karn (11/11/82)
Special to AMSAT Satellite Report
de Hank Magnuski, KA6M 12 October 82
(minor editorial revisions de W3IWI)
AGREEMENT ON PACKET RADIO STANDARDS!
As part of the October 10th, 1982 AMSAT (Radio Amateur Satellite
Corp.) general meeting, president Tom Clark, W3IWI, invited a number of
leaders of active packet radio groups to come to a special working group
meeting on Friday, October 8th, to discuss the future developments in
packet radio activities; specifically, this meeting was intended to discuss
AMICON (the AMSAT International Computer Network) for the Phase-3B space-
craft and to discuss a new project (tentatively called PACSAT) involving a
low-altitude orbit, all digital store-and-forward transponder. To Tom's
pleasure and surprise, the meeting was well supported and a number of
people showed up, representing these groups:
New Jersey - Phil Karn, KA9Q;
Tucson Area Packet Radio (TAPR) - Den Connors, KD2S and Lyle
Johnson, WA7GXD;
St. Louis Area Packet Radio (SLAPR) - Pete Eaton, WB9FLW;
Washington (AMRAD) - Paul Rinaldo, W4RI, Dave Borden, K8MMO,
Terry Fox, WB4JFI and Eric Scace, K3NA;
Pacific Packet Radio Society (PPRS) - Hank Magnuski, KA6M;
AMSAT - Tom Clark, W3IWI, Jan King, W3GEY, John DuBois, W1HDX
and Bob Carpenter, W3OTC.
Even more surprising is that this diverse group managed to agree on
some things. In fact, the agreement reached on adopting a common link level
protocol may prove to be extremely significant in forming the foundation
for U.S. packet radio networking. The meeting which originally was supposed
to iron out some AMICON details managed to have much more of an impact.
Here's some background on what happened:
A year ago, in conjunction with the '81 AMSAT general meeting, Paul
Rinaldo organized the first Packet Radio Networking Conference under the
sponsorship of AMRAD, ARRL and AMSAT in Washington (some excellent con-
ference proceedings are still available). Many ideas and some real and
paper networks were discussed. In the months which followed quite a few
people got their packet radio controllers running and had a chance to
experiment, read, discuss and think about various problems involved in
implementing packet radio networks. The situation became discouraging. It
seemed like a new protocol was proposed for each new set of Terminal Node
Conroller (TNC) hardware that came on the air. Each group started heading
off in a somewhat different direction. The promise of compatible systems
was growing remote. But in this dispersion of effort, people also found out
how difficult it was to implement private protocols and how difficult the
interconnection would be if common ground wasn't found soon. The different
groups also came to realize that there were a common set of problems to be
solved and that one area's solution couldn't ignore the requirements of
other users. The summer doldrums saw very little activity and not much
progress, and provided the background for the AMSAT meeting. This inevit-
able diversity during the R&D phases for each of the groups had to be
reconciled with the impending launch of the Phase-3B satellite in February,
1983.
The AMICON Network
------------------
Three major areas of concern filled the agenda of the AMICON meeting:
the usage of the Special Service Channel (SSC) earmarked for digital
experimentation and called "AMICON" in AMSAT planning documents; the
recommended modulation methods and bit rates to be used on the Phase-3B
SSC; and the detailed link protocol to be used for linking ground stations
via the satellite.
There was general agreement that standards accepted today must be
regarded as developmental. It is to premature to ordain any single scheme
at this time. The AMICON concept is not yet a detailed network design, but
rather an opportunity to develop a new service for amateur radio. In the
developmental phases we may well see many diverse techniques being tested.
The meeting discussed the concept that the use of high-altitude satellites
for packet radio would be sufficiently complex that it was unlikely that
many individual users would be able to muster the resources for individual
access. Rather, AMICON would probably evolve as a channel for linking local
"concentrator" nodes around the world.
There was a lot of discussion on suitable modulation methods and bit
rates that could be supported by Phase-3B. Den Connors and Lyle Johnson of
TAPR, Paul Rinaldo of AMRAD, and Tom Clark and John DuBois of AMSAT
presented their research findings. Many different modem types were re-
viewed and international requirements were discussed. After all the
debate the following conclusions were reached:
1. The AMICON SSC usage should be restricted to 5 kHz bandwidth (at the
-26 dB points). Modem performance must be a primary consid-
eration for any ground station, and the modem used will probably
be of a rather advanced design.
2. The use of 202-type modems using NBFM-AFSK will not work.
3. The use of 202 modems using SSB-FSK will produce marginal results.
4. A 400-1200 bps channel speed is probably optimal, as this speed
satisfies a variety of different constraints and requirements.
5. The PSK modulation techniques developed for Phase-3B telemetry
should be explored at both 400 and 1200 bps.
6. The AMRAD and TAPR groups are going to pursue development of a modem
which employs Minimum Shift Keying (MSK). MSK and PSK are con-
sidered the most promising methods at this time.
7. The SSC will have to be open for different experimental approaches
until there is general agreement and experimental validation of
"the" optimum method.
Considering all the alternatives which were discarded, the above
resolutions represent a significant narrowing of focus. If that wasn't
enough, more was yet to come!
Eric Scace reviewed the work which had been done by the NJ & AMRAD
packeteers to adapt the international X.25 protocol to amateur needs. The
amateur subset has been documented by Terry Fox, and is now called AX.25.
Hank Magnuski distributed a tutorial document on connectionless protocols
and described work which had been done to develop Revision 4 of the TIPM
and LIPM software. This is the first implementation of TNC software util-
izing only amateur calls (and not hard-wired assigned addresses) for
addressing.
By the evening of the October 8, the group began to recognize that
only very minor differences separated the AX.25 and LIPM.04/TIPM.04 ap-
proaches. The TAPR protocol users were willing to implement an interface
based on the recommendations which would come out of the meeting. The
differences were resolved and led to the unanimous adoption of a subset of
an internationally recognized link level protocol (specifically it is
called ANSI X3.66 ADCCP-HDLC BA Class, with options 2, 4, 7, 8, & 11).
Why is this important? First, this link level protocol allows two
AMICON ground stations to construct a packet pipeline. This same protocol
can be used between two stations in a terrestrial backbone net. The same
protocol can also be used for terminal-to-terminal connections, whether
direct or via a simplex packet repeater. It represents only a slight
extension of the CCITT X.25 LAPB link level protocol. And finally, it does
not restrict future development of ISO Level 3 virtual circuit or datagram
protocols.
The key feature of the new design is the adoption of a scheme for
using call-sign addressing in a packet. The packet format looks like this:
-----------------------------------------------------------------
| FLAG1 | TOCALL | FMCALL | CTL | PID | INFOFIELD | FCS | FLAG2 |
-----------------------------------------------------------------
where:
FLAG1 Standard opening HDLC flag byte (01111110)
TOCALL The destination call-sign (6 char's plus sub-station ID code)
FMCALL The source call-sign of (6 char's plus sub-station ID code)
CTL Standard HDLC control byte
PID Protocol Identification byte for information frames.
INFO Information field in information frames
FCS Frame CRC check sequence
FLAG2 Standard closing HDLC flag byte
Frames to be repeated by a local area network simplex packet repeater
also contain a third address field following the FMCALL field. This third
field is the callsign of the repeater. More details on this protocol will
be published in the revised AX.25 specificiation document by WB4JFI et al.
Collectively, the group considered that nearly every link design
starts out with a statement something like this: "Holy 807's, look at all
the overhead bytes you have in using call signs as addresses. I can do it
with just 4 bits." We have found that these other schemes have their own
defects and that the penalty for the call sign overhead is relatively small
in comparison to other delays on the link and in view of the other bene-
fits. In addition, both amateurs and their regulatory authorities are
particularly defensive that the individual's call sign is sacrosanct and
tantamount to being a personal name.
The AMICON session ended with promises from various representatives to
try to implement the required versions of this new protocol. In the in-
terim, LIPM.04/TIPM.04 will be available for use by AMICON ground stations
until the new software is ready.
PACSAT
------
The other main topic of the meeting concerned the possibility of a
dedicated packet radio satellite. This is discussed in a companion news
release from KD2S and W3IWI.