Three magic wishes...

Andre andre at pe1rdw.demon.nl
Wed Apr 29 17:27:14 PDT 2015


Actualy APRS is linear because collisions are not only not retried but
activly used to reduce trafic load.
A kludge has been tought up to set all digipeaters to slottime 0 making
them all transmit as soon as the channel is clear almost garanteeing
that they all transmit at the same time theoreticly allowing only the
strongest digipeater to be heard and decoded.
APRS messages are offcourse the exeption but how many trackers or even
home stations use aprs messaging?

Also APRS can function perfectly with multiple paralell networks because
with the internet gateway configuration messages will find their way to
the other frequenties and the internet side can be used for home
stations and phones to see who is about regardless of frequentie. only
for direct RF tracking you need to be on the same frequentie and I don't
think that is a big problem as that is usualy limited to a group and
they can agree on the frequentie before hand.

with enough bandwith available you can even stream a full feed from
internet to a 23cm or higher frequentie, as it is a simple text stream
any protocoll can be used for that.

73 de Andre PE1RDW

Op 30-04-15 om 01:09 schreef Jeff Francis™:
>   I'm about 3000 miles from home this week, and don't have any of my
> references available, but I certainly seem to remember that a CSMA
> network with hidden nodes (ie, APRS) worked out as roughly an N^2
> problem for reliable communications.  It's most certainly not linear,
> even if it's not quite N^2.  Certainly straight CSMA isn't that bad, and
> something that's timeslotted is straight linear.  But 1200baud packet
> isn't either of those.  As far as 70cm goes, that's all well and good,
> but I was responding to Bill's wish regarding HF digital modes.  When
> you go wide on HF (wide meaning more than a couple hundred hz), you have
> to have more power, or you might as well not bother.  Your power is
> spread too thin.  And I have serious doubts that you can do much that's
> useful with CDMA on HF.  Your symbol rate is so low with HF, you might
> have to wait tens of seconds for your slot, assuming you want slots that
> will pass at least 100 bytes.  Your overall throughput would be a
> pittance, and it's hard to imagine what you could use it for (beyond
> maybe HF APRS or propagation studies).  You'd also need every
> participant to sync with a good GPS clock to get the slots right.  The
> speed of light even becomes a factor.  It takes roughly 70 milliseconds
> for a radio wave to travel to it's antipode.  You have to build enough
> slop into your timeslots to account for that, plus a small amount of
> clock skew (getting GPS time data via NMEA-0183 vs. raw binary data can
> be a 100ms skew right right there, due to the way NMEA-0183 is so poorly
> specc'd - we ran into this when I worked on GPSD years ago).  You
> certainly wouldn't want to try to pass an email over CDMA on HF.  It
> would be quicker to print it and send it via US Mail.
> 
>   I'm all for innovation and less government, but I'm not convinced that
> we're bumping up against "too much government" as our limiting factor in
> the HF digital arena (though it does account for most of the other
> problems we face in this country).  I, in fact, have been pleased and
> impressed with the strides we've made in the HF digital world in the
> last ten years.  We've gone from crappy 300baud packet to the amazing
> variety of modes available in fldigi (not to mention some of the wicked
> fast proprietary solutions like Pactor 3 and Pactor 4).
> 
>   And while I don't agree that turning 70cm into a digital ham cell
> network is the best use of the spectrum, I do agree that if we don't
> start making more and better use of it, we're gonna lose it.
> 
> Jeff N0GQ
> 
> 
> On Wed, Apr 29, 2015 at 10:46 AM, Peter N. Glaskowsky <png at ideaphile.com
> <mailto:png at ideaphile.com>> wrote:
> 
>     > Date: Wed, 29 Apr 2015 06:24:56 -1000
>     > From: Jeff Francis? <jfrancis at gmail.com <mailto:jfrancis at gmail.com>>
>     > To: All things digital and fun <seatcp at wetnet.net
>     <mailto:seatcp at wetnet.net>>
>     > Subject: Re: Three magic wishes...
>     >
>     > 1.  Can't scale, can it?  Assume for the moment that 1% of hams do APRS,
>     > and further assume that 10% of that 1% are on the air at any given time
>     > (obviously, I don't know that these are accurate, but even if they're off
>     > by an order of magnitude, it still can't work).  You're looking at a
>     > potential of 1000x the current APRS activity at certain times of the day
>     > (granted, it'll never be quite that high, but we're talking order of
>     > magnitude numbers).  In the programming world (as well as the networking
>     > world), we call it an N^2 problem.  Resource usage goes up as the square of
>     > the number of participants.
> 
>     I’m not sure what principle you’re referring to here, and I’m not
>     familiar with any networking situation where utilization goes up as
>     the square of participants. However, in the traditional definition
>     of the “network effect” the cost of a network is proportional to the
>     number of participants N, while the value of the network is
>     proportional to N^2, which is a very good thing.
> 
>     Of course that was just a conceptual definition. In practical terms,
>     congestion can follow different rules. In shared-media
>     implementations where collisions lead to retries (like APRS), there
>     will be increasing utilization to some characteristic point followed
>     by congestion-induced decay, but that decay isn’t an N^2 type of
>     function.
> 
>     But APRS isn’t the only protocol in the world, and there’s no reason
>     to stick with it if some other protocol will give us better scalability.
> 
>     > Doubling the hams requires four times the
>     > spectrum (assuming you want them all to communicate and not be segregated
>     > by frequency).  Gatewaying the data from each frequency to each other
>     > frequency is it's own N^2 problem.  It would take up most of the 2M/70CM
>     > bands just to send all of the traffic, and that's assuming you don't
>     > gateway the data from one frequency to another (ie, most of the hams
>     > couldn't communicate with most of the others without know which freq they
>     > were on and switching first).  Which still wouldn't work, because everybody
>     > would try to switch to the frequency most likely to contain other people
>     > and overwhelm it.  I'm not seeing value here, unless part of the magic is
>     > for everybody to keep their stations turned off until they magically sense
>     > the need you have to communicate with them and turn it on just in time to
>     > get your message, then off again.
>     >
>     > 2.  What would doing this get us that we don't have already?  We have
>     > automated stations, we have a nearly infinite supply of modes that work
>     > under all sorts of different conditions, we have more than enough spectrum
>     > for current usage and needs.  About the only thing we can't do that some
>     > people would like to try (at least the only thing I can think of) is the
>     > very wide digital modes, but those disperse the energy so broadly that you
>     > have to have a 1kw amp to use them effectively (not to mention are subject
>     > to huge interference problems).  Even in places where they're legal,
>     > there's not much use of them (like Pactor 4, for example).
> 
>     There are technical solutions like spread-spectrum modulation
>     schemes (CDMA, for example) that allow efficient channel sharing
>     without collisions. And when there are more stations participating
>     in an RF network, the solution is to reduce power, not increase it,
>     thereby improving spatial reuse. The remarkable market success of
>     “very wide digital modes” like LTE at very low power levels is proof
>     of this. The same technology could be applied to amateur radio in a
>     peer-to-peer manner rather than in the base station/mobile device
>     structure of the cellphone network— still with cells, but with
>     dynamic distributed management protocols instead of a single network
>     operating authority.
> 
>     In real-world terms, these solutions would be extraordinarily
>     effective for us. A metropolitan-area digital ham network using,
>     say, 10 MHz of the 70 cm band could carry an aggregate bandwidth of
>     gigabits per second, not the few megabits that a single high-power
>     radio could transmit— or the nearly-zero throughput that would be
>     achieved in an overloaded network of many such transmitters.
> 
>     Frankly I think it would be a vastly superior use of the 70cm band
>     to devote it entirely to digital cellular communication. It could be
>     every bit as functional for us as the cellphone network is for the
>     general population, but without any dependencies on infrastructure
>     we don’t control or vulnerabilities to single points of failure. It
>     would be ready to go in the event of local emergencies, trivial to
>     recreate at a new location if needed, and readily interoperable with
>     other systems through fairly straightforward gateway devices.
> 
>     One of my concerns here is that if we don’t do this, someone will
>     take this spectrum away from us so THEY can do it. In the AWS-3
>     auction, the FCC sold 65 MHz of spectrum for over $41 billion. Our
>     30 MHz of the 70 cm band— which is intrinsically MORE valuable
>     because it offers better structure penetration— is therefore worth
>     more than $20 billion, and what are we using it for? Almost nothing
>     of any practical value. The value of the spectrum will only increase
>     over time, and we’re not increasing our use of it. I believe it’s
>     inevitable that it will eventually be repurposed.
> 
>     So basically I don’t agree that there’s an N^2 problem here, that
>     wide digital modes are a problem, or that we can’t get hugely
>     valuable capabilities that we don’t have now.
> 
>     .                 k4png
> 
>     _______________________________________________
>     Seatcp mailing list
>     Seatcp at wetnet.net <mailto:Seatcp at wetnet.net>
>     http://wetnet.net/mailman/listinfo/seatcp
> 
> 
> 
> 
> -- 
> -=jeff=-
> 
> 
> _______________________________________________
> Seatcp mailing list
> Seatcp at wetnet.net
> http://wetnet.net/mailman/listinfo/seatcp
> 



More information about the Seatcp mailing list