[OLSR-users] Fast Roaming Nodes

Jaime Vargas (spam-protected)
Thu Apr 6 01:31:20 CEST 2006

I doubt than a doppler shift of less than 0.00004% will become a
problem on locking the signal.

Imho, most of the problems will arise from delays at  the MAC layer,
especially during the handshake, that is in the authentication and
association phases.  However, I know this can be optimzed down
to 20ms, which leave 980ms to forward packets.

Of course the routing delays become important. Has anyone
charectirezed OLSR regarding route change propagation?

We are considering putting the AP inside the train and use EVDO
uplinks. But the negotiation process with the train operator will be
easier if we don't modify their carts.

-- Jaime

On Apr 5, 2006, at 5:53 PM, Weber, Michael J. ((US SSA)) wrote:

> John,
> I think you may need a new envelope...
> I get 816.67Hz of doppler shift for 2450MHz at 100m/s closure using  
> the
> equation F' = Fo * (1/(1+(v/Co))), where v is closure speed and Co is
> speed of the wave in the medium in question. For our purposes, v is  
> 100,
> Co is 3*10^8, and Fo is 2.45*10^9. If you plug common X band radar
> frequencies in instead, you get the familiar ~10Hz/mph shift at X band
> that we hear on the audio monitor of traffic radar. This equation is
> good only for non-relativistic speeds. :-) I got it here:
> http://imagine.gsfc.nasa.gov/YBA/M31-velocity/Doppler-shift-2.html
> Still, I expect it will be quite a challenge to route any train-bound
> traffic when the attachment point changes once per second. I might
> suggest a pair of yagis on a pole near the track, one pointing up the
> track, one pointing down the track, to give longer "hold times." An
> antenna external to the train would also be a requirement. It will
> probably be easier to get data from the train to the fixed
> infrastructure via UDP, but TCP sessions will be difficult.
> Good luck,
> Mike
>> -----Original Message-----
>> From: (spam-protected)
>> [mailto:(spam-protected)] On Behalf Of John Gorkos
>> Sent: Wednesday, April 05, 2006 2:26 PM
>> To: (spam-protected)
>> Subject: Re: [OLSR-users] Fast Roaming Nodes
>> I'm no expert, but my back-of-the-envelope calculations for
>> 100 meters/second
>> give me a doppler shift of 31.81 MHz at 2450Mhz at 0 degrees
>> offset.  There's
>> not a chipset in the world that can find a spread spectrum
>> signal 31MHz off
>> center frequency in <1 second, especially given that it's
>> going from 31.81
>> MHz UP as it approaches to 31.81MHz DOWN as it goes past.
>> Unless you're ONLY
>> using channel 6, you're going to be out of band on your
>> transmitter.  If you
>> do only use channel 6, you're going to self interfere.
>> At 900Mhz, doppler is still 11Mhz, and Canopy isn't that
>> good. Even if you stood the antennas a good 500 meters off
>> the tracks and used a 60
>> degree sector antenna, you've still got a delta of 15MHz.
>> Plus, 802.11 can't
>> associate and disassociate to the radios that fast.
>> Try going and talking to the NASCAR guys or the Indy car guys
>> about the
>> techniques they use to get 802.11, and full motion video off
>> the race cars.
>> Hint:  if the cars weren't going in a rough circle around the
>> antenna, it
>> wouldn't work.
>> John Gorkos
>> On Wednesday 05 April 2006 11:41, Jaime Vargas wrote:
>>> Hello,
>>> I am considering using olsr for a commuter trail corridor.
>> Our plan is
>>> to placed routers every 500 feet along the tracks. My
>> question is can
>>> olsrd handle user moving on the mesh at 230 miles/hour (337
>>> feet/second).
>>> That means that user will be hopping from router to router every
>>> second.
>>> Has anyone tried such setup? We also expect about 40
>>> users doing this roaming at the same time. We plan to
>>> do some trials, but before that I will like to know any possible
>>> shortcomings.
>>> Thanks a lot for the help,
>>> Jaime
>>> --
>>> Jaime E. Vargas
>>> Why Wire, Inc.
>>> 601 Birch St. SW
>>> Vienna, VA 22180
>>> T. 703-766-0939
>>> _______________________________________________
>>> olsr-users mailing list
>>> (spam-protected) https://www.olsr.org/mailman/listinfo/olsr-users
>> _______________________________________________
>> olsr-users mailing list
>> (spam-protected) https://www.olsr.org/mailman/listinfo/olsr-users
> _______________________________________________
> olsr-users mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-users

Jaime E. Vargas
Why Wire, Inc.
601 Birch St. SW
Vienna, VA 22180
T. 703-766-0939

More information about the Olsr-users mailing list