On a customer site, the protocol analyzer shows repeated beacons and CTS frames of an AP ( which is not under my control ) that supports 802.11b/g. The strange thing about these CTS is the durations are fixed values of 3957, 4245 and 19,665 microseconds.
Under IP, the biggest wireless data frame is 1544 ( with wep) , and with the transmit speed of 2Mbps, the duration will be around 163 microseconds and less than 350 microseconds for 1Mbps.
Can I draw the conculsion that that AP is broken or misfunctioning ?
Check your clients on this WLAN. In addition to having RTS/CTS turned on, it sounds like you are potentially having intermittent hidden node problems and the retry counters and timers are going to their maximums.
Also, the duration information in this frame is a measure of the time required to complete the 4-way handshake for RTS/CTS. The value is time it will take to send the data or management frame plus the ACK plus one SIFS interval (page 42 of the 802.11 Handbook, 2E).
These may be CTS-to-self protection frames from the ERP (802.11g) access point.
If your analyzer has a HR/DSSS (802.11b) NIC it will see the beacons and protection frames and miss the ERP data frames.
The CTS duration field carries the time to complete the current frame exchange sequence (the IEEE Handbook authors call this a handshake; read at your own risk). Two examples of frame exchange sequences are RTS, CTS, Data, Ack, and CTS-to-self, Data, Ack.
A data frame of about 1500 octets is about 12000 bits and would transit a 1 Mbps medium in about 12000 microseconds, so your numbers are in the ballpark. You might want to sanity check your math.
And remember, protocol analyzers are notorious liars. Even when telling the truth we are easily mislead by them.
I hope this helps. Thanks. /criss
Hi Joel and Criss,
I messed up my calculation. I have a spreadsheet file which I put all frame formulas there ( as a way to practise ) but it was absolutely wrong after I inserted a column and didn't update the formula. I was so bad that I didn't notice that the numbers I got so wrong that are unacceptable.
Thanks for your advice.
Guess you had to do some "Layer 0" troubleshooting to figure that one out.