[OLPC Networking] RSSI value questions

Hal Murray hmurray at megapathdsl.net
Wed Apr 2 18:49:38 EDT 2008


> The recommended configuration for mid-to-high-end school servers has 3
> active antennaes attached, and our recommendation is that they are
> placed well apart. They can be up to 10m apart due to USB cable lenght
> limits, and Wad mentioned 2m minimum recommended distance. If the
> distance is enough (in relation to the granularity of timers in the
> antennaes) then telling the XS about relative location of the
> antennaes could provide enough info. 

I assume the 10 m above is 5m for each antenna.  5m is the nominal limit on 
USB cables.  I think you can get longer than that by using hubs/repeaters.  
I've got some 1 port hubs that are built into the connector blob on a 5m 
cable.  I found a web page that said there is a limit of 5 hubs but I haven't 
tried it.


What sort of timer and/or time stamper does the active antenna and/or WiFi 
gear in the XO have?


I think there are two approaches that might be interesting.

If all you have is 2 antennas listening to the same packet, then you need 
more than good granularity on the timers.  You also need to synchronize the 
timers.

If you have the relative time of arrival of the signal at 2 antennas, you can 
compute the direction the signal came from.  The scale factor is the speed of 
light between the two antennas.  That's 1 ft/ns in air.  10 m is (rounding) 
50 ft, so we need time stamps accurate to a (small) fraction of 50 ns.  
That's the right ball park.

That gives you direction, no distance.


The other approach requires help from the XOs.

Take a pair of systems.  Exchange a pair of packets.  Grab the time stamps, 
both transmit and receive.  That's enough information so you can calculate 
the time/distance between the units and the clock offsets.  That pattern and 
calculation is the core of NTP.  I'll say more if anybody wants.

That gives you distance, no direction.


If you had a handful or systems and lots of distance measurement pairs, you 
might be able to make a map.  I think you need to know the location of a 
couple of units.  Without that, flips of the map over X or Y (or any other) 
axis also give you a valid answer.  The other antennas on the XS might be 
good enough.

This needs timestamps with the granularity of how good you want the location 
to be.  If you want the locations within 10 feet you need (handwave) 10 ns.  
You might get some more info by averaging several samples.


Is this a 2D or 3D problem?


-- 
These are my opinions, not necessarily my employer's.  I hate spam.






More information about the Devel mailing list