[aprssig] APRS low-power-local ALT input channel

[Robert Bruninga]

>>> HamLists at ametx.com 9/27/04 10:19:04 AM >>>
>Ok, time to debunk your numbers, Bob. 
>...For trackers... maybe 60 [per hour] would be more 
>accurate,... Weather stations... every 15 minutes...
>...Add on to that all of the objects generated by many 
>fixed stations, and the number far exceeds 4 packets 
>per hour per station per hour.

True, but then a 1 Watt mobile is not going to be in my
DIRECT range for more than about maybe a few minutes
Hence, I *wont* hear 60 per hour, I may only be able
to hear and avoid collisions with about 4 per hour...
So, OK, how many of those LOCAL DIRECT stations
are sending an object?   Maybe one...  Ok, ill change
my premis from 4 per hour to 4.4 per hour... 

>Second, you are using a number of how many stations 
>you can DECODE via direct.  My station... sees many 
>more stations than it can decode...

True.  But the software CD in all modern digis does not
respond to "noise" on channel (fringe packets) but
true AX.25 bits that ARE decodable.  Else on today's
digis NOTHING would ever get digipeated...  Also the range 
equation goes up by distance squared so including those 
detectible but error packets would probably about double
the number of packets you could avoid with CD...
Yes, Ill agree to a factor of two.  Lets make it 8.

>And all of those other stations (300 per hour according 
>to your count at your QTH)?  You are not competing with 
>them for transmission on RF, only with their packets that 
>are being digipeated by your near-by WIDEn-n which you 
>do see.

Yes, and they are all the ones that *wont* be there on 
the alt-input channel, and so they are not invovled in
CD on the alt channel... nor involved in this debate.

>So, you are incorrect trying to state that CSMA only 
>prevents about 1% of collisions at your location.  It prevents 
>in excess of 80% collisions because you are directly 
>seeing more stations than you are decoding AND
>you are only competing with the other stations in your 
>area for the WIDEn-n digipeater.

Again, your issue was requireing that users who TX on
144.99 must have a second receiver on that freq to
avoid collisions with other locals.  Hence, my response 
only addreesses the probability of colliding on that channel 
with other local users on that channel.  I maintian it is 
exceptionally low.   Based on your comments I will allow 
to double my 1% to 2% to account for packets that could 
be heard but dont decode as you suggest.

But I totally disagree with your 80% number.  That has
nothing to do with collision avoidance with *other*local*
users*of*the*local*digi.  Which is the topic of discussion.
And is the main thing that significantly impacts a stations
ability to be reliably heard or not...

>...If you have only 9 stations on the frequency within sight 
>of the digipeater you want to use, beaconing at an average 
>rate of 3 minutes (most are mobile, according to your
>comments), you end up with 3 beacons per minute.  Seems 
>like that leaves 57 seconds wide open per minute. 
> Well, yes and no.  The odds of a collision are actually 
>much higher...

Two problems here.  The mobiles use Mic-E format which is
only about ½ second, not the full second I allowed for fixed
stations.  Secondly, the probability of colliding during the
length of a packet is statistically half the length of a packet.  
So I'll agree to increase the collision rate to say 4.5s out of
a minute, but because you are talking ½ second mobiles
(so you can justify the large number of packets) then we
must cut the figure in half.  So I'll grant you then 2.25
seconds of collision probability per minute or about 4%.
So I Agree to up the number as you suggest to 4%.

>The reason they are higher is because that over time, 
>....  Unfortunately, once a collision  does occur, it is also 
>likely that the collision will continue ... as there is no ability 
>in  APRS to... transmit again at a random interval (a key
>requirement within the original ALOHA tests, by the way).

AND a *fundamental* aspect of APRS!  APRSdos from day
one has always added randomness to EVERY timer.
Not only does it decay every period, it also adds at least
4 seconds of random delays as well to very packet...  I 
insisted that ALL APRS applicaitons MUST add randomness 
to their rates, but most of the other authors and hardware 
builders just ignore this fundamental principle!

>Now, those 9 stations... would see each other...
>in one form or another.... All-in-all, there is a large 
>opportunity for carrier detect to work.

I guess it depends on how you define "large".  To me
4% is not large.  In fact, it is quite small.  And is certainly 
not worth the added requirement to force every user
of every radio that uses a +600 offset to transmit to
have two receivers, one to listen to the output and one
to listen to the input before transmitting...

And again, your example  you hypothesize above is a far 
larger example than the average station would experience.  
Do a PHG plot of all the stations around an average
low level digi.  I doubt that the *average* digi will find 9 
mobiles in range.  In the dense east between  Baltimore 
and Washington, it is more like about a dozen stations
with say 8 being fixed (4/hr each), 2 are WX (6/hr) and
two are mobiles that were in view for say 10 minutes
(10/hr).  Yes, Ill grant you that in this case the typical
potential for collision of the locals on the input may be
as high as 4% or higher.  But to me, that is still a SMALL
number...

>Finally, let's look at your [ultimate] digipeater design...
>... Every time the WIDEn-n side transmits, your local
>receiver is desensed and cannot copy anything on the 
>"local" side. Talk about needing cavities. 

This is a red hearing.. Yep, just like every existing digi, 
they cannot listen while they are talking...  it is no different
than normal APRS on 144.39.  The users are still "listening" 
on the 144.39 output so they KNOW it is transimtting and 
will NOT chose to transmit then.. whethere they are going
to transmit on 144.39 or 144.99 (if available as an alt channel
on this digi)...  That is the part of CSMA that does
work...

>Also, as pointed out before, the PocketTracker only 
>supports 144.39 and 144.34.  You have excluded them 
>from your configuration with promoting 144.99 for
>a frequency.

Yes, which is what is driving this whole thread!    Im trying
to do a little pre-planning to avert a future problem I see...

because if 
we continue to mass produce low power trackers on the 
national ATV simplex voice channel (which is OK for the rare 
balloon launch, but not OK for thousands of trackers)...
we will have a civil-war and a lose-lose situation on our hands.  

Why do this when we could look to see if 144.99 is available
in most areas and therefore suggest that pocket trackers not
intended for Balloons should be orderd with the 1444.39 and
its +600 TX offset for general use on the ground.  So far, not
one person has taken the time to investigate if 144.99 is
available in their area, or at least taken the time to inform us...

>When we have run events on 144.39 with no problems, 
>... Because we have well positioned WIDEn-n digipeaters 
>which are not running high power that allow for CSMA 
>to work. 

The power output of the digi has nothing to do with colliding
with a low-power tracker on the input...

>Yes, there are collisions.  But much fewer than if CSMA 
>was not used.  We have also run events on 144.34 ...
>with a digipeater covering the event area... Yes, we have 
>ATV in our area, but we have found that we can work with 
>that group on an event-by-event basis.

I agree, that this works well on an event-by-event basis,
but as the dozens of low-power tracking devices explodes
to the thousands with everyone of them comming XTALED
on the overloaded 144.39 and the ATV simplex channel, 
I think the ATV'ers will become less and less enamoured
with these little bleeps on their voice channel..

>Do as you please, Bob....understand that the basis for the 
>need for a blind ALT input channel is flawed. 

Well, I guess we disagree.  I have adjusted my numbers upwards
based on your comments, and now they are up to a more
realistic 4% or so...  If there is any flaw in logic, it is your
claim that it should be any where close to 80%.

>You try to say that you are in ALOHA competition for 
>144.39 with 300 other stations when you are only
>in competition for 144.39 with the WIDEn-n digipeater 
>and what it sees directly.

Yes, we agree here.  The only issue is what is the local
input of the local digi contending with as far as collisions
are concerned.  I maintain that on an alt input channel
it is very low.    Hence it is an ideal place to encourage
users of low power tracking devices or other locals
needing an input channel free of QRM from the other
300 out of area packets...

>I am done with this thread. 

AH, thats good.  What I would  like to hear from is some
other APRS users who have explored the use of 144.99
in their area.  We need to start seeing if this is practical.

I can vouch for Eastern Maryland as being free of any
144.99 activity, either by on the air monitoring, and
searching of all known freqeucny coordinatttion records.

Others please invest in your APRS network and do a
llittle local investigating on the possibilty of using 144.99
in your area... thanks...

de WB4APR, Bob