7

Vague Question: If one access point can do the job adequately is there still an advantage to using more than one?

Scenario: 1 access point at the center of your network required coverage area should adequately cover the area. There will always be less than 50 wireless clients and 0 wired clients. Most of the clients will move throughout the required coverage area while in use and some would like to go just beyond the perimeter; but don't have to. A gigabit backbone is available for access points to be placed anywhere. There are neighboring wifi networks, but only horizontally.

Specific Question: In this scenario is there an advantage to using multiple (say 2 ~ 4) access points and placing them near the perimeter of your required coverage area?

My 2 Hypotheses:

Hypothesis 1: With 1 access point at the center of the required coverage area broadcasting on dual bands with 1 channel each it will be easier to find clear channels. Clients will not have to migrate from access point to access point as they move throughout the covered area; possibly dropping the connection momentarily, or just failing to migrate at all. By maintaing a smaller perimeter there is less chance of interference from other neighboring wifi networks and equipment operating in the same band.

Hypothesis 2: With multiple access points near the perimeter of the coverage area more channels should be used; 1 channel for each access point (or should 1 be used for all since it is the same network?) possibly limiting it to 3 non-overlapping channels and 3 access points. Clients may migrate well from access point to access point while moving. They can leave the required perimeter somewhat. Everyone has a stronger signal, a lower chance of dropped packets, etc. Interference may be a bigger problem, especially with neighboring wifi networks.

NOTES: They will all have the same SSID and WPA2 Passphrase, SSID will be hidden and MAC filtering will be used, 802.11n or 802.11ac will be used. I have a clear understanding of networking and I'm not asking how to do this. I'm asking which is better based on scientific evidence. I don't have the budget to a-b test this.

Maybe this is really a question of how well will the clients migrate on their own, or maybe it it's more than that.

5

In the scenario you describe, you should definitely be looking at multiple access points, preferrably dual band APs.

While coverage may be sufficient, coverage alone is no longer the primary consideration when deploying a wirelss network. Client capacity, channel utilization, signal quality, and reliability are much more important and multiple access points will help with all of these.

By using 3 (or more) APs on multiple channels (1, 6, and 11), you will in effect triple the amount of airtime (bandwidth) available on your wireless network.

Additionally, proper placement of the APs will provide clients a closer AP with stronger signal, which will be more resistant to noise in the RF environment. This will allow better signal-to-noise (SNR) ratios which will translate to the use of higher data rates and this results in more data transmitted per "timeslot".

I would recommend placing them 2/3 to 3/4 of the way from the center to the perimeter, spaced roughly evenly. Try to get them in or as close to the highest user denisity locations as possible (i.e. conference rooms, etc).

Finally, the additional access points will provide increased reliability. With a single access point, if it were to fail or reboot for any reason, this would create a disruption in service. Having multiple access points should allow for coverage to overlap, allowing service to remain (if degraded) when you have an access point down.

  • 1
    I think my only remaining concern is: Client associates with ap1; client is relocated to where the signal from ap1 looks okay to the client, but packets are starting to be dropped; ap2 has an excellent signal and if the client would just migrate it would not be dropping packets. How bad are clients at holding on to a poor signal ap for too long? My experience has been they hold on until the signal about dies unless you cycle your wifi off and on. – David O'Brien Apr 1 '14 at 13:20
  • 1
    @DavidO'Brien, unfortunately, this can be largely dependent on the client software/driver itself. Most have gotten significantly better about making roaming decisions over time. There are also vendor solutions that help to mitigate these problems as well depending on access point selection (for instance Aruba has a feature called ClientMatch to help prevent "sticky" clients). – YLearn Apr 1 '14 at 13:29
2

There are two interrelated issues here: coverage area and channel bandwidth/utilization. Multiple access points will extend the coverage area and improve throughput, because signals will be stronger and therefore your clients can operate at higher speeds.

Even if your coverage is perfectly adequate with one AP, having multiple APs allow you to use multiple channels, and this will reduce channel contention and increase throughput. With one AP, all your clients are competing for airtime on one channel; with 3 APs on different channels, for instance, you have three times the channel capacity, and your clients will have greater throughput.

If you have a modern wireless network with a central controller, it can influence how and when your clients chose a particular AP. Ultimately though, the client makes the decision which AP to associate with, and when to roam.

  • I think my only remaining concern is: Client associates with ap1; client is relocated to where the signal from ap1 looks okay to the client, but packets are starting to be dropped; ap2 has an excellent signal and if the client would just migrate it would not be dropping packets. How bad are clients at holding on to a poor signal ap for too long? My experience has been they hold on until the signal about dies unless you cycle your wifi off and on. – David O'Brien Apr 1 '14 at 13:21
  • @DavidO'Brien The decision when to switch APs is made by the client. There is no standard, and every manufacturer does it differently so they can say they're better than the competition. In other words, the algorithm is proprietary. But I would agree that they'll hang on until the error rate at the lowest possible speed gets too high. By limiting the low data rates, you can force the clients to roam before things get really bad. – Ron Trunk Apr 1 '14 at 15:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.