[1st-mile-nm] Intel's Long-Distance Wi-Fi

Thu Mar 20 20:36:05 PDT 2008

The quoted message isn't exactly correct. The standard 802.11 MAC layer 
starts to have serious problems above ~10 km.When we first started La 
Canada Wireless ~5+ years ago, we had a person connecting from the ridge 
east of Cerrillos to our first (solar) AP above the Eldorado country 
store. IIRC, that was a 20 km shot. The bandwidth obtained was rarely 
above 100 kilobits, while we could see 2-3+ megabits on locally shorter 
links.

The problems are related to timings in the protocol, primarily ack and 
slot. To get longer distances reliably, these MUST be adjusted. See 
http://madwifi.org/wiki/UserDocs/LongDistance for an unusually lucid 
presentation of the issues.

Otherwise, spot on: the Intel announcement is underwhelming and 
misleading. We buy 802.11a equipment (5.8 Ghz) for our backhaul links. 
At present, our longest 11a link is ~10 miles (from Las Lagunitas to 
Vail mountain) and my co-worker there regularly gets well over 3 
megabits over the link. Were we to put higher gain antennas on both ends 
(parabolics, ~4x gain of existing), I have no doubt we could double or 
triple that. Aesthetics are, unfortunately, in play there.

I have gotten 20+ megabits out of a  7 mile point-to-point link using 
this equipment. At my home, I am two hops from the DS-3 upstream. The 
hops are about two miles each. Depending on the site to which I'm 
connected and the load on our router, I've seen over 10 megabits 
sustained on some large transfers to my laptop, which is on my home 
wireless 11g network..

The equipment we use is COTS @ ~$225+, depending on how it is provisioned.

Not fiber speeds, but as  FTTH out here is probably a pipe dream for the 
foreseeable future, I'll suffer. At my day job, BTW, it is rare that 
I'll see 50 megabit speeds to my desktop, even though the intervening 
infrastructure to the backbone is all gigabit+.

FWIW.

Bob

Richard Lowenberg wrote:
> In response to Tom's posting, I've been hoping for the past couple of days that
> no one would post to the 1st-Mile list about the Intel long distance wifi
> transmission announcement.   It is less than meets the eye.   Following is an
> informative posting from another list, about this.   Be careful of tech-hype.
> rl
>
> ----- Forwarded message from sebastian at less.dk -----
>     Date: Thu, 20 Mar 2008 09:55:43 +0200
>     From: sebastian buettrich <sebastian at less.dk>
>
> Dear all,
> i am normally an interested but quiet reader on this list,
> but this time around i would like to comment, since i am specializing in
> community wireless networking
>
> On Wed, 2008-03-19 at 05:35 -0400, Ian Thomson wrote:
>   
>> Hi All,
>> This is a very confusing press release. Many WIFI practitioners are today
>>     
> getting distances of 30 to 60 kms with parabolic antenna. What is so special
> about the Intel development?
>   
>> If it is just that is drops requests to resend packets, then this is doubtful.
>>     
> Any radio engineer will tell you that radio is bad for dropping packets and
> resending them is necessary to recover the full data.
>
> It is absolutely correct -
>
> wireless practicioners have been doing up to 300+ km already and the
> current world record is approx 380 km
>
> http://www.eslared.org.ve/articulos/Long%20Distance%20WiFi%20Trial.pdf
>
> Distances up to 100 km have never been *that* great challenge and can be
> done and are frequently done by wireless practicioners - and a lot
> cheaper than USD 1000 - from around USD 200-300 in wireless gear.
>
> For the tech minded, some detail:
>
> Going above 100 km, the shortcomings of the MAC 802.11 layer have been
> the challenge - in simple words, the "hello" and "acknowledge" timeouts
> between the wireless nodes.
> This is what the Intel product addresses, not by dropping the ACL
> altogether, but by using TDMA - if you are interested in more techie
> details
>
> The statements made in the article,
>
> ""If you take standard Wi-Fi and focus it," Galinovsky says, "you can't
> get past a few kilometers.""
>
> and in the Intel video can only be called misleading and simply wrong.
> So, it s more of a PR cloud, misleading people, than "a huge step
> forward".
>
> hope this helps,
> best to all,
>
> sebastian
>
> ------------------------------------
> dr. sebastian buettrich
> sebastian at wire.less.dk
>
> a free book on wireless networking in the developing world - 2nd edition out
> now!
>
> http://www.wndw.net
> http://wire.less.dk
> http://thewirelessroadshow.org
> ------------------------------------
>
> Richard Lowenberg
> 1st-Mile Institute
> P.O. Box 8001, Santa Fe, NM 87504
> 505-989-9110;   505-603-5200 cell
> rl at 1st-mile.com  www.1st-mile.com
>
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> _______________________________________________
> 1st-mile-nm mailing list
> 1st-mile-nm at mailman.dcn.org
> http://www2.dcn.org/mailman/listinfo/1st-mile-nm
>   