- Image via Wikipedia
Most networks today (wireless or wired) need to support the following.
- High availability
- Pervasive non-stop coverage/access
- High density areas
- Mission critical applications
- Demanding SLAs
- Ease of management
There are two prevailing WLAN architectures – channel layering and adaptive channel/power, both of which attempt to maximum the use of all bandwidth available to WiFi devices. To do this effectively, co-channel interference (WiFi or non-WiFi) must be effectively managed.
Adaptive Channel/Power
Classic WLAN vendors have long used the approach of trying to avoid co-channel interference by adapting the channels and/or power levels of APs. While this technique has certainly provided tangible benefits in the past, it comes with a number of major shortfalls when trying to meet the needs of today’s WLAN networks.
click to zoom
One of the most common issues with the adaptive channel approach is the tradeoff made between having less co-channel interference but with coverage holes, as seen on the left of the image above (click to zoom in), or having fewer coverage holes but with greater levels of co-channel interference, as seen on the right in the image.
The adaptive approach is by nature, unstable and unpredictable. When a network reconverges at the RF level, client service is interrupted. Interference in one area can cause reconvergence system wide. Local performance problems can, and often do, affect system wide performance.
To illustrate the point of how unpredictable and unstable a network can become, a history of channel and power assignments for a single AP, for period of approximately half of a day, is shown below.
ARM instability (Click to Zoom)
Anyone who has ever managed a complex system can appreciate how much easier the task becomes once the manager knows the system well. Anyone that has managed a WLAN knows that the RF layer can be the source of many difficult to diagnose problems. Managing an adaptive channel system is made all the more difficult due to its dynamic nature.
Additionally, channel and power adaptation algorithms are based on idealized (i.e., spherical shaped cells). Multipath in 802.11n helps boosts throughput and range; however, it makes coverage patterns much more unpredictable. Microcells, which were never uniform to begin with, are now even more irregular since the arrival of 802.11n. With 802.11n, adaptation algorithms struggle in an attempt to provide pervasive coverage while paradoxically minimizing co-channel interference, i.e. they increase power to compensate for coverage gaps but increasing power increases interference range so they, therefore, need to backoff causing instability.
The illustration below shows 802.11n coverage patterns. In this example, each color represents a channel. Coverage holes are visible, and where like colors overlap, there are high levels of co-channel interference.
Click to zoom
Referring again to the figure above, if we were to make the cells larger, to eliminate coverage holes, there would be more co-channel overlap/interference resulting poorer performance. Therefore, the tradeoff is, pervasive coverage with poorer performance or something short of pervasive coverage with adequate performance.
Adaptive systems cannot scale to support very high-density deployments. While it is true that adding APs to an area to support slightly higher client density is possible, as more APs are added there is a point of diminishing returns due to the effects of co-channel interference. Add multiple floors to a single building and this instability comes from three dimensions complicating matters even worse.
Channel Layering
In contrast to the adaptive channel approach, the channel-layered approach is more stable, as radio channel and power levels are held constant while traffic is managed, and scalable, as adding a channel layer increases the capacity almost linearly. It is these two qualities of the channel-layered approach that enables support for demanding SLAs, mission critical applications, pervasive non-stop coverage, high availability, high density and ease of management.
With two or more channel layers, support for high availability is provided. In the event that there is interference on a channel in particular area, non-stop coverage is still available in that area, as the other channel(s) are not affected. No network reconvergence is required (as opposed to the microcell situation where a single AP changing power has a ripple effect across the whole WLAN).
Even higher client densities can be accommodated by adding more channel layers. Channels can be layered pervasively or as needed as shown in the illustration below.
Click to zoom
The channel layering approach allows the network to add capacity in areas as needed; either for adding clients to an area, supporting legacy clients and application in conjuction with newer 802.11n clients, or even dedicating an entire channel to an advanced application like voice or video. This stable, controlled approach to adding capacity allows the IT manager get back to the business of managing a network and its applications without needing to manage radios and radio frequency communications.
It should be noted that deploying channel layers does not equate to a greater number APs, than would be used in an adaptive approach, as the channel-layered approach allows all APs to operate at full power, resulting in more coverage per AP.
In contrast to the adaptive approach, which constantly changes the RF layer, causing the IT manager to understand “what did the RF look like when the problem occured”, the RF layer remains constant with channel layering. This eases the task of management, as there is no need to relearn the RF layer before troubleshooting a new issue.
Bottom Line
The adaptive approach, which was at one time a clever workaround, is no longer necessary or adequate to meet the needs of today’s wireless networks and can actually complicate matters. At a time when new technologies (802.11n) are making WLANs a viable alternative to wired LANs, IT managers require more stability at the RF layer, not less. To get that they’re going to have to “Think Different”.
