Radio amateurs like to experiment a lot, but have limited financial resources. So they are using (and sometimes inventing or developing) many different ways of doing far-reaching things over radio that fit the style (especially budget-wise) of missionaries and others who need affordable communication in remote areas. These experiments of course include various methods used to transport digital information on the air, that are used also by other types of user groups.
If you are planning a missionary style of data communications network and the local government (which has to permit any wireless operation in practically any part of the world) gives you the freedom, you have a choice of various frequency ranges and protocols. As with all choices of methods, there is no "best" one - it depends on what you are going to do with it.
I want to express special thanks to Pat Snyder of the JAARS radio department for a lot of update information he gave me from his huge first hand experience with digital radio communications for missionaries' email and automatic flight following (tracking an aircraft to know where to search for it in case of trouble when the pilot has to bring the down the plane safely and no time to talk on the radio).
"Shortwave" (or "High Frequency") means frequencies of 3-30 MHz, corresponding to wavelengths of 10-100 meters. The naming is from back in the 1920s or so, where this was in comparison to longwave / low frequency (frequencies of 30-300 kHz, wavelengths of 1000 to 10000 meters) and mediumwave / medium frequency (frequencies of 300-3000 kHz, wavelenthg of 100 to 1000 meters).
These frequencies can cover really long distances - at least hundreds of miles, even thousands if you select the right frequency range for the time of the day and conditions aren't too bad. I know of regular amateur data connections between Germany and Central Africa or South America.
Licensing may be difficult, because some governments don't like the perspective of international communication without direct control. Typical restrictions include daytime-only use of frequencies that might enable long range (international) communications at night.
Unfortunately, the far reaching propagation, combined with the scarceness of frequencies, together with atmospheric noise (if 10 watts of intended transmission can be heard halfway round the globe, so can a lightning storm ...) and very variable propagation conditions ("fading" can blank out the signal every few seconds even if it's as close as 10 miles) make shortwave a very noisy channel. Without special error-correcting protocols, a bit error rate of 10E-2 (one erroneous bit out of every hundred transmitted ones, compared to 10E-4 for a noisy analogue telephone line, let alone 10E-9 for a good fibre optic link) is considered a good connection. Errors tend to occur in bursts (fading holes, lightnings), so the chance of a group of bits ("packet") arriving correctly is not as bad as standard probability theory suggests, but it still makes you think that the (somewhat outdated) SITOR protocol (maritime teletyping) used to send just 3 characters (of 5 bits each, baudot telex code...) at 100 bits per second and then waited for confirmation.
Data speeds are limited to about 50 to 500 bits per second, depending on the protocol, but you have only half duplex (need to switch between transmit and receive mode) and lots of errors (unless conditions are very good). Expect a short mail message to take a minute or so of airtime.
Cost of radio equipment is relatively high. Expect 1500 US$ plus modem cost (200-500$) for anything capable of decent digital communication. You will need unobstructed space for an antenna wire of 20 to 100 meters long and at least 10 meters above ground. Terrain is not important, as propagation is by reflection at high (80-400 kilometers) atmospheric layers, the ionosphere. Antennas are mostly non-directional.
This refers to frequencies of 30 to 300 (VHF) to 3000 (UHF) MHz. Propagation is quasi optical, so radio ranges are limited to at most 100 miles, highly depending on terrain. Radio connections are mostly an "all or nothing" affair. Either you get clear contact or none at all. So once you have a link, it's clear and almost free of data errors. Frequencies are available plentyfully, so a repeater network may be an option to extend range.
Typically, licensing is easier than with shortwave, because many useful and non-suspicious enterprises (taxi drivers, construction companies, ...) need short range radio permission. I know a missionary, who, in the absence of telephone lines, used VHF police radio to communicate to the state capital, 100 kilometers away.
Radios are cheap, starting around 150 US$. Antennas are small (like TV antennas), but need to be placed as high as possible (you should be able to almost visually see the destination). Directional antennas are usual, but not necessary for short distances.
Brief overview about the OSI layers from Niagara University, Canada.
This is the amateur radio variant of X.25 (you might have heard this before), a well known protocol in (slightly older) public data networks (mainframe links for dumb terminals, banking machines and the like). As an OSI-layer 3 protocol it provides for end-to-end routing across repeaters (amateurs call them digipeaters), which is used in nationwide amateur networks. On top of AX.25, various higher level protocols can be carried, from simple BBS access (can anyone give a more precise definition of KISS and NetROM?) to full blown TCP/IP - the amateurs even got their own class A IP network (address 44.x.y.z, netmask 255.0.0.0), which unfortunately is not linked to the public internet for legal reasons (it must not be possible for non-amateurs to trigger amateur radio transmissions).
Technically, on OSI layer 1 for the slow speed variants (300 baud on HF, 1200 baud on VHF/UHF) the modulation is simple FSK (frequency keying) with 170 Hz (audio into an SSB transmitter, so it's effectively RF FSK) on shortwave and 850 Hz (audio into a standard narrow-band FM transmitter) on VHF/UHF. For 9600 baud VHF/UHF operation special modulation schemes are used, some of which include GMSK (narrow band FSK with specially formed non-binary pulses).
On OSI layer 2, the AX.25 protocol contains error checking and automatic retransmissions, but unfortunately no forward error correction, so it needs a pretty clear channel to avoid very frequent retransmits (a packet should get through after about 8 attempts, otherwise the connection is considered broken). This makes it not very efficient on shortwave. The standard speed of 300 bits per second (bps) is so fast (for shortwave standards) that even the smallest noise pulse is enough to kick around a few bits and thus render an entire packet useless. The classroom assumption of error free acknowledgements definitely does not hold here, even 64 byte packet size can be too much (5-10 retransmissions before a single packet gets through), so you can imagine what happens to TCP/IP with 40 bytes of header information per packet...
As bumpy as it's on shortwave, as great it's to use AX.25 on VHF/UHF. Standard amateur speed is 1200 bps with the upgrade to 9600 bps almost finished, so small file transfers (Email, small software applications), interactive BBS access and slow speed TCP/IP is very feasible (and actually done by amateurs and mission organisations like MAF). This means effectively stretching X.25 beyond the telephone lines.
Equipment is cheap. You can get a TNC (Terminal Node Controller, the equivalent of a modem, plugs into a serial port of any PC and into the microphone and speaker jack of the radio) for as little as 300 US$. If you a taxi (cab) company that recently got new radios, you might get Together with a simple VHF radio (former taxi equipment)
In contrast to Packet Radio, PACTOR is a protocol designed especially for noisy shortwave links. It (currently) comes in two levels: PACTOR (might be called PACTOR-I) and PACTOR-II.
PACTOR is (as far as I understood it, corrections are welcome) basically a combination of the radio teletype protocols (*TOR) and packet radio. The name could be expanded as PACketized Teletype Over Radio.
On OSI layer 1, PACTOR-I uses binary FSK modulation with 100 or 200 baud, auto-selected depending on channel conditions, as determined by the number of retransmission requests.
OSI layer 2 of PACTOR-I includes error checking (CRC) and automatic retransmissions (ARQ), but also forward error correction (FEC), basically an extended checksum so that a limited number of errors can be corrected without retransmission request. These FEC algorithms usually work best, if errors are evenly distributed. But on shortwave, errors occur in bursts, so PACTOR does interleaving - the data is divided in subpackets (2 or 4 bytes, I don't remember exactly), the the first bit of every subpacket is transmitted together, then the second bits and so on. So an error burst will flip only one or two bits within every subpacket and that can be corrected by the FEC algorithm.
There is no OSI layer 3 within PACTOR, but due to the error correction requirements, shortwave is point-to-point anyway. Once the link level (OSI layer 2) is reasonably reliable, end-to-end routing (layer 3) can be done by computers, that don't need to know about the oddities of shortwave (exept perhaps for the packet size...)
PACTOR-I is good for shortwave and in widespread amateur use as a teletype replacement, e.g. for BBS access - but if you want something really great, you should take a serious look at PACTOR-II.
As the name suggests, PACTOR-II is PACTOR-I plus something added. This something are additions on OSI layers 1 (to increase performance on good links) and interaction between layers 1 and 2 (to enable connections at all on marginal links).
On layer 1, a more sophisticated modulation technique, Phase Shift Keying,
is empolyed. If link quality permits, you can get up to 800 bps data rate,
which an impressive thing to have on shortwave. Of cause the bit rates are
adapted automatically to link conditions.
If the frequency of the shortwave gear drifts (not an uncommon event given the varying temperatures in tropical environments and the varying supply voltages of rural power lines or car batteries, combined with cheap equipment), PACTOR-II automatically adjusts the audio frequency. Thus you get near automatic operation with affordable crystal controlled radios. Adjusting audio level is very important. Linearity and stability are more important that output power (quoted from MORE ON COMPARISONS OF HF DIGITAL MODES).
The most impressive improvement of PACTOR-II comes by intermixing OSI layer 1 and 2. This is not considered mathematically clean style, but shortwave channels aren't mathematically clean either. The trick is, that if retransmission (a layer 2 protocol event) occurs (either by explicitely requesting them or by not receiving a positive acknowledgement) the same data is transmitted again (and again and again ...). The receiving PACTOR-II modem knows this and sums up all the repetitions in an (digitally approximated) analogue fashion (a layer 1 technique). After receiving enough copies of the same noisy data, eventually the noise averages out and the data becomes clearly readable (aided by forward error correction).
You must have seen (or heard, respectively overheard) the results to believe how great it is. Pat Snyder says that the JAARS flight following system using PACTOR-II can "pull out data that is 10 to 15 db below the noise level". This means that signals virtually inaudible for the human ear are good enough to make a data connection. When visiting my aunt (DL9YL) in Hamburg, one early afternoon (central european time) I accidentally hit the memory recall button of a canadian amateur station's frequency (20 meter band). In the speaker my ears noticed nothing but the usual no-activity noise, so I thought nothing was going on, but the PACTOR-II display showed a data connection to Canada. On the same evening, when conditions were better, data throughput reached 300 bps.
You might object, that it's nothing special to do that on the Internet - but the Internet uses a multi billion dollar infrastructure, while shortwave uses just a few thousand miles of God's creation ...
The only modems I know of that implement the full PACTOR-II protocol are the famous PTC-2 radio modems, developed and produced by Special Communication Systems GmbH (Ltd.), located in Hanau, about 20 miles east of Frankfurt in the state of Hessen, central Germany, about 90 minutes by car or train from where I live. Prices are around 800 US$ per piece
Very fast on good links, but not as fast as PACTOR-II on bad links.
Equipment is only available from HAL Communications and it's representatives.
Not just a protocol, but also a series of radio modems (modem and radio combined in one box) made by Codan Pty Ltd from Autralia. Good but expensive.
50 baud (amateur 45,45) binary FSK, 5 bit Baudot code, The only thing available in pre-computer days. Impressively noisy mechanics, impressively high error rates. A good experience to see the characteristics of the bare bone shortwave channel.
For standard PCs with sound cards there had been even a software modem available (RITTY).
AMateur Teletype Over Radio. SITOR is the maritime equivalent of this.
Basically 100 baud RTTY plus automatic retransmission. Packet size: 3 characters (15 bits).
A continuation of AMTOR with some kind of packet structure.
If you can get the local government to permit you bypassing their non-existent telephone network, the cost is moderate:
Then you just need the person to manage all that...
Home | Christian Missions | Christustreff Marburg | Pictures of Marburg | Job | Remote Communications | Linux OS | Psion page | Contact