Orb;

The simple answer is that VARA FM doesn’t do any collision avoidance.

However, when there is an active VARA FM session between two station, any additional stations attempting to connect on the same channel will not see a response from either of the connected stations. Data communications between the connected stations will probably be impaired until the interfering station exhausts its connection retries, but the session normally resumes at that time.

Also, VARA FM supports the busy-channel detection that is implemented in the Winlink Express VARA FM session… if there’s traffic on the channel, the operator trying to connect must click the “Connect Anyway” button.

Keep in mind that the actual best-case?throughput for a single 1200bps connection is on the order of 500bps… under optimal signal conditions. Typical throughput on mildly impaired radio circuits is on the order of 300bps. The per-user throughput is a fraction of that best-case value when multiple users share a channel. This difference between the 1200bps raw data rate and 300-500bps throughput is primarily due to the overhead imposed by the AX.25 protocol.

In contrast, unlicensed VARA FM NARROW throughput is 5Kbps over moderately impaired radio channels where 1200bps packet connections are not even practical. Licensed VARA FM NARROW throughput is at least twice the unlicensed value under similar conditions. Double those numbers again for licensed and unlicensed VARA FM WIDE operation.

Realistically, minimum VARA FM NARROW throughput, for unlicensed use, is 10-15 times higher than single connection 1200bps packet… and 40-60 times higher for licensed VARA FM WIDE operation.

I believe that, compared to 1200bps packet operation, the throughput advantage of VARA FM (WIDE or NARROW), even for unlicensed use, more than makes up the lack of a collision detection.

Mark - AD7EF

On Jul 27, 2023, at 4:18 PM, Mark Davis <markad7ef@...> wrote:

Orb;

The simple answer is that VARA FM doesn’t do any collision avoidance.

However, when there is an active VARA FM session between two station, any additional stations attempting to connect on the same channel will not see a response from either of the connected stations. Data communications between the connected stations will probably be impaired until the interfering station exhausts its connection retries, but the session normally resumes at that time.

Also, VARA FM supports the busy-channel detection that is implemented in the Winlink Express VARA FM session… if there’s traffic on the channel, the operator trying to connect must click the “Connect Anyway” button.

Keep in mind that the actual best-case?throughput for a single 1200bps connection is on the order of 500bps… under optimal signal conditions. Typical throughput on mildly impaired radio circuits is on the order of 300bps. The per-user throughput is a fraction of that best-case value when multiple users share a channel. This difference between the 1200bps raw data rate and 300-500bps throughput is primarily due to the overhead imposed by the AX.25 protocol.

In contrast, unlicensed VARA FM NARROW throughput is 5Kbps over moderately impaired radio channels where 1200bps packet connections are not even practical. Licensed VARA FM NARROW throughput is at least twice the unlicensed value under similar conditions. Double those numbers again for licensed and unlicensed VARA FM WIDE operation.

Realistically, minimum VARA FM NARROW throughput, for unlicensed use, is 10-15 times higher than single connection 1200bps packet… and 40-60 times higher for licensed VARA FM WIDE operation.

I believe that, compared to 1200bps packet operation, the throughput advantage of VARA FM (WIDE or NARROW), even for unlicensed use, more than makes up the lack of a collision detection.

Mark - AD7EF

Using G8BPQ software, not uncommon, at 1200 baud AX.25 with NETROM over top of it, the typical 1200 baud throughput is 15 to 25 bytes of payload per second, plus callsigns and flags and whatnot, and that is assuming that PPERSIST is set to some high number, like 200.?

The interesting thing, to me, is what happens when four stations are on the same frequency, attempting two different 1 on 1 conversations, each pair thinking they are alone on the channel. ?

KR3RAL <—> KW1BRB <—> K1DT <—> KE1AST?

KW1BRB in the west suburbs is talking to?KR3RAL, a few miles further out. ? ?

Then later K1DT, in downtown, is talking to KE1AST in the East suburbs.

And then KW1BRB and?KR3RAL?resume. ?But where?K1DT?can hear KW1BRB and KE1AST, the only station KE1AST can hear is?K1DT. ? So?KE1AST?is pinging away trying to get?K1DT?to answer, and?K1DT?is waiting for?KW1BRB to become silent before?K1DT?can transmit. ?Once in a while K1DT sees the channel quiet enough and goes and sends a frame to KE1AST, trashing the frame?KR3RAL?was sending. ? That’s what happens using AX.25. ? What does VARA FM do in this scenario?

The 1200 baud packet exercise, with this scenario, would be massacred pretty badly by bit-errors, plus when a station goes to transmit, it would usually end up waiting until the other stations it can hear are all un-keyed. ?It’s very dynamic. ?If 1200 baud AX.25 is set up with PPERSIST set to a low value on all stations, they’d be really slow in the first place, but the collisions may not be as horrific. ? It’s fun to watch, if one doesn’t take it too personally. ??

The much worse case is when there are a couple of conversations and the stations are using on-channel relays, perhaps more than one relay. ?That’s very common in AX.25 1200 packet. ? The throughput goes from intolerably slow to dropped call.?

I’m a big fan of self-coordinated dedicated point to point link frequencies for VHF or UHF FM data links, where the frequency WIKI has enough channels for people to not be on the same channel as another link within range. ? But this is only useful (in an automated real-time network) if there are more than a couple of radios+modems at each site to cross link the channels. ?See the maps on the ncpacket.net webpage. ?

VARA FM sounds great and I’d love to have something like it that we can build a real-time network out of. ?I wonder how long it will be before one could get a licensed copy in a device that costs under $200 or so, in order that one could build a bunch of multi-port network stations. ?This would put AX.25 to bed finally, maybe? ? ?The $40 NinoTNC on a FM voice radio is good for 2400bits per second, around 40 bytes per second on G8BPQ with PPERSIST=200 and with NETROM taking up time. ?That’s pretty slow compared to what Mark was talking about. ?But a half dozen NinoTNCs can be attached to a G8BPQ network node with half a dozen radios, if one was so motivated. ?In our local network, ncpacket, we have a large percentage of 4800 and 9600 baud capable radios, when matched with a $40 NinoTNC. ?So we’re getting 60 bytes per second to 110 bytes per second on many links. ?But the network stack does not make bit-rate or processing adjustments to compensate for fading links. ?So it gets pretty bad when we have multi-path phasing. ?

We need a network protocol that is more efficient than NETROM, more forgiving of fading links than NETROM, and compatible with VARA FM. ?

I’m partial to the Raspberry PI as a computing platform, but we’ll see how the market for VARA FM compatible embedded solutions develops. ?Using a set of Intel NUCs doesn’t seem completely ridiculous, and getting less and less ridiculous as time goes on? Maybe there is already a better deal than Intel NUC for a single port VARA FM instance of a multi-port node. ??

Tadd KA2DEW in NC. ? ? ?

?

On Jul 27, 2023, at 4:18 PM, Mark Davis <markad7ef@...> wrote:

Orb;

The simple answer is that VARA FM doesn’t do any collision avoidance.

However, when there is an active VARA FM session between two station, any additional stations attempting to connect on the same channel will not see a response from either of the connected stations. Data communications between the connected stations will probably be impaired until the interfering station exhausts its connection retries, but the session normally resumes at that time.

Also, VARA FM supports the busy-channel detection that is implemented in the Winlink Express VARA FM session… if there’s traffic on the channel, the operator trying to connect must click the “Connect Anyway” button.

Keep in mind that the actual best-case?throughput for a single 1200bps connection is on the order of 500bps… under optimal signal conditions. Typical throughput on mildly impaired radio circuits is on the order of 300bps. The per-user throughput is a fraction of that best-case value when multiple users share a channel. This difference between the 1200bps raw data rate and 300-500bps throughput is primarily due to the overhead imposed by the AX.25 protocol.

In contrast, unlicensed VARA FM NARROW throughput is 5Kbps over moderately impaired radio channels where 1200bps packet connections are not even practical. Licensed VARA FM NARROW throughput is at least twice the unlicensed value under similar conditions. Double those numbers again for licensed and unlicensed VARA FM WIDE operation.

Realistically, minimum VARA FM NARROW throughput, for unlicensed use, is 10-15 times higher than single connection 1200bps packet… and 40-60 times higher for licensed VARA FM WIDE operation.

I believe that, compared to 1200bps packet operation, the throughput advantage of VARA FM (WIDE or NARROW), even for unlicensed use, more than makes up the lack of a collision detection.

Mark - AD7EF

On Fri, Jul 28, 2023 at 07:31 AM, Steve Stroh wrote:

Check this sub-article in my newsletter Zero Retries -?

Hi Steve;

I'm a Zero Retries subscriber... keep up the good work! ?

I apparently overlooked the "Mercury software-defined modem" item.

I checked out the git-hub site, and looked for other references on the 'net. Very interesting, and somewhat promising.

I can see where (with a considerable amount of further development) the Mercury software designed modem might result in a low-cost "VHF/UHF FM radio-modem "appliance" that could compete with VARA FM. However, I believe there needs to be a good deal more "machinery" to make it an effective HF modem.

In either case, though, it's very unlikely that either a 'VHF/UHF FM' version or 'HF' version would be compatible with the corresponding VARA system.

Mark - AD7EF

On Thu, Jul 27, 2023 at 06:03 PM, Orv Beach wrote:

Wow, that's pretty ugly, Tadd.

NCPACKET is not so bad when you consider it in terms of a Maker-space hands-on? project.? The network has to follow procedures that are brought down to the level of the least-common ham.? While the sophistication level of AX.25 is pretty high for somebody who has never cared about digital data at all, the diagnostic tools make it pretty easy to follow.? Every ham can visit any site (digitally over the ham-radio-only network) and inspect plus run the diagnostics tools.? When all the links are working (about 80% of the time) we can do a 35-person live chat across the entire 9 county network.?

There is a huge divide between running WinLink and VARA FM on 2m from home, compared to building microwave links at a locked remote site.?

I tried to drum up interest in financing a (free) 600' location in Raleigh on a commercial tower for a Ubiquiti based system and, while there was quite a bit of interest, there weren't enough people willing to pony up to buy the remote hardware and an insufficiency of the interested hams would have been in range of that one site.? I was hoping to use it as a stepping stone to link to other remote sites using 3.3Ghz.? (that was before the 3.3Ghz ham allocation changes) ? I think ARDENmesh must be having its success in places with higher hills, drier climates, and sparser trees.?

The NCPACKET network is also built out of people's houses in an area where the highest mountain is only 1000' above average terrain and most of the region the network is built in doesn't even have that big of a hill.? 900Mhz is almost a lost cause here because of the big trees, high humidity and lack of high places to run from.? 2.4Ghz is much worse.? To do a 5 mile wide 2.4Ghz network in the northern suburbs of Raleigh, where I live, would take about a dozen participants.? From what I've seen, 2.4Ghz radios and antennas are quite different from 2m and 75m radios and antennas. Any ham can start, build, and operate a TARPN network.? They could get a 40 mile wide network with no towers or hills for about $3000 in a thick forest of 100' trees.? In the NCPACKET network, walk-up servicing is available with no security arrangements even in the time of a national emergency but almost all of the nodes are at the house of the operator, so no walking is even required.?? All of this is run with FM transceivers between 51 and 450Mhz. ??

It's a different world, doing things in people's houses where the users own the network.?

So my curiosity, and? the reason I watch this email reflector, is that I am hoping to see VARA-FM evolve into something our local data network can take advantage of.? I think it will as the cost of embedded PCs comes down.? I bought a used Intel NUC at the Shelby hamfest for $75 and the guy had several of them.? MSWindows10, 32GB SSD and 2GB of RAM.? It's a pathetic MSWindows10 computer, but are we getting close??

--KA2DEW?

On Thu, Jul 27, 2023 at 02:44 PM, Tadd KA2DEW in NC wrote:

VARA FM sounds great and I’d love to have something like it that we can build a real-time network out of. ?I wonder how long it will be before one could get a licensed copy in a device that costs under $200 or so, in order that one could build a bunch of multi-port network stations.

Being able to use existing VHF/UHF radios certainly has advantages, but they are voice radios, with audio bandwidths in the 6KHz - 7Khz range, when set for "9600bps data". With this limitation, physics gets in the way of moving enough data for a workable data network, particularly with the challenges of terrain, and foliage (particularly on UHF bands)

It's a different approach than you're proposing, but AREDN is absolutely terrific for building a real-time, high speed (TCP/IP), amateur radio network over large regional areas. It is a true mesh-topology network that is distributed, with substantial redundancy. It is really highly developed in Southern California, and we're rapidly moving in that direction here in the Oregon Willamette Valley (check out?

While this is a data-only network, it does a great job of supporting voice communications through VOIP phones. Using the email and "post office" facilities of the Winlink are a snap, through a Telnet Winlink connection.

The buy-in, including an AREDN transceiver, antenna and mounting hardware, is probably 3 - 4 times the entry point you'd like to see. Another constraint is that anyone who wants to implement a node needs at least one genuine line-of-sight path to another existing, active node. From the standpoint of mesh-network redundancy, a line-of-sight path to two or more nodes is preferred.

I know that both in Southern California and in our local region, the amateur radio EMCOMM community has convinced served agencies of the value of these networks, and have received substantial financial assistance from those served agencies.

Mark - AD7EF