The following information is to highlight the negative impact that occurs when AP’s are installed next to each other <3m, as It is not uncommon in today’s wireless deployment, to see these types on installations. As the old saying goes a picture paints a thousand words, so I have tried to minimise technical jargon, by using pictures.
Figure 1, shows the spectral mask (Shape) of an OFDM modulation pattern, which will be represented throughout. An OFDM spectral mask, is approximately 20MHz wide, and is based on the centre channel. In this example it is based on channel 6.
Figure 1: OFDM Spectral mask
The Spectral mask of any 802.11 modulation contains certain key areas,
- Peak power
- Shoulders
- And where it tapers off into background noise
All have values which are described as decibels relative to peak power (dBr) of the centre frequency
Good design practice is to ensure only non-overlapping channels are used. In the 2.4GHz band to be consider non-overlapping it must be separated by 5 channels or 25 MHz, in the 5GHz band it is 20MHz separation from the centre frequency.
Not only is having non-overlapping channels critical to any good wireless design, so is the importance of AP separation. If AP’s are not physically separated >3m or have some form RF isolation method than interference will occur. This also applies to AP placement when located next to objects, but sometimes this is unavoidable, so correct antenna selection is crucial.
To demonstration this I have setup the following equipment in a small lab scenario as shown in Figure 2, AP1, 2 & 3 at approximately less than 30cm apart, configured on static channels 1, 6 & 11, at transmit power of 20dBm This testing is relative to the equipment used) Spectrum analyser sitting next to the AP’s to view the layer 1(RF) information
Not shown in this picture are test clients, 3x 1×1 Apple IPad mini and Apple MBP that will be used for throughput testing and capturing information.
Figure 2: Lab setup
Shown below in Figure 3. Is the current 2.4 GHz band utilisation.
Figure 3: 2.4GHz band before testing
AP1 is enabled on Channel 1 with a client device conducting a throughput test, as shown in figure 4.
Pay particular attention to;
- The shoulders of the OFDM spectral mask, notice how it bleeds over into adjacent, and non-overlapping channels.
- Notice the channel utilisation, for adjacent and non-overlapping channels.
Figure 4: AP1 on Channel 1 with client running throughput test
I will now disable AP1 and repeat the same test again this time with AP2 on channel 6
Figure 5: AP2 on Channel 6 with client running throughput test
This time leaving AP 2 still enabled, I will re-enable AP1. Notice that the shoulders of both the spectral masks have bleed together.
Figure 6: AP1 on channel 1 & AP 2 on Channel with throughput testing being conduct with Clients
Now compare the channel utilisation from figure 5 to figure 6, notice the increase that has occurred.
AP 2 is disable and AP3 is enable on channel 11 also with a client conducting a throughput test
Figure 7: AP1 & AP3 with throughput testing being conduct on each channel
Figure 7 shows utilisation occurring on channel 6 regardless of any stations. Take note on the channel utilisation in figure 7 and compare it figure 5. The utilisation on channel 6 in figure 5 is almost the same as figure 7.
This time I will re-enable AP2 with client. Take note of the overall increase in 2.4GHz band utilisation and the shoulders of the individual spectral masks.
Figure 8: AP1, 2 &3 enabled with clients running throughput tests
Now let’s compare the difference when AP’s are physically separated.
AP1 and AP3 are separated by 5meters, the spectrum analyser is located in the middle
Figure 9: AP 1 & 3 separated by 5m
Notice the spectral mask shoulders and overall channel utilisation of the band compared to previous test shown in figure 6.
This time I have changed the channels but the distances remain the same.
Figure 10: AP 1 & 3 enabled with throughput test
Compare figure 10 against figure 6. Massive difference. All 3 AP’s are now separated by distance of 5meters.
Figure 11: AP1, 2 & 3 enabled with throughput test an AP’s
Notice the difference in overall utilisation when figure 11 is compared to figure 8.
Hopefully this shows that physical separation is extremely important to minimise interference, however testing should always be performed as the distance required is dependant to the AP, antenna and EIRP selected. To determine the required separation those factors must be account for.
As mentioned this was a small test lab, if this was in production the impact would have been magnified.
