1. Introduction
2. Technical aspects of Packet Radio
3. Media requirements for Internet Services
4. Cost extimation for Packet Radio
5. Conclusions
6. References

Introduction

Technical aspects of Packet Radio

What is Packet Radio?
Packet Radio is a particular digital mode of amateur radio communications which corresponds to computer telecommunications. The telephone modem is replaced by a terminal node controller (TNC); the telephone is replaced by an amateur radio transceiver, and the phone system is replaced by the radio waves. Packet Radio takes any data stream sent from a computer and sends that via radio to another amateur radio station similarly equipped. Packet Radio is so named because it sends the data in small bursts, or packets.

Data packet technology was developed in the mid-1960's and was put into practical application in the ARPANET (the mother of Internet) which was established in 1969. Initiated in 1970, the ALOHANET, based at the University of Hawaii, was the first large-scale packet radio project. Amateur packet radio began in Montreal, Canada in 1978, when the first TNC was developed by the Vancouver Amateur Digital Communication Group.

Amateur Packet Radio revolves around a protocol that has become known as AX.25. This reference is a derivative of a commercial protocol that is X.25, the prefixed A denotes amateur. So to get at the root of how Packet Radio has got to where it is today we need to look at X.25. This is a CCITT reference for a protocol that is used in global networks allowing multinationals such as Couriers and Airlines to build vast and robust networks.

The radioamateurs took the details specifications for this protocol and embarked upon the process of converting it for use by themselves for constructing an amateur network using the already present commercial system as a model. In the event the process of making it suitable for their purpose was more straight forward than they believed and with a few additions of their own that suited the new environment AX.25 is very much a close reflection of the commercial system from which it takes it name.

A number of basic principles apply to all networks and AX.25 is no different to any other. Before we examine them what you must appreciate is that in a commercial system a session to the other side of the globe can be established in a sub second time frame. As you might expect the amateur equivalent is a many fold slower system so the capabilities are much less but still quite powerful.

1. All 'devices' on a network must be uniquely identifiable.
   The obvious choice for amateur radio is the operators callsign. A commercial network will instantly reject a device if it tries to enter the network with a network address that already exists. This equates to a callsign being 'pirated' in the world of amateur radio and whilst the system cannot reject anyone as such confusion will reign and sessions between station will very quickly start to be lost.

2. All traffic must have a specific source and destination.
   Any data that enters the network must know where it is going and where it has come from in case the destination needs to send a reply back. This is one of the addition to AX.25 that dosent exist in X.25 that a station can send a transmission to a arbitrary destination address, this is referred to as a beacon and is sometimes used to satisfy license restriction on station identification.

3. The system must support error checking of the data it is carrying.
   Network that span great distance and have many potential routes can be prone to data being lost or corrupted. A system called a cyclic redundancy check is used to ensure safe delivery. This is a number that is computed by the sender and is included in the frame as it is sent, the receiver computes the data at the other end and if the same number is not produced that frame is reject and requested again.

4. Packet radio is suitable for large networking.
   Since amateurs use radios to transmit their data, their range of communications is limited to approximately line of sight. An average packet station talks in a radius of about 10-30 miles. Packet networks allow amateurs to widen the area of communications past their line of sight, by having a series of packet stations linked by radio, that can be used to get their packet messages to where ever the network goes. Much like the telephone system, networks provide long distance service outside the local area. There are a number of amateur networks which allow amateurs to travel from one area to another. These networks are typically built by a local or regional group that allows packet operators to get outside of their area.

5. The system must support dynamic routing.
   As mentioned above, global networks (much like the Internet) can develop into very complicated systems where there may be many routes between any two points and data between he two points may travel any one or more of these routes. Firstly the system needs to be able to construct the data at the far end into the form it was in when it was sent and also in the event of any failures the missing data needs to be request again. When a portion of the network does fail the system will automatically choice the 'next best' route to work around the hole and then carry on as normal. This will typically happen in the blink of an eye and with no knowledge or intervention of the users.

6. Packet Radio equipment is well standardized.
   The current TNC standard grew from a discussion in October of 1981 at a meeting of the Tucson Chapter of the IEEE Computer Society. A week later, six of the attendees gathered and discussed the feasibility of developing a TNC that would be available to amateurs at a modest cost. The Tucson Amateur Packet Radio Corporation (TAPR) formed from this project. The project progressed from these first prototype units to the TNC-1 and then finally to the TNC-2 which is now the basis for most packet operations worldwide.

7. Packet Radio can be implemented even in Satellite Communications.
   Many of the amateur radio satellites in orbit contain computer systems that provide packet capability. Most packet satellites provide BBS-like functions for messages to be passed to anywhere in the world within 24 hours. Several contain CCD cameras, which allow amateurs to download images of the earth and some allow users to retrieve data from the onboard experiments. Most satellites use AX.25 with special software developed for satellite communications. DOVE, Digital Orbit Voice Encoder, can be received with any normal VHF/FM 2-meter packet station, but most of the packet satellites use SSB and require more complex equipment in order to operate them.

(illustration of a typical station setup with a schematic diagram of a station wiring)

• TNC (terminal node controller):
  A TNC contains a modem, a computer processor (CPU), and the associated circuitry required to convert communications between your computer (RS-232) and the packet radio protocol in use. A TNC assembles a packet from data received from the computer, computes an error check (CRC) for the packet, modulates it into audio frequencies, and puts out appropriate signals to transmit the packet over the connected radio. It also reverses the process, translating the audio that the connected radio receives into a byte stream that is then sent to the computer.

  Most amateurs currently use 1200 bps (bits per second) for local VHF and UHF packet, and 300 bps for longer distance, lower bandwidth HF communication. Higher speeds are available for use in the VHF, UHF, and especially microwave region, but they often require special (not plug-and-play) hardware and drivers.

• Computer or terminal:
  This is the user interface. A computer running a terminal emulator program, a packet-specific program, or just a dumb terminal can be used. For computers, almost any phone modem communications program can be adapted for packet use, but there are also customized packet radio programs available. A dumb terminal, while possibly the cheapest option, does have several limitations. Most dumb terminals do not allow you to scroll backwards, store information, upload, or download files.

• radio:
  For 1200/2400 bps UHF/VHF packet, commonly available narrow band FM voice radios are used. For HF packet, 300 BPS data is used over single side band (SSB) modulation. For high speed packet (starting at 9600 bps), special radios or modified FM radios must be used. 1200 bps AFSK TNCs used on 2-meters (144-148Mhz) is the most commonly found packet radio.

The number of available channels in these frequencies is limited, but an important feature of Packet Radio is that many user can all use the same channel.
Packet Radio, unlike voice communications, can support multiple conversations on the same frequency at the same time. This does not mean that interference does not occur when two stations transmit at the same time. What 'same time' means in this sense is that multiple conversations are possible in a managed, time shared fashion. Conversations occur during the times when the other conversations are not using the channel. Packet radio uses the AX.25 protocol to accomplish this shared channel.

AX.25 specifies channel access (ability to transmit on the channel) to be handled by CSMA (Carrier Sense Multiple Access). If you need to transmit, your TNC monitors the channel to see if someone else is transmitting. If no one else is transmitting, then the TNC keys up the radio, and sends its packet. All the other stations hear the packet and do not transmit until you are done. Unfortunately, two stations could accidentally transmit at the same time. This is called a collision. If a collision occurs, neither TNC will receive a reply back from the last packet it sent. Each TNC will wait a random amount of time and then retransmit the packet. In actuality, a more complex scheme is used to determine when the TNC transmits.

AX.25 is considered the defacto standard protocol for amateur radio use and is even recognized by many countries as a legal operation mode. However, there are other standards. TCP/IP (the Internet protocol) is used in some areas for amateur radio. Also, some networking protocols use packet formats other than AX.25. Often, special packet radio protocols are encapsulated within AX.25 packet frames. This is done to insure compliance with regulations requiring packet radio transmissions to be in the form of AX.25.

During the early days of amateur packet radio, it became apparent that a packet network was needed. To this end, the following packet network schemes were created:

• Digipeaters:
  This one was the first implemented networking scheme for Packet Radio. Digipeaters would simply look at a packet, and if its call was in the digipeater field, would resend the packet. Digipeaters allow the extension of range of a transmitter by retransmitting any packets addressed to the digipeater. This scheme worked well when only a few people were on the radio channel. However, as packet became more popular, digipeaters soon were clogging up the airwaves with traffic being repeated over long distances. Also, if a packet got lost by one of the digipeaters, the originator station would have to retransmit the entire packet again, forcing even more congestion.

• KA-Nodes:
  Kantronics^TM improved on the digipeater slightly and created KA-Nodes. As with digipeaters, KA-Nodes simply repeat AX.25 frames. However, a KA-Node acknowledges every transmission at each link (node) instead of over the entire route. Therefore, instead of an end-to-end acknowledgment, KA-Nodes allow for more reliable connections with fewer timeouts, because acknowledgments are only carried on one link. KA-Nodes therefore are more reliable than digipeaters, but are not a true network. It is similar to having to wire your own telephone network to make a phone call.

• NET/ROM:
  NET/ROM was one of the first networking schemes to try to address the problems with digipeaters. A user connects to a NET/ROM station as if connecting to any other packet station. From there, he can issue commands to instruct the station to connect to another user locally or connect to another NET/ROM station. This connect, then connect again, means that to a user's TNC, you are connected to a local station only and its transmissions do not have to be digipeated over the entire network and risk losing packets. This local connection proved to be more reliable.
  NET/ROM doesn't use all of the AX.25 protocol. Instead, it uses special AX.25 packets called Unnumbered Information (UI) packets and then puts its own special protocol on top of AX.25. This is again used to increase efficiency of its transmissions. NET/ROM nodes, at regular intervals, transmit to other nodes their current list of known nodes. This is good because as new nodes come on-line, they are automatically integrated in the network. However, if band conditions such as ducting occur, ordinarily unreachable nodes can be entered into node lists. This causes the NET/ROM routing software to choose routes to distant nodes that are impossible. This problem requires users to develop a route to a distant node manually defining each hop instead of using the automatic routing feature.

• ROSE:
  ROSE is another networking protocol derived from X.25. Each ROSE node has a static list of the nodes it can reach. For a user to use a ROSE switch, he issues a connect with the destination station and in the digipeater field places the call of the local ROSE switch and the distant ROSE switch the destination station can hear. Other than that, the network is completely transparent to the user.
  ROSE's use of static routing tables ensures that ROSE nodes don't attempt to route packets through links that aren't reliably reachable, as NET/ROM nodes often do. However, ROSE suffers from the inability to automatically update its routing tables as new nodes come on-line. The operators must manually update the routing tables, which is why ROSE networks require more maintenance.

• TCP/IP:
  TCP/IP stands for Transmission Control Protocol/Internet Protocol. TCP/IP is commonly used over the Internet wired computer network. The TCP/IP suite contains different transmission facilities such as FTP (File Transfer Protocol), SMTP (Simple Mail Transport Protocol), Telnet (Remote terminal protocol), and HTTP (Hyper Text Transfer Protocol). The KA9Q NOS program (also called NET) is the most commonly used version of TCP/IP in packet radio. NOS originally was written for the PC compatible. However, NOS has been ported to many different computers such as the Macintosh, Unix, and others. TCP/IP based amateur networks are becoming more common each day.

• BBS Message Transfer:
  Many of the BBS (Bullettin Board System)programs used in packet radio allow for mail and bulletins to be transferred over the packet radio networks. The BBSs use a special forwarding protocol developed originally by Hank Oredsen, W0RLI. Besides full service BBSs, many TNC makers have developed Personal BBS software to allow full service BBSs to forward mail directly to the amateur's TNC. This allows operators to receive packet mail at night and avoid tying up the network during busy hours.

Media requirements for Internet Services

(da fare) Speed, bandwidth and related problems.
(da fare) Definitions of the terms.

Internet is based on the TCP/IP protocol and it can be implemented directly over a Packet Radio network (for example with Linux) as seen above (the so-called "incapsulation of TCP/IP into AX.25"), but sometimes this technique can be highly inefficient, expecially in low-speed situations. This happens because the strict timing requirements of TCP/IP that in conjunction with AX.25 CSMA mechanism will generate frequent retrasmissions of packet with longer and longer delays (backoff).
Protocols different from AX.25, like for example DUAL, has been developed to try to solve partially this problem, but these did not became standards, so the problem still remains. A more reasonable solution for simple low-speed radio networks can be to avoid the use of TCP/IP and to limit the services to the ones that can be implemented directly over a slow AX.25 link (like e-mail and file transfer). An other interesting application is www4mail that allow an efficient web browsing by using Netscape^TM and e-mails.

A rough estimation of the possible protocols and services at different speeds can be done with some basic assumptions for few scenarios:

• Point-to-Point or Point-to-Multipoint connection between ISP and served user(s):
  This first case can be a remote university department with a small number of networked computers that is connected to the mail campus network through a router and a Packet Radio link.
  The second case can be for example a cheap implementation of a campus LAN (many departments that share a Packet Radio network) or a regional network for providing (high) schools with basic connection to Internet through a central access point.
  
• Number of simultaneous users on the served local network:
  it can be lower than the number of access terminals, because not all users access the network at the same time.
  

Point-to-Multipoint Packet Radio network between ISP and some remote sites

	1200-2400 bps	9600-19200 bps	higher speed (38400 bps or more)
2-3 remote sites	no TCP/IP AX.25 (PMS mails, small file transfer)	TCP/IP (telnet, e-mail, ftp) AX.25 (PMS mails, file transfer)	TCP/IP (telnet, e-mail, ftp, web browsing)
5-7 remote sites	no TCP/IP AX.25 (PMS mail)	TCP/IP (telnet, e-mail) AX.25 (PMS mail, small file transfer)	TCP/IP (telnet, e-mail, ftp)
10-20 remote sites	no TCP/IP no AX.25	no TCP/IP AX.25 (PMS mail)	TCP/IP (telnet, e-mail) AX.25 (PMS mail, file transfer)

Assumptions: only one user at the same time for each remote sites, all sites working simultaneously.

Scenarios:

1. Academic Campus or Research Institution without Internet connection.
   Services that can be implemented:
   
   • Local services from a basic internal e-mail system up to a local Web facility for spreading information inside the campus.
     
   • Regional services like e-mail connection for urban and rural (high) schools and/or remote campus sites for remote consulting of people and archives of the University.
     
2. Academic Campus or Research Institution with e-mail only connection to Internet.
   
   • Local services from e-mail to limited Web access for spreading and retrieve information.
     
   • Regional services like e-mail connection for urban and rural (high) schools and/or remote campus sites .
     
3. Academic Campus or Research Institution with full Internet connection.
   
   • Local services from e-mail to Web access for spreading and retrieve information.
     
   • Regional services like e-mail connection (up to limited Web services) for urban and rural (high) schools and/or remote campus sites .
     

Cost extimation for Packet Radio

All costs provided are for 1 unit, one link means at least 2 units.

1. Low speed Packet Radio (1200-2400 bps)
   
   • TNC : from zero (using the PC audio card) up to 200 US $ for a complete TNC;
   • radio: from 150 US $ depending of power and features (fixed or variable frequency, band, etc.);
   • antenna: around 50 US $ for medium directivity-medium gain antennas;
   • cables and power supply: around 100 US $

2. High speed Packet Radio (9600-19200 bps)
   
   • TNC : from zero (using the PC audio card, max. speed 9600 bps) up to 500 US $ for a complete TNC;
   • radio: from 200 US $ depending of power and features (fixed or variable frequency, band, etc.);
   • antenna: around 50 US $ for medium directivity-medium gain antennas;
   • cables and power supply: around 100 US $

Conclusions

(in preparation)

References

• Packet Radio:
  http://www.tapr.org/tapr/html/pkthome.html
  http://www.stack.net/~victor/hamradio/packet/packet.html
  http://www.cam.org/~radio/packetintro.html
  http://itg-pc1.acns.nwu.edu/~odenbach/classes/c96/pktfaq.html
  http://www.cam.org/~radio/packet_biblio.html
  
• The Internet in Developing Countries:
  

Version 1.01 ©1999 Carlo Fonda - ICTP Radio Unit
for comments contact the author cfonda@ictp.trieste.it