VoWLAN troubleshooting with spectrum analysers

RF interfering with VoIP on your WLAN? Spectrum analysers can help root out the problem.

RF interference can play havoc with your wireless LAN (WLAN), especially when you're running VoIP on it. Wireless protocol analysers are great for diagnosing link, network and application layer issues that block or break connections. But in some cases, the culprit lies beneath all of these protocols at the physical layer. That's where spectrum analysers can be indispensable, allowing you to get your VoWLAN back on track before the conversation dies.

Clearing the air

LAN administrators know that the first thing to rule out is physical disconnection from the network card, wall jack or switch port. Doing so for wireless is harder because you can't simply tug on an Ethernet cable. Instead, you must depend on indicators that reflect the state of the airlink you can't see, like Wi-Fi connection status and signal strength.

For some problems, you'll work your way up the stack, debugging mismatched 802.11 parameters, IP addressing and routing issues and, eventually, applications (see Wireless network troubleshooting: Connectivity). Before you embark upon that laborious process, take a minute to check the "air quality" between the user and access point (AP) with a spectrum analyser.

Reading between the lines

WLAN protocol analysers can only capture 802.11 packets. By inspecting 802.11 packets, tools like WildPackets OmniPeek or AirMagnet Handheld Analyser can easily identify common WLAN problems like co-channel interference. They can also show the impact that non-802.11 transmissions have on Wi-Fi associations and applications.

But a protocol analyser can't tell you what's causing non-802.11 interference. For that, you need a spectrum analyser like Cognio Spectrum Expert, MetaGeek Wi-Spy or BVS BumbleBee. Cognio's analyser is also sold by AirMagnet, Fluke Networks and WildPackets.

Spectrum analysers are portable tools that combine a PC card or USB fob with laptop or PDA software. That specialised hardware listens to energy in the 2.4 GHz and 5 GHz unlicensed bands shared by WLANs, measuring the power (amplitude) of radio transmissions and pulses that occur at discrete frequencies. The associated software uses statistical analysis to plot spectral usage, letting you visualise and quantify air quality.

In addition, some spectrum analysers can uniquely identify and classify energy sources and help you determine their approximate locations. With this kind of help, you can take steps to shield, replace or eradicate those troublesome interferers that degrade enterprise WLAN operation and performance.

Troubleshooting RF interference

Non-802.11 devices that generate RF energy can have varied impacts on your WLAN. Some operate intermittently for brief periods (e.g., microwave ovens), while others transmit continuously (e.g., analogue video cameras). Some devices emit low power over a narrow frequency band (e.g., Bluetooth), while others have far greater impact on Wi-Fi (e.g., 2.4 GHz TDD cordless phones). When troubleshooting interference, your goal is to identify the source, assess its impact and determine the right course of action.

Step 1: Start by inspecting obvious symptoms exhibited by the Wi-Fi client using connection status and common network debugging tools like Ping (to measure network reachability, latency and loss) and iPerf (to measure application throughput).

Figure 1. Examine WLAN client symptoms. (Click image to enlarge)

Figure 1 illustrates what you're likely to see when you ping a reachable destination during a brief burst of wideband RF interference. Although the Wi-Fi signal still looks strong, latency increases and packets are lost. This degradation can slow file downloads and cause streaming video or packetised voice dropouts. If the interference persists, the connection's data rate may drop; eventually, sessions and associations will become disconnected.

Step 2: If the Wi-Fi client can send traffic but experiences this kind of poor performance, break out your WLAN protocol analyser. Capture 802.11 traffic near the client, looking for time periods where latency increases, throughput declines or sessions break.

First, rule out interference from other Wi-Fi devices operating on the same or adjacent channels. Co-channel interference can have a big impact on AP performance and is fairly easy to spot with a protocol analyser. If another AP or ad-hoc node is competing with your AP for airtime, you might change your AP's channel or remove the interferer.

Figure 2. Investigate symptoms using WLAN analyser.

Otherwise, if your AP has exclusive use of the channel, drill down into the client's traffic, examining AP/client parameters, channel utilisation and 802.11 traffic counters. Figure 2 shows AirMagnet Laptop watching client 00:08:7D:1A:5D:12 during wideband RF interference. Here, we can see wide fluctuation in signal to noise ratio (SNR), accompanied by a burst of retransmissions and CRC errors. This coincides with the period of packet loss shown in Figure 1. If you were troubleshooting a voice problem, you might also use a VoFi analyser to see call quality metrics like MOS and R-Value.

Step 3: Given these symptoms of RF interference, it's now time to determine the cause. You could hunt around for an obvious source, like a nearby microwave oven. But to save time, and to find more subtle interferers, use a spectrum analyser to record, pause and replay RF energy observations. By looking at spectral usage graphs before, during and after the affected time periods, you can find significant RF activity.

Figure 3. Use spectrum analyser to visualise airwaves.

Figure 3 shows a recording made by MetaGeek Wi-Spy, played back through Chanalyzer. Here, we can see RF energy (measured in dBm) being plotted in real time. At the bottom, a planar view shows the max, average and current power levels observed at each frequency in the 2.4 GHz band. A pair of Wi-Fi APs can be seen operating at channels 6 and 11, centred at 2.437 and 2.462, respectively. But the blue max area shows that energy was also at some point emitted across the entire band.

To see why, look at the top graph. This swept spectrogram plots RF energy as it occurs over a period of time. Here, the horizontal green band indicates that a burst of wideband RF energy spanned the entire spectrum for roughly ten seconds. Before and after that burst, we see "normal looking" Wi-Fi activity on channels 6 and 11.

Devices that emit RF energy often end up producing waveforms that are easily recognised. Spectrum analysers come with a few sample captures that can help you learn what to look for. These graphs also let you see which frequencies are being impacted, helping you choose alternate channels to avoid that interference. Unfortunately, wideband interference impacts every channel -- including those occupied by our APs. So we dig a bit deeper.

Step 4: Experienced administrators can become familiar with well-known RF energy patterns -- the signatures associated with particular kinds of radio devices. But you might prefer to use a spectrum analyser that automates that process by classifying transmitters based on the frequencies, modulation, pulse types (continuous or burst), operation types (analogue or digital) and protocol framing methods they use.

Figure 4. Identify and fingerprint the interference source.

Figure 4 gives an example of the device classification performed by AirMagnet Spectrum Analyser when troubleshooting our wideband interference problem. We can see more than a dozen devices emitting RF energy near the spectrum analyser. The analyser correlated transmissions to individual source devices, in some cases supplying identifiers carried in link layer protocols (e.g., MAC addresses, Piconets). The analyser also came up with a probable device type for each source -- here we can see several 5 GHz cordless phones, a microwave oven, our own AP and a generic wideband interferer.

Note that not all devices are actively transmitting at any time, so the analyser lets us see both active devices and a historic list of devices seen in the past. We can use controls to focus our search by narrowing frequencies, amplitudes and time periods or by tracing a particular device. We can also use manual classification to tag known/unknown APs, letting us analyse how other interferers are impacting our AP and how their energy level compares to our AP's own power and background noise.

Step 5: To eliminate an interference source, you must be able to find the transmitter. A spectrum analyser can speed that process by providing tools that let you view the strength of the RF energy emitted by a specific interferer in real time.

Figure 5. Locate the device causing interference.

Figure 5 depicts the AirMagnet Spectrum Analyser Device Finder tool. Here, we focus on the energy being emitted by our generic wideband interferer. We could use the signal strength graph to walk around the client, performing a hotter/colder search. But, in this example, we divided the search area into quadrants and used the finder to record signal strength in each quadrant. We then repeated that process, performing a second search in the quadrant with the strongest signal (lower -dBm reading) until we isolated the interferer's location in the upper left corner of our search area.

Step 6: Of course, the final step in this troubleshooting process is to take appropriate action. That may involve removing or reconfiguring the interferer. When those steps are not feasible, you might end up relocating your AP, changing its channel or even migrating from the 2.4 GHz band to the less crowded 5 GHz band.

Spectrum analysis can be performed on an as-needed basis. However, larger distributed WLANs may need a more comprehensive perspective, assisted by integrated spectrum analysis tools.

Figure 6. Use integrated WLAN / spectrum analysis tools.

For example, Figure 6 depicts integration between AirMagnet Laptop and AirMagnet Spectrum Analyser. This lets yo

Read more on Collaboration software and productivity software