Question Mesh WiFi and Bandwidth?

[oguruma]

I'm not sure I understand how available bandwidth is affected using Mesh setups.

Suppose I have a 4x4 MIMO dual-band WiFi 5 AP. The theoretical maximum bandwidth is 1700Mbps.

What happens if I have 3 of those all meshed together with a wireless backhaul? What would the theoretical bandwidth be then? (1700 / 2 ) / 2? Assuming they're all position in a straight line, and AP3 has to talk to AP2, which talks to AP1 (also the router)?

 

[hotaru.hino]

Bandwidth is as much as the narrowest pipe in the chain. If they're all theoretically 1700 Mbps, then that's your bandwidth. But if any one of the nodes has issues maintaining that speed, then it drops to that speed assuming traffic goes through the node.

 

[oguruma]

Bandwidth is as much as the narrowest pipe in the chain. If they're all theoretically 1700 Mbps, then that's your bandwidth. But if any one of the nodes has issues maintaining that speed, then it drops to that speed assuming traffic goes through the node.

How does that work if they are meshed together with a wireless backhaul (without a dedicated backhaul channel)?

If each AP has 1700Mbps bandwidth (4 433 Mbps chains), but they have to talk to the clients, and also talk to each other, how would it be 1700Mbps back to the router, if the device is connected to the third AP in the chain?

 

[hotaru.hino]

If each AP has 1700Mbps bandwidth (4 433 Mbps chains), but they have to talk to the clients, and also talk to each other, how would it be 1700Mbps back to the router, if the device is connected to the third AP in the chain?

Any sort of bandwidth prediction over a wireless medium is going to be theoretical at best. The actual performance depends on a lot of factors. But again, the maximum bandwidth that transfers between two computers is capable of over a network is whatever the slowest piece in the chain is.

If you're trying to optimize your wireless network setup, there are some RF simulators that try to predict coverage (but it doesn't really work with bandwidth). Otherwise, you're just going to have to set a budget, buy the best things you can afford, and experiment with placement and RF settings.

 

[bill001g]

It appears you have already figured out why mesh/repeaters work so poorly.....the router manufacture want you to think this can magically create wifi bandwidth.

As you suspect running a repeater without a dedicated backhual will reduce the bandwidth by 1/2. It likely is much much more since wifi is half duplex and it will be likely the signals collide and damage each other. For half duplex to work all the devices must be able to hear each others transmissions but if you can hear the main router from your end device you don't need a repeater.

This is made even worse because of all the lies they tell about wifi speeds. For most people unless you sit on top of the router you are not going to get much over maybe 300mbps maximum. You start adding repeaters to that and it drops the speed even farther.

Even if you were to use dedicated backhual device you now have a new problem. Assuming you live in a country that has 2 80mhz radio band you now have doubled your chance of taking interference from neighbors. In many cases a single house is attempting to use every radio band on the 2.4 and 5g so we have massive interference on all wifi.

Pretty much I say you only use a repeater when any signal is better than no signal at all.

 

[oguruma]

It appears you have already figured out why mesh/repeaters work so poorly.....the router manufacture want you to think this can magically create wifi bandwidth.

As you suspect running a repeater without a dedicated backhual will reduce the bandwidth by 1/2.

What is the effect of multiple meshed-together APs on bandwidth? Suppose I am trying to get WiFi coverage down a long hallway, and I have 3 APs all in a straight line. Assuming they all support 1700Mbps (theoretical), does it go: 1700 > 850 > 425?

Meaning, it's not just "halved" but it's halved for each "hop"? Is that how it works?

 

[bill001g]

First this only applies if you are talking repeaters not AP.

A AP has a ethernet cable to it so it has the full gigabit back to the main router. Now if you were to daisy chain the AP rather than run a seperate cable back all the AP would share 1gbit BUT unlike a repeater the data is only sent 1 time.

The repeater problem is far more complex that just a reduction in the bandwidth. Very technically the 3rd repeater would drop it by 1/3. You now have 3 copies of the data being send into the same fixed radio bandwidth. ie router--r1, x1-x2, x2-end device.

The extra bandwidth used is not actually the major issue. The problem is that you have more than 1 data packet going at the same time. So lets say the first packet has gone between the router and the r1 and then from x1 to x2 and is now ready to transmit to the end device.

At this exact time the router sends another packet to x1. So x2 also starts send data to the end device BUT since it is wifi it also send the same signal back to x1. x1 would know this is not his signal and ignore it but the energy is still in the air. So now you have a signal from the main router and the repeated signal coming back from x2. x1 has 2 signals interfering with each other and many times the data is dropped. So the data must be retransmitted which can cascade until you get session drops. This is only a simple example think of data going both direction from mulitple end devices and then you have interfering signals coming in from your neighbor.

 

[oguruma]

First this only applies if you are talking repeaters not AP.

signals coming in from your neighbor.

Right, that does sound wildly inefficient. There are quite a few consumer-grade 2-3 satellite mesh kits on the market, how do those typically perform?

 

[bill001g]

My standard comment is you use mesh/repeaters when any signal even a bad one is better than no signal. It is a very last option I would recommend when any other form of network will not work.

In general they work ok but not fast. Most people are clueless and don't know how to test the speed. They also install them incorrectly putting them in the remote rooms rather than placing them 1/2 between. In many cases they only "think" it works better when there machine is actually connecting directly to the main router rather than the repeater in the room.