UBA-RST TCP/IP
TaskForce information
Bringing the
Internet 'On The Air'
The
members
UBA-RST members participating in TCP/IP developments
are:
The
list of people involved does not end here. Several SysOps exchange technical
information regularly through the newsgroup
ampr.modes.tcpip. Articles in this newsgroup are mainly
technical discussions on TCP/IP.
Back to top
The goal of our
project
Our main goal is to bring the Internet "on the air". We
want to achieve this for several reasons.
- Currently, the user is
often enforced to use some specific software to obtain certain information
(especially for retrieving binary data like programs: depending on your BBS, you should use 7+, YAPP, SP, ...).
- Partly as a result of
the problem mentioned just before, simply using Packet Radio is as user-un-friendly as you can get. Being forced to use
several applications generally means that the user has to learn as many
different command-sets (while the majority of those applications still are
console-applications, with very complex configurations and keyboard
command-sets).
- Not only has the user to
work with different command-sets, he also has to do repetitive tasks himself
by hand: finding a route to certain stations, checking if new mail has
arrived, looking if new interesting articles have arrived at the BBS,
deciding which application to use, ... Usage of a protocol-stack alleviates
the users of all these problems: the network becomes transparant; the user
doesn't even have to know he is communicating with other computers.
- This is a philosophical one: radio-amateurs are
people who develop and experiment a lot. In certain areas, they are ahead of
the industry. Keeping this in mind, it is unforgivable that we don't use any
protocol-stack (but just a single data-protocol). Our choice to use the
TCP/IP protocol-stack is mainly motivated by its widespread usage (cf. the
Internet) and especially its openiness (protocols are described in RFC's,
which are available for free through the Internet). Introducing a
protocol-stack on our frequencies will not bring us ahead of the industry
itself. It will, however, provide us a development-platform which is
interoperable with the industrial communication technology.
Initially,
only FlexNet nodes allowed TCP/IP traffic, making it hard for us to link TCP/IP
server systems. Recently, XNET software was installed at all major Belgian
nodes, providing TheNetNode, FlexNet and IP routing. This allows us to introduce
dynamic routing (using RSPF, for example) in a near future.
You may also download the paper we
provided for our participation in the International Packet Meeting in Darmstadt,
Germany (april 2000). It provides a good overview of our project.
Back to top
Current
developments
Introducing a protocol-stack on top of AX.25, means introducing
more overhead. In our opinion, it is therefor not advisable to simply add TCP/IP
to the classical Packet-radio at 1200 baud. This results in a 3-fold project:
-
Whether TCP/IP will be used or not,
data-communications at a speed of 1200 baud simply is too slow. So, it is
clear that high-speed TNC's should become available. The development of such
a TNC is one part of our TCP/IP project. We have developed a high-speed TNC
(19200 baud and faster) using microcontrollers. The hardware has been
developed by Walter (ON4AWM); a range of firmware products have been
developed by Joachim (ON1DDS). For now, 2 kinds of firmware are distributed:
the classical KISS-firmware and SLIP-firmware. The SLIP-firmware will make
the TNC behave as an ordinary modem, used on land-lines. This way, the local
TCP/IP server will be presented as an Internet Service Provider to the
computer's Internet software. With minimal effort, any modern computer can
be linked up to our wireless Internet (using Internet software, which is
delivered together with the computer's Operating System). The product is
extremely flexible: firmware can be downloaded into the device at any time.
New versions or even other low-level protocols can be downloaded easily.
Generally, we do not limit the usage of the developed piece of hardware to
TNC-usage.
-
Transceivers, which are currently available, are
not able to transmit/receive data rates, faster than 9600 baud. We're
currently experimenting with high-speed (first 76k8, currently 153k!) equipment on 70 cm. More info
on the
website of Walter (ON4AWM).
-
We have re-organized the distribution scheme for IP
addresses in Belgium. Furthermore, we introduced the use of authoritative
zones for domain names. We hope for the worldwide acceptance of this
authority, so that we can start experimenting with DHCP and DDNS. More
information can be found in the updates page.
Initiatives were taken by Gert (ON4BLU).
-
Finally, TCP/IP services have to be made available.
Originally, these services were provided by the JNOS package. JNOS - a
monolithic application - lacks stability and performance, however. For this
reason, we moved to a Linux server with the AX.25 protocol compiled into the
kernel. This provides us the ability to use native UNIX TCP/IP protocols and
utilities. Major advantage is the free availability of nearly ALL
application protocol servers/daemons. We use Red Hat Linux.
Most of the services have actually been configured, we mainly need
to improve our mail gateway to the classic Packet Radio BBS network. We are
currently coordinating the setup of additional servers, to start an actual
network. These developments are done by Gert (ON4BLU).
Meanwhile, we have been telling the story at many
locations. Here's the material used when spreading the news:
Back to top