Packet radio has its roots in the 1960s in research conducted by organizations such as the Rand Corporation and DARPA. The best known project of that era was ALOHANET, a research network which went into operation in 1970, linking computers in different parts of the Hawaiian Islands by radio links. This work eventually attracted the interest of amateur radio operators.
In 1978, regulatory changes in Canada paved the way for amateur operators to begin experimenting with packet radio techniques, leading to the first design for a TNC (Terminal Node Controller: a device for interconnecting a computer or data terminal to a radio) the following year. Interest soon spread to the US; and, in the early 80s, several new TNC designs appeared, accompanied by work on standardizing packet radio protocols. The boom in amateur packet radio began in earnest in 1985 when a group called Tucson Amateur Packet Radio (TAPR) introduced a new TNC called the TNC-2 and made it available as a kit. The design was also licensed to several manufacturers, and TAPR has gone on to become the pre-eminent international amateur packet radio organization.
The TNC-2, and other designs which followed includes a built-in 1200 bps modem. Although slow by computer networking standards, this bit rate was chosen because inexpensive modem chips were available, and because the modem can be interfaced to virtually any voice radio transceiver via a simple hookup to the microphone and speaker connectors. Transmitting at higher bit rates using radios which were designed for voice transmission is much less straight forward. Thus, while experimentation began with bit rates up to 9600 bps, 1200 bps operation predominated in packet radio, and this remains true even today. Many packet radio users are primarily interested in applications such as email and keyboard-to-keyboard chat, and 1200 bps generally does an adequate job of supporting those. Others, however, were interested in large file transfers, digital voice and similar bandwidth-intensive applications and began looking for ways to push the packet radio envelope.
This means that a 1200 bps audio modem can easily be connected to a voice radio via external connectors on the radio. For higher it rates, you must go in and make internal connections and possibly other modifications (such as changing the modulator from phase modulation to true FM). Most voice radios work poorly at 9600 bps due to slow turn around time and distortion from their internal filters, and operation at greater than 9600 bps is out of the question.
In 1987, an amateur radio operator in Georgia (USA) named Dale Heatherington announced that he had designed a packet radio modem that would run at 56 kbps and was making parts kits available. A few of us in our group of experimenters in Ottawa acquired kits and, by mid-1988, had our first 56 kbps link running between my home and another about 5 km away. We used the KA9Q TCP/IP software in our XT-class PCs; and, within minutes after establishing the link, we had FTP file transfers running while we chatted. A third station appeared within a few weeks, and we had a network! However, the limits of this radio-based topology quickly became apparent: I could contact either station by aiming my antenna (which was equipped with a rotator) in the appropriate direction, but I could not communicate with both at once.
Our next project was, therefore, to establish a central network node in a high location, a tall building on one of the local college campuses. This was the beginning of our packet radio MAN (Metropolitan Area Network), with a "star" topology in which all of the 56 kbps stations were set up with fixed antennas aimed toward the hub site. This provided connectivity between all of the stations, but there remained a problem with contention for the radio channel. All stations could hear the hub but not necessarily the other stations in the MAN. Access to the radio channel was managed by CSMA (Carrier Sense Multiple Access), but this system breaks down when stations cannot detect each other. They may begin transmitting at the same time causing a "collision" of their respective packets; this usually results in both packets being lost, requiring them to be retransmitted. This can drastically reduce the throughput and general effectiveness of the MAN.
We addressed this problem by using two different radio channels and constructing a real-time repeater, located at the hub site, which retransmitted all of the transmissions from the network stations. All stations transmit on the repeater input frequency and receive on the repeater output frequency, and each station can now effectively hear all of the others, making CSMA perform the way that it should. The Ottawa MAN repeater went on the air in January 1990, the first of its kind in the world, and it continues to perform well. It has since been supplemented with repeaters for 1200 bps and 9600 bps and, in 1991, became connected to the Internet. Most of the 56 kbps stations also act as gateways to other packet networks. The coverage of a 56 kbps MAN such as the one deployed in Ottawa is highly dependent on the topography of the surrounding countryside. In our case, it covers a radius of about 50 km, but the distance would be considerably less in hilly areas.
What does it take to put a 56 kbps packet radio station together? Until recently, this was strictly the domain of the experimenter, largely due to the effort involved in constructing and adjusting the modem. However, this impediment has now been removed as a wired and tested version of the modem is now available from a commercial source. This new modem is also capable of functioning as a repeater. In addition to the modem, the basic elements of the 56 kbps station are an antenna, a transverter and a modem-to-PC interface. The antennas are no problem to find commercially. Unlike most lower-speed modems, a 56 kbps modem is an "RF modem", which means that its input and output are directly at some radio frequency. The transverter is a device which translates this frequency to another one which is more appropriate for packet radio transmission (most likely in the 440 MHz band). There are several sources for these units. Lastly, there is the interface. When we started with 56 kbps, we used TNC-2s as modem-PC interfaces, but we found that they performed poorly at this rate -- the effective throughput was only 19.2 kbps. In 1991, the Ottawa group designed a high-speed modem interface card for the PC which overcame this problem, and we continue to sell the card as a means of raising funds for our activities in packet radio. The card has drivers for the KA9Q software under DOS or Windows and for the native TCP/IP stacks in Linux and OS/2. There are other suitable interface cards available as well.
As a rough estimate of the cost of a 56 kbps station, the modem and transverter are each about $350 US and the interface card $125 US. For the antenna, a simple whip antenna may suffice in some cases, but a directional antenna such as a yagi is preferable. Adding the antenna and the coaxial cable to connect it to the radio equipment will bring the total cost to about $900 US.
The packet radio systems described above are based upon "narrowband" technology. Similar technology is used in the commercial packet radio systems such as CDPD and ARDIS. One thing nearly all of these systems have in common is that data rates seldom exceed 56 kbps. There are, in fact, practical limits to how fast you can go with narrowband modulation, particularly in multiple-access MAN situations. The limits are imposed by distortion of the radio signal by multipath propagation (the radio signal travels from transmitter to receiver, not just by the most direct path, but also by one or more reflected paths). Commercial radio relay systems avoid multipath by using microwave frequencies, highly directional antennas and clear line-of-sight paths; but this is not practical for low-cost packet radio networks.
A more affordable alternative to achieving higher data rates in packet radio has appeared in recent years in the form of wideband "spread spectrum" systems. Without delving into technicalities, suffice it to say that these systems overcome multipath distortion by virtue of their wide bandwidth, which provides "frequency diversity".
In the context of packet radio, the spread spectrum systems which are of most interest are the so-called "wireless local area networks" or WLAN products. As in the narrowband packet radio systems described above, a WLAN unit consists of a PC interface, modem, radio hardware and antenna. In this case, however, these functions tend to be more tightly integrated.
The most extreme example of this is a WLAN product from Xircom in which the entire unit, including antenna, is on a PCMCIA card. This product provides very short range transmission and is intended for use within buildings, but there are other WLAN products which can be used with directional antennas to span considerable distances (a few km, or more if good line-of-sight paths exist). There is also a wide range of data rates available, with the high end units providing rates of 2 Mbps or more. Prices also vary greatly from about $400 for the Xircom unit to $5000 US for a system intended for long-range wireless bridge applications. However, there are a variety of interesting products in the $700 to $800 US range which offer 1 to 2 Mbps data rate. Most of these are ISA bus or PCMCIA plug-in cards and come with software drivers that make them appear like ethernet cards to network software.
A further attractive aspect ofhe WLAN products is that, in many countries, they (or at least some subset of them) can be used without acquiring a radio license. In order to qualify for this exemption, they must meet certain requirements with regard to power output, antenna gain, etc. This has its downside, however -- there will be greater potential for interference from other users of the same radio spectrum than is the case with the licensed narrowband systems since the license carries with it some degree of exclusivity in the frequency allocation it covers. Amateur radio operators are experimenting with the spread spectrum techniques used in the WLAN products, and this will likely eventually lead to higher-speed packet radio systems in the amateur bands with fewer restrictions than is the case with the unlicensed systems. This is where the future of high-speed packet radio lies.
Packet radio provides an interesting alternative to conventional telecom services in the delivery of distance education. It can bypass expensive tariffs in wireline data transmission and has the potential to break the bandwidth barrier imposed by POTS (Plain Old Telephone Service). Both narrowband and wideband packet radio technologies are available. Narrowband, whether deployed in the amateur radio bands or in commercial bands, requires that radio licenses be obtained (there are a few exceptions for very low-power systems). Wideband can be operated without licenses in many instances and offers higher data rates, but there is also greater risk of interference problems. The wideband WLAN systems really shine in applications involving high mobility and relatively short distances such as wireless notebook computers whose users roam within a building. They also are widely used as wireless bridges, for example, to link the LANs in widely separated buildings on a campus. Their use in MAN situations, such as described above in the context of 56 kbps narrowband packet radio, is less proven. This is an area that is ripe for research and is currently being pursued both in the academic and amateur radio communities.
The major impediment to the use of packet radio technology in applications such as distance education is the knowledge and effort required to establish the initial infrastructure. It is not as simple as connecting a modem to a telephone line and running some software (which many can attest is not always a trivial proposition!). The knowledge needed to properly site antennas, test radio link performance, install networking software and deal with the unexpected problems that inevitably crop up can only be gained through experience. Equally important is the need for education of the local users so that the wireless system can continue to be maintained and expanded. This is challenging, but worth the effort!
Barry McLarnon
VE3JF/VA3TCP
Ottawa Amateur Radio Packet Working Group
Email: bm@hydra.carleton.ca
Ottawa, Ontario, Canada
Return to GLOSAS News Contents for this issue.
URL: http://library.fortlewis.edu/~instruct/glosas/pacrad61.htm
March, 1996
GLOSAS NEWS was orinally posted to the WWW at URL: http://library.fortlewis.edu/~instruct/glosas/cont.htm by Tina Evans Greenwood, Library Instruction Coordinator, Fort Lewis College, Durango, Colorado 81301, e-mail: greenwood_t@fortlewis.edu, and last updated May 7, 1999. By her permission the whole Website has been archived here at the University of Tennessee server directory of GLOSAS Chair Dr. Takeshi Utsumi from July 10, 2000 by Steve McCarty in Japan.