[pskmail] Re: integrating Winmor TNC
- From: karel Fassotte <karel.fassotte@xxxxxxxxx>
- To: pskmail@xxxxxxxxxxxxx
- Date: Fri, 28 Jan 2011 08:32:04 -0500
Hello all, Rein,
To make real comparison with PSKmail we have to consider the wave form and
the way the waveform choise is handled.
The WINMOR protocol uses basic OFDM (Orthogonal Frequency Division
Multiplexing) modulation and a number of modulation modes and error
correction
schemes to adapt to changing channel conditions.
There are currently 2 operating bandwidths of 500 and 1600 Hz (@ 26 db
below peak power output:
-500 Hz BW 2 carriers 46.875 Baud 4FSK or 93.75 baud PSK using TCM
4PSK, 8PSK or 16PSK
-1600 Hz BW 8 carriers 46.875 Baud 4FSK or 93.75 baud PSK using TCM
4PSK, 8PSK or 16PSK.
I tested winmor in a NVIS link, without any interferece in Ecuador and 2.5KB
message was transfered in 80 seconds. No only once
but concequently. Without compression. In my opinion the OFDM multicarrier
PSK waveform is more effective than the BPSK, PSK mail is using.
What I am asking is to consider integrating these waveforms in PSKmail, to
make it even more robust and faster.
The specifications of winmor are known, the software is closed code, but may
the binary of the TNC can be bounded to PSKmail as FLDIGI modem does.
OK go ahead.
greetings
HC1AKP
2011/1/28 Rein Couperus <rein@xxxxxxxxxxxx>
> Maybe I can throw something in on this subject, as I have some
> hands-on experience with pskmail....
>
> It really looks like PSK500 is the fastest PSK mode which can be effective
> on HF,
> the setup we have now is often too optimistic, and the link falls back to
> PSK500R
> unless the channel is absolutely clean. When this happens, the net
> throughput
> goes up to max. 1600 bytes/minute (max. measured 1692 bytes/min) including
> ARQ
> overhead. In case of file transfers this is compressed, so actually the
> throughput is more in the order of 2kB/min, which is the value Karel needs.
> We are seeing this in the seldom case that we have a clean channel
> As long as there is no requirement for mode change, RSID is not used.
> Binary attachments are integrated in pskmail.
>
> When we are looking at clean links (these don't exist in EU) we could get
> some more
> throughput by allowing frames of more than 8 blocks, which would increase
> the
> throughput slightly.
>
> And of course if somebody can supply us with an open source fast modem for
> fldigi
> we could get talking...
>
> 73,
>
> Rein EA/PA0R/M
>
>
>
>
>
> >Hola Karel,
> >
> >My reply to your comments between {} below.
> >
> >Regards,
> >
> >John
> >
> >On Fri, Jan 28, 2011 at 11:10 AM, karel Fassotte
> > wrote:
> >> Hello John,
> >> I will comment below.
> >>
> >> 2011/1/27 John Douyere
> >>>
> >>> Hola Karel,
> >>>
> >>> There are three key design objectives in the Pskmail project that
> >>> relate to your questions:
> >>>
> >>> 1. Open source (right its no open source9 {I am not sure what you mean
> here but ALL components of Pskmail ARE open source: Fldigi, the server in
> Perl and the Client in Java. Therefore anybody with special requirements can
> get into the code and modify as In my case - I came in a few years after the
> start of the project and have contributed a number of items}
> >>>
> >>> 2. Narrow band (less than 500 Hz). (why ?) {so that we do not use
> unnecessary bandwidth which has been a key complaint by Ham radios with
> wider systems and are able to use all the narrow segment of the bands
> (including the very important 30M for Europe)}
> >>>
> >>> 3. Solid link even in the presence of QRM/QRN (NVIS links are very
> stroung
> >>> little interference) {I don't know in Equador but here we have
> thunderstorm activities, mainly in summer, that create significant QRN even
> when the signals are S9+10 as is common in NVIS conditions here}
> >>>
> >>> The Winmor protocol is available as a software TNC (like Fldigi) but
> >>> unfortunately only under Windows. So that excludes for the moment
> >>> having it linked to Pskmail. (right)
> >>>
> >>> On the speed aspect there would be some work required to increase the
> >>> speed further with say 2PSK or 4 PSK (e.g. 2 or 4 PSK500 or PSK500R
> >>> modulations in parallel). That would not be that hard to do in Fldigi,
> >>> but the timing will become more critical in Pskmail as it was more
> >>> designed to accommodate very different (and some slow) digital mode
> >>> timings rather than a fixed and precise timing between client and
> >>> server. (that could be a solution. However I liked the very ffast and
> >>> efficient turn arround of Winmor)
> >>>
> >>> So probably the issue at present is more to understand the need for
> >>> faster speed as the design objectives has served us well until now I
> >>> believe. (yes that might be the case but link build up and conecting
> >>> speed, addaptive is of great advantage for a quick handling of mail.
> Speed
> >>> is also an issue) A mail of 1KB should be transfered and handled in a
> >>> minute. {What happens if it is delivered in two minutes? What
> percentage of the total response time is that?}
> >>>
> >>> Please note that Pskmail (at least in recent versions) will
> >>> automatically adapt it's speed depending on the conditions, from THOR8
> >>> to PSK500 and in my personal experience has been able to establish
> >>> links several times when I could not with Winmor from my same portable
> >>> setup. (I dont have the same experience. I have no problem linking
> Winmor
> >>> on my links)
> >>>
> >>> Of course it could be a difference in the server's setup since they
> >>> are separated by several kilometers and don't run the same antennas,
> >>> but that is my experience.(My conexions are NVIS links up to about
> 1000km)
> >>>
> >>> What happens often is that the link is not symmetric, either because
> >>> the portable station is running low power and/or has compromise
> >>> antennas, or the server (which normally runs high power - say 50 watts
> >>> - and has good antennas) has local QRM. (maybe an item)
> >>>
> >>> That is why Pskmail will TX and RX in different modes and adapt each
> >>> individually to the link quality.(I understands but that takes time and
> >>> time is critical for emergency communications)
> >>>
> >>> I have several time run QRP power (FT-817) with a dipole in NVIS
> >>> conditions (which I believe represents the most common situation for
> >>> disaster conditions) and had very good downlink from the server and
> >>> poor uplink to it, but Pskmail did it's job nicely almost every time.
> (I
> >>> hope it can be faster and better)
> >>>
> >>> When the link is good the PSK500 modulation will provide an 800 words
> >>> per minute raw speed, and taking into account the compression of text
> >>> messages (about 2 x for text) and the overhead of the Pskmail protocol
> >>> we should see net exchanges of about 200 to 400 words per minute or
> >>> above 70Kb of data per hour. That is a lot of data exchanged I feel.
> (for
> >>> bigger files I would use another protocol 188-110, this is slower
> bulding up
> >>> the link but has much faster data troughput uo to 2Kb/s) {If speed is
> that critical, have you looked at RFSM-8000 which uses MIL-STD 188-110A /
> MIL-STD 188-110B modulation. It runs normally under windows but I have been
> able to run the older version - RFSM-2400 under Wine on a Ubuntu PC, so it
> may work with RFSM-8000 too. You can get a free trial license from their web
> site, if this project is still active}.
> >>>
> >>> In my experience also, the psk modulations (as used in Fldigi and
> >>> Wiinmor) are not the most robust due to phase shifting and selective
> >>> fading that is characteristic of that propagation mode. In that case
> >>> the fsk modes (MFSK and THOR for example) provide much lower error
> >>> rates. (Winmore is fast adaptive and uses 5Khz. bandwith or 1,6Khz.
> >>> Bandwith using multi carriers OFDM PSK)
> >>>
> >>> I have seen many times a slower but more reliable link (in MFSK for
> >>> example) providing a higher net data rate than a fast, lest reliable
> >>> link with many repeats. (Looking for a solution for a good NVIS link,
> not
> >>> for links wit a lot of QRM, QSB, speed should be maximum 1300b/s)
> >>>
> >>> What format do you expect your emergency network to take: NVIS or long
> >>> distance, what type of stations (fixed, high power, good antennas OR
> >>> portable low power, compromise antennas), how many stations, how many
> >>> messages, of what size, what type of message content (text or binary
> >>> data). (NVIS uo to 1500 km, 50 watts of power 100% duty cycle, a good
> NVIS
> >>> dipole on several frequencies. About 20 stations, 1K, max. 2KB
> messages.
> >>> Messages usualy are text based but could contain binary attachments
> (small)
> >
> >{what volume of messages? as that will give you the expected channel
> >usage and also the average latency of the messages. I propose you
> >start from there and then see what the numbers tell you in regards of
> >the speed you need, rather than starting from the speed itself}.
> >
> >{In the digitalradio Yahoo group there are a few people who have setup
> >such networks and can give you good feedback. I noticed that speed
> >what not their main criterion, but they focussed on ease of use,
> >reliability of communication, ability to broadcast messages to
> >multiple stations etc. Maybe you can ask some questions there too.
> >They have used Fldigi with and Flwrap and Flmsg. Instead of relying on
> >ARQ for safe delivery of their messages they use modes with strong FEC
> >like MT63 and Olivia together with the tools above and that seems to
> >work well for them}.
> >
> >>>
> >>> These are all factors that are important in your selection of the best
> >>> solution.
> >>>
> >>> Hope this helps.
> >>>
> >>> Regards,
> >>>
> >>> John (VK2ETA)
> >>
> >> Please go on suggesting, I might not the only one thats looking for this
> >> solution.
> >> Greetings
> >> and thanks
> >> Karel (HC1AKP)
> >>>
> >>> On Fri, Jan 28, 2011 at 3:03 AM, karel Fassotte
> >>> wrote:
> >>> > Hello all,
> >>> > I have been investigating for some time now solutions for a emergency
> >>> > network in Ecuador I also have been testing pskmail. I think it is
> >>> > allright
> >>> > but I would like to have more and faster modem waveforms included.
> >>> > FLDIGI is
> >>> > a nice multiwave form software modem, but for mespeed could be
> faster. I
> >>> > have been testing the WINMOR TNC of Winlink.
> >>> > My results are that the WINMOR TNC is very adecuate and hast handling
> >>> > maximizing overall troughput. This is of big interest for an
> emergency
> >>> > network. Many small (1KB) messages should be handled.
> >>> > The RMS message terminal is also very easy to use.
> >>> > This is all closed software, only windows untill now. I am not
> >>> > interested in
> >>> > windows, but I am interested in solutions for open code with maximum
> >>> > funcionality and easy use.
> >>> > Can this WINMOR TNC be integrated in pskmail? This would realy be a
> >>> > great
> >>> > improvement over the existing FLDIGI modem.
> >>> > Do exists other soundcard modems that have the same specs as the
> WINMOR
> >>> > TNC?
> >>> > Please let me know.
> >>> > greetings, 73
> >>> > Karel Fassotte
> >>> > HC1AKP
> >>> >
> >>> >
> >>> >
> >>>
> >>
> >>
> >
>
>
Other related posts:
