Chapter 6

Designing Radio Management

This chapter covers the following topics:

Understanding RRM: This section describes the algorithms that can monitor and adjust radio frequency parameters automatically in a wireless network.

Localizing RRM with RF Profiles: This section explains how RRM can be configured globally or customized for specific areas or buildings.

Optimizing AP Cell Sensitivity with RxSOP: This section covers a Cisco feature that can reduce AP cell size to increase performance and lessen co-channel interference in highly dense wireless networks.

This chapter covers the following ENWLSD exam topics:

2.0 Wired and Wireless Infrastructure
2.3 Design radio management
2.3.a RRM
2.3.b RF Profiles
2.3.c RxSOP

In Chapter 5, “Applying Wireless Design Requirements,” you learned how to size access point (AP) cells appropriately by changing the transmit power levels and leveraging appropriate antennas. You also learned how important a proper channel layout is to promote efficient roaming and minimize co-channel interference. You probably also realized how difficult these tasks are when you have to tune the radio frequency (RF) parameters manually across a large number of APs.

In this chapter, you will learn about Radio Resource Management (RRM), a flexible and automatic mechanism that Cisco Wireless LAN controllers can use to make your life much easier.

“Do I Know This Already?” Quiz

The “Do I Know This Already?” quiz allows you to assess whether you should read this entire chapter thoroughly or jump to the “Exam Preparation Tasks” section. If you are in doubt about your answers to these questions or your own assessment of your knowledge of the topics, read the entire chapter. Table 6-1 lists the major headings in this chapter and their corresponding “Do I Know This Already?” quiz questions. You can find the answers in Appendix D, “Answers to the ‘Do I Know This Already?’ Quizzes and Review Questions.”

Table 6-1 “Do I Know This Already?” Section-to-Question Mapping

Foundation Topics Section	Questions
Understanding RRM	1–9
Localizing RRM with RF Profiles	10
Optimizing AP Cell Sensitivity with RxSOP	11

1. Which one of the following protocols is used by APs to learn of other nearby APs?
1. TPC
2. NDP
3. 802.11 neighbor announcement frames
4. 802.11 beacon frames
2. Which two of the following statements are true about NDP messages?
1. NDP messages are sent only on an AP’s operating channel.
2. NDP messages are sent on every channel that an AP supports.
3. NDP messages are transmitted using the AP’s normal transmit power level.
4. NDP messages are transmitted using the AP’s maximum supported transmit power level.
3. When RRM collects AP neighbor data from NDP messages that are received, it organizes APs into RF neighborhoods. Which one of the following is the correct criteria for forming an RF neighborhood?
1. Any two APs appearing in each other’s neighbor list at any RSSI value
2. Any two APs appearing in each other’s neighbor list at RSSI greater than −80 dBm
3. Any two APs appearing in each other’s neighbor list that are in the same RF group
4. Any two APs appearing in each other’s neighbor list but not in any other neighbor lists
4. The TPC algorithm is used for which one of the following purposes?
1. To adjust the transmission control protocol rate
2. To detect problems in transmission perimeter coverage
3. To adjust the transmitting primary channel
4. To adjust the transmit power level
5. If RRM decides to change an AP’s transmit power level, which one of the following correctly identifies the action to be taken?
1. The AP’s transmit power level will be changed immediately to the desired value the next time TPC runs.
2. The AP’s transmit power level will be raised or lowered by one 3 dB increment as needed each time TPC runs.
3. The AP’s transmit power level will be changed immediately to the desired value the next time DCA runs.
4. The AP’s transmit power level will be raised or lowered by one 3 dB increment as needed each time DCA runs.
6. If the DCA algorithm detects that an AP is experiencing interference or excessive noise, what might it do to mitigate the problem?
1. Increase the AP’s transmit power level.
2. Decrease the AP’s transmit power level.
3. Change the AP’s channel number.
4. Direct the client to a different band.
7. Which one of the following runs the DCA algorithm?
1. RF group leader
2. Master controller
3. Each controller
4. Cisco Prime Infrastructure or DNA Center
8. Suppose the 5GHz radio in one of several APs in a building has failed. Which one of the following algorithms should be able to detect the failure?
1. CCA
2. DCA
3. Dead radio detection
4. Coverage hole detection
9. Which two of the following choices are the correct criteria used by the FRA algorithm?
1. 2.4GHz coverage overlap
2. 5GHz coverage overlap
3. An AP’s COF value
4. An AP’s transmit power level value
10. Suppose you have designed a wireless network to leverage RRM for automated tuning. You have assembled a list of RRM, TPC, and DCA settings that will be applied to all APs in the RF group. However, the APs in one building on your campus need to use only a specific set of channel numbers and a different lowest mandatory data rate. Which one of the following strategies would be best for you to follow?
1. Manually configure each AP in the one building with the specific RF settings needed.
2. Create an RF profile that is tailored for the one building and then apply it to an AP group containing the building’s APs.
3. Disable RRM on APs in the one building.
4. The one building cannot receive customized RRM settings, so it must conform to the global parameters used everywhere else.
11. Suppose you discover that there are some issues in high-density areas of your network caused by high channel utilization and co-channel interference between APs using the same 5GHz channel. Which one of the following design decisions could you take to alleviate the problem?
1. Raise the TPC maximum transmit power level limit value.
2. Disable the channels that show co-channel interference in the global DCA channel list.
3. Disable more 5GHz channels and try to steer clients toward the 2.4GHz band instead.
4. Use RxSOP with a carefully chosen threshold.

Foundation Topics

Understanding RRM

Suppose you need to provide wireless coverage in a rectangular-shaped building. For simplicity, assume that the building has one floor and no interior walls or other objects that would affect RF propagation. Using the information you have learned from this book, you decide to use six APs and locate them such that they form a staggered, regular pattern. The pattern shown in Figure 6-1 should create optimum conditions for roaming and channel use. (The building dimensions have not been mentioned, just to keep things simple.)

Images — **Figure 6-1** *A Hypothetical AP Layout*

So far, you have considered the layout pattern and an average cell size, but you still have to tackle the puzzle of selecting the transmit power level and channel number for each AP. The transmit power level will affect the final cell size, and the channel assignment will affect co-channel interference and roaming handoff. At this point, if all the APs are powered up, they might all end up transmitting on the same channel at maximum power (100 mW, for example). Figure 6-2 shows one possible scenario; each of the AP cells overlaps its neighbors by about 50 percent, and all the APs (and their clients) are fighting to use channel 36!

Where do you begin to prevent such mayhem? Because the AP locations are already nailed down, you can figure out the transmit power level that will give the proper cell overlap. Then you can work your way through the AP layout, choosing an alternating pattern of channel numbers. The example with six APs might not present a daunting task, but a large building with many APs on many floors is an entirely different situation.

Do not forget to repeat the transmit power and channel assignment tasks for both 2.4 and 5GHz bands, as most APs have dual-band radios.

Also, if you plan on using 802.11n or 802.11ac with channel widths greater than 20MHz, do not forget to reserve the extra channels needed for that. Remember that only the 5GHz band is capable of supporting wide channels. Also remember that your choice of channel width also affects the available number of non-overlapping channels you can assign to the APs.

Suppose you happen to notice one day that an AP has failed. You could always reconfigure its neighboring APs to increase their transmit power level to expand their cells and cover the hole left by the failed AP.

One day in the future, you might identify an area where a higher density of users begins to gather. If you decide to introduce additional APs to distribute the client load, you will need to revisit the entire configuration again to make room for new cells and channels. As a result, you will probably need to rework the channel assignment on all of the APs to accommodate the new APs and their channels.

Did your life as the wireless LAN administrator just become depressing and tedious? Cisco Radio Resource Management (RRM) can handle all these tasks regularly and automatically. RRM consists of several algorithms that can look at a large portion of a wireless network and work out an optimum transmit power level and channel number for each AP. If conditions that affect the RF coverage change over time, RRM can detect that and make the appropriate adjustments dynamically. The sections that follow explain each of the mechanisms and algorithms used by RRM.

Discovering the RF Neighborhood with NDP

As you work through a wireless design, you have the ability to see the AP locations on a map and interpret the spatial relationships between each AP and its neighbors. You can even draw the cell boundaries or view them in a site survey map to gauge the cell overlap. RRM cannot use maps and drawings to make its calculations. Even if it could, the maps might not accurately depict the dynamic RF conditions in an area. Instead, RRM must collect real RF data from one AP to another and infer how each AP is situated with respect to its neighbors.

RRM uses the Network Discovery Protocol (NDP) to advertise each AP’s presence. If an AP’s advertisements are received by other APs, those APs must be in proximity to each other. In Figure 6-3, AP-1 is transmitting NDP messages to announce itself. Each of the other APs is able to receive AP-1’s advertisement and measure its received signal strength—one of the components necessary for RRM calculations.

NDP advertisements are sent to the multicast address 01:0B:85:00:00:00, which is recognized by all other Cisco APs. The messages are transmitted at the highest power allowed for the channel and band. By using the maximum power level, RRM can always know the strength of the signal as it leaves the AP’s antenna. Then when that signal is received by other neighboring APs, RRM can use the RSSI to gauge the free space path loss between the transmitting and receiving APs. If the currently assigned Tx power level (lower than the maximum) was used instead, RRM could have a difficult time interpreting the results. In addition, relatively distant APs might not be able to receive the NDP frame at all.

NDP advertisements are transmitted using the lowest data rate possible in the band— regardless of whether or not that data rate has been enabled for use. For example, 1Mbps is always used in the 2.4GHz band and 6Mbps in the 5GHz band. By using the lowest data rate, the advertisements are more likely to be intelligible farther away from the transmitting AP and in noisy environments.

Now consider that each AP transmits its own NDP advertisements, and each AP listens to receive advertisements from any neighboring APs. Over time, each AP can collect the advertisements it receives and report the results to the wireless LAN controller (WLC), as long as the advertisements come from APs that are members of the same administrative group as the receiving AP. RRM can then compute any adjustments to the transmitting AP’s power level to tune its cell size appropriately. Figure 6-4 shows example results gathered by the APs in an area; for simplicity, only AP-4 and AP-6 are shown.

So far, this description of the NDP process has considered only received signal strength, which assumes that every AP is transmitting on the same channel. Hopefully that is not the case in an actual wireless network! Instead, RRM must evaluate the interaction between APs on the channels they are using. Then it can compute and make changes to both transmit power levels and channel assignments.

To reach neighboring APs that might be operating on any arbitrary channel, each AP must transmit its NDP advertisement on every channel it is configured to support. It does this by waiting for an idle period on the current channel and then quickly tuning to a different channel and sending the NDP frame there. It must work through the entire set of channels over a period of 180 seconds. DFS channels are the exception; an NDP frame will be sent on a DFS channel only if the AP is currently the channel master and has determined that no radar signals are present.

Each AP must include its normal operating channel number in its NDP advertisement so that other APs will know its channel assignment. Figure 6-5 shows an over-the-air packet capture of NDP advertisements that were transmitted by a single AP. Notice the Channel column and how the AP has cycled through the channels sequentially over time. It even cycled through the 2.4 and 5GHz bands.

Figure 6-6 illustrates the scenario from Figure 6-4 with channel information added. Now each AP is able to build a list of neighboring APs complete with received signal strength and channel number information. Notice how the neighbor lists also contain data that can indicate co-channel interference. For example, AP-4 is operating on channel 52 and its neighbor list shows AP-3 also operating on channel 52 with a reasonably strong signal strength. Once the APs send their neighbor lists to the WLC, RRM can compute any adjustments needed to form a more effective channel assignment layout.

Each AP maintains a list of NDP advertisements received from up to 34 neighbors. To maintain some stability in the data, entries are automatically pruned and removed if no NDP message has been received from a neighbor after 15 minutes has elapsed.

NDP advertisements contain the following information about the sending AP:

Radio ID: Designates which radio (2.4GHz, 5GHz) sent the frame
Group ID and hash: Designates the logical group name where the AP is a member
Encryption: Key information if the NDP message is encrypted
IP address: The address of the WLC where RRM algorithms are running
AP channel: Normal operating channel
Message channel: Channel used to transmit the NDP message
Message power: Transmit power level (dBm) used to transmit the NDP message
Antenna pattern: The transmitting antenna pattern used

Note

Every Cisco lightweight AP is expected to send periodic NDP advertisements. If a beacon frame is received from an AP that has not sent an NDP advertisement, the wireless intrusion detection system (WIDS) running on the Cisco WLC will flag that AP as a rogue device.

RF Groups

RRM works by monitoring a number of APs and working out optimal RF settings for each one. The APs that are included in the RRM algorithms are under a common administrative control and are considered members of a single logical RF group. For example, all of the APs that belong to an enterprise can be contained in a single RF group. Figure 6-7 shows an example of an RF group called “Enterprise” that contains nine APs.

By default, an RF group contains all the APs that are joined to a single controller because only one RF group name can be configured on it. If you have multiple controllers, you can include all of their associated APs in the group by configuring the same RF group name on each controller. All of the controllers must be able to communicate with each other through the normal wired network infrastructure.

An RF group can span multiple controllers, but only one of them can run the RRM algorithms for all of the APs involved in the RF group. That means one controller must be elected as the RF group leader. The group leader can be elected automatically or through static configuration. In an automatic election, the WLCs exchange information about each other. The WLC with the highest-performing platform and the greatest AP license will become the group leader. If there is a tie among identical controllers, the one with the highest IP address will win the election.

Note

A single RF group can contain up to 20 WLCs. Depending on the WLC platform, a single RF group can contain up to 6,000 APs.

Because there is no RF interaction between AP radios on the 2.4GHz band and radios on the 5GHz band, RRM can treat each band independently. In fact, each band has its own separate instance of RRM and its own RF group leader. After the group leader elections, you may see a single controller acting as group leader on both bands, or one controller as group leader on 2.4GHz and another as group leader on 5GHz.

Each AP in the RF group will then send its AP neighbor list updates to the RF group leader, where the RRM algorithms will be run. RRM bases its calculations on the degree that APs will interact or interfere with each other. Therefore, it has to identify all of the APs that might be affected by a transmit power or channel change. If a change is made to one AP, other neighboring APs could be impacted and need adjusting, too, causing their neighbors to be impacted, and so on. In other words, any RF change can have a cascading effect across all APs in a geographic area.

RRM organizes all APs contained in an RF group into RF neighborhoods, or sets of APs that are in close RF proximity to each other. The criteria is simple: Any two APs that appear in each other’s AP neighbor list will become members of the same RF neighborhood, as long as the RSSI is −80 dBm or greater. An AP is removed from the neighborhood only if the RSSI of its received NDP messages drops below −85 dBm.

The idea is to develop an RF neighborhood by adding APs that are heard by other APs. The neighborhood expands from one AP to another until the NDP messages are too weak to meet the criteria. An RF group can expand across floors in a building, as long as an AP’s signal can propagate through the floor or ceiling and be received by another AP located there.

Figure 6-8 illustrates the RF group and neighborhood concepts. All nine APs are part of the RF group “Enterprise” because they are all under the same administrative control. AP-1 through AP-6 all become members of RF neighborhood “A” because each one can be heard by another member of the neighborhood at an RSSI of −80 dBm or above. AP-7 and AP-8 form a separate RF neighborhood; they have a close RF proximity to each other but not to any members of RF neighborhood “A.” Likewise, AP-9 becomes the sole member of RF neighborhood “C” because it is not close enough to be heard by any APs in the other two neighborhoods.

Once RF groups and RF neighborhoods have been defined, RRM can proceed with its analysis and computations. Each RF neighborhood can be handled independently because any transmit power or channel changes made to one will be too far away to affect any other.

Transmit Power Control (TPC)

The RRM algorithms are designed to keep the entire wireless network as stable and efficient as possible. The transmit power control (TPC) algorithm is one facet of RRM that focuses on one goal: setting each AP’s transmit power level to an appropriate value so that it offers good coverage for clients while avoiding interference with neighboring APs that are using the same channel.

Figure 6-9 illustrates the TPC process. APs that were once transmitting too strongly and overlapping each other’s cells too much are adjusted for proper coverage, reducing the cell size more appropriately to support clients.

Likewise, if any AP cells are too small and cannot effectively overlap their neighbors’ cells, TPC will attempt to increase the transmit power levels to expand the cell size.

Controllers have no knowledge of the physical location of each AP. By looking at Figure 6-9, you can see that the APs are arranged in a nice evenly spaced pattern, but the controller cannot see that. When an AP joins a controller, it advertises only its MAC address, IP address, and some basic information to the controller. The AP is able to know its own transmit power level and can build a list of other neighboring APs as it receives NDP advertisements from them. From the AP’s point of view, it can gauge how other APs are impacting its own cell but not how it is impacting others.

However, the TPC algorithm is able to compute the impact that each AP has on its neighbors because it runs on the RF group leader. Recall from the “Discovering the RF Neighborhood with NDP” section that the group leader maintains a database of the list of neighbors that each AP has overheard. To find out how strongly one AP is being received in other AP cells, the group leader can search the neighbor lists for the AP and respective RSSI values.

If an AP is being received too strongly in other AP cells, the TPC algorithm can configure a lower transmit power level on that AP. Likewise, if other APs have measured its signal as too weak, TPC can raise the AP’s power level appropriately. Cisco has designed TPC to choose the neighbor that received the third strongest RSSI from an AP as the criteria. With NDP advertisements transmitted at the maximum transmit power level, the third strongest RSSI should be −70 dBm, as illustrated by Figure 6-10.

To gauge how far AP-1 is capable of reaching into other AP cells, the neighbor list of AP-5 is used because it has the third strongest RSSI recorded for AP-1. The goal is for AP-1 to be received at −70 dBm at AP-5’s location, but it has been received at −66 dBm instead. That means AP-1’s transmit power level can be lowered by 4 dBm from the maximum.

This same calculation is performed for every AP in the RF group, using the following formula:

Tx_Ideal = Tx_max + (Threshold – RSSI_3rd)

Suppose AP-1’s maximum transmit power is 15 dBm. Then the ideal transmit power would be 15 dBm + (–70 dBm – –66 dBm), or 15 dBm minus 4 dBm. From Figure 6-11, you can compare the maximum power cell boundary at AP-5 with TPC’s ideal boundary, shown as a shaded circle, where AP-1’s transmit power level has been reduced to 11 dBm and the cell is bounded where the signal strength is −67 dBm.

Tip

Be aware that the maximum transmit power an AP can use depends on which band it is currently operating in. The regulatory domain determines the maximum value that can be used. For example, in the United States, Wi-Fi devices are limited to 15 dBm in the U-NII-1 band (channels 36–48), 17 dBm in U-NII-2 (channels 52–64) and U-NII-2 Extended (channels 100–140), and 23 dBm in U-NII-3 (channels 149–165). Even though TPC does not consider channel numbers, the maximum transmit power values used in its calculations actually depend on which channels (and bands) the APs are using at the time the TPC algorithm runs.

This discussion has focused on the TPC algorithm, but do not forget that you must still begin with a thorough design that specifies AP locations and transmit power levels that place the −67 dBm cell boundaries where you need them to be. As an example, suppose AP-1 is located correctly in the design of Figure 6-11, and its transmit power level is set at 12 dBm. This value has been chosen because it is half (3 dB less than) the maximum power level of 15 dBm, leaving some flexibility to adjust greater or lower if needed. The design goal cell boundary is also shown in the figure. Fortunately, TPC’s notion of the ideal transmit power is very close to the designer’s notion, producing similar cell boundaries.

Keep in mind that a good design with appropriate AP locations and spacing affects RRM’s TPC results. As long as APs are located with a reasonable space between them, TPC can produce expected results that are also reasonable. RRM will try to adjust AP transmit power levels to compensate for any holes or excessive overlap if it can. However, if you begin with APs that are located too far apart, where their cells might not overlap even at maximum power, TPC can only do so much to compensate for the design problems. After all, APs have a limited range of transmit power level values. An AP cannot transmit any greater than its maximum power level or any less than its minimum. In other words, even though RRM is automatic, it is not magical.

Once the ideal transmit power has been calculated for each AP in the RF group, TPC can begin making the necessary adjustments. Rather than make large, sudden changes in AP transmit power levels, TPC tries to maintain stability by making only incremental changes of 3 dB each time it runs. By default, TPC runs every 10 minutes. Suppose an AP’s transmit power level needs to decrease 12 dB to reach the ideal value. TPC will lower the level by 3 dB the next time it runs, followed by another 3 dB during each of the next three cycles.

Tip

To add more stability to the wireless network, you can configure the TPC interval to a greater value. For example, some environments that depend on critical services, such as mobile voice communication, might benefit from keeping RRM changes to a minimum during work shifts. In that case, you could configure the TPC interval to be 8 or 12 hours instead of the default 10 minutes.

Why does TPC use 3 dB increments to raise or lower AP transmit power? Cisco APs use a power level index that correlates to an actual value in dBm, as listed in Table 6-2. A power level of 1 always denotes the maximum allowable power level, which varies by band. Notice that some bands use eight power levels, while others use only six or five. Each increment of the power level index changes the power level by 3 dB, effectively doubling or halving the power.

Table 6-2 Example Cisco AP Transmit Power Level Value Correlation

Power Level	2.4GHz Ch 1–13	U-NII-1 Ch 36–48	U-NII-2 Ch 52–64	U-NII-2 Ext Ch 100–140	U-NII-3 Ch 149–161
1	23 dBm	15 dBm	17 dBm	17 dBm	23 dBm
2	20 dBm	12 dBm	14 dBm	14 dBm	20 dBm
3	17 dBm	9 dBm	11 dBm	11 dBm	17 dBm
4	14 dBm	6 dBm	8 dBm	8 dBm	14 dBm
5	11 dBm	3 dBm	5 dBm	5 dBm	11 dBm
6	8 dBm	—	2 dBm	2 dBm	8 dBm
7	5 dBm	—	—	—	5 dBm
8	2 dBm	—	—	—	2 dBm

By default, TPC determines the total transmit power level change needed for each AP in the RF group and then carries out those changes incrementally. You can also configure upper and lower limits to govern TPC’s results. By default, TPC uses a maximum limit of 30 dBm and a minimum limit of −10 dBm. Notice that those values are outside the range that an AP can use, so the limits are effectively disabled. If you find that TPC is setting some APs with a transmit power level that is too high, you can configure a specific maximum value. Likewise, you can set a lower limit to prevent any APs from using a level that is weaker than expected. For consistency, remember to configure TPC identically on all controllers that might be members of the same RF group.

Tip

TPC can be a difficult concept to grasp. You can review its operation in greater detail in Appendix C, “RRM TPC Algorithm Example,” which works through a step-by-step explanation of the algorithm in action in a real-world, high-density scenario.

Dynamic Channel Assignment (DCA)

When multiple APs are used in a wireless network, proper channel assignment is vital for efficient use of airtime and for client mobility. If neighboring APs use the same channel, they can interfere with each other. Ideally, adjacent APs should use different, non-overlapping channels. Working out a channel layout for many APs can be a difficult puzzle, but the DCA algorithm can work out optimum solutions automatically for all APs in an RF group.

Whenever a new AP first powers up, it uses the first non-overlapping channel in each band—channel 1 for 2.4GHz and channel 36 for 5GHz. Consider a simplistic scenario where all APs are new and powered up for the first time. You would end up with a building full of overlapping cells competing for the use of 2.4GHz channel 1. The DCA algorithm works to correct this situation by finding a channel that each AP in the RF group can use without overlapping or interfering with other APs. The basic process is shown in Figure 6-12. Like TPC, DCA works out one channel layout for the 2.4GHz band and another layout for the 5GHz band.

By default, DCA uses the complete list of channels available to a regulatory domain, with the exception of channels subject to DFS restriction. You can choose to include DFS channels as well. DCA also considers the channel width as it plans a channel assignment. By default, it will use 20MHz, but you can configure a different fixed channel width. As well, DCA can leverage Dynamic Bandwidth Selection (DBS) as an additional criteria to maximize throughput by changing the channel width in a dynamic fashion. DBS examines the mix of clients on an AP and decides if they and neighboring APs are compatible with a wider channel and could benefit from additional throughput.

DCA does not just solve the channel layout puzzle once for all APs. The algorithm runs every 10 minutes by default, so that it can detect any conditions that might require an AP’s channel to change. APs in the RF group are monitored for the metrics listed in Table 6-3 that can influence the channel reassignment decision.

Table 6-3 Metrics Affecting DCA Decisions

Metric	Default State	Description
RSSI of neighboring APs	Always enabled	If DCA detects co-channel interference, it may move an AP to a different channel.
802.11 interference	Enabled	If transmissions from APs and devices that are not part of the wireless network are detected, DCA may choose to move an AP to a different channel.
Non-802.11 noise	Enabled	If excessive noise is present on a channel, DCA may choose to avoid using it.
AP traffic load	Disabled	If an AP is heavily used, DCA may not change its channel to keep client disruption to a minimum.
Persistent interference	Disabled	If an interference source with a high duty cycle is detected on a channel, DCA may choose to avoid using it.

Each of these metrics is combined into a single cost metric (CM) that reflects the potential performance that is possible on each channel. Notice that each metric is related to something that either interferes with or competes for airtime on a channel. The CM is something similar to an SNR measurement, with the addition of an interference component. A low CM reflects poor channel performance. The best performance is possible when an AP and a reasonable number of clients have exclusive use of an otherwise quiet channel.

The DCA algorithm tends to look at each AP individually to find the ones with the lowest CM and the worst RF conditions—ones that might be improved by moving them to a better channel. Changing the channel of even one AP can affect many other APs if there are no other viable alternative channels available. Suppose the worst AP is moved to a channel that shows an improved CM, but that action ends up forcing another neighboring AP to a channel that is worse than its original. Such a channel change was probably not wise after all because it did not result in an overall improvement.

To take a more holistic approach, DCA also considers how a channel change to one AP might affect all of that AP’s nearest neighbors, as found in its NDP neighbor list. Channels might need to change for some of those neighbors as well, as a result. DCA also considers the effect of the channel change on the neighbors of the AP’s neighbors, but none of those will be allowed to undergo a channel change. The idea is to gauge any improvement in the local area around the worst AP but move forward with a channel change plan only if the greater area around the AP would see an improvement or stay the same. DCA’s goal is to make improvements, if possible, without impacting all APs in the entire RF group.

Figure 6-13 gives a simple view of the process, where the worst AP is located at the center, and shaded areas contain the AP’s neighbors and neighbors of the neighbors. The darkest shaded area shows the extent of the channel decision, based on the CM of the single AP at the center. Notice that there are many other APs outside the shaded area that are also part of the same RF group but are not considered. Once DCA works on the worst AP, that AP and its neighbors are removed from further channel decisions. DCA then moves on and alternates between other random APs and APs with the worst CM until all APs in the RF group have been examined.

The end result of DCA is a channel layout that takes a variety of conditions into account. The channel layout is not just limited to the two dimensions of a single floor space in a building; it also extends to three-dimensional space because the RF signals from one floor can bleed through to another. As long as the APs on different floors belong to the same RF group, co-channel interference between them should be minimized.

The DCA algorithm uses a sensitivity threshold to keep channel changes under control. Some conditions such as the interference or traffic load on an AP can vary widely and quickly over time, causing the CM values to increase or decrease. Such abrupt changes could cause DCA to make frequent channel assignment changes. To prevent that from happening, the difference between the proposed CM and current CM must be greater than the sensitivity threshold before a change can be recommended by DCA. You can configure the sensitivity to low, medium (the default), or high.

Channel layout is a puzzle that may require several iterations to solve. For this reason, the controller that is the RF group leader will undergo an RRM startup mode after it is elected. During startup mode, DCA ignores the sensitivity threshold and any checks of neighbors of AP neighbors to calculate CM values. The assumption is that the RF group leader has just come online and needs to build the channel assignments from scratch. The startup mode consists of 10 DCA iterations at 10-minute intervals, or a total of 100 minutes before the channel layout reaches a steady state.

From then on, DCA will continue to run at regular intervals (default 10 minutes), but you can configure a different interval or run it manually on demand, if needed. For example, environments that depend on wireless phones might need to maximize network stability with a DCA interval of 8 or 12 hours. You can also configure an anchor time to fix the DCA schedule around a specific time of day. Keep in mind that channel changes can be disruptive to real-time applications.

Note

After APs have been added or removed from a network, or if the AP channel width has been changed globally, Cisco recommends a best practice of manually putting DCA into its startup mode. This will allow the RF group leader to aggressively build a fresh channel assignment without carrying over knowledge of previous conditions.

Event-driven RRM (ED-RRM) takes this a step further; DCA can be automatically triggered based on RF events that occur in real time and can take action with a channel change within 30 seconds. The Cisco CleanAir feature provides the triggers for ED-RRM and can be based on detected signals from an interfering transmitter or a rogue AP. If ED-RRM is enabled and an interfering device is detected on a channel, DCA is triggered and will try to avoid the channel in question. Once ED-RRM has flagged a channel, DCA will avoid using that channel again for 3 hours.

Coverage Hole Detection

The TPC algorithm normally adjusts AP transmit power levels to make cell sizes appropriate. Sometimes you might find that your best intentions at providing RF coverage with a good AP layout still come up short. For example, TPC can gauge appropriate coverage where there are many neighboring APs throughout an area but not many neighbors toward the edge of a coverage area. If clients venture out to the edge, they might find an AP that has its transmit power level set too low, which TPC cannot easily detect. You might also have an AP radio that happens to fail, causing a larger coverage hole. How would you discover such conditions? You could make a habit of surveying the RF coverage often. More likely, your wireless users will discover a weakness or hole in the coverage and complain to you about it.

RRM offers an additional algorithm that can detect coverage holes and take action to address them. Coverage hole detection mitigation (CHDM) can alert you to a hole that it has discovered and it can increase an AP’s transmit power level to compensate for the hole.

Coverage hole detection is useful in two cases:

Extending coverage in a weak area
Rapidly healing a coverage hole caused by an AP or radio failure sooner than the TPC algorithm can detect and correct

The CHDM algorithm does not run at regular intervals like TPC and DCA do. Instead, it monitors the RF conditions of wireless clients and decides when to take action. In effect, the algorithm leverages your wireless users who are out in the field and tries to notice a problem before they do.

Every controller maintains a database of associated clients and their RSSI and signal-to-noise ratio (SNR) values. It might seem logical to think that a low RSSI or SNR would mean a client is experiencing a hole in coverage. Assuming the client and its AP are using similar transmit power levels, if the AP is receiving the client at a low level, the client must also be receiving the AP at a low level. This might not be true at all; the client might just be exiting the building and getting too far away from the AP. The client might also have a “sticky” roaming behavior, where it maintains an association with one AP until the RSSI falls to a very low level before reassociating elsewhere.

Coverage hole detection tries to rule out conditions that are experienced by small numbers of clients and signal conditions due to client roaming behavior. The process begins by discovering individual clients whose RSSI falls below a threshold for a minimum of 5 seconds, causing a number of failed or lost frames. Voice clients must fall below −75 dBm and data clients below −80 dBm. Because single clients and a short duration are used as criteria, these events are called pre-coverage holes, as they may be precursors to holes having a greater impact. Each event is reported to the controller so that it can be logged, tracked, and located.

A full-blown coverage hole is detected when some number and percentage of clients, all associated to the same AP, have RSSI values that fall below a threshold for a longer duration—all while staying associated to the same AP. By default, the following conditions must all be met for a coverage hole to be detected:

The condition must affect at least three clients associated to a single AP
The affected clients must make up at least 25 percent of clients on the AP

Unlike TPC and DCA, which operate on the entire RF group, coverage hole detection runs on a per-AP radio basis. Also, each controller runs CHDM independently. Once a coverage hole has been detected, CHDM can mitigate the effects of the hole by incrementing the AP’s transmit power level by one step. This doubles the power level of the AP and keeps the power level change minimized to avoid creating co-channel interference and working against the efforts of the TPC algorithm.

Flexible Radio Assignment (FRA)

With a properly designed wireless network, the TPC algorithm can tune the transmit power level of each AP to optimize cell coverage and minimize co-channel interference. Recall that TPC runs on the 2.4 and 5GHz bands independently. Wireless designs usually determine AP placement based on the 5GHz AP cell boundaries because the 5GHz band offers more non-overlapping channels to use in high-density areas. Once an AP has been mounted, each of its radios (and internal antennas) has a coverage cell that is centered around the AP’s location. If the AP has one 2.4GHz radio and one 5GHz radio, you might think the two cells would be superimposed to cover the same area, assuming the same transmit power level is used for each. However, each signal is subject to free space path loss, which is frequency dependent. That means at the same transmit power level, a 2.4GHz signal will reach its −67 dBm boundary quite a bit farther away from the AP than a 5GHz signal will.

If a design is based on 5GHz cell sizes, the end result is properly overlapped 5GHz cells with greatly overlapped 2.4GHz cells. Notice the difference between 5GHz and 2.4GHz cells in Figure 6-14. Because of the excessive overlap, some of the 2.4GHz cells can be considered to be redundant, as they add no unique coverage. Instead, they can make co-channel interference worse. Figure 6-15 shows overlapped coverage of several 2.4GHz radios. Consider the highlighted AP cell in the left portion of the figure and then notice the RF coverage in the right portion with that AP removed. Even with the AP absent, the area is still almost adequately covered. That means the 2.4GHz radio could be turned off or disabled with no noticeable impact.

RRM offers the Flexible Radio Assignment (FRA) algorithm, which can detect redundant 2.4GHz radios and repurpose them in the 5GHz band instead. FRA works with Cisco APs that offer one 5GHz radio and one flexible (XOR) radio, which can operate as a 2.4GHz radio, a 5GHz radio, or a monitoring radio.

FRA uses the NDP neighbor list data to compute a Coverage Overlap Factor (COF) for each 2.4GHz AP radio. The COF indicates percentage of an AP’s cell area that has overlapping signals from other APs that are −67 dBm or stronger. If the COF meets or exceeds a threshold defined as low (100%), medium (95%), or high (90%), the radio is flagged as redundant and can be reassigned to a different role to make it more useful. Once a radio becomes redundant, FRA will not allow any of the other radios that overlapped it to become redundant too.

A redundant 2.4GHz flexible (XOR) radio will normally be reassigned to 5GHz duty, where it will overlap the AP’s other 5GHz radio cell. The new overlap will not be destructive because the two 5GHz radios will be assigned to different channels so that they can share the load of associated clients. The additional 5GHz coverage is especially beneficial in high-density coverage areas. The cell size of each 5GHz radio depends on the AP model; some models support a macro/macro arrangement, where the two radios produce similarly sized cells. Other models support a macro/micro scheme, where one 5GHz radio produces a smaller cell footprint than the other, in a cell-within-a-cell arrangement. If the 5GHz coverage is already sufficient, then the redundant radio will be reassigned to a monitor role instead.

Note

Once a flexible (XOR) radio has been reassigned from 2.4GHz to 5GHz, it will continue to operate that way. You can manually configure it to revert back to 2.4GHz operation if needed. Also, if the CHDM algorithm detects a 2.4GHz coverage hole at the AP’s location, it will immediately revert the radio back to 2.4GHz operation.

When two AP cells overlap each other on different channels, it might be difficult to get wireless clients evenly distributed across the two radios. FRA can use the following three different methods to influence which BSS clients are able to join:

802.11v BSS transition—If a client is 802.11v-capable, the AP will send the client a neighboring AP list containing only the target BSS, followed by an 802.11 deauthentication message. This effectively forces the client off its previous BSS so it can attempt to join the only advertised neighbor BSS. This method is enabled by default.
802.11k site report—If a client requests an 802.11k site report to determine available BSSs, the AP will respond with a site report containing only the target BSS. This method is enabled by default.
802.11 probe suppression—When probe requests are received on both 5GHz BSSs, the AP can respond from the target BSS and stay silent from the non-optimal BSS. This method is disabled by default.

In a macro/micro cell scenario, FRA also monitors client RSSI so it can steer clients toward the macro or micro BSS based on signal strength thresholds.

You can influence FRA actions through Client Network Preference, choosing to prefer either connectivity or throughput. Preferring connectivity will tend to keep the network stable, by preventing redundant radios from switching bands if more than three clients are present at the time DCA runs. Preferring throughput will allow FRA to reassign redundant radios to the other band regardless of the number of clients present.

Localizing RRM with RF Profiles

RRM can be enabled and configured on a WLC globally and independently for the 2.4 and 5GHz bands. Global configuration allows you to quickly make adjustments to all APs in an RF group at once. In large enterprises or geographically dispersed networks, you may need to tune RRM operation in one area differently from others. On a global scale, there is no easy way to accomplish that.

A common practice is to create logical AP groups on a WLC and assign specific APs to their respective groups. For example, a large campus might consist of many buildings. If each building has its own AP group, then RRM parameters can be applied on a per-group or per-building basis. AP groups can also contain a unique list of WLANs to be offered. Even within a building, you might have auditoriums or large classrooms where high densities of clients are expected. You could map those APs into their own AP group and treat them separately from other APs in the building.

You can also tailor RRM parameters to specific needs by configuring them in RF profiles, which then get applied to AP groups. This gives great flexibility for localized RF tuning and policy definition. An RF profile contains the following parameter definitions:

802.11 data rates and MCS support
TPC power level limits and thresholds
DCA channel list and channel width
CHDM thresholds
RxSOP threshold (covered in the next section)
Client density and distribution options

By defining RF profiles and applying them to AP groups, you can customize TPC, DCA, and CHDM algorithms for APs that are members of an AP group. Figure 6-16 shows an example of an enterprise RF group that uses AP groups and RF profiles to tune RRM for each building. The global 5GHz RRM parameters have been configured to use one mandatory data rate of 12Mbps. All possible channels and 5GHz bands have been enabled for use by DCA. The TPC transmit power level limits have been left at the defaults of 30 and −10 dBm. The global RRM values will be applied to all APs that are not members of a specific AP group. Also, the global settings define the values that are inherited by all AP groups. The APs in two buildings have been mapped to an AP group for each building, MainBuilding and BuildingA, while APs in large auditoriums are members of an AP group called MainAuditoriums. A specific RF profile has been defined and applied to each AP group. Notice the settings listed in bold type that are unique in each RF profile and customized for each AP group.

Note

As you work out a wireless design, keep in mind that every RF profile must first inherit its parameters from the global RRM settings. For example, that means you can only make changes to channel numbers in an RF profile that have been enabled globally. Parameters that have been disabled globally will not be available to use in any RF profile.

Optimizing AP Cell Sensitivity with RxSOP

In Chapter 5, you learned how neighboring APs can interfere with each other if they share the same channel and their signal strength is above a threshold. The Clear Channel Assessment (CCA) threshold is one mechanism used by wireless stations to determine if a channel is busy. If a station has a frame to transmit, it must first check to see if the channel is busy. A signal received above the CCA threshold of −82 dBm (2.4GHz channel) or −85 dBm (5GHz channel) indicates that the channel is busy, so the station about to transmit must wait until the channel is clear. The more this happens, the greater the channel utilization will be, using up more and more of the airtime.

Ideally, a wireless design should space APs far enough apart to have an appropriate cell size and cell overlap, but neighboring APs assigned to the same channel should not be able to receive either other’s signal above the CCA threshold. In practice, this is not always possible, especially in high-density areas. Even with low transmit power levels and directional antennas, the APs might still be close enough to contend for airtime unnecessarily.

Beyond the edge of an AP’s cell, conditions can become difficult for client devices too. Clients normally roam from AP to AP as they move about. When a client leaves a coverage area, there are no more APs to roam to, so communication continues until either the AP or the client has an RSSI that falls below the receiver sensitivity. In other words, clients can try to use an AP’s cell past the planned boundary.

Cisco offers the Receiver Start of Packet Threshold Detection (RxSOP) feature, which can solve both of these problems. RxSOP applies an RSSI threshold as wireless frames are received by an AP. If the signal strength is above the threshold, the frame is received and demodulated normally. If it is below the threshold, the AP simply ignores the rest of the frame as noise.

RxSOP can be useful in the case of distant APs sharing the same channel. Where the 802.11 CCA threshold would normally prevent an AP from ignoring weak transmissions from a distant AP, RxSOP defines a second-higher threshold that can filter them out. In effect, RxSOP puts “earmuffs” on the AP’s ears to dampen out signals that are almost too weak to use. The end result is increased efficiency because the AP no longer has to wait for weak signals to finish transmitting before it can use the channel. Where channel utilization was once high because of weak competing devices, RxSOP filtering can open up more airtime and lower the utilization.

Figure 6-17 illustrates the effect of an RxSOP threshold on received signals at an AP. AP-1 and AP-3 share the same channel 36, even though they are separated by some distance. The graph shows how AP-1’s signal strength attenuates over the distance from AP-1’s location to a point past AP-3. Suppose AP-3 needs to transmit a frame but performs the mandatory CCA check to see if channel 36 is busy. If AP-1 is already transmitting, its signal strength stays above the CCA threshold of −85 dBm even past AP-3, forcing AP-3 to defer and wait. If an RxSOP threshold of −78 dBm is added to the scenario, AP-1’s signal strength falls below that threshold just past AP-2. Therefore, if AP-3 uses RxSOP, the transmission from AP-1 will fall below the threshold at AP-3’s location. AP-3 will be free to drop AP-1’s frame and begin to transmit, provided no other stronger signals are received and in progress.

The RxSOP threshold also serves to reduce AP cell size, addressing distant clients that do not undergo a clean break as they travel outside the RF coverage. Once client signals fall below the RxSOP threshold, the AP no longer spends time processing the client’s frames. This tends to force clients off the Wi-Fi network sooner, giving them a better chance to switch over to cellular service instead, if they are able.

By default, RxSOP is disabled on all APs in an RF group. You can enable and tune RxSOP globally for an entire RF group, but it is usually best to leverage it more locally through the use of an RF profile and AP groups. Through the WLC GUI, you can set the RxSOP threshold as Low, Medium, High, or Auto, on a per-band basis. Table 6-4 lists the settings and their corresponding RSSI values in dBm.

Table 6-4 RxSOP Threshold Settings and RSSI Values

Band	High	Medium	Low	Auto
2.4GHz	−79 dBm	−82 dBm	−85 dBm	Radio default
5GHz	−76 dBm	−78 dBm	−80 dBm	Radio default

Note

You should use caution and careful planning if you decide to implement an RxSOP threshold in a network. If you set the threshold too high, AP cell sizes could become much smaller than expected. That can open up coverage holes between APs and leave wireless clients stranded with no connectivity. It is best to start with a low threshold, test the results, and then decide if the threshold needs to be increased further.

Summary

This chapter described the main concepts needed to configure and manage individual AP radios. More precisely, you have learned the following:

How managing radios manually can become an administrative burden
How controllers and APs can discover their RF neighborhood and other nearby APs
How radio management can be organized through RF groups
How the transmit power levels of multiple APs can be calculated and set automatically with the TPC algorithm
How channels can be assigned to APs automatically through the DCA algorithm
How RF coverage holes can be automatically discovered
How RF profiles can be used to make localized changes to the RRM algorithms
How AP cell sensitivity can be adjusted and optimized with the RxSOP feature

Exam Preparation Tasks

As mentioned in the section “How to Use This Book” in the Introduction, you have a couple of choices for exam preparation: the exercises here, Chapter 18, “Final Preparation,” and the exam simulation questions in the Pearson Test Prep Software Online.

Review All Key Topics

Review the most important topics in this chapter, noted with the Key Topic icon in the outer margin of the page. Table 6-5 lists these key topics and the page numbers on which each is found.

Table 6-5 Key Topics for Chapter 6

Key Topic Element	Description	Page Number
Figure 6-3	Measuring the RSSI of NDP messages from an AP	115
Figure 6-8	An RF group and its RF neighborhoods	120
Figure 6-9	Basic concept of the TPC algorithm	121
Figure 6-10	TPC and the third strongest NDP message	121
Table 6-2	Example AP transmit power level numbers	123
Figure 6-12	Basic DCA operation	124
List	Coverage hole detection criteria	128
Paragraph	FRA coverage overlap percentage criteria	129
List	RF configuration parameters available in RF profiles	131
Figure 6-17	Effects of setting an RxSOP threshold	133

Define Key Terms

Define the following key terms from this chapter and check your answers in the glossary:

cost metric (CM)

Coverage hole

coverage hole detection mitigation (CHDM)

coverage overlap factor (COF)

dynamic bandwidth selection (DBS)

dynamic channel assignment (DAC)

event-driven RRM (ED-RRM)

Flexible Radio Assignment (FRA)

neighbor discovery protocol (NDP)

radio resource management (RRM)

transmit power control (TPC)