**CORRECTION** :-( Re: [MLB-WIRELESS] 802.11g Starts Answering WLAN Range Questions

[John Dalton]

Hi all,

I'm not sure how you draw a red faced smiley.  The numbers
were wrong in my last post.

Specifically, the following conclusions were wrong:

> ...54Mbit/s is a funny number?

and

>  Saying 802.11a is 54Mbit/s is like saying 802.11b is 22Mbit/s.

Correcting: 54Mbit/s is not a funny number, in that the MAC can see
a data rate of 54Mbit/s under ideal conditions.  Saying 802.11a is
54Mbit/s is like saying 802.11b is 11Mbit/s.  Apologies for the error.

It turns out I understated the physical bit rate and an optional
rate 3/4 error correction code is available.  My explanation of
how it works was right. That'll teach me not to be lazy and post
without checking and quoting references.

Here is the REAL story from the IEEE802.11a standard [1].

802.11a has 64 subcarriers in a 20MHz band.  The
period of the FFT/IFFT is 3.2us.  0.8us cyclic
extension is added, so the total symbol period is
4us, giving 250000 symbols per second.

There are 64 subcarriers in an OFDM symbol.  The subcarrier
corresponding to zero Hertz is unused.  The highest frequency
5.5 subcarriers are unused, forming an upper guard band.  Similarly
the lower 5.5 subcarriers are unused forming a lower guard band.
A further 4 subcarriers are used for pilot tones, used for
frequency and time synchronisation.  This leaves 48 subcarriers
for data.

The number of data bits per symbol (hence the data rate) varies according
to the modulation and type of code used.  The choices are summarised in the
table below (use a fixed width font).  The 6, 12 and 24 Mbit/s modes are
mandatory.  The 9, 18, 36 and 54Mbit/s modes are optional.

Advertised  Modulation   Coding  Bits per    Coded bits per   Data (uncoded) bits   Data Rate (Mbit/s)
Data Rate                Rate    Subcarrier  OFDM symbol      per OFDM symbol       (4us symbol period)
(Mbit/s)

6           BPSK         1/2     1           48               24                    6
9*          BPSK         3/4     1           48               36                    9
12          QPSK         1/2     2           96               48                    12
18*         QPSK         3/4     2           96               72                    18
24          16QAM        1/2     4           192              96                    24
36*         16QAM        3/4     4           192              144                   36
48*         64QAM        2/3     6           288              192                   48
54*         64QAM        3/4     6           288              216                   54
* = optional mode

These data rates do not account for losses due to the MAC.  I have no hard numbers on MAC
efficiency, but I gather there is a substantial loss. Users will not get 54Mbit/s.
If I get the time, I will figure out some numbers om MAC efficiency, but again
no promises.

In summary:
The reason 54Mbit/s is missing from the plot the Drew originaly linked to
(http://img.cmpnet.com/commsdesign/csd/2003/jan03/zyren-fig10.gif)
is NOT due to coding reducing the data rate.
Possible reasons are:
1) The 802.11a MAC is less than 40% efficient (so the user only gets 20Mbit/s).
2) The propagation conditions could not support the 54Mbit/s mode at any range.
3) The measurements were invalid.

Regards
John


References:

[1] IEEE Std 802.11a-1999
    Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications
    High-speed Physical Layer in the 5 GHz Band

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