| |
VE2HAR > MT63 11.03.05 21:04l 289 Lines 13965 Bytes #-7344 (0) @ WW
BID : 6464SENTTO
Read: GUEST
Subj: Re: [MT63] 20 kHz wide Digital Proposal (Was: ARRL MUST GET
Path: DB0FHN<DB0THA<DB0ERF<DB0ROF<DB0ACH<DB0MKA<DB0LJ<DB0RES<ON0AR<VA2HAR<
VE2HAR
Sent: 050311/1807z @:VE2HAR.#MTL.QC.CAN.NOAM Laval #:41369 $:6464sentto
> > > Walt said:
> > > You might see 56KBPs data but it would only be on-the-air for a short time.
> > > The "channel" would be shared.
> >
> > Are you suggesting that a TDMA system be used to share a 20KHz wide channel
> > at HF? How would you allocate timeslots? Would you just use CS/MA? I'm not
> > even sure where to begin to explain how poor of an idea that would be.
>
> Well I'm not sure either but that can be a goal.
>
> I suppose we could use a GPS clock. I do know that some hams are working on GPS
> sync. and I think that perhaps the SDR-1000 radio by FlexRadio uses this.
So, to TDMA a channel that propogates around the planet--even with everyone in
perfect GPS sync--you still have signal propogation time between people being
able to transmit. So, every 'turnaround' of the channel costs you as long as
it takes for a signal to get all of the way around the planet--bouncing up and
down as it does so. Even then, you're left with the problem of who arbitrates
timeslots and how do you ask for one?
What seems to make more sense is a bunch of smaller channels feeding into a
set of BBS like systems which could coordinate message flow. The BBSs would
need to have a different set of channels to pass traffic amongst themselves.
TX/RX could be on different bands so that it could be simultanious, even.
Ad hoc networks are a nice idea, but they just don't scale. A well thought
out system will beat them every time.
> > But, until there is a *need*, why do we 'need' to do this?
>
> Well different strokes for different folks...I want to be able to receive a
> message from stations in the middle of a hurricane who are running HF on battery
> power and a NVIS antenna laying on the ground or in close proximity to the
> ground. Even CW can't do this...you need a QRO signal of more than 500 watts to
> do this. Or, a mode that will give you 10 db of gain with no increase in noise.
> So this is where modes like MSFK16 and MT63/Olivia come into play.
I'm curious now, you say NVIS doesn't work with less than 500W? Could you
explain why CW doesn't work here? NVIS should be a good propogation channel
as there is only once bounce off of the ioniosphere to cause doppler.
> > What we have here is a proposal by a small group of people for a large chunk of
> > an internationally shared (and contested) resource.
>
> I keep hearing this but to tell you the truth, I talk to the EComms hams in the
> UK and other parts of the EU and they are thinking the same way...more/faster
> throughput and more robust signals and realize that you cna't do this with a 3
> KHz bandpass.
EComms hams in Europe? The last time we had this 'need an international standard
for HF EComms' discussion, the european hams looked at us funny because in many
countries in europe, hams are expressly prohibited from EComms. Has that changed?
> > Also, your throughput vs signal level vs bandwidth thinking needs to be clarified.
>
> Yes, we need a standard.
>
> I have proposed that the minimum throughput of 200 WPM, achieved at a SNR of -5
> dB on a poor CCIR channel and within a 2.7 KHz bandpass. The first GOAL should
> be 400-800 WPM at a -5 dB SNR on a poor CCIR channel within a 6 KHz bandpass.
> The second goal should be 800-1200 WPM at a -5 dB SNR on a poor CCIR channel
> within an 8 KHz bandpass.
Okay, we're back to the problem that I was mentioning. Those channels aren't
similar. You have a signal at the same level VS noise in changing bandwidth.
Either you're cranking up TX power or you're looking at a path that is shorter.
> Currently with the types of signal detection used, a ~45 baud modulated signal
> is about all you can deal with due to intersymbol interference...remembering
> that with each additional "hop" and propagating through more than one
> ionospheric layer, your intersymbol interference increases. Part of the
> solution is controlling you radiation pattern. If you want to talk out to 600
> or so miles, make sure that you antenna's take-off angle is between 45 and 89
> deg. I see lots of success on MT63 betweem the Central U.S. and Europe from
> stations using beams aimed correctly...they are creating the fewest hops between
> stations. Stations using other antennas may find that they don't do as well as
> neighboring stations using beams.
Quite true. The less of the 'sky' you illuminate means you're taking a smaller
'sample' of the ioniosphere and that can be expected to be more uniform than
a larger chunk.
So, that leads to the question: are you looking for a mode that will require
staitons to be using directional beams?
> > This leads to a few design decisions:
> > 1) baud rate should be <300 and >30
> > 2) Fewer carriers at a higher baud rate beats more carriers at slower buad rate
> > 3) More carriers are more resistant to fading than fewer carriers
> > 4) TX power is *finite*
> > 5) Spectrum is scarce
>
> I recently attended a end of project briefing at Southwest Research Institute
> which indicated that current data detection methods of HF propagated PSK signals
> limit the transmitted baud rate fo no more than 30 Baud to detect a clean bit
> constallation. However, the techniques which they have developed allow for
> error free detection of sifnals and a good constallations with transmitted baud
> rates of over 100 baud. The "kicker" here is that for each transmit time
> period, there is a detection time of 4 times the transmit time before received
> data signals become usable.
The limitations for baud rate are for 'dumb' receivers. Several times, it has
been proposed (TAPR HFSIG mailing list) that some kind of adaptive channel
equalization be applied to the systems in use--or to a newly designed systems.
The problem always falls back to the methods being pattent encoumbered.
> There ain't no free lunch.
No, indeed, it probably requires a hefty bit of DSP to acheive, as well, right?
> Develop a goal and/or standard and then develop you modem, whatever it turns out
> to be, to meet the goal or standard.
>
> What do we want?
Start there. Define the problem you want solved and the constraints on the
solution--your hurricane scenario, for instance. Then publish that and let
hams chew on it for a bit.
> > So, MT63 (63 carriers of BPSK and heavy FEC) in a 1KHz BW at -5db (in the same
> > BW?) SNR competing with a similar signal in 6KHz of BW will be -12.8db SNR (in
> > a 6KHz BW). Do you expect to be able to decode that?
>
> Sure, you can right now with software you can decode a signal at -15 dB SNR on a
> poor CCIR channel at ~50 WPM. We call than mode MFSK16.
Yes, that's because it's carrying *less* information, not more as your hypothetical
6KHz signal is. MFSK16 can do that because it encodes that little information with
a lot of FEC.
This is why I hate SNR for these kinds of discussions, it can be misleading. The
proper term is Eb/N0. The energy given to a bit of information divided by the
spectral power density. Eb/N0 describes the payload data, not the modulated
data. You can do anything you want to the payload data and the Eb/N0 of the
payload stays the same. See the SS discussion in the email to Karl for more info
on that.
You can then transmit log2 (6Khz/1KHz) = 2.6 times more bits in that bandwidth.
So, we're looking at 520 WPM if we can keep all other things equal.
> > What I'm getting at is that adding bandwidth isn't a wonderful solution to your
> > problems.
>
> I don't know...do we have a problem? We can pump out 1 KW of signal at 200 baud
> in a 2 KHz bandwidth and overcome lots of propagation problems...its called
> Pactor III and QRO. Or, we can run 400 WPM at 100 watts PEP (10 watts average)
> in a 6 KHz channel with a mode that allows several (Ok 2 or 3) signals to work
> in the same 6 KHz channel at the same time.
>
> Your choice.
This gets me back to 'what problem are we trying to solve'? Are we talking about
a guy in a bunker in the middle of a hurricane? How much traffic does he need to
pass? He's head down in a bunker, it's not like he can go walking around and
collect H&W info. :) For 'after the storm' H&W traffic you'll be in quite a
different boat.
> > I'll make a back of the envelope guess and say that olivia with 8 or 16 carriers
> > in a 2kHz BW is fairly close to ideal for amateur HF use. Any wider and you'll
> > need more power. Any faster and you break the 300 baud limit. And more carriers
> > and you cut throughput.
>
> Ok, I haven't studied the specs on Olivia so don't know what it's theoritical
> limits are.
It's just a flexable multi carrier/flexable bandwidth experamental digital mode.
> > > How many CW QSOs running at 20 WPM will it take to produce 1200 WPM error free
> > > copy at a signal level that MT63 can copy? These are signal levels that even
> > > the best CW operator can't copy a signal at. So the answer is that NO number
> > > of CW QSOs can produce the throughput of 1200 BPS, near error free, at a -5 dB
> > > SNR on a poor CCIR channel.
> >
> > This is a poorly constructed straw man. For one, your use of SNR is poorly
> > defined. -5db *in what bandwidth*? For CW, the necessary bandwidth is much
> > less than tha of the mt63 signal--2KHz vs <200Hz. That alone gives the CW
> > signal (or PSK31) a 10db head start--it's got 1/10th the noise to fight
> > against. Also, if the CW signals are generated by different stations, total
> > power for the CW operation increased with each staiton, so the CW stations
> > could fit 10 transmissions in the space of one MT63 transmission. Where the
> > single MT63 TX would have one unit of power, the CW operators get one unit
> > of power per station, so they get 10 units of power. Now they're up 20db vs
> > the MT63 signal. Are you saying that 10 CW signals at +15 db SNR (in the
> > appropriate BW) are going to copy less data than one MT63 transmission?
>
> Ok...-5 dB SNR on a poor CCIR channel which I believe is a 3 KHz channel with
> defined fading and doppler.
>
> Wail...all the CW signals must come from the same operator and the data must all
> be in context. Otherwise you are mixing apples and oranges.
You said 'how many CW QSOs' which was somewhat misleading. :) I don't think this
comparison is meaningful, anyway, so it's pretty acedemic.
> > > All this IS possible if we are willing to step out of the box.
>
> > The 'box', yes, but you really do need to stay in this universe.
>
> The commercial folks ARE doing this. Are they out of the universe?
They have resources that we as hams just don't have--large budgets, access
to propriatary techniques, no/little limiation on modulation scheme, more
bandwidth, etc. They live with a completely different set of constraints.
> > > We're also looking at doing much of this work using a computer sound card so
> > > separate controller hardware is not needed.
> >
> > That is a good usage of existing COTS hardware, but never forget that it does
> > limit the flexability of such a system vs the CW competator. A simple CW rig
> > and a pad of paper and pencil are pretty easy to come by.
>
> Certainly...but my computer and filtering can receive 30-40 WPM CW ar signal
> levels that no CW operator can hear....but there are also errors.
Sure, and the message passing protocol that goes with message traffic handling
is meant to deal with that--just like the FEC and ARQ of the digital modes.
The only reason I brought up this point was to make it clear that you're requiring
a good deal of equipment be operational affter a disaster. A littl CW radio
and enough wire to make an antenna can be stuffed into a padded ammo can and
chained to a cement block. That'll survive anything that doesn't suck it up
and take it with. Doing that with a PC+monitor+generator is harder.
> The idea is error free data at high speeds. How many messages can an NTS CW net
> pass in one hour? How many characters do they send?
You can pack in many more of them in 8KHz than you can pack in this large digital
signal. They can each be passing different infrmation to different people.
The goals for EComms seem to be:
1) survive the event with an operable station
2) some traffic must get through
3) the more that gets through, the better
> > Out of curiousity, is there anyone on the HSMM WG who plays devil's advocate
> > at your meetings? From the stuff coming out of it, I get the impression that
> > the dreaming goes fairly unchecked.
>
> Well, we don't have meetings, we have an internal group mail list and a publis
> maillist which you can join by going to URL
> http://listserv.tamu.edu/archives/arrl-80211b.html
> Select join the list. Or go directly to URL:
> http://listserv.tamu.edu/cgi/wa?SUBED1=arrl-80211b&A=1
> and fill in the blanks.
Hmmm, no wonder it wasn't obvious where this discussion was coming from. :)
No mention of HF nor EComms in the name there. 802.11b? A mode designed
for the multipath/LOS microwave environment? This starts to put things in
perspective. :)
> And yes, we have several who paly the devil's advocate and some of them are also
> on this list.
Good. A bad idea squashed early can save a lot of pain later.
Cheers,
David n0ymv
------------------------ Yahoo! Groups Sponsor --------------------~-->
Create your own customized LAUNCHcast Internet Radio station.
Rate your favorite Artists, Albums, and Songs. Skip songs. Click here!
http://us.click.yahoo.com/r4oloD/xA5HAA/kkyPAA/CPMolB/TM
--------------------------------------------------------------------~->
<< Try MT63 on 80m - great fun!>>
- The MT63 Reflector -
MT63@egroups.com
(To unsubscribe. send email to
MT63-unsubscribe@onelist.com)
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/MT63/
<*> To unsubscribe from this group, send an email to:
MT63-unsubscribe@yahoogroups.com
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
Read previous mail | Read next mail
| |
