HT Duplicate (MCS 32) and non-HT DuplicateBy CWNP On 12/05/2011 - 14 Comments
Just when you think you pretty much understand 802.11n, you turn the corner and realize there’s more to learn. This time, the topic was non-HT duplicate and HT duplicate. I’d read before about the non-HT Duplicate format, but I had never realized that there is also an HT Duplicate format, which is the use of MCS 32. I'd like to document how they work and what they're for, even if I'm the only one who cares.
HT Duplicate and non-HT Duplicate do the same basic thing, but for different reasons. They both copy the same basic frame in both 20 MHz halves of a 40 MHz channel.
Non-HT Duplicate Protection
Let’s first look at non-HT duplicate with an example of an AP and a 40 MHz channel. When the AP has a 40 MHz data frame to send to a 40 MHz client station, the AP may need to protect its transmission from 20 MHz stations in either (or both) the upper or lower 20 MHz portion of the 40 MHz channel. So, it could send a CTS to the lower 20 MHz and then send a CTS to the upper 20 MHz; or it could just send them both at the same time, which is obviously more efficient. The idea here is simply to send two legacy-formatted CTS protection frames at the same time in both 20 MHz halves. This sets the NAV of all stations in both halves to “pave the way” for the 40 MHz transmission. Easy enough, right?
Duplicate HT (MCS 32)
There is also such a thing as an HT Duplicate transmission, though it isn’t called that in 802.11n. It’s actually just known as MCS 32. The Wi-Fi Alliance has an optional certification for this feature, called HT Duplicate (MCS 32). MCS 32 serves up the lowest MCS rate for 802.11n, and surprisingly enough, it’s also a 40 MHz transmission. But wait…I thought 40 MHz was for high bandwidth, throttle to the floor, 802.11n sweetness! You were right, except for this exception.
MCS 32 is an 802.11n frame format that duplicates a data frame in both halves of the 40 MHz channel. By duplicating the same frame in both halves of the channel, the reliability is improved at the receiver. This is especially useful for 40 MHz receivers when a normal 40 MHz data transmission at a low MCS rate becomes unreliable. MCS 32 uses BPSK modulation with 1/2 rate coding and 48 data subcarriers (per 20 MHz). In other words, MCS 32 is a 6 Mbps OFDM (802.11a/g) transmission. But, it’s duplicated in both halves, creating diversity gain for the 40 MHz receiver.
At 6 Mbps, the data rate is not exciting, but the added reliability is the primary goal. The drawback of MCS 32 is that it still occupies 40 MHz of the channel, even though it is a low data rate. The advantage of MCS 32 is that for a simple 802.11n receiver with a 1x1 radio, the diversity gain can improve reliability such that the packet error rate (PER) is considerably lower than it would be for MCS 0. When you compare the data rate of MCS 32 (6 Mbps) to MCS 0 (6.5 Mbps), it’s a nominal difference, but the added reliability of MCS 32 at range is always welcome.
Final Comments and Suggestions (FCS)
Most enterprise APs aren’t using MCS 32, but a few are. Client products and SOHO APs are more likely to support it—based on Wi-Fi Alliance product certificates. You could make the argument that channel reuse is less of a concern (40 MHz is a given) at home and you’re more likely to have clients at the edges of good service where range improvements are more important. But, it’s more likely that MCS 32 support in SOHO APs is just a byproduct of the chipsets used for SOHO products, which attempt to implement as many features as possible. It’s somewhat unlikely that you’ll run into this feature in the enterprise, but if you do, you’ll know what it is. I find it interesting that 802.11n was so densely packed with new, important features that the minor ones like non-HT duplicate protection and MCS 32 Duplicate frames are almost completely ignored…almost.Tagged with: 802.11n, wi-fi alliance, MCS 32, Duplicate HT, non-HT Duplicate, Reliability, Diversity gain