Wi-Fi channel planning is often a difficult problem. There are only so many channels and neighboring networks use the same channels. You should set the power level to match user devices. There are many kinds of obstructions which will affect the coverage. Luckily vendors offer a seemingly simple solution to this complex problem: let the automation take care of it. Practically all systems have an auto setting which is supposed to find the optimal solution. What does it really do?
In the simplest case the auto setting just means that the access point will choose the least occupied channel at start up after a short scanning period. Transmit power is typically “full on”.
The problem with the simplest approach is timing. If you are like me, most of the installations happen in the evening or over the weekend. You turn on the device when all the neighboring offices are closed and their Wi-Fi networks idle. It will be a totally different environment when all the offices are fully manned during work days. However, simple systems won’t change the channel once it has been selected. The change would break all active network connections and the access point can’t know when it would be a good time for it. You could be stuck with a very poor channel choice until next restart.
Some systems can schedule channel reselection. However, if you set it to occur at night, the environment won’t reflect the actual use case during the day. If you schedule reselection during daytime you will risk user connections.
Multi access point systems are vulnerable to a special case with electrical black-outs. After a black-out all access points are started exactly at the same moment. They will scan the channels at the same time. As a result, they will often choose the same channel. The net result is the worst case, since the access points will need to alternate with their transmissions.
Selecting the proper transmit power level is also a difficult task. Full power is the least risky choice for the vendors, since client devices will show full bars as the signal strength. The devices may struggle with data transfer, however, since Wi-Fi is always bidirectional. The client devices need to transmit to the access points and especially phones have much lower transmit power than access points. You will end up with poor transfer rates even though the signal appears to be great – in one direction.
Larger Wi-Fi systems usually have a centralized management controller. The controller collects data from the access points and can see the big picture. You can get much better results with centralized systems. Cisco WLC is a good example of a smart system.
Actually centralization is not the key. Many low-end centralized systems leave the channel choice to the access points. They are no better than simple systems described above and have the same problems. On the other hand, Aruba Instant is a distributed systems (without a central controller) but the access points will negotiate with themselves and get the big picture. Actually Aruba’s automatic channel management is one of the best, even when it is distributed. Aruba has a centralized controller, too, if you have dozens of access points.
A smart system will know the transmit power of each access point. It will also collect information on how the access points are receiving each other. With transmit power levels and remote signal levels the system can create a network map, which will show how the APs are located relative to each other. With this information the system can choose channels for the APs so they are not interfering with each other. The system can also deduce that at the midpoint between the APs the signal is twice the strength the other AP is receiving. The system can therefore set the transmit power levels so that there is required signal level between the APs.
This all requires dense enough access point placement. In theory the APs wouldn’t need to receive each other at all. It suffices if the the coverage reaches the midpoint between APs. However, in that case the system can’t build the network map to base its choices on. However, with a denser AP placement the system can compensate for an AP that is malfunctioning or being upgraded for example. The system will just turn up the transmit powers so the neighboring APs will cover the hole.
A smart system can also adapt to changes in the environment. When an access point finds a new transmitter on its channel it will report it to the controller. The system can then change the channel plan to accommodate the intruder.
System stabilization will take some time since the process is iterative. A new Wi-Fi network may need a couple of days to reach a stable state. Occasionally automatic adjustments can lead to an unstable state, where the channels are constantly being changed. For example, an office building with a central elevator shaft may cause the channel changes to circulate around the floor incessantly.
Fortunately transient transmitters like personal hotspots or passing trains or buses, are typically on 2.4GHz. 5GHz band tends to be more stable.
Manual channel planning yields better results than simple automation.
Smart system are useful, at least when the environment is stable enough. You just need to verify that all useful channels are being utilized. Many systems will only use non-DFS channels on 5GHz, even though they default on 40MHz channel width. That will leave you with only two channels and inevitable interference problems.
Even the smart systems tend to set the transmit power levels quite high. The aim is to please customers with “good coverage”. In many systems you can set the maximum and minimum power levels. However, in practice the system will use the set maximum power level for all access points. The down side of limiting the automation is that the system can’t exceed the set limits even in an equipment failure case.
The conclusion is that smart system will help with Wi-Fi management, if you have familiarized yourself with the system and its settings. No automation will completely remove the need for skill and knowledge of the environment. Smart systems are especially good at coping with unexpected changes in the environment.