[Xastir] Feature bloat?

[Wes Johnston]

It would seem that the backup WIDE digi would have to not have heard the
primary for X number of minutes.  The way to do this is to think in terms of
10 minutes, not each instance of a packet.  If the wide node ID's itself
every 10 minutes, then allow two iterations of that time with no ID heard.
So the worst case here is 20minutes with no wide node.  Back the ID time
interval up to 5 or 7 minutes and have digined (or xastir in the intent of
my origninal message) wait for 10 to 14 minutes before becoming a wide.  As
soon as the main wide digi ID's once and is heard by the secondary digi, the
2nd digi once agian assumes the role of RELAY instead of WIDE or WIDEn-n.

Summary:  The swapping of primary/backup roles is done with hysteresis... it
takes more to make the digipeater move up to primary, and it only takes a
little to make it go back to secondary status.

In order for the backup digipeater to assume the role of primary it must
*not* hear the primary digi  or not see it's callsign in the path of another
station for > id_period*2

In order for the backup digipeater to stop playing the part of primary and
back off it must only hear the callsign of the primary WIDE within a packet
or an ID from the primary wide node.

Wes

----- Original Message -----
From: "Henk de Groot" <henk.de.groot at hetnet.nl>
To: "Steve Dimse" <k4hg at tapr.org>; "Wes Johnston" <wes at johnston.net>
Cc: "Xastir" <Xastir at krypton.hscs.virginia.edu>
Sent: Sunday, June 15, 2003 5:11 PM
Subject: Re: [Xastir] Feature bloat?


> Hello Steve,
>
> At 09:49 15-6-03 -0400, Steve Dimse wrote:
> >Actually, it'd probably be better to do this on a packet by packet
basis...if
> >any particular packet is heard but not digied, then digi it. I think
digi-ned
> >has this capability, but not sure.
>
> No, it has not. This has been suggested a number of times but the problem
> is, I don't know how to do it! There is no solid way to detect if the
> nearby digi digipeated the signal since its station call is not included
in
> the packet most of the time. For example:
>
> Assume I only receive K4HG>APRS,RELAY*,WIDE3-2 but nothing else. It does
> not tell me much. The backup digi may have missed the original packet,
> which is likely if the backup has a smaller footprint than the main-site.
> Our backup can not see if it got this packet from a far-away digi or from
> the main-digi since there is no identification which station altered
> WIDE3-3 into WIDE3-2.
>
> If the backup digi has the same range as the main-digi then I could
receive
> K4HG>APRS,RELAY*,WIDE3-2 and then K4HG>APRS,RELAY*,WIDE3-1. Does that mean
> that my nearby main-digi heard it? Since the digi should see a few of the
> surrounding digis it could just as well be a second copy from another
> distant digi.
>
> The only packets I can positively identify as comming from the "main" digi
> are the ones in which its call is included. This is however only a small
> portion of the packets, I think this is one of the huge diadvantages of
> WIDEn-N; you can't see who transmitted the copy.
>
> What is missing is a signal strenght indication in the packet. If it was
60
> dB over S9 then I would know that a very local station send that packet,
> e.g. the main digi.
>
> There are 3 other solutions I can think of that technically work:
>
> The most advanced solution is that the main and backup-digi would have
some
> background communication with eachother to tell about the packets
received.
> When one goes down, the other completely takes over. But with a background
> link between the stations a much easier solution is to have 2 receivers
> coupled via the digipeater to one transmitter! Of course that is very easy
> to do with DIGI_NED, just make a crossband configuration with 2 receivers
> on the same band and only 1 transmitter. If you want redundancy include a
> signalling wire to activate the transmitter in the backup digi when the
> main-site's transmitter is down.
>
> A second solution is to equip the main-digi with a 70cm second band. Then
> accros town erect a recever station that transmits all received packets
via
> 70cm to the main-digi. The main digi will ignore packets received on 70cm
> as duplicate if it got the original direct on 2 meters and digipeat the
> packet when it missed the original. You can configure it such way that you
> don't loose an extra hop. You can even make the extra receiver station
> analog, just pass all received data from 2m onto 70cm in audio...
>
> The last option I can think off is just to have enough digipeaters in the
> grid, if you miss a packet then you will see another copy from another
digi
> in your view. I think this was the original idea. With overlapping digi's
> you don't loose much coverage if one goes down. Of course this eats extra
> bandwidth...
>
> There is no free lunch in any of these solutions. All these three
> alternative solutions are technially feasible however.
>
> For a client program which originates packes, like Xastir or even a HamHud
> device, it is much easier to detect if they got digipeated. They just
> broadcast the original themselfs and only have to look for a single echo
of
> the packet to see if they got digipeated.
>
> If somebody comes up with a solid algotithm how to do make a working
backup
> digi I will be happy to implement it to try it out. Up to now I didn't
find
> a working method.
>
> Kind regards,
>
> Henk.
>
>