Wireless Network Utilization

[IT Services]

Relatively simple question, I am just having a tough time figuring out analyze the data.

Anyway...

We use the Tamograph Wireless Survey utility and with this utility you are able to perform passive, predictive and active surveys.

With the passive\predictive you can calculate the SIR (Signal to Interference ratio) expected. You can change the "average network utilization" as a variable to formulate the SIR values.

I am having a tough time correlating our controller\AP data with that value.

When I look at the AP's I see Load Statistics. Would I be concerned with TX, RX or Channel Utilization or ALL when it comes to 802.11 interference ratios?

[George Stefanick]

Im not familiar with SIR. SNR, yes. 

I focus on channel utilization personally. This tells me how bust the radio / channel is and Im able to correlate this to problems or slowness. 

[IT Services]

From the application support documentation:

"This visualization shows the signal-to-interference ratio (SIR) measured in dB. SIR is a measure to quantify by how much the signal level of an AP (interfered AP) exceeds the interference level. The interfering signal is the signal being transmitted by other APs (interfering APs) that may or may not belong to your WLAN and that use the same or one of the adjacent 802.11 channels. In low SIR zones, client devices may experience low throughput. SIR is shown for the AP that experiences the worst interference in the given map area among the APs selected for analysis. You can deselect one or several of the selected APs to see SIR values for the APs that experience less interference."

"SIR is best illustrated with an example. Consider an area where the AP signal strength is -50 dBm, and the AP works on channel 1. In the same area, a -70 dBm signal from another AP that works on the same channel can be seen. If WLAN utilization is 100% (i.e., if the APs send radio waves all the time), the SIR value would be 20 dB. However, real-world WLAN utilization is almost never that high, which decreases the interference and increases SIR. If the interfered and interfering APs have the same signal strength, the SIR value would be 0 dB. In classical, non-digital radios, a SIR value of 0 dB makes signal reception impossible, but 802.11 devices use a technology that allows them to operate despite a zero or even negative SIR value, which sounds counterintuitive."

On network Utilization -

"Average network utilization – This setting defines how heavy the interference is from the interfering APs. If the interfering signal strength is high, but the network utilization is low, the interfering AP does not create much interference. A typical office WLAN has a network utilization of between 10% and 25%. Adjust this setting to match the actual value for your WLAN. "

After reading all of this would you say that Channel utilization is the best value to look at on the Cisco WLC that correlates to the "Average Network Utilization"?

I am going to check with the software vendor as well. In the meantime what is a good, expected Channel utilization value? Is there a range?

[George Stefanick]

One thing Ive learned over the years. Virtual surveys or better known as predictive modeling is as good as the information you enter. You can get a pretty good model in most cases. 

SIR must be a Tamo feature to give you a predictive model or with passive data a grading of network health. On the end of the day it a model. So many things can throw that model off. How many 11ac clients vs 11a clients for example. 

Calculating channel utilization is done generally the same by many of the vendors on APs. It monitors how long the radio is in the busy state (CSMA/CA) for a period of time and comes up with the %. I normally see on channel aps with 45%+ channel utilization start to feel slowness with realtime applications. For voice, thats why CAC is used, for example. Ive seen basic data applications still work ok at 70% (burst traffic) web browsing ...

Hope this helps ..

[IT Services]

George you are dead on with how Tamograph is using the SIR values.

I received some information back from them and they explain Network utilization in this manner:

I don't know the Cisco's definition of channel utilization, so it's hard to tell how it correlates to network utilization, but our definition is: ( bandwidth actually used by the clients ) / (maximum channel bandwidth). For example, if all the clients connected to the given AP consume 100 Mbps on the average and the AP uses a 40 MHz channel in the 2.4 GHz band, which means that the maximum channel bandwidth is 300 Mbps, then we can say that the average network utilization is 33%.

When looking at the Cisco AP load statistics, what do the TX and RX utilization values essentially mean? Do you have a general idea?

[George Stefanick]

Oh that is so not the right way to calculate channel utilization. You can pass low amounts of bandwidth or no bandwidth and have high channel utilization. 

Let me explain. This all goes back to CSMA ca. Ap radios are no different from client radios. They both use csma ca with Dcf to manage who should tx. The ap radio uses a calculation how often the radio is busy. In other words how often a frame is transmitted and the radio can't talk . This could be a frame at qam256 or a 1 PHY frame sent very slow at DBFSK. You see both take time to tx. One is sending a crap load more bits and both could, I use could lightly, take the same amount of time. 

Does that make sense ? 

[IT Services]

Thanks Geroge. It does make sense to me which is why I was confused on their reponse as well. Knowing what they "think" they classify NETWORK UTILIZATION as for the passive\predictive surveys, do you think Channel Utilization is a good value to watch on our AP's to guestimate the NETWORK UTILZATION.

I am ok with your best guess as well because this is certainly only for predicting health of the existing or potential networks. I am really just looking at mocking up power levels and seeing how much impact Co-Channel interference "could" potentially have in our environment. The busier the network the more likely a poor SIR. Do you think Channel utilization is the best value to use?

Also under AP load Statistics. What is the TX and RX values mean?'

Again. Much appreciated, you have always had great posts in the Wireless realm.

[George Stefanick]

You have to base line something that is consistent even if something isn't accurate if it's consistent you can expect "this" to happen when "that" happens. 

Channel utilization is something my group uses regularly when testing amd troubleshooting. Just yesterday I had 60 people on 2 aps and channel util was up at 80%. They complained all the way to the pop VIPs that the guest network pipe was used up. Fact was, it had nothing to do with the pipe and had everything to do with channel utilization. All along I seen at max 2 MB throughput .. Here again the SIR model would have been way different. So whatever measuring stick you use test it and vaildate it to the real world. 

As as for the Rx and tx let me check my notes .. That evades me for the moment ..

[IT Services]

Thanks George. I think a lot of avoiding poor SIR is in proper AP placement to begin with. Trying to stick with the 20% radio overlap as much as possible is likely the best way to go.

I will await to hear what you are able to find about the RX and TX values.

Thanks again.