HF Packet vs WinLink

[N5PVL]

HF Packet, as it stands, is both functionally and technically superior to WinLink in every particular.

Both systems are being used to move 'store and forward' type messaging, the very nature of which is not time-critical. By definition, a wait of an hour or more is typical before these kind of messages may be retrieved by any method. Longer waits are not unusual by any means. So while it is true that PACTOR III is faster than HF Packet, ( while taking up 6-7 times as much spectrum ) this additional speed is not the critical factor it is made out to be.

Both systems are designed to serve a large number of locations... HF Packet, being an AX25 application allows a dozen or more stations ( locations ) to share a single frequency slot at the same time in what is known as 'vertical stacking'. Compare this sprectrally efficient system with PACTOR ( and any other TOR mode ) which cannot stack vertically and so must utilize 'horizontal spreading' instead, with each station located beside ( or spread out from ) the next one, taking up a great deal more spectrum in order to serve the same number of locations.

Add PACTOR's significantly wider bandwidth to this equation, and it is easy to see why the WinLinkers are trying to gain access and control of virtually every bit of available HF spectrum while the PACKET networkers are content to efficiently operate within the narrow automated sub-bands, as they have done for decades with no problems.

Then there are issues of reliability and technical advancement to consider... The WinLink 'network' is designed around the STAR network topology that became obsolete for anything but home LANs in the late 1970's. WinLink's central server provides a single failure point that can ( and does ) bring down the entire network in the event of a minor difficulty. Last year for example, the entire WinLink 'network' went belly-up for a day or two as the result of a common email virus corrupting thier user base, which had to be re-established from scratch.

The Packet network utilizes the more modern and secure PARTIALLY-MESHED network topology that the Internet also uses, and because of this, there is no single failure-point within the network which can affect operation of the network as a whole. It is more technologically advanced than what WinLink utilises and it is also much more reliable.

Do a web-search on NETWORK TOPOLOGIES, then compare what is said about WinLink's STAR topology with what is said about Packet's ( and the Internet's ) PARTIALLY-MESHED network topology. - You'll see what I mean in short order.

So you have two networks attempting to do the same thing:

Packet is more advanced, more efficient, works just fine within the narrow autiomated sub-bands, is more reliable and historically causes no problems.

Compare that with WinLink's use of a long-obsolete network topology that is far less efficient, unreliable in the extreme and which has already garnered WinLink a widespread reputation as a spectrum hog and QSO-crasher. - A reputation which will rapidly worsen if they are allowed to spread all over the phone bands and crash QSO's as the ARRL proposes.

That's why I say that HF Packet, as it stands, is both functionally and technically superior to WinLink in every particular.

When you consider the additional speed that Q15x25 mode brings to the Packet net ( a virtual 2500 baud ) and the even greater increment of speed that the development of AMP ( Amateur Multicast Protocol ) will give the Packet net for distributing messages addressed to "ALL", then it is easy to see why the WinLink 'network' has a pretty firm grasp on the 'hind titty' position among ham radio store and forward messaging systems.

In other words, Packet is cool - and WinLink sux. #

Charles Brabham, N5PVL
Director: USPacket.Org
http://www.uspacket.org

[kb3mng]

Quote[/b] (N5PVL @ Jan. 28 2006,05:41)]HF Packet, being an AX25 application allows a dozen or more stations ( locations ) to share a single frequency slot at the same time in what is known as 'vertical stacking'.

Without a central control, how does HF Packet overcome the inherent throughput limits that Aloha and CSMA networks experience when there are multiple nodes that want to transmit?

[N5PVL]

They don't... HF Packet is a carrier sense multiple access network, and polling schemes like DAMA have been found to be unworkable and impractical in the HF environment.

It is just as well, as a central DAMA master would introduce the same type of vunerability that WinLink suffers from, by introducing a single failure point that would effect a large part of the network. It would not endanger the entire network as with WinLink with its obsolete STAR network topology, but it would still be a weak link that need not exist.

Since the information being sent is of a "store and forward" nature as mentioned above, ultimately speed is of much less importance than reliability.

It is thought that the introduction of AMP ( Amateur Multicast Protocol ) stations to the so-called "Packet" net will bring on a significant speed increase across the board, much more so than by simply doing the same kind of point-to-point networking faster as with PACTOR III and Q15x25.

Q15x25 is capable of "vertical stacking" and is currently coming into use within the "Packet" network, but this is not by any means the most significant advancement in the cards for the Packet net right now.

There are practical limits to how much speed one can reasonably expect to achieve on HF, and those limitations are not such a problem with AMP, which gets its speed increase by more clever routing and distribution instead of by spreading out to hog up more and more bandwidth, providing a bigger, fatter target for the various flavors of interference and signal degradation to be found on HF.

In fact, AMP delivers faster distribution while taking up less, not more spectrum. From the standpoints of spectral efficiency and rapid distribution of data over a wide area, no point-to-point system can ever hope approach it - ever.

By removing the load represented by messages of any type that are addressed to "All", AMP with make the HF Packet net much faster and more efficient than it presently is. It will introduce a speed increase across the board.

WinLink is locked into a self-imposed "technology jail" and having painted itself into that corner, cannot advance in this way at all. There is no real prospect of WinLink improving it's speed and most especially not its already dismal efficiency in this way at all. - All they can really hope for now is to be able to hog up more and more spectrum until the amateur radio community ( and the FCC ) finally loses patience with them altogether.

That's why the Packet net is cool, and the WinLinkers have hitched themselves to a falling star - one that sux.

[N5PVL]

Here is a bandwidth comparasin between a single HF PACKET signal and PACTOR III.



As you can see, a single PACTOR III station takes up just over five times as much bandwidth as an HF PACKET station does.

If there was only one of each type signal on the air, the above illustration would tell the whole story... Unfortunately this is not the case though, as both HF PACKET and PACTOR III are being used in networks intended to transport store-and-forward type messaging for many locations.

An essential and significant difference between HF PACKET and PACTOR III in this respect is illustrated below:



Here we are looking at a comparasin between five HF PACKET stations and five PACTOR III stations... Note the difference between PACKET's ability to "stack vertically" as opposed to PACTOR III's "horizintal spread".

If there was only five of each type signal on the air, the above illustration would tell the whole story... Unfortunately this is not the case though, as both networks ( HF PACKET and WinLink ) serve many more locations than that.

There is not enough space here to illustrate this, but a dozen or more HF PACKET stations can "stack vertically" and operate with the width of a single signal ( 500 Hz. ) but to have a dozen PACTOR III stations "spread horizontally", more than 31200 Hz. of spectrum would be occupied.

( Some space between the 2600 Hz. wide PACTOR III stations must be there, so they will not interfere with each other. )

Assuming that no more than a dozen HF PACKET stations will tend to "stack up" at a time, then we can take the 31200 Hz. figure for a dozen PACTOR III stations, divide it by the 500 Hz. that a dozen HF PACKET stations take up, and see how much more space PACTOR III takes up than HF PACKET in a general sense...

A network utilizing PACTOR III takes up 62.4 times the spectrum HF PACKET does.

Since both networks carry store and forward type messages that are not time-critical, what we amateurs must ask ourselves is whether the speed increase that PACTOR III offers is really worth taking up 62.4 times as much spectrum, making it necessary to "horizontally spread" them all over the phone bands as opposed to HF PACKET, which works just fine within the narrow automated sub-bands and doesn't crash anybody's QSO.

Is WinLink really worth it?

[AC0EA]

Hey guys, check it out! The guy that runs the USPacket organization is saying that HF packet is better than WinLink! Who'da thunk?

I don't use either, nor do I plan to. I might use the "post long-winded explanations" in my forthcoming essay: "Water - It's Wet!" however!

[N5PVL]

It's true... I generally give my fellow amateurs more credit than I would to non-hams, and do not write to "the lowest common denominator" but instead assume a certain level of technical competence and interest.

The short version is: " Packet is cool - and WinLink sux. " - but saying just that would bring on calls for an explanation anyway.

By the way; It really won't matter whether you use HF Packet or WinLink - when WinLink starts crashing your HF QSO's on phone, PSK31, CW, or what-have-you.

This assumes that you have HF priveleges or plan to someday... #Most hams do.

[AC0EA]

Quote[/b] (N5PVL @ Feb. 02 2006,16:45)]This assumes that you have HF priveleges or plan to someday... Most hams do.

Bwahaha...you may safely assume that a 2x2 call starting with "A" has HF privileges. I'm even working my way up to 20wpm CW so I won't have to endure endless hours of OFs calling me a "low-code Extra." But whatever, man, you gotta get your kicks in somewhere...

Anyhoo, being anti-WinLink and anti-EchoLink is your pet project, so I'll stop poking you about it. Odds are, if I *ever* get a QSO stomped on by a WinLink transmission, I'll switch to a different band, or *horrors*, just turn off the radio and go do something else. Being mad at WinLinkers won't make them go away, and cross-posting this rant (which I believe was originally on your blog) to eHam and QRZ probably won't change too many minds. I just don't care, that's all - maybe we could ask you why you care so much?

[N5PVL]

With vanity calls, it's hard to say who has what priveleges or where they might be located. - I was too lazy to look you up, so I gave a generic answer, applicable to all.

Read the posts about the two "bandwidth segmentation" proposals currently before the FCC and still open to comments from amateurs. - If either one of these goes through, PACTOR III QRM-bots will be given a "green light" to spread out all over the HF bands.

As it is, they are restricted to certain areas and can only do so much damage to other ham's QSO's.

As a digital weenie, I don't spend much time doing phone or CW QSO's, but I look out for all of my fellow hams, as long as they are doing the same. - It is only the self-absorbed "to heck with everyone else" types I have a bone to pick with.

[kb3mng]

Quote[/b] (N5PVL @ Feb. 02 2006,15:59)]A network utilizing PACTOR III takes up 62.4 times as much spectrum as HF PACKET.

To make a really compelling argument, you have to address a few more questions. It seems you have a good grasp of the material, so it should not be difficult to clarify.

How long is that spectrum occupied? Three separate instances of 2 minute pactor sessions is arguably using less of a scarce resource than one instance of a half-hour session containing 6 stations.

Also, what actual throughput would you expect if there are another 11 stations on the frequency with you? I would expect that collisions and backoff would eat up quite a lot of your channel capacity. During backoff, the station isn't transmitting any bits, but from the perspective of any other mode, it is still occupying the frequency. (By analogy, you wouldn't say a CW signal is not occupying the frequency during the space between words.)

It would be especially interesting if you can give actual measured throughputs for the example scenarios you are citing. Otherwise, the best a reader can do is make educated guesses from similar (but not identical) technologies. e.g. When I was in school, the faculty were all happy about Slotted Aloha because it could use 14% of the network capacity. I've never heard of anybody building a Slotted Aloha network, so it isn't necessarily interesting for this discussion.

[N3HGB]

Sorry - I have never tried anything as useless as HF packet. I remember it taking a half hour of retries to download the shortest message from a BBS on 20 meters. Back then the other choice was AMTOR and it worked so much better. You could use the FEC mode for roundtable converations or broadcasts. NAVTEX uses FEC SITOR (same thing as AMTOR) and it works pretty well. The real measure of efficiency would be bandwidth X time per message. So what if packet is narrowband if it takes all day to get a coherent mesasage through. Right now I am loving PSK31 - how much narrower can you get? Why don't you just stick to the real issue - Winlink needs to not spam the whole entire phone band.