I can answer the second thing quickly, “stacking” the ssids as you say, makes the inference into what we call “co-channel” interference. Most of the wireless headers are unencrypted, though your payload is encrypted (the data being transferred) but not so much for the headers. Because of this, and the fact that each ap is talking on the same frequency, there’s a small amount of collaboration that can occur between wireless networks. If someone starts a wireless multimedia (WMM) session that will last 8.2ms, then all radios on channel will know that the channel is occupied for the next 8.2ms, and basically go idle for that much time. If you’re on differing channels, but still interfering, aka adjacent channel interference, then those messages may not be understood, causing a lot more collisions. Collisions being when two radios transmit at the same time and the channel needs to clear and everyone backs off, and you try again (usually happening on the order of milliseconds, and possibly several times per second). Collisions will negatively impact your performance more than the channel simply being busy. The protocol in use for wireless collision avoidance is CSMA/CA or carrier sense multi access with collision avoidance, which is an amended version of CSMA/CD (collision detection) used in half duplex wired communication.

How’s that for a “short” answer?

For channel conditions, I’m looking at walls, building materials, open air distances, appliances, furniture… Anything that may attenuate, reflect, refract, or otherwise degrade signal strength whenever I start to assess an area for wireless. This is important so I know how many access points I need and how close together they need to be to overcome the obstacles placed in the environment. Once I have a rough idea of how many access points I need and how close together they should be, my next consideration is the expected client density and the objectives of the network. Something made for a busy stadium will have more access points than something made for a local cafe. If I’m doing a large number of access points my focus will be on maximizing how many clients can be connected, and driving that number as far down per access point/radio, as possible. Fewer people on a channel means more free airtime for their traffic, which equates to faster speeds. I’d be looking at using most of the 5ghz spectrum on the smallest channel width and have each radio be on its lowest power setting. You’ll have clients moving between access points a lot, but you won’t end up with more than a couple dozen per radio. I’d look into directional antennas, to minimize the broadcast range so I can reuse channels closer together. In such a high density space I would want to have some kind of Wi-Fi blocking or attenuation tech installed in the exterior of the building to prevent outside signals from coming in and inside signals from going out. Both for security and control over the airspace. Fewer things to interfere with; you only have to worry about what’s inside that perimeter. Then it’s a matter of setting up the channels for use in accordance with local laws, and letting the system handle channel assignment. With a huge number of access points, manually setting the channels is impractical. So everything I’ve said about it until now isn’t even for channel selection, it’s all things that support channel selection.

For small networks, especially in high density scenarios where the density is due to neighbors, whether that’s commercial neighbors in a plaza or mall, or residential neighbors if you’re in a suburb, an apartment, or a condo; for this, you want to pay careful attention to not only what other networks are around, checking from multiple points not only inside but outside of the premise as well, but what channels they’re on and what their relative signal strength is. If signal strength is low then not a lot to think about. Avoid the channel if you can, but if you can’t, there are worse selections. I’m also looking at the attenuation obstacles here, environments with large obstacles will benefit from lower band channels, either 2.4ghz or UNII 1 for 5ghz, and environments with a lot of radios on the 5ghz spectrum, may benefit from enabling the UNII 2 DFS channels (dynamic frequency selection). A lot of cheaper gear can’t operate in the UNII 2 DFS space because they haven’t bothered to implement DFS, which is a legal requirement for anything operating in that band. So the guys in the apartment next door that are using an off the shelf, cheapo router on sale from best buy probably won’t have the ability to even select those channels for use, and you’ll be free to use them with little to no interference… Unless the DFS triggers that is… For less dense areas I want to tend towards UNII 1 and 3 for stability, and only have enough 2.4ghz to cover the area. 20 MHz wide channels on 2.4ghz, 40 wide on 5/6Ghz. Should net about 400mbps or so per radio, and unless you have gigabit + Internet, with everything on Wi-Fi, some remarkably clear airspace, and only a single access point, going to 80mhz channel width is usually unhelpful. I’m looking at not only the channels with low/no occupancy, but I also want to look at how busy those channels are, but this aspect usually requires monitoring over a duration of time, with specialized hardware. I would choose to overlap with a dormant network with a stronger signal, than overlap with a network that is much weaker in signal strength, but very busy all the time. I also prefer channels 1/6 on 2.4 GHz because channel 11 is near the upper limit of 2.4ghz, and just above that limit is the frequency used by microwave ovens. If any microwave ovens don’t have perfect shielding and you’re on channel 11, you’re going to have a bad time. In environments with more than one access point on 2.4ghz, I don’t worry too much about it since any affected client can hop to another access point when interference ramps up.

There’s more but my brain is tired today.