[MLB-WIRELESS] 802.11g Starts Answering WLAN Range Questions

[John Dalton]

Technically, the correct name for the OFDM modulation used
in 802.11a is COFDM.  COFDM stands for Coded Orthogonal
Frequency Division Multiplexing.  Coding plays a central
role in how 802.11a works.  The theory goes something like this...

The 802.11a system contains an FFT (Fast Fourier Transform).  Think
of the FFT as something which divides the frequency band of the
802.11a signal into lots of smaller frequency bands, like radio
stations next to each other (there are 64 of these 'radio stations'
in 802.11a).  The transmitter distributes the bits to be transmitted
across these 64 different narrow channels (each corresponding to a
'radio station'), which then zing off through the antenna to the
receiver.  On the way a significant percentage of these signals
get trashed by the environment in which the radio signals travel.
It's a tough world out there for radio signals.  The end result
is the receiver does not receive the data from some of the 'radio
stations', so we have missing data.  We need some way to fill these
gaps.

This is where coding comes in.  To code a stream of bits we pass
it through a black box which introduces REDUNDANCY.  This means more
bits come out then go in.  If this coding is done properly, we can
remove some of the bits in the output from the black box, then
use the remaining bits to recover the data which went into the black box,
effectively 'filling in the holes'.  (The thing which
does this recovery is called a decoder.)  In the first 802.11a
prototypes, the coder produced twice as many bits as it received.
(the final standard allows other rates as well).  This meant the
data rate going into the coder was 27Mbit/s while the date rate
coming out was 54Mbit/s.

So what we do is pass the 27Mbit/s data from the user through a coder.  This
doubles the number of bits so the bit rate goes from 27Mbit/s to 54Mbit/s.
We then use the FFT to transmit the date.  In the process a number of bits
may get lost.  We then pass the received data through a decoder, which
drops the bit rate back to 27Mbit/s and in the process fills in the 'holes'.
There are other 'inefficiencies' in the system, but this is the main one.

See now how 54Mbit/s is a funny number?


Compare this with 802.11b...

Here we have a data stream running at 11Mbit/s.  We pass this
data stream through a device called a spreader.  This makes the
data take up the same amount of frequency spectrum as a 22Mbit/s
signal.  We then transmit the signal.  In the receiver we pass the
received signal through a thing called a despreader (duh?) which
reduces the data stream back to the original spectrum (equivalent
to 11Mbit/s) and in the process reduces the 'crudiness' due to the
environment through which the radio signal had to pass to get
from the transmitter to the receiver.


Anyway to summarise.  Saying 802.11a is 54Mbit/s is like saying
802.11b is 22Mbit/s.  (and I'm not referring to '802.11b+', which
in itself is a misleading use of terminology as such equipment
is non-standard.)


A disclaimer:  This is the best plain English explanation I could
write in the time available.  There are lots of holes and questionable
simplifications in it.  If you recite the above in your information
theory exam you will fail.  Having said that I think it's an okay introduction.
Feel free to improve on it under the terms of the Free Documentation License.
For the full story, go and download the 802.11a standard from the
IEEE's web site (and get a good book on communications theory from
the library).

Please let me know what parts of this don't make sense, so I can rewrite it.
I don't mean to insult anyone's education with the above, but I tried to
keep it simple so everyone could understand it.

For the technically minded, from memory the standard specifies
either rate 1/2 or rate 2/3 codes.  I think these are convolutional
codes.  I can't recall the waterfall curves from memory.  If I get
a chance I'll try to post some numbers from the standard, but no promises.
As you can gather from the rates, these is some serious error correction
going on here, compared to what you have on wires.

Coding is distinct from framing.  There are further losses in 802.11a
(and 802.11b) due to framing.

Regards
John Dalton


Some keywords for the interested to read up on:

Forward Error Correction
Multipath Fading
Fast Fourier Transform
Direct Sequence Spread Sprectrum
Quadrature Amplitude Modulation


To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message