Answer 2: gets a bit trickier. The situation you described is most likely a camera inside of a host... you would need to be inside before you can spot it and interact with it. By inside I mean have hacked access. The other option is to actually physically be with your team and you can direct connect to said camera.
That part emphasized above is really a bit irky, as MercilessMing pointed out as well.
If the camera is connected to the host via cable: Yeah, sure. You need to get inside the host to find it. But
if the camera is wireless-active, there
should still be some kind of signal presence. Ergo, it should at least be possible to find out that it´s there
and to find out out the next step to hack the thing without too much guesswork. Meaning: Finding the host and getting access.
My tipp as a GM in this case: Treat the camera like a hidden icon, defending with the host rating - and its severely underused Sleaze rating! - against Matrix perception tests (
the host masks the camera signal for protection, yadda yadda). Success gives you the information that there´s indeed a wireless device which is secured
inside a host (which you will also find out with a success on that action) and that you need to access the host if you want to hack it.
Note that this is not RAW/RAI, but an eyeballed houserule. Works pretty well, tho.
Also note that this setup (wireless cameras slaved to a host) is probably not the most common for a security setup anyways - precisely because of the reasons detailed above. However, wireless cameras are faster to employ for improvised perimeters and some security devices (f.i. drones for example) are obviously hard to employ with cables attached to them. So this really is a sitation you will run into rather often. Better have a clear path for hackers to handle this.
"Direct connection": i left thus a bit fuzzy on purpose with the intent to build on it when I much more word space in a matrix supplement, but in general... if the device is wireless you do not anything special other than being in close proximity... I would say 10 meters or less but thus is just so you can justify being able to pick its signal out without difficulty at that point (I really see no point in overthinking things and having a situation of ... but I can physically see the camera why can't I detect it? Worse case maybe make a matrix perception check) but the point is you can bypass host A at that point ... if not wireless you need a datatap of some type.
Short-range wireless direct connections: Now
that would really be awesome, would love to see that!

A little suggestion to tie that all together (and also stitch
this little mess back up nicely):
- The range for Short-range wireless connections is ithe device rating in meters. If you only have a Meta Link, keep your stuff nearby

- A Network can have a an unlimited amount of Short-range wireless direct connections (SRWDC). All of these are protected by the PAN. Basically, it´s the 6th world equivalent of Bluetooth.
- A Network also can have a (theoretically) unlimited amount of "traditiona" wired direct connections. These have the additional benefit that hackers can´t find them in the Matrix without getting access to the Network.
- For longer ranges, you need to slave the device to be still considered part of the Network (would that be called a WAN at this point? Probably.) and get the proper protection. Think of slaves as the "outposts" of a Network. The limit for the number of "real" slaves is based on the Data Processing Attribute of the Master (if its not an RCC).
- This also means that you can enhance the overall reach of your Network (and its SRWDCs!) through clever slaving and daisy-chaining. F.i., a hacker could slave the commlink of the Infiltration Expert to his/her own Cyberdeck. The Infiltration Expert could then sneak the Commlink on site to establish a (shortrange or wireless) direct connection to a secure device.