Myth #1: The only interference problems are from other 802.11 networks.
It is true that there are a tremendous number of 802.11 devices out there (projections are for 100 million units sold in 2005), and that these other 802.11 networks can cause interference with your network. This type of Wi-Fi interference is known as co-channel and adjacent channel interference. But since other 802.11 devices follow the same protocol, they at least tend to work cooperatively – i.e. two access points (APs) on the same channel will share the capacity of the channel.
But there are many other types of devices emitting in the unlicensed band, and some of them are selling in volumes that dwarf the number of 802.11 devices. These devices include:
These non-802.11 types of interference typically don't work cooperatively with 802.11 devices, and can cause significant loss of throughput. In addition, they can cause secondary effects of Wi-Fi interference such as rate back-off – where retransmissions caused by interference trick the 802.11 devices into thinking that they should use lower data rates than appropriate.
Myth #2: My network seems to be working, so interference must not be a problem.
The 802.11 protocol is designed to be somewhat resilient to interference. When an 802.11 device senses an interference burst occurring before it has started its own transmission, it will hold off transmission until the interference burst is finished. If the interference burst starts in the middle of an ongoing 802.11 transmission (and results in the packet not being received properly) then the lack of an acknowledgement packet will cause the transmitter to resend the packet. In the end the packets generally get through.
The result of all these hold-offs and retransmissions, however, is that the throughput and capacity of your wireless network are significantly impacted. For example, microwave ovens emit interference on a 50% duty cycle (as they cycle on and off with the 60 Hertz AC Power). This means that a microwave oven operating at the same frequency as one of your 802.11 APs can reduce the effective throughput and capacity of your network by 50%. So, if your network was designed to achieve 24 Mbps everywhere, you may now be reduced to 12 Mbps in the vicinity of the microwave when it operates.
If your only application on the WLAN is convenience data networking (ex. Web surfing), this loss of throughput may not be immediately obvious. But as you add capacity and latency sensitive applications such as VoWi-Fi to your network, controlling the impact of interference will become a critical issue.
Myth #3: I did an RF sweep before deployment. So I found all the interference sources.
One of the most troubling issues about interference is that it is often intermittent in nature. The interference may occur only at certain times of day (when someone is operating the device – ex. a cordless headset), or on certain days of the week. So, unless an initial sweep is done for an extended time, it's very easy to miss sources of interference.
And even if the sweep was extensive (for example, making measurement in each area for 24 hours), things change over time. It's very easy for someone to introduce one of the many devices that operate in the unlicensed band into your environment. No amount of periodic sweeping can truly guarantee that you have an interference free environment.
Myth #4: My infrastructure equipment automatically deals with interference.
Many of the newer switch-based WLAN infrastructure products claim to manage RF interference problems. In the end, they are somewhat limited by the capabilities of the 802.11 chipsets they are based on, and the 802.11 protocol itself.
With their 802.11 chipsets, the infrastructure providers can detect the presence of non-802.11 signals. And in response to detection, they can try to change the 802.11 channel of the APs in the area of the interference.
The problem with this approach is that it doesn't solve many of the problems that are out there. Some interfering devices (ex. Bluetooth, cordless phones, 802.11FH) are frequency hopped, so it's not possible to change channels away from them – they are everywhere in the band. And even for devices that operate on a static frequency, it can be quite a challenge to manage channel assignments in a large cell-based network.
In the end, it's critical that you be able to analyze the source of interference (i.e. identify what the device is, and where it is located) in order to determine the best course of action to handle the interference. In many cases, this "best action" will be removing the device from the premises. In other cases, the response may be to move or shield the device from impacting the network.
Myth #5: I can overcome interference by having a high density of APs.
The inexpensive nature of 802.11 access points makes it tempting to deploy them with very high density. For example, some networks are being deployed with an AP in every room. This type of deployment has the benefit of greatly increasing the capacity of the network, by allowing "spatial reuse" of the spectrum. It seems intuitive that by having more APs spread around, it's more likely that a client will be able to operate successfully even when interference is present.
Unfortunately, when deploying a dense network of APs, it's necessary to reduce the transmit signal power of each of the access points. If you don't reduce the power, the access points generate interference to each other – this is known as co-channel interference.
The reduction in the transmit power of the AP exactly offsets the potential benefit of interference immunity. So, in the end, the interference immunity of a network with a dense deployment of APs is not significantly better than that of a less dense deployment.
Myth #6: I can analyze interference problems with my packet sniffer.
802.11 packet sniffer products suffer from the same problem as WLAN infrastructure equipment. They can only see what the 802.11 chips tell them. So, they can tell you about secondary indicators of interference such as increased retransmissions and lower data rates. But they can't analyze interference problems, determine the cause of the interference, and help you find where the interfering device is located.
Note that a second problem with the data from 802.11 chips is that power measurements are un-calibrated. So, the data you receive from an 802.11 chip about the signal strength of an AP (or other device) can not be expressed in absolute dBm units. This makes it very difficult to put meaning on the numbers they report.
Myth #7: I have a wireless policy that doesn't allow interfering devices into the premises.
Having a wireless policy is a good first step in tackling the interference problem. But no policy is effective without enforcement.
One of the great attributes of unlicensed band wireless devices is that they are inexpensive and widely available. As a result, it's very easy for employees to purchase these devices and bring them to work. In many cases, these employees are not even aware that a particular device may cause interference with your wireless network.
And some devices like cordless headsets and microwave ovens may be a necessary part of your business, so can't be completely disallowed.
Summary: You have to expect that interfering devices will sneak onto your premises.
Myth #8: "There is no interference at 5 GHz.
It is generally true that there is less interference devices currently present at 5 GHz as compared to 2.4 GHz. But this will change over time. Just as everyone moved from 900 MHz to 2.4 GHz to avoid interference, the "band jumping" effect will catch up with 5 GHz.
Some devices that already exist at 5 GHz include: Cordless phones, radar, perimeter sensors, digital satellite.
Myth #9: I'll hire a consultant to solve any interference problems I run into.
If you have been running a WLAN for some time, you will know that there are frequent instances where your network doesn't operate perfectly. Without having your own visibility into interference, you are left conjecturing on whether or not interference is the problem. Lack of visibility is an issue for IT personnel – especially when the CEO is asking why he was having trouble yesterday connecting in the conference room.
And beyond the control issue, it's expensive and time consuming to bring in a consultant to debug these kinds of issues. A single visit and trip report can cost on the order of $5-10K.
Myth #10: I give up. RF is too hard for me to understand.
Don't despair. Tools are now available that makes RF easier to understand, even for those who consider themselves wired network specialists – not wireless experts.
Cognio products classify the sources of your interference, so you don't need to read the "wiggly lines." And after we've identified the interference, we help you find and eliminate it. For more information, go to www.cognio.com.
About the author
Neil Diener has 20 years of experience in the architecture and design of reliable communication systems and embedded software. At Cognio, he is responsible for leading the company's technology and product strategy. Prior to becoming CTO, Neil was director of software technology. Before joining Cognio, he held a series of director level positions at companies including Telogy/Motorola, Sun Microsystems, Xerox, and Ask Jeeves. Neil holds a BS in Electrical Engineering from MIT and an MS in Computer Engineering from USC.
This was first published in January 2006