Comments on: How does 802.11n get to 600Mbps?

By: Michael

Michael — Sun, 24 Jan 2010 13:28:00 +0000

Here’s a discussion of interframe spaces that might help: http://www.cwnp.com/pdf/802.11_arbitration.pdf.

By: Ron

Ron — Wed, 13 Jan 2010 14:53:31 +0000

I just finished reading “Next Generation Wireless LANs” a Cambridge book. This article was extremely good and helped summarize what I just finished reading. In the book it discussed that the use of RIFs in place of SIFs added significant data throughput (I think Greenfield only). Can someone comment on this?

By: Michael

Michael — Thu, 10 Dec 2009 15:33:48 +0000

802.11n doesn’t include HARQ because:
1. It’s too expensive.
2. And it wouldn’t work anyway

HARQ works by retaining your record of a packet that was received with error and combining it with a retry packet in such a way as to reduce errors. You can only do this if you retain a sample-by-sample record of the packet.

802.11 PHYs are stateless in the sense that they don’t care who spoke to them previously, using what rate or PHY mode; each packet is fully independent.

The cost arises in HARQ because the PHY has to keep a copy of previous packets.

It doesn’t work in 802.11 because it breaks the stateless nature of the PHY, i.e. the processing of the next packet depends on the previous one. In an unlicensed band you don’t have full control of this. Furthermore it breaks the layering because now the PHY needs to know which packets came from which STAs before attempting HARQ recombination.

So HARQ can work in systems like WiMax because:
1. It’s a licensed band - the FCC can shut down anybody who doesn’t follow the rules.
2. The base-station controls exactly who talks and when
3. The cost / complexity is loaded into the base-station to improve uplink (where signal is weak), not downlink (where you don’t need it due to high tx power).

By: Wei

Wei — Thu, 10 Dec 2009 14:04:34 +0000

Is there a reason for not includng Hybrid ARQ in 802.11n ? New standards such as WiMax has included it already, why not 802.11n?

By: Daniel

Daniel — Fri, 04 Dec 2009 03:28:04 +0000

Very good review. Quick, to the point, yet quite technical and clever.
Thank you

By: Michael

Michael — Wed, 02 Dec 2009 21:51:30 +0000

The available spectrum varies from country to country. The Gartner comment probably referred to 2.4 GHz spectrum in the USA, which is usually too crowded to be usable for 40 MHz channels anyway. Here is a diagram of the 5 GHz spectrum in the USA that shows that you can get 23 non-overlapping 20 MHz channels, which equates to 11 non-overlapping 40 MHz channels:

By: key

key — Sat, 10 Oct 2009 17:45:42 +0000

Good overview of 11n improvments…
Can anybody clarify number of non overlapping channels with 11n in 5.4 GHZ spectrum with Channel Bonding (40Mhz)

I am confused as a Gartner document says there are only 3 non overlapping channels with 40 MHZ

By: TBerry

TBerry — Sun, 27 Sep 2009 15:22:44 +0000

Regarding the problem with the “concrete walls”, my suggestion is to use a directional antenna between APs.

By: Jorge

Jorge — Sat, 18 Jul 2009 06:30:08 +0000

Quoting Matt: “..I can tell you that you’re not going to get a 600mb file in one second over 11n..”

Remember that the unit Mbps. DOES NOT stand for MegaBytes per second but for MegaBits per second. It has nothing to do with file sizes.

1 byte=8-bits, Mb=MegaBits, and MB=MegaBytes.

If it was truly a 600Mbps. connection, the actual speed translated to MB/s would be around 73MB/s.

By: DS2 Blog

DS2 Blog — Sun, 26 Apr 2009 18:17:36 +0000

How fast can G.hn be?…

“How fast is G.hn?” That’s a question I usually get from people interested in the standard promoted by HomeGrid Forum, and until recently it was difficult to provide an exact response. The reason is that the actual throughput provided by……