New Technology Or Existing Technology

[ky5u]

Digital Bandwidth: How Much is Needed?

In a previous discussion, we tried to explore the question of whether the ARRL Bandwidth Petition went far enough in allowing adequate bandwidth for digital applications. The conversation that followed, we digressed into a discussion of the merits of the proposal itself. This article attempts to refocus discussion on possible digital uses for the future and the amount of bandwidth required. How can we know if a petition is adequate without knowing what is really needed for digital applications?

We will not focus on applications using less than 500Hz bandwidth, as they are well known and have not been an issue on HF in the past. Beyond the need for these narrowband applications to do better with detecting other modes, the technology is mostly non-intrusive and mature.

Wide Band Digital

Using three well known digital signal transfer methods/protocols as examples it can be seen that digital over a 20kHz analog channel in FM yields a sync speed of 9600bps and a throughput of about 4800bps. GSM related data modes (GPRS/Edge) use a 200kHz channel for a 128kbps sync speed and an effective throughput of about 70kbps. 802.11 devices use spread spectrum modulation methods that can consume over 20Mhz for throughputs over 100 megabits per second. These numbers are "ballpark" numbers based on published data by watchers of data services across the country. From a user standpoint, a 9600bps connection can be used for slow email delivery, simple data entry, and text based applications. Internet Access at 9600bps is excruciating.

The 128kbps connect rates of cellular GPRS/Edge allow internet browsing, email delivery, file transfer and more. The "user patience" file size (size of file transfer where user begins to notice the lack of speed compared to a WAN) is about 1.5MB. This is not the case with 802.11 devices that become mere wireless extensions of a WAN (wide area network). Files transfer and email is delivered effortlessly. So with a choice of required bandwidths including 20khz, 200khz, or over 20Mhz, what is the performance factor needed for Amateur Radio?

Stated Purpose

The ARRL states the following in their petition as reason for change:

Quote[/b] ]In order to encourage the implementation of new technologies in the Amateur Radio Service, the rules must be modified to more flexibly accommodate use of such technologies.

...

There is a pronounced trend in the Amateur Service toward digital communications, without necessarily replacing analog modes. It is apparent therefore that both analog and digital modes will be used in the same bands at the same times for the foreseeable future.

...

..this Petition seeks to facilitate and encourage the development, refinement and use of new digital technologies without the regulatory remnants developed at a time when the principal emissions used in the Amateur Radio Service were Morse telegraphy and single- or double-sideband amplitude-modulated telephony.

The ARRL goal to provide for the "implementation of new technologies" is an important distinction. As stated in the previous article, Don Rotolo, N2IRZ, is quoted in an article in the December 2005 issue of CQ Magazine as saying:

Quote[/b] ]N2IRZ: ...a flat bandwidth maximum of 3.5kHz is proposed (with an exception for AM), raising quite a ruckus in the digital community.

...

Considering the service that the emergency response community wants and needs, which is a relatively fast data channel than can span one to three hundred miles, such a narrow bandwidth does not meet the need.

With some "back of the napkin" calculations, Rotolo arrives at the requirement of/for 25kHz to meet the stated needs (at the end of the CQ article). If there is a distinction to be made here, it is that the ARRL seeks to implement new modes, while many members of the "digital community" seem intent on moving existing digital technology to the Ham Bands. One can only guess that the ARRL did not specify higher bandwidths (like 25kHz) because they were counting on more efficient digital modes in the future that could move data in less spectrum.

New Technologies

What's on the horizon for digital development and Amateur Radio? What are the current limitations for data rates in the proposed 3.5kHz requested in the ARRL Petition? What are our goals in developing "new technologies"? What is considered "state of the art" in digital encryption and packing?

Existing Technologies

Do we acknowledge the need for larger bandwidths? With a total of 350kHz on an HF band, what is the implication of assigning up to 200hkz of that spectrum for a single network connection? Are the estimates of 25kHZ made by Mr. Rotolo valid? Should we even be discussing existing technologies on HF with the amount of legacy mode users? Which is more important, legacy mode use or development of HF with existing technology? Does ECOMM need 25kHz applications and why?

Discussion

Rather than bash the ARRL Petition, this article attempts to open discussion on the basic needs of digital services on Amateur Radio. The fundamental question is: Should we be looking for new more efficient digital technologies, or should we be implementing existing digital technologies in Amateur Radio. Balancing the realities of needed bandwidth for existing technologies (considering bandwidth limitations on HF), and the need to break new ground in Amateur Radio (in the digital arena), what is the right thing to do?

Your thoughts are welcomed.

[K9FV]

Are we talking about freqs ABOVE 1gig Hz?? If so, then.... have at it. It might be ok to play around up there, but NOT IN THE HF BANDS!!!! No way, no how.

Ken H>
K9FV

[W3MIV]

Quote[/b] (AG4YO @ Jan. 04 2006,15:18)]Should we be looking for new more efficient digital technologies, or should we be implementing existing digital technologies in Amateur Radio.
OK, I'll bite.

We should clearly be doing both, but the reality is that moving large amounts of information requires lots of bandwidth. So far as I know, nobody has come up with a way around that cumbersome limitation. If you are out there, Scotty, put down the dilithium crystals long enough to give us a hand here.

The problem, in my view, is that far too many people are trying to "justify" ham radio by forcing it into a mold of their own ambition. Events of this past year show clearly that we can make substantial contributions when needed using what we already have and know how to deploy.

The HSMM Committee and its sycophants (all you NCTs go look that up) notwithstanding, we do not need to try to play "military-commo-wannabes."

We can do what we do very well, and it has been demonstrated time and againg to be more than sufficient to meet real needs. We are that Dutch boy with his thumb, and our role has been to provide the stopgap that can save lives and provide some comfort and assistance to strained facilities. But we are not, and should not be seen to be, a fixture for the long-term or the heavy lifting.

[n5rfx]

Charlie,

I certainly think that HF is not the appropriate place to provide real-time Internet access. Using the Internet as a backbone for amateur email is appropriate. When I say email I am talking about non-real-time text messaging, like we have done with AX.25 for years.

As far as the real time QSO, the modes that allow keyboard-to-keyboard communications have not increased the character per second throughput as much as they have improved reliability. In fact the character per second rate on average has decreased. In the past to achieve a high level of text accuracy on HF, ARQ was the only game in town. Now we have FEC modes that can provide accuracy nearly as good as ARQ in a faded environment. Also the ability to copy weak signals has greatly improved with algorithms developed over the years.

When it comes to file transfers, the most common types are image files. These modes have been given the name digital SSTV. They exist in the phone/image subbands because that is where the rules allow them. The one difference between the real time text modes and digital file transfer is that the text modes are authorized only in the RTTY/Data subbands of the 80 through 10-meter bands, and digital file transfer is authorized in the phone/image subbands of the 80 through 10-meter bands. The RTTY/Data subbands have a 300 Baud limit, the phone/image subbands have no Baud rate limit, only a requirement that no non-phone emission shall exceed the bandwidth of a communications quality phone emission of the same modulation type. What ever that means. Most developers appear to have made self imposed bandwidth limits of 2.5 KHz for whatever reason. To increase data throughput, FEC is not as robust, but there is the ability of these modes to ask the sending station for a repeat of the blocks that were received in error.

The point has to be made that all of these advancements were made under the current regulations. I cannot agree that our current rules DO NOT facilitate and encourage the development, refinement and use of new digital technologies. This is simply not true. There is not one modulation scheme appropriate for the HF band that is not authorized.

While there are no unreasonable prohibitions with respect to digital modes, one unreasonable aspect is that the current rules do not allow us to mix data and image in the 80 through 10-meter bands. There have been schemes developed that use digital signals to start synchronization, then switch to an analog image mode (analog fax) that is asynchronous. This is a very fast and accurate way to send images, but amateurs in the U.S. are not authorized to use this method. Of course this is an example of mode mixing that requires a bandwidth of less than 500 Hz and falls outside of your discussion.

Another problem I see with the today’s rules is selective prohibition of mixing analog and digital modes. Analog and digital modes are allowed to mix in the phone/image subband in the 80 through 10-meter bands, but we cannot mix analog and digital modes in the RTTY/Data subbands. As I pointed out above, there are modes that mix analog and digital emissions that have greatly improved the spectral efficiency of sending images, but we in the U.S. are not allowed to use them.

I agree that HF is not the appropriate band for digital emissions greater than 3.5 KHz. Advancements will be made to improve the reliability and throughput of digital modes. I don't see where the current rules will inhibit that progression. What will be inhibited is the participation of U.S. amateurs in general communications advancements where switching back and forth between digital and analog emissions is a requirement.

73,

Mark N5RFX

[W5HTW]

Quote[/b] (W3MIV @ Jan. 08 2006,17:47)][
The problem, in my view, is that far too many people are trying to "justify" ham radio by forcing it into a mold of their own ambition. Events of this past year show clearly that we can make substantial contributions when needed using what we already have and know how to deploy.

The HSMM Committee and its sycophants (all you NCTs go look that up) notwithstanding, we do not need to try to play "military-commo-wannabes."

We can do what we do very well, and it has been demonstrated time and againg to be more than sufficient to meet real needs. We are that Dutch boy with his thumb, and our role has been to provide the stopgap that can save lives and provide some comfort and assistance to strained facilities. But we are not, and should not be seen to be, a fixture for the long-term or the heavy lifting.
Precisely. There is this ever-increasing desire by the military/cop wannabes to pretend they are what they are not. I completely believe we amateur radio operators are capable first-responders, but under current conditions, we do not need to hang around and provide long-term communications. Under future conditions we may not even be needed as first-day/hour responders, as new military and local government technology is coming to a theater near you. Dropped-in vehicles, in the form of Humvees, boats, Jeeps, vans, tracked vehicles, etc., with complete National Guard or DOD commo stations, with VLF to EHF capabilities are on the way, and on the way soon. It isn't a matter of technology; it is a matter of money. We hams can return safely to 20 meter DX. Or if you prefer, shooting skip.

Even now, though, we do not need long term wide bandwidth data handling capabilities for emergencies. We get the job done, as we have ALWAYS done, with quick and efficient voice. Then we get out of the way and let a myriad of professional communications systems take over.

Consequently the reason for digital exploration must be for the fun of it, not be play soldier. We have an army where you can do that. And even get paid for it. To me, that means keep it where it does not wipe out the rest of ham radio. While I still believe ham radio could play a role in the development of new technologies, simply taking what is already available on UHF commercial and government operations, and putting it into the HF ham bands is nothing resembling "development." Moving email via ham radio is not "development." All too often we hams tend to adapt commercial technologies to our hobby and then pretend we have created something.

I am afraid there are enough internet and email enthusiasts, though, their vocal position on turning ham radio into another ISP is going to override the desires of the ones who like radio to be radio, not simply the internet. Many of us have no need for an ISP, but we love the hobby that has been amateur radio for many decades, and that is being turned into the internet.

Ed

[wa4dou]

Why do you think FM is not permitted below about 29 mhz? No digital application is ever going to be permitted to use 25 khz., much less 200 khz. of a 350 khz. wide amateur band in the HF spectrum. Its idiotic to even think about.

[ky5u]

Reality is that there are portions of the "Digital Community" upset because they won't have a 25kHz bandwidth segment on each HF band (and are upset because the ARRL let AMers have 9kHz). It takes us back to a simple question: What do they need 25kHz for? Can anyone answer this? The CQ article by N2IRZ simply alludes to the need for higher data rates for Ecomm.

If there is a valid need, shouldn't we consider why it is not in the ARRL proposal? My thought was that the ARRL was looking to the future and new modes not yet discovered that could squeeze great performance out of 3.5kHz rather than providing bandwidth for existing network hardware needing bandwidths above 3.5kHz.

Are we ready to agree that data modes using more than 3.5kHz don't belong on most of HF? It doesn't sound like we are.

[K9FV]

Perhaps I should make a bit more clear my feelings - HF is NOT for broadband internet access - not even narrowband access. I do like e-mail via my HF rig - winlink is very good at this.... as long as there are not too many mailbox stations on the band, and the bandwidth is not excessive. I don't mind it being a bit on the slow side - that will keep too many folks from trying to use winlink as their "main" email service. BUT for us boaters and RV'ers who have no other reasonable (and how does one define "reasonable"??) access, it is nice.

No, sailmail does NOT provide the same weather info that is available on winlink.

Ken H>

[ab9lz]

""The fundamental question is: Should we be looking for new more efficient digital technologies, or should we be implementing existing digital technologies in Amateur Radio.""

...the later, many of the existing protocols are downright ancient. Hams have made great strides in cooking up protocols that are de-rigeur for the problem at hand, witness psk31, a super effcient (bandwith wise) implementation of IRC (chat). And PUA43 for the amazing weak signal EME work some folks are doing. (to name a few). A 3.2 khz bandwidth for HF raises the bar, anything more is a criminal waste. Folks WILL step up to meet the challenge.

73.
Mark.

[kb3mng]

Broadband over HF is nearly impossible, so we don't need to debate it.


Amateurs have a total of 1400 khz below 15 mhz. Assume Pactor III is a reasonably efficient protocol to use as a model. It achieves 2722.1 bps in 2.4 khz under ideal conditions. 1400 / 2.4 * 2722.1 = 1.588 mbps, or about the capacity of a single T1, or two mid-range DSL (768kbps), or six low-end dsl (256 kbps).

We might reasonably assume that somebody might want to use the amateur HF bands for something other than broadband. http://www.qrz.com/iB_html/non-cgi/emoticons/smile.gif We also might reasonably assume that conditions are rarely ideal. Consequently, the possible number of "broadband" connections is tiny.


The general public seems to view the internet as web, email, and file sharing. I think this is what people are thinking of when we talk about the idea of "internet over HF". (I doubt many readers are thinking about whether it is a good idea to run NTP over HF.)

This "general public" view of the internet mostly requires broadband capacities. Consequently, I doubt that "internet over HF" (using the general public view of "internet") can be made usable. The bandwidth just isn't there.


b.t.w. Winlink is not the internet. It is a separate email system that uses it's own protocols on HF. It just happens that after your mail reaches a Winlink node, Winlink can forward it to an Internet machine. It is important to keep this distinction in mind, because it makes a lot of difference to the implications you can assume from it. The fact that Winlink may handle email messages does not imply web browsing, for example.


I'm pointing this out now, because Charlie asks the right question (the one that should be asked before proposing new bandwidth limits), and it seems only reasonable that we try not to get sidetracked by non-issues.

[w1jeq]

Missouri ARES has the right idea on EM-COMM,
we are setting up AX-25 packet network using 2m ports on
nodes scattered across the state. the nodes then talk to each
other and delever traffic from node to node on a 6m freq. Hf is
left for voice and cw and some winlink. my view is that HF
does long distance voice and cw and narrowband digital y well.
note that for optimum throughput on any digital mode requires
strong non fading signals, cw and voice can get through at
much lower signal levels and conditions. vhf and up is the place
where digital shines and hf is for long-haul comms. another note
is that voice can have fairly quick throughput. i have tried a
popular voice to text app and found with few errors (<5) i could
send 10 NTS format radiograms with noise and levels fading
faster then i could ever type them. and cw is one of the most
reliable and quickest ways to send traffic. why? the bandwidth
can be <500hz and as long as you hear dit or dah you can
tell whats being sent.
high speed digital is not appriate on our hf spectrum, why
clutter up our long haul capable spectrum with modes that
may not being able to take advantage of that becasue they
require too much signal then amateurs can garuntee on a
given path.
we need to focus on things such as making our uhf+ spectrum
more asscesable for amateurs (more equipment) protecting our
spectrum (BPL), cleaning up the airwaves (unlicensed
operators on our bands and those that don't follow the rules)
and helping to gain spectrum for users in other ITU regions
and other possible long comms on LF)

More then all lets make sure that everyone erregards to
what facet of the hobby they presue, has spectrum to enjoy
what they like, i couldn't provide any assistance to katina efforts
but why should i give up my access to radio. thier is enough
spectrum to do all that.

Conclusion: Use Wisely Protect what we have Promote
Allow other hams access Preserve for the future

[kh6ty]

Quote[/b] (kb3mng @ Jan. 09 2006,18:59)]Broadband over HF is nearly impossible, so we don't need to debate it.


Amateurs have a total of 1400 khz below 15 mhz. #Assume Pactor III is a reasonably efficient protocol to use as a model. #It achieves 2722.1 bps in 2.4 khz under ideal conditions. #1400 / 2.4 * 2722.1 = 1.588 mbps, or about the capacity of a single T1, or two mid-range DSL (768kbps), #or six low-end dsl (256 kbps).

We might reasonably assume that somebody might want to use the amateur HF bands for something other than broadband. http://www.qrz.com/iB_html/non-cgi/emoticons/smile.gif We also might reasonably assume that conditions are rarely ideal. #Consequently, the possible number of "broadband" connections is tiny.



I agree!

The most advanced error-free high-speed mode currently available for multimedia is Pactor-III, which should blaze along at 225 characters per second, but, according to Winlink statistics, Pactor-III only achieves an average of 15 characters per second on HF in actual use, but takes up five times the bandwidth of Pactor-II, which should blaze along at 50 characters per second, but only averages 10 characters per second in actual use on HF, taking up a bandwidth of 500 Hz, all compared to a much more efficient ARQ mode, PSKmail, which achieves 5 characters per second in a bandwidth of only 100 Hz. However, PSKmail is obviously not a "high-speed" system!

So, Pactor-III uses up a bandwidth that is 25 times as much as PSKmail, but only achieves a speed advantage of 3 times over PSKmail.

"High-speed multimedia or broadband" on HF? The conditions encountered on HF, such as QRM, QRN, and QSB require so many repeats that even Pactor-III and Pactor-II cannot achieve but a small fraction of their potential speed. Data transmission requires error-free reception, and so far, nobody to date has developed a system at high speed that can do that on HF, regardless of the bandwidth consumed. Hams in casual communication can just ignore an occasional error or let their minds make corrections based on context, but in messaging, one wrong number received in a phone number renders the entire message useless.

Even increasing the baud rate does not help, and instead of developing digital modes with a baud rate higher than 300, the latest trend of mode designers is to keep the baud rate below 100 and use multiple tones, so the current limit on HF of 300 baud is entirely adequate, but not for reasons of restricting emitted bandwidth by limiting the baud rate, but because of unavoidable HF propagation characteristics. This point is conveniently absent from the ARRL petition in their attempt to sell us on the need to change the HF bands from being used for communications between people to being used primarily for multimedia data transmission, mostly in delayed time, in the nature of an answering machine.

Before any regulations are forced upon those of us who use the HF bands for communication, instead of email "messaging" let's see a workable system under an STA to prove high speed data on HF can be made to work in the real world. That is the safeguard provided by the current FCC approval system and it should not be abandoned.

Skip KH6TY

[w6em]

How much bandwidth for that new, yet-to-be-conceived digital mode?

Enough speculation.

Apparently, the ARRL thinks it knows.

Why change anything? #Certainly the ARRL knows that experimentation can be done at low enough power levels to preclude the necessity of a license. #And, if more power is needed, an experimental license can be had on just about any band or frequency, with some exceptions.

The ARRL must know something we don't: Perhaps a manufacturer has already conceived a new "whiz-bang" digital mode and they simply *must*, for whatever influence the dollar will buy, have the flexibility to market it.

Look how long, for example, it took Kenwood to repeatedly ask the FCC to allow Skycommand &reg;.

Add speculation to speculation. #I say, if the need exists, (for wide-band accomodation) reveal the technology. #If it doesn't, whistle Dixie.

BTW, welcome to the Heart of Dixie. #I assume you are now in the Birmingham area as I am.

Lee
W6EM
Leeds, AL

[KC6MYO]

PACTOR III is not an efficient use of bandwidth giving only about 1 bit per Hz.
802.11 uses 20MHz bandwidth channels on the 2.4GHz band
801.11b has a maximum throughput of 11Mbps
#that gives an efficiency of about 0.5 bits per Hz
802.11g has a maximun throughput of 54Mbps
#that gives an efficiency of about 2.5 bits per Hz
802.11n now in development has a maximun throughput of 384-512Mbps
#that gives an efficiency of about 25 bits per Hz

What does that mean in practical bandwidth terms?

100Hz # x #25 bits = # #2.5kbps #good enough for text modes
500Hz # x #25 bits = #12.5kbps
2.5kHz #x #25 bits = #62.5kbps #better then dialup speed
25kHz # x #25 bits = 625kbps #DSL speed

For the sake of argument 6MHz (ATV) bandwidth x 25 bits = #150Mbps
That's faster than Fast Ethernet!

What we need is more efficient use of digital bandwidth.

In rare cases where 25KHz is needed, a STA should be used, rather than a rule change. But usually the 62.5kbps of 2.5kHz would be enough, and allow so many more to use the available spectrum.

[ky5u]

Quote[/b] (K9FV @ Jan. 09 2006,13:03)]Perhaps I should make a bit more clear my feelings - HF is NOT for broadband internet access - not even narrowband access. I do like e-mail via my HF rig - winlink is very good at this.... as long as there are not too many mailbox stations on the band, and the bandwidth is not excessive. I don't mind it being a bit on the slow side - that will keep too many folks from trying to use winlink as their "main" email service. BUT for us boaters and RV'ers who have no other reasonable (and how does one define "reasonable"??) access, it is nice.

No, sailmail does NOT provide the same weather info that is available on winlink.

Ken H>
I understood your position, Ken, but there are Amateurs out there who don't agree. They want higher data speeds on HF and are proposing something in the order of 25kHz as what they need. To you and to MNG's point, I don't see concern from them that they might be asking for more bandwidth at the expense of other users, or at the least they believe the digital uses supercede other users in importance for our future. Not editorializing here, just stating what I see.

But the ARRL plan does not call for 25kHz bandwidth on most HF bands, right? But, the other petition given a RM number by the FCC from the "Think Tank" has no bandwidth limitations that I can find at all as well as no mode restrictions. It would allow a 300kHz signal provided nobody else was using the bandwidth at the time. Again, no editorilization, just emphasis to make the point that every Amateur needs to have an opinion on this and not bury their head.

So does ANYONE here make the case that we should have more than 3.5kHz for data signals on most HF bands?

[K9FV]

I just checked your bio Charles and you are just "next door" to me here in Mobile (temp location). Kudos on your shack and career occomplishments... maybe we can work CW some night since you say you'll "slow down" to my speed {grinning}

You and I agree on so many things. All I really want is for the rules to allow experimentation for the folks who really enjoy - BUT let's keep the modes narrow on the HF bands - 3K is good. AND we don't need any "automatic robots" anywhere on HF!!! The "semi-automation robots" (winlink) can be used, BUT they really do need to be confined to a set band segments. I like winlink and use it - don't wish to see it go away, but I do feel winlink needs some "guidance" on limits. The winlink group does make a real effort to assure the users do check freqs BEFORE transmitting.

I think perhaps a section of the band for the digital modes would be good - it's just too easy for someone using the computer modes to slip up and interfere with an existing QSO.

EVERYTHING on the HF bands should be fairly easy to monitor - hey, "copying the mail" makes for good background noise sometimes when doing project work.

Up in the gigahz range, perhaps some encription would be ok??? Not sure about this, but if it's needed for some mode, then I'm sure not knowledgeable enough to make a rule there. There is so much room up there, maybe even internet surfing?? Heck, I don't know. My dream is for everyone to be polite and just get along - Gentlemen's agreement type of thing.... BUT that requires Gentlemen. I'll try to be a gentleman, and everyone else try also and I'll just bet it would work.

73 de Ken H>
K9FV

[W3MIV]

Quote[/b] (AG4YO @ Jan. 09 2006,14:54)]If there is a valid need, shouldn't we consider why it is not in the ARRL proposal? #My thought was that the ARRL was looking to the future and new modes not yet discovered that could squeeze great performance out of 3.5kHz rather than providing bandwidth for existing network hardware needing bandwidths above 3.5kHz.
Just because a committee of pipe dreamers has proposed some new Wunderwaffe for HF does not mean that there is much (or even any) support for it on the BOD. These guys have been going where no man has gone before for a long time now, and they are long overdue to check in.

You will, however, find some highly disgruntled "HiFi SSB" types who feel strongly abused by the 3.5kHz bandwidth maximum, feeling that all of the basso buffo tones in their mellifluous tonsils must receive equal time. I believe that this was a primary reason for the move from the original 3kHz of the proposal put our for comment.

[NY7Q]

WHAT THE ARRL IS PROPOSING IS JUST ONE LINE THIS SIDE OF BEING PURE COW MANURE.

[ab0wr]

Quote[/b] (KC6MYO @ Jan. 10 2006,08:12)]PACTOR III is not an efficient use of bandwidth giving only about 1 bit per Hz.
802.11 uses 20MHz bandwidth channels on the 2.4GHz band
801.11b has a maximum throughput of 11Mbps
#that gives an efficiency of about 0.5 bits per Hz
802.11g has a maximun throughput of 54Mbps
#that gives an efficiency of about 2.5 bits per Hz
802.11n now in development has a maximun throughput of 384-512Mbps
#that gives an efficiency of about 25 bits per Hz

What does that mean in practical bandwidth terms?

100Hz # x #25 bits = # #2.5kbps #good enough for text modes
500Hz # x #25 bits = #12.5kbps
2.5kHz #x #25 bits = #62.5kbps #better then dialup speed
25kHz # x #25 bits = 625kbps #DSL speed

For the sake of argument 6MHz (ATV) bandwidth x 25 bits = #150Mbps
That's faster than Fast Ethernet!

What we need is more efficient use of digital bandwidth.

In rare cases where 25KHz is needed, a STA should be used, rather than a rule change. But usually the 62.5kbps of 2.5kHz would be enough, and allow so many more to use the available spectrum.
Hmmmm....

How do you propose to get 25bits/hz on a noisy, fading circuit like you would find on HF? What modulation scheme will you use?

tim ab0wr

[wb5yiw]

Ok, I admit I don't know a whole heck of a lot about all these digital modes, but from what I can glean from this discussion, it seems that most of the concerned parties want access to email from about anywhere, and want to do that on HF, right? Well, 99 percent of the email I get is text. A few pics and cartoons, but by far the majority is text. While I admit that I'm no programmer, it shouldn't be that difficult to write a program that would access your text only email from the internet, turn the text into high speed CW and send it. Presto, no bandwidth problem and if you're a real CW man (or woman) you wouldn't even need a computer at your end to do it. You could even send a reply with nothing more than your radio and key if you write the program for your robot at home correctly. Or get really adventurous, and have the computer record a sound file of your reply in your own voice and send that file via email. You could even have the robot play it back to make sure it recorded properly before sending it. Sure, some of this might require some rule changes, but if we're changing them anyway, why not?

If you have need for some sort of security, (don't want others reading your mail), don't send it over radio. Use SSTV for the pics of Little Johnny catching his first fish (or your daily Dilbert cartoon), or wait till you get back to civilization and connect to a wireless hotspot and down load the pics then.

Want to send pics or video out from disaster area? Wait a few minutes, CNN or some other news source with lots of money will be there shortly. They are far better equipped to do that than we are and they don't eat up half of a band at a time doing it.



73's

Bryan

[ky5u]

Quote[/b] (ab0wr @ Jan. 10 2006,21:08)]Quote[/b] (KC6MYO @ Jan. 10 2006,08:12)]PACTOR III is not an efficient use of bandwidth giving only about 1 bit per Hz.
802.11 uses 20MHz bandwidth channels on the 2.4GHz band
801.11b has a maximum throughput of 11Mbps
that gives an efficiency of about 0.5 bits per Hz
802.11g has a maximun throughput of 54Mbps
that gives an efficiency of about 2.5 bits per Hz
802.11n now in development has a maximun throughput of 384-512Mbps
that gives an efficiency of about 25 bits per Hz

What does that mean in practical bandwidth terms?

100Hz x 25 bits = 2.5kbps good enough for text modes
500Hz x 25 bits = 12.5kbps
2.5kHz x 25 bits = 62.5kbps better then dialup speed
25kHz x 25 bits = 625kbps DSL speed

For the sake of argument 6MHz (ATV) bandwidth x 25 bits = 150Mbps
That's faster than Fast Ethernet!

What we need is more efficient use of digital bandwidth.

In rare cases where 25KHz is needed, a STA should be used, rather than a rule change. But usually the 62.5kbps of 2.5kHz would be enough, and allow so many more to use the available spectrum.
Hmmmm....

How do you propose to get 25bits/hz on a noisy, fading circuit like you would find on HF? What modulation scheme will you use?

tim ab0wr
Yep, his numbers are wrong. Just noticed the error. Using his formula, D-Star at 150kHz should have a data rate of over 6mbps. It's actual rate is 128kbps.

I believe the correct figure would be 2 bps per Hz. So for this table, the correct figures should be:

100Hz=200bps
500hz=1.0kbps
2.5kHz=5.0kbps
25khz=50.0kbps
200kHz=400kbps

These are optimistic numbers. In reality I see 20kHz yeild about 9.6kbps and 200kHz about 128kbps. Different modulation schemes and compression will move the numbers around. But no way are you going to get 625kbps from a 25kHz connection without a leap in technology.

[ab0wr]

Quote[/b] (AG4YO @ Jan. 12 2006,13:39)]Quote[/b] (ab0wr @ Jan. 10 2006,21:08)]Quote[/b] (KC6MYO @ Jan. 10 2006,08:12)]PACTOR III is not an efficient use of bandwidth giving only about 1 bit per Hz.
802.11 uses 20MHz bandwidth channels on the 2.4GHz band
801.11b has a maximum throughput of 11Mbps
#that gives an efficiency of about 0.5 bits per Hz
802.11g has a maximun throughput of 54Mbps
#that gives an efficiency of about 2.5 bits per Hz
802.11n now in development has a maximun throughput of 384-512Mbps
#that gives an efficiency of about 25 bits per Hz

What does that mean in practical bandwidth terms?

100Hz # x #25 bits = # #2.5kbps #good enough for text modes
500Hz # x #25 bits = #12.5kbps
2.5kHz #x #25 bits = #62.5kbps #better then dialup speed
25kHz # x #25 bits = 625kbps #DSL speed

For the sake of argument 6MHz (ATV) bandwidth x 25 bits = #150Mbps
That's faster than Fast Ethernet!

What we need is more efficient use of digital bandwidth.

In rare cases where 25KHz is needed, a STA should be used, rather than a rule change. But usually the 62.5kbps of 2.5kHz would be enough, and allow so many more to use the available spectrum.
Hmmmm....

How do you propose to get 25bits/hz on a noisy, fading circuit like you would find on HF? What modulation scheme will you use?

tim ab0wr
Yep, his numbers are wrong. #Just noticed the error. #Using his formula, D-Star at 150kHz should have a data rate of over 6mbps. It's actual rate is 128kbps.

I believe the correct figure would be 2 bps per Hz. So for this table, the correct figures should be:

100Hz=200bps
500hz=1.0kbps
2.5kHz=5.0kbps
25khz=50.0kbps
200kHz=400kbps

These are optimistic numbers. In reality I see 20kHz yeild about 9.6kbps and 200kHz about 128kbps. #Different modulation schemes and compression will move the numbers around. #But no way are you going to get 625kbps from a 25kHz connection without a leap in technology.
Most of the 802.11n stuff I have seen either require 40Mhz channels or use of two separate 20Mhz channels. It will also only provide about 100Mbs throughput. That's actually 100E**6/40E**6 or 2.5bits/hz.

That's pretty close to your estimate of 2bits/hz. And I agree that is an optimistic estimate, especially on noisy HF channels.

It's just one more piece of evidence of how far the digital bigots will go in trying to convice people that "digital is the coming thing"!! Tell the "big lie" often enough and people will begin to believe it.

tim ab0wr

[ky5u]

Tim,

Intrestingly enough the ICOM D-Star ID-1 radio shows 150kHz bandwidth for HS Data at a rate of 128kbps. Would be interested to see tests that support that. In reality I see 128kbps at 200kHz bandwidth (just barely).

I know it depends on software, compression, etc.... but it looks fishy.

[kb3mng]

Shannon's law describes the maximum information you can transmit in a particular bandwidth:

C = W log2(1 + S/N )

C is bits per second
W is bandwidth in hertz
S is signal power
N is noise power
(notice that S/N is NOT the SNR in db - it is the ratio of the powers)

This is the form that applies to additive gaussian white noise. (There is an integral you can use for some other noise distributions.)

This is a theoretical maximum, but it doesn't tell us how to construct an error correction scheme that achieves this efficiency. If you do a web search, you'll find claims that an encoding that reaches 50% of Shannon's limit is about as good as it gets. That used to be true.

There are now two classes of error correcting codes that closely approach Shannon's limit. They are "Turbo Codes" and "LDPC" (Low Density Parity Check). Both require some pretty intense computation. I have not heard of any amateur applications of either.

But what is the theoretical limit?

2^(C/W)-1 = S/N

For 2 bps/Hz, I get S/N=4 (6 db)
For 10 bps/Hz, I get S/N=1024 (30 db)
For 25 bps/Hz, I get S/N=33554432 (75 db)

As you can see, it is theoretically possible to get 25 bps/Hz if you have a big enough transmitter... http://www.qrz.com/iB_html/non-cgi/emoticons/smile.gif

[ab0wr]

Quote[/b] (kb3mng @ Jan. 13 2006,11:06)]Shannon's law describes the maximum information you can transmit in a particular bandwidth:

C = W log2(1 + S/N )

C is bits per second
W is bandwidth in hertz
S is signal power
N is noise power
(notice that S/N is NOT the SNR in db - it is the ratio of the powers)

This is the form that applies to additive gaussian white noise. #(There is an integral you can use for some other noise distributions.)

This is a theoretical maximum, but it doesn't tell us how to construct an error correction scheme that achieves this efficiency. #If you do a web search, you'll find claims that an encoding that reaches 50% of Shannon's #limit is about as good as it gets. #That used to be true.

There are now two classes of error correcting codes that closely approach Shannon's limit. #They are "Turbo Codes" and "LDPC" (Low Density Parity Check). #Both require some pretty intense computation. #I have not heard of any amateur applications of either.

But what is the theoretical limit?

2^(C/W)-1 = S/N

For 2 bps/Hz, I get S/N=4 (6 db)
For 10 bps/Hz, I get S/N=1024 (30 db)
For 25 bps/Hz, I get S/N=33554432 (75 db)

As you can see, it is theoretically possible to get 25 bps/Hz if you have a big enough transmitter... http://www.qrz.com/iB_html/non-cgi/emoticons/smile.gif
Yep, you hit it on the head.

A couple of notes on this.

S is the signal power that is received during the transmission.
N is the noise power contamination that is received during the transmission.

So the total power received is Signal Power plus Noise Power.

If you assume a load resistance of 1 ohm then the rms value of the received signal is sqrt(S + N). The rms value of the noise power is sqrt(N).

An input signal variation less than sqrt(N) will not be able to be distinguished.

So, the number of different levels that can be distinguished without error is # sqrt(S+N)/sqrt(N)

With enough difference between the noise power level and the signal power level you can send increasing levels of information per transmitted pulse.

On HF, however, especially with the limitations of amateur equipment, 2 bps/Hz would be a lot better to plan on than 25bps/Hz!

Shannon's law gives you a theoretical error-free channel capacity. This should be reachable without the crutch of error correcting codes. If it is not then it is because of impairments that are not being accounted for in the noise power measurement. Since error coding supplants actual information bits, the use of error codes will actually prevent ever actually reaching Shannon's limit. That doesn't mean they aren't useful. They provide a "leapfrog" for fixing sequences where variations in the sqrt(N) kept the receiving equipment from being able to differentiate between information carrying levels in the received signal.

tim ab0wr

[ky5u]

Thanks, Tim. I spent the last almost 15 years dealing with digital voice and data from vendors who have their data testing (and heads) in a vacuum.

[K2PI]

This seems too much like recreating an 802.11x type wide area network, and it strikes me as an inappropriate use of the spectrum we are allocated. #I've always suspected that many of our newer Hams don't appreciate that our hobby is fundamentally different than the Internet, and that it is something to be cherished for that very reason. #I'm not saying this is the ARRL's goal, but I hope their idea isn't to meld the two interests in the hopes of attracting more new Hams.

I sincerely hope not; I can't imagine what type of changes that would bring, but not for the better if the League's track record is any indication.

Then again, maybe they can bring some spelling and grammar checker software along with them, because I gotta tell ya, there is some real need for remedial english classes judging from these internet forums.

73.

Harv
K2PI / M0DUO http://www.qrz.com/iB_html/non-cgi/emoticons/biggrin.gif

[NL7W]

All,

Given the relatively narrow bandwidths and low signal-to-noise ratios available on our HF amateur radio bands, doesn't the Shannon-Hartley theorem severely limit the maximum throughput to just tens of kilobits per second?

One can only squeeze so many bits per hertz out of a limited number of kilohertz on HF bands.

Discussion?

73.

[NL7W]

Quote[/b] (ab0wr @ Jan. 12 2006,14:03)]Quote[/b] (AG4YO @ Jan. 12 2006,13:39)]Quote[/b] (ab0wr @ Jan. 10 2006,21:08)]Quote[/b] (KC6MYO @ Jan. 10 2006,08:12)]PACTOR III is not an efficient use of bandwidth giving only about 1 bit per Hz.
802.11 uses 20MHz bandwidth channels on the 2.4GHz band
801.11b has a maximum throughput of 11Mbps
that gives an efficiency of about 0.5 bits per Hz
802.11g has a maximun throughput of 54Mbps
that gives an efficiency of about 2.5 bits per Hz
802.11n now in development has a maximun throughput of 384-512Mbps
that gives an efficiency of about 25 bits per Hz

What does that mean in practical bandwidth terms?

100Hz x 25 bits = 2.5kbps good enough for text modes
500Hz x 25 bits = 12.5kbps
2.5kHz x 25 bits = 62.5kbps better then dialup speed
25kHz x 25 bits = 625kbps DSL speed

For the sake of argument 6MHz (ATV) bandwidth x 25 bits = 150Mbps
That's faster than Fast Ethernet!

What we need is more efficient use of digital bandwidth.

In rare cases where 25KHz is needed, a STA should be used, rather than a rule change. But usually the 62.5kbps of 2.5kHz would be enough, and allow so many more to use the available spectrum.
Hmmmm....

How do you propose to get 25bits/hz on a noisy, fading circuit like you would find on HF? What modulation scheme will you use?

tim ab0wr
Yep, his numbers are wrong. Just noticed the error. Using his formula, D-Star at 150kHz should have a data rate of over 6mbps. It's actual rate is 128kbps.

I believe the correct figure would be 2 bps per Hz. So for this table, the correct figures should be:

100Hz=200bps
500hz=1.0kbps
2.5kHz=5.0kbps
25khz=50.0kbps
200kHz=400kbps

These are optimistic numbers. In reality I see 20kHz yeild about 9.6kbps and 200kHz about 128kbps. Different modulation schemes and compression will move the numbers around. But no way are you going to get 625kbps from a 25kHz connection without a leap in technology.
Most of the 802.11n stuff I have seen either require 40Mhz channels or use of two separate 20Mhz channels. It will also only provide about 100Mbs throughput. That's actually 100E**6/40E**6 or 2.5bits/hz.

That's pretty close to your estimate of 2bits/hz. And I agree that is an optimistic estimate, especially on noisy HF channels.

It's just one more piece of evidence of how far the digital bigots will go in trying to convice people that "digital is the coming thing"!! Tell the "big lie" often enough and people will begin to believe it.

tim ab0wr
AB0WR and KB3MNG did hit on something here.

I didn't delve into the ealier posts as carefully as I should have, but I do believe Tim's assessment of 2.5 kHz of bandwidth yielding about 5 kbps is awfully close to being on-the-money. With selective fading and the usual propagation traits found on less-than-optimal HF circuits (which occur most of the time), I don't believe wider bandwidths could be efficient use of our precious spectrum -- considering today's availble Internet Protocol (IP) transport alternatives in most areas of the world.

The means of efficiently and cost-effectively transporting IP voice, video, and data to most anywhere in the world exists today – including the most remote places in Alaska, like St. Paul Island, which is in the middle of the Bering Sea between the Russian far east and the west coast of Alaska (I happen to have a government IP over satellite connectivity project happening out there). IP connectivity via the Public Switched Telephone Network (PSTN), satellite IP service providers, microwave circuits, and upcoming high-speed wireless means such as 802.16 and 802.20, offer far greater capabilities than any unreliable and constrained HF circuit.

I’m sure its fun to push the state-of-the-art regarding constrained digital data over HF radios. But let’s face it; digital via HF circuits is not conducive to the speeds of any of today’s transport means mentioned above.

Now, why is it important for digital data folks to lay claim to massive amounts of bandwidth on Amateur Radio?

Please enlighten me.

73 de Steve, NL7W

[ab0wr]

Quote[/b] ]
nl7w:
"Now, why is it important for digital data folks to lay claim to massive amounts of bandwidth on Amateur Radio?"




It isn't important except for a few. Those few see the saving grace of amateur radio as being a "free" telecom provider for emergency agencies, both governmental and non-governmental organizations. This will somehow "bless" amateur radio and prevent its demise.

The problem is that amateur radio was never meant to be a free telecom provider, the HF bands don't have enough bandwidth to provide telecom service, and the assumption that amateur radio is somehow dying never seems to quite pan out.

The present view that the ARRL has somehow been deluded into believing is that "digital" will *save the day*. Tie amateur radio and computers together and the young people will just absolutely flock to amateur radio (and grow the coffers at the ARRL!).

It's a bunch of malarky. My kids don't care at all how their cell phone works or their computer works, but they use both to communicate -- because they fill a *need*. Amateur radio doesn't fill a *need* for them and they could care less about it.

Amateur radio filled a *need* for me. It gave me something my dig into to satisfy my curiousity and it gave me a way to communicate with people outside the little circle I knew in the rural area I grew up in.

Today the kids with that *curiousity drive* find things like genetic engineering, robots, and on-line gaming more interesting. They can communicate with members of their "guilds" inside on-line games, those members can come from anywhere.

What amateur radio needs to find is a way to make it fill a *need* for young people. I still think that is going to turn out to be some kind of synergistic evolution of robots and ham radio. It's probably going to involve VHF/UHF links, not HF links.

It's a shame that we still have a bunch of old fuddy-duddy's stuck in the 1990's thinking ham radio and the internet is going to be a saviour of anything. They are almost six years behind the curve and falling further behind every day.

73,

tim ab0wr