Forum

Yeah it was a silly one "Hey everyone Cisco dont do N". Brilliant, dont let me do the marketing!!!!!!!!!

Yes Keith brings up a totally valid point. You will never see 300 or 450 in the real world. My own personal tests I have gotten 80 with iperf but the real point is, most people just look for the little '300' in their task bar and ignore the fact they don't really get that. I am very careful to explain to clients why they will never see 300 throughput for all the reasons Keith mentioned.

In the future, I anticipate more sophisticated techniques at the PMD layer. possibly more bits per baud [ in the RF sense, not Cisco serial ] of signal change. Years ago it was thought 16 QAM would never be practical due to phase noise etc. However, time went on and improvements in electronics have brought us up to 64 QAM.

A lot of experience is being gained with spatial streams and real-life recordings of actual signal propagation. More sophisticated beam-forming techniques will come into play.

An analagous situation occurred in the development of VOIP. Originally we had voice running at 64 kbps [ from Nyquist?s Theorem and full 8 bit sampling ]. Experience on the TASI system gave engineers a good understanding of the psychology and physiology of speech. [ About half way down the following ]:

http://www.google.com/imgres?imgurl=http://www.atlantic-cable.com/Article/1968Lenkurt/Lenkurt01.jpg&imgrefurl=http://atlantic-cable.com/Article/1968Lenkurt/index.htm&usg=__O44nfm6Tu4PtMQqW-sFBSzgMTCs=&h=1000&w=645&sz=103&hl=en&start=3&um=1&itbs=1&tbnid=DokxJSsjFIoBBM:&tbnh=149&tbnw=96&prev=/images%3Fq%3Dtime%2Bassignment%2Bspeech%2Binterpolation%2Bundersea%2Bcables%26um%3D1%26hl%3Den%26rlz%3D1T4ADBS_enUS249US249%26tbs%3Disch:1

ADPCM followed running at 32 kbps then 16 kbps. They then started looking at how the vocal chords produce sound and what information is embedded within the voice pattern for word recognition [ not emotion or nationality ]. Before we knew it, we were on to Code-Excited Linear Prediction and the extremely low data rates involved there.

802.11n will go from strength to strength and will be the ?test-bed? for much higher rates in the future. Then we will trully be able to get 300 odd Mbit/s of throughput !! [ albeit with much higher physical data rates, as all the old millstones of contention and gaps and acks will be around for a long time to come ].

DSP?s have come a long way, and the mathematical algorithms that perform their amazing duties will improve as time goes on.

RF propagation modelling will continue to be extremely important.

Gigabit Wireless [ One day it will be Gigabit THROUGHPUT wireless !! ] in the home is only a matter of time.

Dave

Dave,

I agree with your prediction, but don't forget that voice streams are by definition "lossy". Not so with html data, for example.

Take a voice capture, throw away a few milliseconds here and there and either the brain or the hardware will splice it together. You can even do that by throwing away some amount of voice data evenly over some time period, and the only effect will be to make it sound like the person is just talking faster. You can't do that with non-lossy data.

--------------------------------------------------------------------------------------------------------------------------------------

My favorite example is hardware that throws away silence periods greater than so many milliseconds - on a voice mail system it really makes the person leaving the message sound like they really have their act together.

That is, as long as they keep their mouths shut while they're thinking and not going "umm" or "ahh".