[aprssig] Local Event using RELAY?

[Henk de Groot]

Hello Pete,

AE5PL Lists schreef:
> #1 - ALOHA is not transmitting in-the-blind.  It is a much more complex
> protocol which is significantly different from APRS.  The term has been
> misused extensively here to justify people using the statistics gathered
> in that project which are not related, in any way, to VHF APRS

Sorry but we are talking about the channel access characteristics only, 
not the full ALOHA network. ALOHA in this context is "pure ALOHA" which 
was the first type of channel access that the University of Hawaii used in 
1972. This type of channel access has a maximum troughtput at 18.4% 
channel utilisation. With less packets the channel is under-utilized and 
with more packets throughput drops due to many collisions.

So this is independent of any retries. For the complete ALOHA network the 
retries by the bighter layes caused the network collapse as soon as you 
went over the 18.4% channel utilisation. From 0% to 18.4% the channel 
thoughput increases and the channel is stable. Above 18.4% the channel 
becomes unstable since more packets are lost, more need to be retried and 
the load just increases so fast that the network comes to a standstill.

So behaviour of the higher layers causes the ALOHA network to breakdown 
once you go "over the top". The transmitters have to back-off. This is 
very different from APRS, APRS runs happely when the channel is 40% loaded 
- only you cannot get nearly as much data as with 18.4% load but it 
doesn't get worse like the first ALOHA network would.

The U of H made serveral improvements to pure ALOHA to enhance maximum 
channel capacity. Two improvements were to use fixed length packets and to 
introduce timeslots. When we talk about ALOHA channel access in 
conjunction with APRS it always refers to "pure ALOHA".

For the term ALOHA for this type of channel access, I don't know a better 
term that is widely known and accepted. If you would stay in line with the 
terminology like CSMA, CSMACD, TDMA, FDMA, CDMA, then pure ALOHA should be 
called "MA". But that's not the way it is in literature...

Now for your APRS area being mostly CDMA. I have to agree for 
home-stations, digipeaters and others with good antenna's and terrain that 
allows you to hear eachother. But if I take my TH-D7 outside with its 
rubber-duck, the TH-D7 can only hear the digi - at least frequently...

You have to be aware that all these small devices like the TH-D7 with 
rubber-duck will actually wait until the big stations in the area (which 
they all can hear) stop transmitting and all these low power/small 
footprint devices are competing for the same periods of silence when the 
big stations shut up. So CSMA increases the probability that these small 
devices are jamming eachother - causing both packets to be corrupted 
instead of one making is to the digi.

While reading your mail I also had another strange feeling. If every 
station can already hear all its neigbours, why do you have a digipeater? 
That only doubles the ammount packets for no good reason. Otherwise your 
90% must be flawed since every digipeater can theoretically transmit at 
most 50% of the time since it needs to receive the packets first. If you 
hear multiple digis then you have definatly too many in too close proximity.

Home stations, digipeaters and stations with a good antenna and/or high 
power should never transmit blind. But I realy do not see much of a 
problem with low power/low footprint devices.

Okay, I'll leave it with that. I think the raised points are clear and I 
even think we don't disagree much. Avoid however to take things for 
granted because you once learned that doing something is a bad thing. Keep 
an open mind and dare to question these kind of general assumptions.

By the way, did you know DAMA also works without carrier detection? In 
DAMA the master (the node) does the listening for you, when the slave (a 
connected station) is polled it has to respond immediately, regardless of 
what's going on on the channel. The idea is that the master controls who 
transmits and when, so the slave just obeys without question. The master 
is responible for polling the slave when the channel is clear, the slave 
is not allowed to add any extra delay in responding; when the response is 
not received immediately the master will poll the next slave. There cannot 
be any uncontrolled stations on the DAMA channel, except for sending SABM 
for connection-setup.

Another example of running without carrier detection is Full Duplex 
connection on a point to point link. Its just assumed that you are the 
only transmitting station on one of the two QRG's of the FD link. All our 
packet interlinks on 23 cm work that way.

Kind regards,

Henk.