Other vendors take different approaches to spectrum management. One, for example, uses an architecture whereby the controller manages both APs and clients and puts all APs on a single channel far apart from one another at full power. The vendor says this approach eliminates most interference. What is your argument against this architecture and in favor of your own approach to spectrum management?
What about other RF management approaches?
Spectrum management involves much more than just maintaining acceptable RF interference levels in the environment. While it does sound interesting that a user can reduce the amount of interference by configuring the wireless devices to operate on a single channel, it needs to be clarified that this only relates to adjacent channel interference and not the overall interference in the RF environment. Interference encompasses multiple sources, and Wi-Fi based adjacent channel interference is just one piece of the puzzle. A large contributor to lowered WLAN performance is interference arising from non Wi-Fi devices operating in the same spectrum band. These are sources that should not be ignored and have the capability of wiping out all signals, not only in the single channel where all the devices may be operating, but also the entire frequency range. These “hidden” or “invisible threats” are present in everyday corporate environments and are commonly implemented inside everyday devices or equipment. They include Bluetooth or zigbee devices, microwave ovens, cordless phones, wireless cameras, motion detectors, etc. Simply moving all devices on to a single channel cannot prevent these non Wi-Fi sources from impacting performance.
In certain deployments, single channel architecture may be required or preferred to satisfy a particular business or technical requirement, but for thorough interference monitoring and network management, “true” spectrum analysis solutions like AirMagnet Spectrum XT, or the spectrum sensors as part of AirMagnet Enterprise that monitor for all sources of interference in the RF environment, must be implemented.
Tool-based solutions can be effective; however they require travel to the source of interference and are highly manual intensive. When spectrum intelligence is fully integrated into the network architecture it operates 24/7, constantly monitoring for interference and air quality issues. This allows IT to take a more proactive approach to spectrum management. Instead of waiting for interference to be reported by an end user and then dispatching a tool to analyze the problem, IT can find interference as soon as it occurs and take immediate action. Having a 24/7 history also makes it possible to look back in time. Using historical data, it's easy to perform analyses of trends over time.
Whether it is tool based or infrastructure based, the only way to effectively manage RF interference in the unlicensed band is to classify and understand the sources. First, understanding the device type is important when the human needs to take action. Second, understanding the source is critical to enforce spectrum security policy. And third, knowing the source of the interference is important in order to be able to implement intelligent mitigation policy based on the type and characteristics of the device. For example, if a device is of the type where it is not mobile, and tends to always operate on the same frequency (ex. Microwave oven), then when I see it once I will tend to see it again. So, knowing a device like this is operating in a certain part of the network can be factored into channel assignments, even when the device is not currently active.
One of the often overlooked aspects of wireless network design is antenna choice and placement. In general, properly placed directional antennas will produce a more robust and reliable network than one based on randomly placed omni antennas. While it takes time to design a wireless network using directional antennas, the resulting network can be less susceptible to outside interference, less likely to cause interference to other networks in close proximity, and much more secure than the simple design based on omni antennas. One metric worth considering is antenna gain. The more directional an antenna is, the higher the antenna gain it exhibits. Antenna gain is the direct result of restricting where the signal does and does not go. Two neat aspects of antenna gain are that it is noise free and that it requires no external power.
While directional antennas are sometimes useful (outdoor and specific indoor use cases), it's impractical to use them en total for an indoor deployment. If you're going to take that approach as a network designer, Ruckus's beam steering solution is a better fit because it's a dynamic, rather than static solution. At least that way, you'd get the benefit of both directional and omni antennas at the same time. Ruckus should now send me a gift basket. :)
Hey Devin... thanks! You saved me a lot of typing. By the way, I did send you a gift basket. It's not my fault you didn't want our cheapest 11b/g AP. :)
GT
Let's face it, if Cisco didn't came up with CleanAir and others wouldn't have followed, we would not have this discussion right now. External interference has always been an issue and unlike a lot of people here want to make you think, it hasn't been the major reason for performance issues: co-channel interference is and has always been.
There were external interference is in 97% of the cases something you can't exterminate, co-channel interference is an issue you can address with a decent single channel solution. So to come back to my inital statement, if there would not be such a big marketing campaign about it, we would all still believe that working in license-free frequencies beholds the risk of getting interference from external devices and live with it.
I have seen and did troubleshooting in a lot of cases with micro-cell infrastructures were people stood underneath an AP and were unable to communicate with it. If they would lower the power, the interference would drop but not all places would be covered as needed...
In all of these cases, the radio management needed to be done manually because the automatic setting, no matter the brand, would always lead to instabilities.
We have always done spectrum analysis during site surveys or debugging/troubleshooting sessions. This is indeed only a snapshot view and yes, it would be handy to have it baked into the system and see what is happening all the time. But it is nothing more than that. It helps, but it doesn't solve things immediately.
Normally, a good site survey should indicate potential interferers in cases were the IT department is not in sync with other dept like the security dept that installed wireless alarm or smoke alarm systems using 2.4GHz.
and even if you detect an interferer, it usually comes from outside and you can't do anything about it except changing channels. And guess what, if the WiFi isn't working, what is the first thing a (home) user will do? Change channel. A professional user of non-single channel solutions would not do it since he would need to review his entire channel plan but believe me, from our questionings amoung our customers, this is the first thing that pops to their heads (since they do the same at home..)
Now, with this technology, perhaps we can get this done automatically. Great, but you can't change channel seamlessly. Moreover, with a non-single channel solution, this means the entire network must change channel! All AP will go down to select a new channel. New channel = new RF envirnment so for automatic RF systems, this means searching new output power etc. If the interference is periodic and broadband (harmonics of elektro motors etc) this might result in a lot of channel changes over short period of times. Is that then what we want?
If it will only do reporting what then will it really offer? The certainty that you see it but can't do anything about it?
This falls in the same space as an WIPS/WIDS system. Of course you would like to have it and of course it is useful but in reality, how much does it add for the customer and is he willing to pay for this? I have noticed a very strong discrepancy between the US and Europe in these cases.
I think that the single channel architecture (SCA) is a neat concept and performs very well on ONE channel. Actually, they perform better than any other vendor on that one channel. One benefit is that the STAs don't perform go through the true roaming process which is an advantage. Also, they can have per STA QoS settings since the APs have a Beacon for each STA. Now, the downsides.
- It's only on one channel. Even with the niftiest protocols, there is only so much traffic that can exist on one channel. The response to that is to add more "blankets" of SCAs. This requires significantly more APs than a standard multiple channel architecture (MCA).
- What if a source of interference on that one channel appears? The entire system would have to change channels (disconnecting all STAs) or remain on the channel and live with the interference.
- Currently, all of this is managed by the controller which means a significant amount of data is flowing through the controller. That makes the controller a single point of failure (I know, redundant controllers) and more importantly, can create a significant bottleneck now that 802.11n can push 200 Mbps per radio.
I don't see that SCA's actually manage the spectrum. They don't combat against sources of interference and can't easily avoid interference if one appears.
GT