Spectral Masks and InterferenceBy CWNP On 09/20/2007 - 59 Comments
The 802.11n draft 2.00 document defines both 20 and 40 MHz spectral masks. A transmit spectrum mask is the power contained in a specified frequency bandwidth, and regulatory bodies such as the FCC (United States) regulate how much power can be emitted from a transmitter at the center frequency and at given frequency points (called offsets) on both sides of the center frequency.
20 MHz channels in 802.11n are the same as specified in clause 17 (previously 802.11a) and later in clause 19 (previously 802.11g). Therefore, 802.11n transmitters using 20 MHz channels use the same spectral mask as 802.11a/g OFDM transmitters. 40 MHz transmitters can use one of two spectral mask formats: 1) 2 x 20 MHz masks or 2) a 40 MHz mask. When a 40 MHz transmitter is using a 40 MHz channel, a 40 MHz spectral mask is used. This mask is identical to that of a clause 17/19 OFDM 20 MHz mask. When a 40 MHz transmitter is using HT duplicate format, two identical 20 MHz masks are used.
Here are illustrations of the spectral masks for 20 and 40 MHz 802.11n channels. First the 20 MHz, then the 40 MHz.
From a practical application standpoint, this matters. Many of today's WLAN designers are looking for a way to have very high user density with a minimum complement of equipment. To do this, they are either using multiple radio APs, locating multiple APs in the same physical area, or both. A new methodology with many names (because nobody has yet stepped in with a standard name for it) is on the rise. Some of its names are "channel stacking", "channel spanning", and "channel blankets". This architecture is generally seen with single-channel architecture (aka Virtual Cell) vendors, but could be implemented in traditional channel reuse scenarios (aka Micro Cell) if carefully planned.
There are a number of things that can cause interference between APs. Two significant ones are:
1) Being on the same channel within range of each other
2) Transmitting at high power with high-gain antennas while being in close proximity
Putting APs on the same channel with a single channel architecture, transmissions must be coordinated at the controller to avoid APs interfering with each other. In a traditional channel reuse architecture, APs on the same channel must be spaced at an appropriate distance with their power adjusted appropriately to avoid co-channel and adjacent channel interference. All of this is no problem in today's deployments. When things become problematic is when you place an AP on channel 6 within only a few feet of an AP on channel 1 (for example) hoping to "stack" channels in the same physical space for the purpose of higher user density. This same scenario would apply both to 2.4 and 5 GHz channels. APs will interfere with each other even though they are not on the same channel when they are co-located this closely. This is the same reason why you don't configure two 802.11 radios within the same AP for 2.4 GHz channels. They're just too close (physically). Why do they interfere? Spectral masks. This interference causes throughput degradation due to the adjacent channel interferece.
For this reason, one vendor has already begun suggesting that if the customer goes with a channel stacking design, they should skip a 20 MHz channel between 40 MHz channels to avoid adjacent channel interference between stacked radios in an area.