Re: [V4Protocol] What's happened
- John,There are several issues here.... For one thing it is literally impossible to tweak, tune or even accurately evaluate the performance of ANY digital mode without a controlled environment such as those on a HF simulator of some type. e.g. Many casually associate the “quality” of the signal with its strength....one factor but path distortions (multipath, doppler etc) are actually more critical than just signal strength. For example I have run both PSK (31, 63 etc), WINMOR and V4 through various channels and PSK modes (including those PSK modes in WINMOR) do not fare well in poor multipath environments such as NVIS, aurora etc. Though most narrow band slow modes (e.g. PSK31) can be made to perform well in weak signal (white Gaussian noise) environments.Optimizing a protocol for general usage environment entails repeated optimizations for factors such as error correction coding, automatic tuning, ARQ reliability (protocol robustness), poor multipath propagation as well as weak signal. For example it is possible to get another 1.7 dB or so weak signal performance by doubling the sound card sample rate and FFT sizes (at the expense of CPU loading) and I will look into that at some time in the future. But I want to get the basic protocol functionally working which includes ARQ before launching off into some of these optimization efforts. One thing ARQ will allow is to lower the detection thresholds (those magic constants inside the program no one sees!) that determine for example if and when a frame sync is detected. For true ARQ the thresholds can be set more aggressively since the ARQ protocol protects against any bad data getting through. This for example will allow better weak signal net throughput yet still allow the 100 % copy that ARQ allows.To put it in perspective...The WINMOR protocol (more complex for sure than V4) took over 2 years to develop and optimize. Pactor 3 (with the PTC II’s massive DSP power) was optimized for over 10 years. These things all take time, real engineering effort and controlled methodical testing. Surely over the air rag chews are fun and provide good feedback on functionality and desired functions but when it comes to really MEASURING performance (error rates vs. S/N over standardized CCIR channels) there is no substitute for using the HF channel simulator and performing hours of testing and statistic collections.The theory and initial tests I have run with V4 suggest it can be made very robust and still fit in a narrow 200 Hz channel and deliver modest 50+wpm typing speeds. It will however always be possible to design and field modes that offer better weak signal performances by using slower throughput (e.g. PSK31, JT65 etc) and/or wide bandwidth (e.g. Domino/Olivia).Rick KN6KB
Happy New Year, Rick. I'm sure I speak for all of us when I convey my
thanks for another year of your hard work programming for the benefit
the digital community of Amateur Radio.
I've been and am a strong supporter, but at the risk of irritating those
who have been excitedly waiting for the release of ARQ functionality, I
going to be bold enough to ask tha instead of diverting focus at this
point toward another set of issues, might it be worth a bit more time to
tweak FEC mode's performance a bit more in weak signal and poor band
I've been monitoring (and working) more often lately, and even with
version 0170 I just can't rid myself of the impression that in poor band
conditions I could be easily making good QSOs with any of a few other
modes, when with V4FEC I get little more than sporadic decodes. With
several of the other soundcard modes when I can just barely hear that
there's a signal on frequency I know I can get usable (often 100%) text,
but in those same conditions I struggle with V4FEC to get frames decoded.
Have we reach the point of diminishing returns with effort on the
performance of V4FEC?
But then again, perhaps I should just quietly wait and see.
73 and thanks again,
- Thanks for again sharing your plans for FEC and ARQ. As I said in
closing, I'll quietly (and gladly) stand by as we try to help your
efforts, however meagerly.
73, John W2KI
- Happy New Year Rick, to you and yours.Yes, the chat FEC seems quite stable to me per se. You don't need to wrapthe ARQ in ribbons or wrap or anything. Just put it where the tree was andthen step back out of the way.Howard W6IDSRichmond, IN Em79NV<tapping fingers, glancing at watch - AGAIN!>----- Original Message -----From: Rick MuethingSent: Sunday, January 02, 2011 10:10 AMSubject: RE: [V4Protocol] What's happened
The current version appears pretty stable. I have been working on the ARQ mode and had to make some significant changes there. Of course the standard holiday, parties etc. Hope to have a rev out this week with ARQ enabled.