Re: HSLS?

At 10:29 AM 3/10/2004, Sascha Meinrath wrote:
>..Our main bottleneck is the DSL line connecting the network to
>the internet -- I suspect that if we had a T1 or the like that this would
>still be the case.  Since 802.11 handles collisions by having each node
>wait a random interval before resending, I think that even with a dozen or
>more nodes, the bottleneck will still be the connection speed to the
>internet.

In 2000-2001 I worked in a fairly busy office; they had approx. 30
people on 2 floors, with 3 APs and it was not really successful
there; perhaps 20% of the time it was just not accessible. And
another 30% of the time it was too slow or intermittent.

Ethernet cables started appearing, taped to the carpet.. 3,4,5..
so that gave me the impression it doesn't fail gracefully when
you get a whole bunch of highspeed, developer workstations.
I think the nature of this problem was the demand for thousands
of object calls, objects within objects wrapped in Microsoft cruft,
wrapped in XML cruft and their experimental webservices.
As well as their code repository, etc.etc. client-server applications.

So, perhaps my fears are not justified. I just had a very concrete,
long-running experience where basically they had to throw out all
the gear.. I still have 3 of the cards from the trash. aironet PCI-340's

>However, the current software will not scale well to scores and scores of
>nodes -- which is why we're working on this problem over the next year.

I think that is not the critical problem and here's why:   First, the main 
reason
people want this, 300 million people in this country is for Internet access.
Second. suppose you had a large suburban area, with cu mesh on
20% of the rooftops.

Then, estimating the number of Internet gateways needed to service
this network is a fairly tractable problem.. isn't it?   People want more
than a megabit. OK, if you could give them a reliable 1 megabit
and usually 2 megabits, I think that could be competitive in the
market for a few years, assuming that the jitter, latency etc. were ok.
Most websites and routes, only pump less than a megabit anyway.

OK.  Now, we know that most suburbs don't have a 3rd, independent
gigabit gateways or backbones. They have expensive telco T-3s etc.
but that's it.

Municipalities are saying they can get fiber laid at 10 cents on the
dollar if they simply wait for the installation of sewers or other work to
piggyback.  So, cities could and should, grow fiber resources although
the location is somewhat random.

Wherever a fiber, conduit, etc. resource reaches within 1000, or 500
meters etc. of residences it is rather a no-brainer that the people
will strongly want to put wireless gateways on them, and that
Internet providers will then trip over themselves to provide Internet
routes. i.e. sell internet services.

so, wouldn't you expect, the people or the city would put as many mesh
radios as the fiber and/or physical circumstances can cheaply allow?  say,
every 200 meters or so?    Thus most of the connections might be
lateral, 90 degrees from the fiber route.  And the number of hops is
not going to be many, nor is the route calculus.  Whenever you get
too complicated with the mesh trying to make it do marvelous things,
it's just going to bottleneck at the fiber gateways anyway.  Isn't it
cheaper to think about lots of AP's on a fiber, like christmas tree lights?
So why bother, with 10-hop radios or 100-node meshes, is my sense
of things.

Furthermore you ALWAYS seem to run into patents, and high prices
when there is some "cool" technology. At some point it seems
cheaper even if you have to pay, to put some fiber up and down
the suburban area even if citizens have to form a utility district to
do it.   It's still way cheaper than paying for phone, cable, internet,
security services, i..e. multiple pipes into the home.

Of course explaining all this to my neighbors costs 10x more hours
than it would cost to just build the freakin thing.

In the city I live, we are paying $50/month for broadband internet
service. (you get it cheaper only if bundled with cable or other
constraints).  That's $600/year. We all have money and
investments. Every family in Kirkland having broadband also has
investments/savings over $1000 that is getting below 5% interest.
The capitalized value of an income stream of $600 at 5% would be
$12,000.  ..   A rational citizen will pay $12,000 to
escape from a recurring cost of $50/month.  If Kirkland's city
council provided leadership and coordination, citizens could
purchase and operate our own broadband network, with off the shelf
equipment far cheaper than $12,000.
These Cu-wireless devices may be the right stuff, for circumstances
where a city has a few fiber facilities and needs a cheap, last-Kilometer
solution... is this reasoning useful at all?

I hope I haven't stepped on any delicate sensitive issues here, I am
just trying to figure out what is going to happen, and what kind of
equipment to recommend people buy....  Instead of waiting
for the telecoms industry to capture us all with the perfect,
lock-in network for the next 30 years.

TOdd

>ETX will help, having multiple connection points will help, using radios
>where we can automatically set transmit power will help, etc. Either way,
>you probably won't see much of a problem until you get many users on the
>network (unless you happen to have access to a T3 or something and want
>that sort of speed ;).
>
>--Sascha
>
> > http://www.cuwireless.net/images/wirelessmap.jpg
> > Not asking for the worst case scenario but really,
> > the scenario you'd almost expect, from bandwidth
> > hungry users in the evening time..
> > Would the throughput
> > 1. crash to zero?
> > 2.  scale down gracefully?
> > 3. become sporadic, randomly, i.e. this is what
> > my instincts tell me, but that's based on Internet
> > and office LANs when they get overloaded..
> > there would be unpredictable latency
> >
> > Just looking for your guess, ol boy.  You know your
> > gear by now... overall, do we get 100Kbps in this situation,
> > or 500kbps or 1Mbps?
> >
> > Many thanks
> > Todd
> >
>
>--
>Sascha Meinrath
>Project Manager & President      *      Project Manager
>Acorn Active Media Foundation   ***     Eggplant Active Media
>www.acornactivemedia.com        *      www.eggplantmedia.com

_______________________________________________
CU-Wireless mailing list
[email protected]
http://lists.cu.groogroo.com/cgi-bin/listinfo/cu-wireless
Project Page: http://cuwireless.ucimc.org