# Do beam-forming devices expect a particular antenna configuration?

Do beam-forming devices expect a particular antenna configuration? For instance, do they expect the antennas to be a certain distance from one another? Or parallel to one another? Etc.

Background Info:

I have a beam-forming capable WiFi router located in a utility room. I would like to detach the antennas and add extender cables, so I can run them to various parts of the attic. I am concerned that whatever calculations are necessary to perform beam-forming may include hard-coded values based on the current configuration, e.g., each antenna is placed exactly 2 inches apart, and parallel to one another.

• Yes. Relative antenna location and orientation are important. The length of the wires (i.e. time delay) to each antenna is also important.
– AJN
Commented Aug 3, 2020 at 13:04
• I suspect the software has to be smart enough to deal with mismatches antenna lengths. People are going to do things like put the unit on a metal shelf, surround it with other metallic equipment, use it in complex spaces with lots of reflection, etc that are going to screw up simple assumptions. Never looked at code for a router, but I bet it's actually doing some kind of optimization on the antenna delay lengths anyway. Commented Aug 3, 2020 at 13:31

Yes, it will be important to keep all antenna extensions equivalent to ensure the propagation delays are equal. You will want to keep any changes equivalent to each antenna to minimise efficiency loss. It’s probably best to keep the spacing the same, if you do increase the spacing, it's probably best to increase it between all the antennas.

I have done this before and found that in the technical doumentation the manufacturer recommended how to reposition the antennas and defined a maximum cable length, so it might be worth looking if your model has the same.

• The beamformer needs to have knowledge of the antenna array's geometry. That geometry by no means has to have equal propagation delays – in fact, in three-dimensional space, that would very much limit the possible geometries, which isn't the case. Commented Aug 3, 2020 at 15:30
• I agree, fundamentally the beam forming comes down to the combined position of the 3 antennas. Therefore, I recommended looking at the technical documentation of the router to see if they recommend how to approach this problem. One of the most likely key characteristics will be to keep the antennas in the same 3-dimensional plane as they are on the router, but to re-mount them at the new position with identical length & type cable with the same connectors to minimise impedance differences. Commented Aug 3, 2020 at 15:36
• there's nothing in here about three antennas. And impedance has really nothing to do with this. Commented Aug 3, 2020 at 15:43
• True, I did assume 3 just because my router has beam forming & has 3 antennas. I understand that this could be any number of antenna's greater than 1, the point still stands. Would it not be a concern when extending an antennas connection that you could mismatch the network enough to alter the characteristics of one node? Or are most routers smart enough to detect this and automatically adjust transmit/receive levels to compensate for it? Commented Aug 3, 2020 at 15:52
• I think "extend" in OP's question is about the size of the array, not of the individual antenna. Commented Aug 3, 2020 at 16:06

There are two classes of beamformer. One type has the antenna configuration hard-coded into its mathematics, and attempts to construct geometrically beams towards wanted sources, and/or nulls towards unwanted sources. The other type uses general complex matrices to maximise SNRs of expected signals, and to the extent that it 'beamforms', the positions of the antennae appear implicitly in the operation of the optimaistaion during channel measurement.

The 'beamformer' in a MIMO WiFi base station is of the second type.

There may be second order effects if the algorithm used expects the antennae to be within a wavelength of each other, as would be the case in a physically small router. Widely separated antennae, although better than close antennae in some respects, have a larger dPhase/dAngle_of_arrival, which may potentially cause some training algorithms to arrive at sub-optimal solutions.

Do beam-forming devices expect a particular antenna configuration?

Well, they need to be designed to work with the particular antenna configuration you have.

But that constellation doesn't inherently have to be a particular one. Sure, some shapes might be advantegous for specific directions, but you'll find beamformer arrays in all kinds of arrangements; from simple antennas in equal distances on a line to honeycomb constellations to satellite constellations in space.

Note that for beamforming behaviour to emerge, i.e. happen "coincidentally", a MIMO precoding (on transmit side) or maximum ratio combining (on receive side) algorithm doesn't need to know anything about the geometry of the antenna array at all - it happens automatically if the channel coefficient are such that there is a strong dominant path from a single direction, and there's enough antennas to deliver the degrees of freedom.