WiFi restoration

[Doug]

On 04/23/2015 06:29 AM, Timothy Murphy wrote:
> One room in my house is at the boundary of WiFi reception,
> and WiFi occasionally fails there.
> When this happens it is nearly always restored by re-booting.
> Re-starting NetworkManager never does the trick, however.
> Is there any other step I could take, short of re-booting?
> I'm running Fedora-21/KDE.
>
WiFi is radio frequency signals. What you need to do is take steps to 
make the signals stronger in that area.
One of the possibilities is making the transmitting antenna higher up,  
or closer to the weak area. The height may make the signals clear 
obstacles which could be blocking--such
things a filing cabinets, refrigerators, other large metal objects--a 
metal-clad fire door, for instance. Closer, of course is obvious, but 
may be inconvenient. If the receiver in
the weak area of the house is fixed in location, then moving that 
antenna to a closer or higher spot may be the answer, but if the 
receiver site is mobile--a laptop, for instance,
then that is not going to be feasible.
If both antennas are or can be fixed in location, then it is 
possible--likely, even--that there is a spot on the "weak" area that has 
stronger reception. Due to reflections, there are
peaks and nulls in signal strength, and if the antenna in the weak area 
is in a null, just moving the antenna a foot or two in some direction 
may be all you need.
Finally, there are repeaters--devices which receive the signal on one 
frequency and retransmit it on another channel. The repeater can be 
placed somewhere in between the
two areas. The repeater will only need AC power and a clear signal 
path--on top of a high piece of furniture, or perhaps on a small bracket 
high on a wall.

Note: I have used the terms "transmitter" and "receiver" for simplicity, 
but in reality, of course, signals are being transmitted and received in 
both directions.

Why does rebooting make it work? Probably because the signal is 
digitally encoded, on the very edge of detection, and the decoding is 
marginal. While it works, the clock is
locked up, and when it fails, the clock is free running, and can't find 
enough signal to lock up again. When you turn off the signal completely, 
during reboot, the receiving
end settles down and is likely to be close to the necessary frequency to 
begin with.  Just a guess, but it's reasonable.

Hope that helps.

--doug, retired RF engineer