HomeAboutMailing ListList Chatter /0/0

Wifi Question

2020-10-14 by: DaWorm
From: DaWorm 
Can anyone tell me how well (or how poorly) a typical Wifi router handles a
large number of low bandwidth connections?  For example, could one typical
Wifi router handle 40 or 50 devices that only needed to send/receive a few
tens of kbits every few minutes?  Or is it the WPA2 overhead that would
bring them down rather than the bandwidth or number of connections?  How
would you handle a large number of Wifi devices, lets say up to 100, with
hopefully a single Wifi router, or can it not be done?


=============================================================== From: Andrew Rodgers ------------------------------------------------------ A good access point should be able to handle that, not likely to find it in a SOHO all in one though. An AP should specify how many simultaneous connections it will support. Interestingly, there's a new standard just for this sort of application called HaLow that will be shipping in the next generation of wifi chipsets. I actually think it's got the best chance of any indoor IoT comms platform, because it will just be built into enterprise WAPs and will get deployed as part of normal lifecycle upgrades, vs a separate special deployment.

=============================================================== From: Dave Brockman ------------------------------------------------------ could WPA2 es, The first problem is, there really isn't any such thing as a typical wifi router. There are dozens if not hundreds of different configurations of hardware, software, and default settings. A 3 or 4 antennae system with MIMO would be your best bet, but would still be less performant than 2 or 3 UBNT Nano units spread across the room/area you are attempting to cover. I might try to put 100 clients on an UBNT AC-HD AP, although 3 AC-Pro units spread would again be more ideal. Routers that have to perform routing, NATing, firewalling would not be the ideal candidate for WiFi provider, I wouldn't think. Their other roles consume additional resources. Dedicated AP only has to do WiFi things. Cheers, -Dave

=============================================================== From: DaWorm ------------------------------------------------------ Unfortunately, in the use case I'm looking at, having multiple devices is cost prohibitive for reasons beyond your normal IT costs. A single all in one Wifi router vs. multiple access points and a wired router to connect them has a large cost difference in both equipment and installation costs, generally meaning having to hire a union electrician to run the wires in conduit, and possibly new outlets to power everything, if not POE. Wifi is used in the first place to eliminate wiring, because running 40 to 100 LAN drops and the switches needed to connect everything is entirely out of the question. The extra cost of separate APs isn't such a big deal in an office environment, and if you're supporting multiple computers, tablets, phones and so forth, it's just part of the cost of doing business. But when this is being paid for out of small value transaction fees in the realm of only a few dollars per day, and where typical break even times on initial investments for everything else besides the networking portion is measured in terms of two to three years, adding $500 to $1500 to the initial investment moves that break even time out significantly. And the needs of the system are far less than the typical office environment in terms of data handling, no pushing videos or transferring large files, just a few small packets every few tens of minutes, where the TLS connection setup overhead dwarfs the actual data payload (and many messages may not need the security of TLS, and as such are usually less than one MTU in size). Firmware updates are the closest thing to a "large" data transfer, and that would be a few 100K to maybe 3M of data, tops, and would be very rare and could be staggered and scheduled for the middle of the night, and if it takes an hour to update instead of a minute, no one really cares. So even if a typical low cost home or entry level business Wifi router would choke hard on 50 PC or phone clients, if it would work for low bandwidth clients like I'm envisioning, it would be good enough. My problem is I don't know if they would choke under those circumstances as well. As you say, there are 100's of models out there, so it might just come down to testing various models to see if they will or they won't work. I was however hoping someone might know about this already and save me the trouble and expense of buying a bunch of different routers to test just to find out if the arrangement can work in the first place. Jeff

=============================================================== From: Aaron welch ------------------------------------------------------ Your explanation makes me wonder if the financial model for your idea isn= =E2=80=99t the problem. I have been guilty more than once of trying to overbuild a solution to a problem that is not worth the effort. If you have a hardware/software solution with a breakeven measured in months, then what is your cost of customer acquisition? n , is N e s y d f he at e ld r Aaron Welch Chief Mechanic @ Geek Ventures 423-505-9999 n2nightfall@gmail.com "Enabling people to do great things with their own ideas."

=============================================================== From: Dave Brockman ------------------------------------------------------ Keep in mind, WiFi is half-duplex (unless your AP and your client have multiple antennae). WiFi is also a single collision domain (think hub vs switch) and is extremely chatty without any clients connected. Would I try to connect 40 extremely low bandwidth devices to an AIO SOHO router, absolutely not. Would I connect 40 clients to a single dedicated AP, sure, I'd give it a whirl. Would I attempt to connect 100 devices to a dedicated AP not designed for high density, absolutely not. I told you what to look for, multiple antennae and MIMO. Give it a shot, the worst that can happen is that you sink the cost of the AIO into the project. I don't know what you're doing, but if $500 extends the break-even time by a term measured in years, I might reconsider the viability of the project. I also don't know your local and org requirements, but cabling can be run without electricians and without conduit -- depending on the physical environment. Cheers, -Dave ngle n ate fice But 0 =C2=A0 g g to a , My s t just and =2E=C2=A0=C2=A0 ample, to t the WPA2 devices, t not l 4 e area n an UBNT be r other do WiFi