802.11n draft 1.10 - A kinder gentler neighbor

In summer of last year I wrote a cynical blog about "How to jam your neighbor's Wi-Fi legally".  I was referring to the fact that "Draft N" products were achieving their high throughput at the expense of annihilating their neighbor's 802.11 b/g Wi-Fi access.  As soon as the draft 802.11n MIMO devices came on line and started transferring data, neighboring legacy 802.11 b/g devices saw their throughput flatten out to zero.  While this may not concern some people, they should be concerned because wireless interference ultimately hurts everyone.

802.11 b/g devices use a block of spectrum in the 2.4 GHz range that is 20 MHz wide.  Here is a break down of how the non-overlapping channels are allocated:

• Channel 1 uses 2402 MHz to 2422 MHz.
• Channel 6 uses 2427 MHz to 2447 MHz.
• Channel 11 uses 2452 MHz to 2472 MHz.

The earlier 802.11n devices used double the amount of spectrum with 40 MHz wide bands centered around channel 6.  Unfortunately there are only three of these non-overlapping channels in the 2.4 GHz space for North America.  This meant that any device that is centered around channel 6 using 40 MHz of bandwidth would run all the way from 2417 MHz to 2457 MHz which is the almost the entire usable spectrum.  This caused interference with neighboring devices operating on every available channel in the 2.4 GHz spectrum.  If that wasn't bad enough, not only was 802.11n using the same radio spectrum but it was also a lot more aggressive in that space because of the nature of MIMO technology transmitting so much more data.  While these original 802.11n draft standard devices were supposed to back off and only use a 20 MHz wide band when their neighbors were using channel 1 and 11, the problem was the definition of "use" meaning they probably had to be 100% active or get bumped off.  Even when the neighboring devices were attempting to hit 100%, they never got the chance to get to 1% operating throughput let alone 100% so they were effectively strangled by the powerful draft N device.

As a result of the early disaster in the first generation of 802.11n draft devices, wireless chipset makers like Atheros (which I interviewed for this blog) began working on "politeness algorithms" that would interact nicer with the communications of in-range 802.11 b/g legacy access points and their clients.  According to Atheros' product manager CTO Bill McFarland, Atheros designed their "politeness algorithm" with a threshold of 30% for neighboring legacy devices.  That means if a neighboring device is using their designated channels to transmit data more than 30% of the time, the Atheros device would back off to 20 MHz operation.  But the IEEE 802.11n standards body were concerned that this would not be friendly enough so they implemented a stiff policy in the new 802.11n version 1.10 draft that ANY attempt by neighboring legacy devices to transmit data (excluding management frames like SSID beacons) would force an 802.11n device to pull back to 20 MHz operation for a period of 30 minutes.  But since it's highly improbable for PCs that are turned to avoid sending any data packets even when they're idle, this effectively means that 802.11n draft 1.10 devices will almost never get to operate in 40 MHz mode if they're operating within range of a legacy device which effectively cuts their speed roughly in half.

Note that half of 802.11n speed is 50 mbps sustained data rates which is still twice as fast as 802.11g which operates at roughly 25 mbps sustained throughput rates.  802.11b which is the oldest common Wi-Fi standard could only sustain a little more than 6 mbps which is one quarter the speed of 802.11g.  Full speed 802.11n can sustain nearly 100 mbps while in 40 MHz mode though some MIMO implementations like Airgo (recently acquired by Qualcomm) managed to go much higher than 100 mbps.  Airgo is claiming 802.11n Draft 2.0 in their latest product even though 2.0 hasn't even been signed off yet but they're claiming sustained data rates of 160 mbps though I haven't seen this new benchmark independently verified.  What I do know is that Airgo has led the MIMO speed race for nearly two years on every independent benchmark I've seen so it's certainly possible but I'm going to try and get my hands on some of this gear to see if it's hot stuff or hot air.

What all this means is that while 802.11n will offer a huge 4x or more performance boost in theory over legacy 802.11g, it will most likely only be able to deliver about 2x in most realistic environments since most people have neighbors with Wi-Fi.  One other change that Atheros told me about was the fact that their latest product centers around channel 3 or channel 9.  What that means is that it can take up the space of channel 1-6 or 6-11 which increases the opportunity to run in 40 MHz mode if there is a single legacy device neighbor on channel 1 or 11.  The problem there is that most devices default to channel 6 which would always prevent 40 MHz operation unless you can convince that neighbor to move to 1 or 11.  This is the whole tragedy of 802.11n acting as a great leap backwards to 2.4 GHz only operation by shunning 5 GHz operation which has at least 3 times the amount of channel space but we'll leave this discussion for tomorrow.

The last bit of warning that I will leave you is that no chipset or wireless gear product maker will guarantee compatibility with the final version of 802.11n for the current draft products.  They will all tell you that their products will be firmware upgradeable but none of them will commit this to writing so you will have to take your chances.