Forum

Using Airmagnet Survey in a warehouse filled with auto parts, I see a lot of retransmissions (50% or more is typical unless I'm very close to the AP).
I'm guessing that most likely this is due to multipath, and diversity antennas are used to try to combat this. I've also done Spectrum analysis with Cisco Spectrum Expert. Other than the occasional Microwave oven there is no non - wifi interference.
My question is this: can retransmissions be caused by co-channel interference?

-Andy

Absolutely. There are mechanisms for sensing the presence of a valid Wi-Fi signal ( or just "other energy" ). If the AP in question ( for example ) detects a signal, that AP will "hold back" and not try to transmit until "the air is clear". There is a back-off mechanism involved as well. If the AP goes through it's procedures and transmits, and another AP on the same channel happens to transmit at the same time, then that introduces the possibility of co-channel interference.

Usually co-channel interference will cause a reduction in throughput due to the carrier-sensing mechanisms operating properly, and APs "simply waiting to transmit while the air is occupied by another AP on the same channel". However it is possible for two more APs to transmit at the same time after the waiting and back-off periods have expired, with resulting collisions and subsequent re-transmissions.

Dave

Sounds like a textbook case of hidden nodes. In a large warehouse with big parts, tall shelves and huge boxes/crates, its easy for a hidden node(s) to not rx signals from other nodes during its L1/L2 channel assesment. So it sends, thinking its clear, when its not. And likewise for the other non-hidden nodes, during their 'errant' CCA's. So it looks like a great lab exercise of implementing and experimenting with the CTS/RTS mechanism. Have fun!!

The only time that hidden node will cause a real problem is when there is significant upstream traffic.

If you can provide the amount of traffic, device type (11a/b/g/n) and direction of traffic that will help.

Do you have external antennas on your Cisco APs?

GT

Andy

If you can get a copy of the CWNA study guide, there is an excellent chapter on troubleshooting which may help with background info while you are working the problem.

Dave

Are you trying to use client devices attached to forklifts?

This can be tough, especially if the drivers carry the devices inside coffee cans attached to their rides. Don't laugh. It happens.

I'm very thankful for all of the helpful responses. The CWNA study guide is invaluable (I've read it 2x and counting)! Thank you Dave, Ronald, GT and Wlanman!

To follow up on your questions:

GT - the clients are 802.11g handheld or mounted computer/barcode scanners using a terminal emulation program. Probably not much traffic in either direction. Usually 1-15 clients per AP. People are complaining about being dropped from the network while roaming and having to restart the computer to get back on. (We have found some of the problem to be 802.1x conversations never finishing correctly and implementing CCKM has helped greatly.) The retransmissions became apparent when doing an active site survey for coverage.

As for antennas - Yes, 3 external dipoles 2.2 dBi directly attached to the aironet 3502e APs. Also have 3 more for the 5 GHz band, which is why I didn't get any higher gain antennas that would have to be mounted separately!

Also, I found your videos on youtube.com/gthillwifi about hidden node/AP and RTS/CTS to be very helpful. The Cisco 11n APs also have transmit beamforming (ClientLink), does this technology help with the hidden AP problem also? Even if the client moving (handheld)? Thanks

Wlanman - Some are on forklifts but so far no coffee cans! the rest are mounted to carts.

Ronald - The CWNA study guide mentions 30 bytes for RTS/CTS threshold in my situation. Does this need to be set on the Clients and Wlan Controllers? If set on Controller, can it be set on a per SSID level?

Thanks again,

-Andy

CWNA, CCNA Wireless

Andy, you should see if certain, specific radios are being cutoff. If so, then implement the RTS/CTS on those, and the threshold value has to be experimented on. Start high, then go lower..then stop when the problems go away hopefully. If all radios are being cutoff, then you might have more serious issues, then RTS/CTS or CTS/SELF can be implemented on the WLC, so the WLC can clear the air. Per SSID? not sure, but at least if you know which groups are having problems, then you might be onto something.