| |
PA2AGA > HDDIG 28.09.99 02:30l 152 Lines 7330 Bytes #-9756 (0) @ EU
BID : HD_99_243F
Read: GUEST
Subj: HamDigitalDigest 99/243F
Path: DB0AAB<DB0PV<DB0MAK<DB0ERF<DB0NHM<DB0SHG<DB0SM<PI8DAZ<PI8GCB<PI8HGL<
PE1NMB<EA7URC<PE0MAR<PI8VNW
Sent: 990927/2015Z @:PI8VNW.#ZH2.NLD.EU #:3615 [HvHolland] FBB7.00g $:HD_99_243
From: PA2AGA@PI8VNW.#ZH2.NLD.EU
To : HDDIG@EU
Received: from pa2aga by pi1hvh with SMTP
id AA20118 ; Mon, 27 Sep 99 19:30:01 UTC
Received: from pa2aga by pa2aga (NET/Mac 2.3.67/7.5.3) with SMTP
id AA00016032 ; Mon, 27 Sep 99 20:42:12 MET
Date: Mon, 27 Sep 99 20:38:48 MET
Message-Id: <hd_99_243F>
From: pa2aga
To: hd_broadcast@pa2aga
Subject: HamDigitalDigest 99/243F
X-BBS-Msg-Type: B
>> end is going to have that are above its noise floor. (It may also
>> cause problems for others using the same, or adjacent, frequencies
>> for other links.)
>
>I never mentioned the length of the links I am proposing to
>build. I only mentioned the length of some EXISTING links.
>But apples and oranges in the same bag here. More than one
>discussion going on.
>
>1) I'm interested in playing with high speed RF network linking.
>Just me. I can own both ends. They can be close. In the same
>house perhaps. Cost is not a serious consideration. Might link
>to Ted next door (a bit less than 1 km away).
Ok, very low power will suffice for this.
>2) I'm interested in replacing a few existing 9600 baud links with
>faster links to improve our local network. They are moderate
>distance, clear path, easy access to the sites. 1W works fine
>for 9600 baud. This one is cost limited.
Again, high power is not warranted, and would unnecessarily
drive up costs.
>3) I'm interested in connecting places far apart using RF.
>These are long haul links. I've done these before, at 1200 baud,
>using troposcatter on 2M. Path lengths are 200 - 400 miles.
>A kilowatt and four long yagis helps to make these long links
>work. See for example the W5XO links that used to exist
>in Texas (high power transmitters, big yagi arrays, tall towers).
>Cost is not a serious consideration, at least at my end of the link.
>(Consider the cost of a "good" HF station).
Now this is a different situation. You're going to have to use much
lower speeds on such obstructed links. So the discussion with regard
to high speed modems, and the low power levels suited to reducing
intersymbol distortion at those rates, is moot. Forward scatter systems
require high power because the path loss is so very high. There's no
way around that. But ISD is very high too, and there's no way around
that either, so you can't use high rates. (I'll have more to say about
the sustainability of low speed nets further down the page.)
>> The correct answer is not to just try to "bull through". The correct
>> answer is to properly engineer the path so that it will work at a
>> power level where multipath will fall below the receive site noise
>> floor.
>
>Gary, you not paying attention.
>More ERP equals more power spread equals longer path.
And more power *spread* is exactly what you want to avoid. That
causes more multipath reflections from more objects further from
the line of sight path. *You* aren't paying attention. The *last* thing
you want to do is spew excessive signal around the countryside.
What you need to do is tightly focus your signal along a well
engineered line of sight path, and keep the power spreading down
to the *absolute minimum possible* to avoid generating excessive
ISD. For properly engineered line of sight paths, low transmitter
power is always sufficient. For poorly engineered paths, excessive
power generally won't help because the ISD, and BER, becomes
unacceptable.
>> >I guess my bottom line is that I'm not at all interested in why
>> >this cannot be done. I'm only interested in how to go about
>> >doing it. Where there are no links, a not-quite-perfect link
>> >is a huge improvement.
>>
>> I've told you how to do it. Do the RF path engineering properly.
>> There are well established procedures and formulas for doing
>> this. AT&T Long Lines, and others who've needed RF links of a
>> specified quality, have been using these procedures for over
>> half a century. They are well proven.
>
>You told me nothing useful. In fact, all you said was "Don't run
>more than 4-10W ever under any circumstances." Well, on most
>of my links I run less (1W or 2W) and use aluminium to make
>up the ERP requirement. Some run more, on both ends, to make
>up the needed power spread.
Keep in mind that my comments apply to high rate systems. We were
specifically addressing 56 kb, and to a lesser extent 9.6 kb. For low rate
forward scatter systems, different concerns apply. For such systems,
high ISD is a given, and the very large path loss makes high power a given
too. We have to deal with the high ISD by reducing the data rate. Since
low rate systems generally aren't of much utility for sustainable networks,
I have very little else to say about low rate over the horizon systems.
>Also keep in mind it is not always POSSIBLE to use the perfect
>path between two perfect stations. This is ham radio. We get
>to play. We do not NEED 6 nines.
It is true that on dedicated single hop paths, reasonably high error rates
can be tolerated. The ARQ protocol will eventually bull through. But you
can't live with that if the path is part of a larger network which also has
moderate to high error rates on the other links the packet must traverse.
The overall path will either quickly degenerate into a cacophony of retries
and timeouts, or, if you use a clever protocol to prevent that, the network
latency will become unacceptably long.
If you're trying to establish a fast network, poor paths can't be tolerated.
If you're content with low rates, we've already done that, almost two
decades ago. It wasn't very interesting on a sustained basis, and most
of the packeteers involved in building them have abandoned such slow
networks. (At times, packet latency on the East Coast north-south
network exceeded 15 minutes. That's ludicrous.)
In other words, once you meet the challenge of doing it, you have to look
at what good it serves as an information utility in order to decide if it is
worth sustaining. A large number of users decided it wasn't worth the
expense and quit supporting it. That led to the breakdown of many sections
of the net, rendering it of even less utility.
Some people have tried to solve that problem by bridging missing or
particularly low performing sections of the network with landline links.
That's what you and Charles seem to rail against so vocally (and so
fruitlessly). It certainly isn't a pure amateur radio network any longer
when that is done, and it is certainly reasonable at that point to ask
why bother with using amateur radio at all? I have no quarrel with that
viewpoint.
I'm only interested in doing amateur radio networks (a point which you,
and particularly Charles, can't seem to grasp). But there's a large
difference between the builder challenge and the every day operation
and maintenance of such complex coordinated systems as packet
networks. For the latter, you have to look at the system in terms of its
utility to its users. If it doesn't meet that test, it won't long survive.
At present, fast RF networks can be viable, but for the most part slow ones
cannot. They just won't deliver the performance that today's user demands.
So I've concentrated my amateur radio networking efforts on fast sustainable
networks. Because of geographical and financial facts of life, those networks
To be continued in digest: hd_99_243G
Read previous mail | Read next mail
| |
