Clear-channel assesment question.

[igorfi]

Good Evening Folks.



I'm reading a book about 802.11n and found a very interesting thing. Here is the quote:



"Energy detection Energy detection is based on a simple rule that a strong signal will block 802.11 transmissions. To address the need to defer transmission in the face of strong interference, 802.11n specifies that a signal that is 20 dB above the minimum sensitivity will also cause the CCA to set the channel busy. That is, energy of -62 dBm is strong enough to cause the medium to be busy, even if no signal can be decoded"



Gast, Matthew S.. 802.11n: A Survival Guide: Wi-Fi Above 100 Mbps (Kindle Locations 1335-1339). O'Reilly Media. Kindle Edition.



The quote seems like applies to the energy from a non-specified channel? So if my AP which is set to ch 1 will sense the -62 dBm signal on any channel (for intstance 11), then it cannot transmit? The book is about 802.11n protocol...

 

[TJ Hooker]

I can only assume it's talking about in-channel energy. If not, then it would basically be pointless to even have different channels in the first place.

I'm pretty sure the AP would tuned to whatever channel it's configured on, in which case I don't think it'd even be capable of measuring power across the entire wifi band.

 

[igorfi]

Hi TJ Hooker.

That's what confused me, because for the 'in-channel' he wrote a different thing:

"Signal detection: When an 802.11n transmission at -82 dBm is received, the CCA is busy. -82 dBm was chosen as the cut-off because it is the minimum sensitivity specified by 802.11n. (Many products are able to substantially exceed that sensitivity level.)

Gast, Matthew S.. 802.11n: A Survival Guide: Wi-Fi Above 100 Mbps (Kindle Locations 1333-1335). O'Reilly Media. Kindle Edition.


So i'm puzzled here. Unless he was talking about the power of the noise floor (in my first post).

 

[TJ Hooker]

Hmm, it might be talking about other wifi interference vs general interference. Note the difference in wording:
"When an 802.11n transmission at -82 dBm is received, the CCA is busy."
Whereas in the first text you quoted, it's just talking about any sort of energy (not necessarily other wifi transmissions). So the CCA busy threshold is lower if it's actually another wifi device trying to transmit on the channel, but has a 2nd, higher threshold for general noise/interference.

That's my take on it, could be wrong.

 

[igorfi]

Hi TJ Hooker. I have re-read this a few times and even returned to a few earlier paragraphs is this book. This is what i found there:

"If the channel is busy, the device must defer transmitting according to the 802.11 protocol rules. CCA rules are built by choosing a level at which the CCA mechanism must detect a transmission, typically in dBm, and stating that any signal stronger than the threshold results in a CCA reading busy. As with previous PHYs, the CCA is based on both signal detection, where a device receives and decodes an 802.11 transmission, and energy detection, where a device receives a transmission whose energy is far enough above the noise floor that it will interfere with an 802.11 transmission. The noise floor is the ambient wireless signal in the area that transmissions must rise above to be heard"


So this means, that if the noise floor is higher than -62 dBm (regardless the channel. This could be a bluetooth signal or a microwave oven), then the AP can consider the meduim busy and refrain from transmitting. Am I right here?

Gast, Matthew S.. 802.11n: A Survival Guide: Wi-Fi Above 100 Mbps (Kindle Locations 1324-1329). O'Reilly Media. Kindle Edition.

 

[TJ Hooker]

I still think the 62 dBm relates to in channel noise. I'm pretty sure that making the AP able to measure out of channel signal strength would need to be a deliberate design choice that would add complexity/cost while seemingly offering no benefit. I just don't see why the AP would care if there's a bunch of interference in some other channel when deciding whether to transmit on its own channel.

So the AP would only look at in channel signals strength, but would have two thresholds for deciding the channel is busy. A lower one if it's seeing another wifi signal, and a higher one for non-specific noise/interference.

Again, this is just my interpretation of the paragraphs you've quoted.

 

[igorfi]

I was thinking the same way up until now. But the Bluetooth radio doesn't operate on a particular channel and occupies pretty wide spectrum within the 2.4 band. But an AP would see this 'noise', but couldn't decode it (but still would avoid from transmision). Hence it was named as an 'Energy detection" as oppsed to 'Signal detection'. This applies to 802.11n protocol specification though. Ohter protocols can have different tresholds. I know it sounds shocking, but the book is very famous (actually three books about 802.11a/b/g, 802.11n and 802.11ac).

I guess not all the APs are using both - a Signal detection and an Energy detection. But then again, it is a 802.11n specifcation. So the chips should comply to these rules.