thousands of underdeveloped world cities. By 'suitable' is meant public or private circuits capable of carrying data with reliability and bandwidths necessary for scientific research between institutions increasingly equipped with advanced in-house computer and even local area networks. US scientific institutions, researchers and students cannot reach by data links, the primary institutions in Mongolia, such as the Mongolian Technical University, the Mongolian Academy of Sciences and the many other educational facilities which reside in the capital city. A poor and obsolete Postal and Telephone and Telegraph government voice telephone system, a holdover from when the Soviet Union controlled Mongolia, is all that exists. Efforts to use even late model error-correcting phone datamodems to link the computers in the various institutions is only partially successful, since the non-digital circuits are so old. It is difficult to maintain even a 14.4 Kbps connection without the line dropping many times in an hour. The PTT there has neither the resources nor the expertise to provide even the lowest speed (56 Kbps) wired circuits to link the numerous institutions to the only outside Internet data link which is the recent PamAmSat satellite downlink over which a 128 Kbps Internet connection provided by Sprintlink and funded by a previous NSF effort. As a result of the first step to provide Internet access to Mongolia, NSF funded the establishment of a costly ($72K/yr) satellite Internet link. This link only reached one set of computers in a single building by direct wiring, in the central part of the city. The 'last mile' problem of connecting up this satellite link to the institutions (which range in actuality from one to ten kilometers from the satellite ground station) which must then distribute the signal to their own wired intranets, had so far defeated efforts to link it by local telephone facilities. In this respect, Mongolia is typical of scores of underdeveloped nations with a poor telephone infrasturcture, too limited to adapt to modern requirements for institutional datalinks to and from other nations, especially the United States. Wireless datalinks provide one possible general solution to this problem, not only in Mongolia, but in many other countries. There were a number of challenges, which made this at once a difficult project, as well as one from which valuable lessons and techniques were learned for application elsewhere. 1. The weather is more severe in Ulaan Baatar than is areas which the our wireless field test project have been implemented. High winds, and average temperatures from -15 to -25 degree Centigrade for entire winter months will subject the radios, connectors, shielding materials, and antennas to extremes of cold and wind. The question was whether or not economical commercial-grade radios and systems could handle the weather and give similar MTBF times to those systems operated in milder climates. 2. Whether the electromagnetic environment from the equipment that exists in Mongolia today, some of which will result from systems developed and deployed by the Soviet Union in the past, in a different technological regulatory environment, will affect the spread spectrum dataradios that we intended to deploy, which were developed in the US and operate in environments controlled by FCC rules. 3. Could the wireless network be integrated into networks made up of a mish-mash of US and foreign computer and peripheral equipment and would interoperability be a problem? 4. Given the level of in-country expertise able to help install, and maintain the wireless network, could they be trained sufficiently in a short period of time to handle the network in the future, with only the assistance of remote expertise to call upon for help. Before we got there The Mongolian government had permitted a group of Russian trained Mongolian engineers to form a private technology company, DataCom Co., Ltd, headed by Dr. Dangaasuren Enkhbat, using facilities at a technical center once occupied by Soviet engineers. It is this company, with limited liquid assets, but in a suitable central facility in the capital, which had been turned to by NSF to provide the critical organizational link between the satellite-based Internet feed , and in-country institutions. In return for being able to eventually provide Internet service to private companies in Mongolia, and being given control of the satellite groundstation, DataCom agreed to provide links to Mongolian public scientific, educational, government, and library institutions for at least two years, at no cost. With initial funding from various sources, and with help from groups such as Sprint, PanAmSat, Comstream, NSF and the US Ambassador to Mongolia at the time, Donald C. Johnson, DataCom! became the first Internet service provider in Mongolia in December of 1995. The top level domain name for Mongolia is MN. DataCom maintains a website at <http://www.magicnet.mn/>. Once the project was approved by NSF, our first task was to perform an analysis in order to determine what would be required for the expedition and what would have to be done once we arrived in Ulaan Baatar in order to deploy the wireless network. We made use of maps, videos and communications with the staff of DataCom, in order to determine exactly what types of radios would work in their environment and with what configurations and physical distributions of sites would be required for distribution of Internet services from the DataCom location. The decision had been made to install a basic wireless MAN that would consist of eight sites. One being the DataCom site and the other being seven sites which were selected by DataCom and were all universities with the exception the US Embassy. Based upon our experience in the WFT project, we had made a preliminary decision to use unlicensed Part 15 dataradios developed and marketed in the US by a small Boulder, Colorado company called FreeWave Technologies, Inc. (<http://www.freewave.com/>). They make a spread spectrum radio that can send data at 170 Kbps over the air and interfaces to a computer via a serial data interface (RS-232) at 115.2 Kbps. The radio operates in the 902-928 MHz part of the spectrum that is one of three bands utilized by Part 15 devices in the US. These radios put out 1 watt of power and using various antenna configurations we have been able to obtain send data with them up to distances of 60 miles. For more information on the use of these sorts of unlicensed spread spectrum radios, check out our WFT project website at <http://wireless.oldcolo.com/>. Our analysis indicated that the 902-928 MHz band was not completely available for our use in Mongolia, since the commercial cellular telephone network there used GSM technology which operated in a portion of the band from 902-915 MHz. So in order to make use of the FreeWave radios in Mongolia, we had to make two things happen. We first got DataCom to obtain from the Mongolian government permission to operate the radios in the 915-928 MHz part of the spectrum. This would avoid the conflict with the GSM cellular network. We next got FreeWave to provide us with a special version of their radio which would operate in this smaller portion of the 902 MHz band. As it turns out, this was not a difficult problem to solve as they had already encountered the same problem when they first entered the Australian market which uses GSM cellular in the same band. In order to connect the radio that was to be deployed at each site to the local LAN we determined that we needed a low-cost Internet router that could be installed and maintained at each location. For the WFT project, we had already developed such a router based upon the widely available Linux operating system and a low-cost PC. We decided to use this router in Mongolian as we already had a great deal of experience with this router and an attached FreeWave radio. In addition to Linux, all a PC needed to function as a special purpose IP router was the addition of a high-speed serial card that would support an interface speed of 115.2 Kbps and enough RAM to allow Linux to function properly (in this case we used 16MB). The next equipment item of importance were the antennas and feedlines that we were going to use for the effort. It was clear that we needed an omni-directional antenna for the DataCom site since it was the point-of-presence (POP) for all Internet services and each remote site would have to connect to it. For the remote sites, we would require a directional antenna that could focus all of the available energy from the radio on the POP. Although the distances involved in the Ulaan Baatar network would be a good deal less than we had been used to in the US (average about 20 miles), we decided to use the same high gain omni and directional antennas that had used here rather then taking the time to find lower gain alternatives. The idea here being that its better that have 'more' then to get there with 'less' and find out that you need 'more'. The 'more' here being antenna gain. We also decided to use the same low loss coax feedline (LMR-400 for those of you knowledgeable about such things) to connect the radio to the antenna that we had been using in the WFT project. Even with all of the up front analysis that we had done with respect to the requirements at each site, we still had no idea as the exact amount of feedline that we would require. As a result, we decided to take three times the amount that we thought that we would require. The final equipment related issue that we had to deal with was electrical power. In Mongolia, they use 220 VAC and we had to obtain the necessary power adapters and connects to insure that the radios and routers could all operate on local AC power. As it common practice with an expedition of this sort to a remote part of the world, we made sure that we had spares for all of the key equipment. This meant that we were going to ship a number of spare routers and radios. When it came to just how we were going to get all of the equipment to Mongolia, the NSF intervened and made arrangements with the US Ambassador in Mongolia that allowed us to ship everything via diplomatic channels (the famous 'diplomatic pouch'). This had the advantage that we were able to avoid the normal problems encountered using commercial shipping firms and dealing with customs procedures. It turns out that spread spectrum radios cannot be exported to countries such as Mongolia and that export required a special export permit from the US Government. Obtaining such a permit can take anywhere from three to six months. As we were trying to get the expedition completed before the harsh Mongolian winter started (December thru March) we were fortunate in being able to bypass that requirement by shipping the radios the way that we did. The downside in using the diplomatic channel for shipping is that once you put your shipment into the system, you only have a rough idea as to just when it is going to arrive at its destination. Thus we had to use estimates on the time it took other items to arrive in Mongolia as the basis for determining just when the expedition team should arrive in Mongolia. As it turned out, we were lucky in that most of the equipment arrived before we arrived in country, with the exception of a few pieces that arrived after we had been in country about a week. After we got there We left for Mongolia at the October of last year. The expedition team consisted of Glenn Tenney of Fantasia Systems of San Mateo, CA and myself. Its fairly easy to get to Mongolia these days. All it took were direct flights from San Francisco to Tokyo then to Beijing and finally to Ulaan Baatar. All in all, about three days total travel time. The only prerequisite is making sure that you have all of the necessary visas. The one for China is easy to obtain, but the one for Mongolia is a bit harder as you have to first have an invitation letter from a person or company in Mongolia to obtain a visa to visit their country. After we had gotten in country and had made the initial visit to DataCom and made an inventory of our equipment that had arrived (except for the few missing items), the first order of business was to make a survey of all of the chosen sites in order to determine just what would be needed to deploy the equipment. We were shocked to find that DataCom was surrounded by high-rise buildings that effectively blocked line-of-sight to all of the sites where installations were to have taken place. Its important for you to understand that one of the key factors in getting this type of low-power radio to work properly is to have a clear light-of-sight path to any location that you wish to contact. Without major obstructions, it is possible to establish communications over fairly long distances, with obstructions it is usually not possible to go more than a few hundred feet. Given what we saw from the DataCom site, it looked like it was time to wrap things up and get on the way back home. However, we decided to make the best of a bad situation and make an attempt to proceed anyway and try to see just what could be accomplished. First, we installed the first radio and router at the DataCom site. In order to maximize the power output to the antenna from the radio, we mounted the radio on the roof of the building just a few feet from the antenna in a small equipment shed. This approach allowed us to use just a short coax feedline to the antenna and then a rather long serial data cable to the router which was on the second floor of a four story building. Since this location was the main hub for the wireless network, we used an omnidirectional antenna. We next setup a mobile radio so that it could be powered by batteries and carried about in a car with one of the directional antennas. We decided that we wanted to travel around town and see if it was possible to receive the signal at any distance from the POP location. To our surprise, we were able to get a good signal at most of the sites where we were to have installed a radio. All of this without having line-of-site to the POP. After some analysis, we were able to determine that even though the town of Ulaan Baatar is populated with a large number of high-rise buildings, the materials that are used in their construction are such that they appear to be effectively transparent at the frequency of the radio emissions that we were using. This windfall allowed us to proceed with installation at each site as planned. The most difficult part of the installation is the proper placement of the antenna and its connection to the radio. The major goal is to minimize the length of feedline used to connect the radio to the antenna. The basic rule being that the longer the feedline, the less power actually gets to the antenna and hence the lower the quality of the signal at the remote location. This is where we spent most of our time during the next two and one-half weeks that we were in country. Each site had to be surveyed and then we had to work with the local people in charge of that site in order to convey to them what needed to be done and make sure that the proper permissions were obtained to do the work. At some sites, nothing major was required as the antenna and the radio/router ended up being in the same room. In other sites, such as the US Embassy, the antenna had to be mounted on the roof and feedline run several floors to get to the radio and the router. Once the site preperation was complete, it was a very simple matter then to install the radio and router. One major problem that developed once we were over the site preperation issues and we actually into testing the radios, was that we found the PC's we selected were unable to have their serial ports run at speeds over 19.2 Kbps without there being data overruns. What this meant was that as soon as we tried to send data to the radios at any speed over 19.2 Kbps, then nothing worked. After a few days spent in investigating the cause of the problem, we discovered that even though the PC's that we using technical specifications indicated that the on-board serial hardware could be run at speeds up to 115.2 Kbps, this was not the case. DataCom has some new Pentium PC's that arrived while we were there that were to be used for a public Internet access center When we installed Linux on those machines and configured it as a router and set the serial interface to the 115.2 Kbps speed, everything functioned properly. As luck would have it, there was someone coming over from the US to Mongolia who we were able to contact and have them bring the necessary number of high-speed serial cards over so that we were able to install them in the PC's that we had and get them to function properly at the higher datarate. This delay however meant that our stay had to be extended from the original planned two weeks for an additional half week. As a result, we were finally able to leave Mongolia with the wireless network installed and good connections to the seven remote sites. What we learned The network that we installed is still up and running today. The staff at DataCom has been able to maintain the network and even extend it with additional sites. Here is a summary of some of the key things that we learned as a result of this effort: o A good site survey ahead of installation is the key to that installation being a success. The specifics of a survey require people with a good deal of expertise in several areas. If you can't get those people to the site in person ahead of time to perform the survey, then the success of the installation can be in serious doubt when you're doing it in such remote locations as Mongolia. o Always make sure that you test the equipment that you're using before you ship it to a remote location. Never believe the specifications for a product until you've tested it yourself!!! o Radio is still something like a Black Art (aka Magic). There is always something new to be learned. Until we had gone to Mongolia and did this installation, we would have never believed that you could have deployed such a wireless network as we did in a dense urban area without having line-of-site to all locations. o Low-cost spread spectrum radio products that are being used in increasing numbers in the US to deliver access to the Internet, can be successfully deployed in developing countries in limited expertise in both radio and Internet technologies. In closing, we'd like to take this opportunity to acknowledge the help and assistance of Steve Goldstein of the NSF, without whose caring and wisdom this effort would not have been possible.