[USRP-users] Wireless switch for MIMO?

Marcus Müller marcus.mueller at ettus.com
Fri Sep 5 12:03:17 EDT 2014


Hi Cochenour,
just throwing a few numbers in:
Assuming perfect transmission, IEEE 802.11n can do a maximum of
realistically let's say 200Mbit/s (given an awesome 4x4MIMO system...
For myself, I have never actually achieved more than about 108Mbit/s
with my notebook) *brutto* overall. assuming you don't want to tx and rx
at the same time, let's, for argument's sake, say we're able to use the
192Mb all by ourselves for samples alone. So, N2x0, I suppose, has a
16bit short complex data format -> 32bit/sample. Thus, you could achieve
up to 6MS/s (next possible actual sampling rate would be 5.88MS/s ==
100MS/s/17, but you usually want to stick to even decimations, thus
making 100MS/s/18==5.56MS/s the rate of choice).
But that would imply unrealistically good WiFi conditions, zero MAC
overhead, zero UHD overhead, no packet loss... Not wanting to dampen
your expectations too much, but I'd say a net rate of 64Mb/s would still
be good -- remember, you'll have to *guarantee* that rate, not only
achieve it on long-time average. That'd be 2MS/s -- not so much,
considering you spend 20 or 40MHz of ISM bandwidth to transfer a signal
of less than 2MHz bandwidth...

There are high rate alternatives to WiFi, of course, as employed by
people that build up cell networks, for example. I've never played with
those, though, but I'm pretty sure it'd be fun :) Now, all regulatory
limits aside, using a highly directive antenna on your stationary PC
tracking the buoy, and an antenna with still some gain on the latter, I
guess you'd get SNRs that would allow for high speed data transmission,
but I don't know of such systems, to be honest.

Now considering the fact that you want to spectrum sense what happens on
a buoy: would "amplify and forward" make sense in your scenario, ie.
just building a dumb relay to forward the RF channel you want to your
stationary PC, maybe even with two mixers, so that you can use two
copies of the same signal to reconstruct it on the stationary PC? That
could be solved in analog completely, but would of course lack the
tunability of USRPs.

Greetings,
Marcus

On 05.09.2014 17:36, Cochenour, Brandon M CIV NAWCAD, 4.5.6 via
USRP-users wrote:
> I've used wired Ethernet switches to combine a bunch of USRP data streams together.  The Netgear models have a plenty big "pipe" in which to carry several streams at "low enough" sample rates back to the host PC.
>
> I have an application though in which I might not be able to get to my USRPs with a CAT-5 cable (a floating buoy!), and was considering an wireless switch (one on the buoy with the USRPs, another back with the PC).  However, it doesn't seem like the wireless devices have the same throughput.
>
> Has anybody tried something like this?  What's the biggest pipe I can expect if I can't connect to my remote USRPs directly with a CAT-5 and two gigabit Ethernet switches?
>
> Thanks!
>
> ---
> Brandon
>
>
>
>
> _______________________________________________
> USRP-users mailing list
> USRP-users at lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com





More information about the USRP-users mailing list