[Xastir] Smart RELAY digipeating

[Curt, WE7U]

On Wed, 5 Jan 2005, Wes Johnston wrote:

> Ummm.... I've been watching this thread and think the idea is to not digipeat
> RELAY by a home station if you have already heard the same packet digipeated by
> the local WIDE.  That's a great idea.  So your station sits on the packet for 3
> to 5 seconds before digipeating.

That's my idea of being a good neighbor.  I think we were talking
about igating though w.r.t. getting everything talking at the same
time.  That way you wouldn't see the same packet gated to RF
multiple times by different igates, but of course you'd have to deal
with a potential problem of being equidistant RF-wise (signal
strength vs distance) from two or more igates, causing you to not
decode the packet at all.  I would think that to be an uncommon
situtation though.


> In the past few days is where I've gotten confused.  There seems to be talk
> about KISS tncs delaying the delivery of a packet to the network.  If what I've
> seen is correct, it is being said that the kiss tnc can cause the delay.  Yes,
> it can, but if the packet is heard digipeated from the WIDE, that KISS mode tnc
> isn't going to care... it's got "data" in queue and is just waiting for access
> to the channel.

Yep.


> The trick to make this work _correctly_ is to use 6-pack TNCs.  They communicate
> the busy/clear status back to the host computer.  Unlike KISS tncs, you can't
> queue a packet in the tnc for minutes at a time... the host must watch for the
> channel to be free, and as soon as it is, the host must start sending data
> forcing the TNC on air.  This of course moves the host computer one step closer
> to the radio and allows us to abort digipeating a relay packet.

Or a soundcard TNC.  Or an SCC card.  Or a KISS TNC with the
parameters set so aggressive as to just blurt out no matter what.


> Sure is a shame that KISS tnc's don't communicate the CD led status via one of
> the rs232 handshake lines.

Perhaps a super-KISS TNC is in order.  Figure out the best of what's
out there, perhaps add in a bit more so that someone can code up the
protocol in the future.

What I'd like to see:


*) CRC checksum on packets going both ways across the serial line.
This is to prevent packets that are corrupted via serial
ports/serial lines from undermining the integrity of the system.

*) RTS/CTS usage (KISS protocol currently specifies that handshakes
not be used at all).  Have the TNC respect the RTS line/stop sending
if inactive.  Have the TNC set the CTS line appropriately as it can
accept data.

*) Have the TNC ignore other lines if not connected, such as DTR,
else if pulled inactive, don't send to computer.

*) Implement DSR out of the TNC as TNC-on/operational:  Set it
active.  TNC-off or not working:  Set it inactive.

*) Perhaps the RF carrier detect status available via handshake
lines as you're talking about.  Heck, if the serial transfer speed
was high enough, you could just have that control the CTS line.
Have the TNC start keying up as data is streaming in.  There's
actually a signal designed for CD though on a DE-9 or a DB-25:  DCD.

*) Multiple RF port capability, MKISS or 6pack protocols.

*) Implement a new KISS set of parameters where the PC can
effectively control the start of transmit instead of the TNC.  If
this special mode isn't turned on, the TNC controls the transmit via
p-persistence/slottime variables.


One should probably check out flexnet and the SMACK and 6-pack
protocols, as well as other slight variations on the KISS protocol
to see what the best features are of each.  Last time I looked there
where three or four variations on the KISS protocol.

One of the nicest features would be some way of auto-tuning to the
RF channel based on how busy it is.  If the TNC could handle that
part of it, networks would have a harder time hitting overload
conditions.

I expect Bob might have some more ideas along these lines too, or
can blast some of mine out of the water.

--
Curt, WE7U.   APRS Client Comparisons: http://www.eskimo.com/~archer
"Lotto:    A tax on people who are bad at math." -- unknown
"Windows:  Microsoft's tax on computer illiterates." -- WE7U
"The world DOES revolve around me:  I picked the coordinate system!"