Spectrum Management
802.11a was originally developed as a standard for the United States only. European radio regulations for the 5 GHz frequency band are stricter, which necessitated the development of special procedures to adapt the 802.11 MAC for use in Europe, which were eventually standardized as 802.11h in 2003.
Transmit Power Control (TPC)
Transmit power control is required by European regulations[*] to ensure that 5 GHz radio transmitters stay within regulatory power limits and avoid interfering with certain satellite services. Developing better control over transmission power brings many other benefits, some of which have been long known in the world of mobile telephony. High-powered client transmissions may cover very large areas. In a densely-deployed network of APs, a single client at high power may have much higher range than is necessary. Long range is not always a plus. Running a radio transmitter at high power decreases battery life. If the range of a transmission is longer than necessary, the extra reach represents "wasted" power. Higher power may also lead to a reduction in network throughput as client devices interfere with each other unnecessarily.
[*] ERC/DEC/(99)23, available at http://www.ero.dk/doc98/Official/Pdf/DEC9923E.PDF.
Figure 8-21. High client power interference with multiple APs
For an illustration of problems caused by high client power, consider Figure 8-21. In the figure, nine APs are shown in a network.[*] Client #1 is associated with the center AP in the top row. If the client computer is configured to transmit at maximum power, the range of the transmission will reach out to the outer circle. However, the transmission to the AP requires power to only the inner circle. The difference between the two circles is the excess range caused by transmitting too high. Running the radio transmitter higher than necessary will drain the battery on a portable device faster than is strictly necessary. (To extend the battery life of phones, mobile telephone networks have incorporated transmit power limitation techniques for many years.)
[*] Note that laying out a nine-AP network using only three non-overlapping channels is not possible, which is a strong argument for using 802.11a in high-density environments.
Controlling the range of transmission also makes the network function more smoothly. All transmissions must gain exclusive access to the radio medium. When excessive power is used, the area covered by a transmission is much larger than it should be. Any station operating on channel 11 within the outer circle must defer to a frame in progress from Client #1. The shaded area between the optimum transmit power and the excessive power represents the area which is blocked because Client #1 is transmitting at power that is too high. For example, Client #2 is blocked from communicating with its nearby AP because it must share the radio medium. Reducing transmission power to the level required to reach only the serving AP may improve communications throughout the network by limiting overlap between nearby APs.
Basic operation of transmit power control
Transmit power control (TPC) is an 802.11 service that attempts to hold transmit power to the lowest possible productive level. It takes into account the maximum power allowed by regulators, as well as further constraints. Although designed to satisfy regulatory requirements, transmit power control may have additional benefits.
The absolute cap on any radio transmission is set by regulatory authorities, and is found in the relevant documents and rules published by the authority. Regulatory maximum power may be configured into an AP or station, or it may be learned from Beacon frames containing Country elements. Radio transmissions may be subject to additional constraints that further reduce the maximum power. European regulations specify a further mitigation requirement of at least 3 dB to reduce interference with satellite services.
Maximum transmission power is specified using the Country element in Beacon frames, and is available to any station wishing to associate to a network. The Country element specifies the regulatory maximum power, and the Power Constraint element can be used to specify a lower maximum transmission power specific to the network.
Before beginning operation, stations must calculate the maximum transmission power that may be used. Typically, this is calculated by taking the regulatory maximum power and subtracting any constraints for mitigation or additional local constraints. For example, network administrators may wish to specify a restrictive local maximum transmission power to reduce range and therefore interference.
Changes to the association process
When a spectrum management-capable station associates (or reassociates) to an access point, it must supply the minimum and maximum transmission power in a Power Capability information element. Access points may incorporate the information supplied by a client into the association process, and are free to use it any manner. How, or even whether, an access point uses the power capability information is not specified by 802.11 standards. Stations that have high power and may violate regulations may be rejected to preserve regulatory compliance. Access points may also reject clients with low transmit power capability; the standard suggests that it may increase the possibility of hidden nodes, although that seems unlikely.
Changing the transmission power
Both access points and client stations may dynamically adjust the transmission power on a frame-by-frame basis. For each frame, the receiver may compute the link margin, the amount by which the received power exceeds the minimum acceptable value. Link margin is the margin of safety. If a station's transmissions were received at the minimum acceptable value, the link margin would be zero, indicating that any detrimental change to the link would interrupt communication. Most stations aim for a small link margin, but the specific calculation of link margin and the determination of a particular desired value are not specified by the standard.
To make informed changes to transmission power, stations may request a radio link measurement. Stations send an Action frame requesting a transmission report. The Action frame sent in response contains a TPC Report element with two descriptive statistics. First, it contains the transmit power of the report frame. Based on the transmission power, the receiver of the report can estimate the path loss of the radio link. Second, the report frame contains a value for the link margin, which informs the receiver of the ratio between the received power and the minimum acceptable power. If the minimum acceptable power were, say, -70 dBm and the signal was received at -60 dBm, the link margin would be reported as 10 dB. If the link margin is "too high," transmit power can be reduced. If the link margin is "too low," the transmit power can be increased. Like many other components of the 802.11 standard, "too high" and "too low" are left to the discretion of the software.
The standard is designed to ensure that the maximum transmission power is not exceeded, but it does not restrict power selection in any way. There is no requirement for advanced functionality. That said, it is conceivable that advanced access points supporting transmit power control may attempt to keep track of the power required to reach each associated station, so that close-in stations require less power than far-away stations.
Dynamic Frequency Selection (DFS)
In addition to the requirement for transmit power control, European regulations require that stations avoid interfering with 5 GHz radar systems, as well as spread the power load across all available channels. Accomplishing this is the task of 802.11's Dynamic Frequency Selection (DFS) mechanism.
Basic operation of DFS
Dynamic frequency selection consists of a number of procedures to enable 802.11 devices to change the radio channel based on measurements and regulatory requirements. It can affect the initial association procedure and ongoing network operation.
When stations first associate to the network, the Association Request frame includes a Supported Channels information element, which communicates the channels supported by the station. Access points may reject the association based on the content of the information element, though such behavior is not specified by the standard. One approach would be to reject stations that support "too few" channels, on the theory that it limits the ability of the access point to switch operation on to different channels because it must move to a channel supported by all associated stations.
Once used on an operational network, DFS will periodically test the channel for potential interference from other radio systems, most notably 5 GHz European radar systems. Testing the channel is accomplished by stopping transmissions on the network, measuring for potential interference, and, if necessary, advertising that the channel will change.
Quieting the channel
To perform tests on the radio channel, quiet periods or quiet intervals are used. Quiet intervals are a time when all stations in the BSS do not transmit, which is helpful when making measurements for potential interference from radar systems. Quiet periods are scheduled by the inclusion of the Quiet information element in Beacon and Probe Response frames, and describe when and how long all stations should cease transmissions. Only the most recently transmitted Quiet information is effective. When multiple Quiet information elements are transmitted, the most recent one supercedes all prior scheduled quiet periods. During the quiet period, all stations set the network allocation vector (NAV) to the length of the quiet period to ensure that the virtual carrier sensing algorithm will defer transmissions.
When an upcoming quiet interval is scheduled, the radio channel still operates under normal rules for access to the radio medium, with the additional rule that any frame exchange must complete before the start of the quiet period. If a scheduled frame exchange cannot be completed, the station relinquishes control of the channel and defers transmission until after the conclusion of the quiet period. Failure to transmit a frame due to an impending quiet period does not, however, increase the transmission count. When the quiet period resumes, all stations must contend for access to the radio channel again. There is no preservation of channel access across a quiet period.
In an infrastructure network, channel quiet scheduling is completely under the control of the access point. Access points are allowed to change the duration of quiet periods, the time between quiet periods, or even to stop scheduling quiet periods altogether. Independent networks choose the quiet period scheduling when the network is created. When new stations take over responsibility for sending Beacon and Probe Response frames, they have no discretion to alter the quiet period parameters, and simply copy the previously used parameters.
Measuring
At any point, measurements of the radio channel can be taken. Stations may request other stations to measure the radio channel. Access points may find measurements from stations to be particularly useful in learning about the state of the radio channel because the reports from stations are likely to come from a variety of different geographical locations. Measurements may be taken in a quiet period or while the radio is in service.
Any station may ask other stations to make a measurement. Inquiries for radio information are sent in Measurement Request frames, described in Figure 8-23. In an infrastructure network, all frames must go through the access point. Associated client stations may only ask the AP for radio information. Access points in an infrastructure network may ask either a single station or a group of stations for a measurement, simply by addressing the request frame appropriately. In independent networks, there is no centralized point of control, and any station may issue a request to any other single station or group of stations. Although requests to a group address field are allowed, receiving stations are also allowed to disregard them.
After sending a measurement request, the standard assumes that the receiver needs time to gather the data for its reply. Following the transmission of a measurement request, a station refrains from sending any further frames.
Upon receipt of a Measurement Request frame, a station must determine how to respond. The Measurement Request frame must always be answered, even if the response is a refusal to perform the requested measurement. To be processed, a request must be received in enough time to set up and take the measurement. Measurement requests specify a time at which the measurement is to occur. If a request is held in a long transmission queue, it is conceivable that it could be arrive at the destination after the requested measurement should begin. Stations are allowed to ignore such "late" requests. The receiver of a measurement request must also be able to collect the data supported in the request. Depending on the receiver's hardware, it may not be possible to support all requests. Although increasingly rare, not all 802.11 hardware is capable of supporting every allowed channel.[*] Many cards on the market today do not support multichannel operation, either, so a request to measure a different channel from the current operating channel cannot be supported. Stations may refuse to perform a measurement for any nonspecified reason, provided the requester has not marked the request as mandatory.
[*] For example, the Cisco CB-20 card, based on the Radiata chipset, could only support the eight lowest 802.11a channels. It would be appropriate for such a card to refuse a measurement request in the high U-NII band.
In addition to the polled operation, in which stations ask for measurements, it is possible for stations to spontaneously report statistics by sending unsolicited Measurement Report frames.
Radar scan
One of the major reasons to quiet the channel is to search for the presence of 5 GHz radar systems in use in Europe. The regulations[*] do not require any particular search method for these radar systems. It only requires that radar be detected when its signal strength rises above a defined interference threshold.
[*] ETSI EN 301 893, available from http://www.etsi.org.
When a radio interface is started, it must search for a radar signal on the channel. No transmissions are allowed until the "coast is clear" and it has been established that there is no radar to interfere with. Radar detection must be carried out periodically throughout operation. Whenever radar signals are detected, the network must switch to another channel to avoid interference.
Spectrum management services enable a network to move to another channel. The decision to move to another channel may be caused by the presence of radar interference, but the generic mechanism to switch channels may be useful for a variety of purposes beyond compliance with European radio regulations. Networks capable of altering their operational channel may do so to minimize interference with other 802.11 devices and optimize the radio plan. Channel switching is designed to move as many of the associated stations to the new channel as possible, but it is always possible that communications with some or all of the associated stations may be disrupted. The standards do not place any constraints on how to select the new channel, but simply say that an access point should attempt to choose a channel supported by as many of its stations as possible. Some regulators have drafted rules that mandate energy spreading across the entire band, so it may be that regulatory requirements force a switch to channels that are not supported by all stations.
In an infrastructure network, the selection of the operating channel is under the sole control of the access point. As part of the association process, access points collect information about which channels can be supported by the associated stations. Access points inform associated stations of the impending switch by using the Channel Switch Announcement information element in management frames, as well as Action frames. To improve the ability of an access point to send an channel switch announcement, the appropriate action frame may be sent after the PCF Interframe Space (PIFS), which gives it a higher priority than any new atomic exchange on the medium. The standard suggests, but does not require, that the channel switch be scheduled far enough in advance that any powersaving stations have the opportunity to become active and receive the channel switch announcement.
IBSS operation
Changing the operating channel in an independent network is significantly more complicated than in an infrastructure network because there is no single logical controller for frequency selection operation. Instead of having the function reside in the access point, independent networks have the DFS owner service, which coordinates multiple stations to run the frequency selection service.
One station in a network is designated the DFS owner, and is responsible for collecting measurement reports and monitoring the channel for radar signals. If any station in the independent network observes a radar signal, it will be reported in the channel map subfield. Upon notification that radar has been detected, the DFS owner takes charge of changing the channel.
The DFS owner is responsible for deciding on the new channel, and sending the channel switch announcement frame. It may not be possible to select a channel that complies with regulatory requirements and is supported by all stations. Independent networks do not have a central point of data collection, so even if there is a channel supported by all stations, there is no guarantee that the DFS owner will be aware of it.
In an independent network, the DFS owner may change. Just like Beacon generation, stations may join or leave the network, including the DFS owner. To cope with the DFS owner going away, stations may enter DFS owner recovery mode. In this mode, multiple stations can become DFS owners, and will schedule the channel switch announcement frames required by the standard. However, the first station to transmit the channel switch frame becomes the only DFS owner, and other stations will drop the role. DFS owner recover is similar in concept to the distributed Beacon generation discussed in Figure 8-20.
Action Frames
Action frames are used to request a station to take action on behalf of another. Spectrum management services use Action frames to request that measurements be taken, gather the results of those measurements, and announce any required channel switches. Figure 8-22 shows the format of the Action frame, which is essentially a category plus details that depend on the category. The action details may vary depending on the category field.
Figure 8-22. Action frame
Category
The category field is set to zero for spectrum management.
Action
All spectrum management frames use the first byte of the action details to specify the type of action being undertaken. Table 8-2 shows the possible values for the Action subfield. Values not shown are reserved.
Elements
Spectrum management action frames carry information in information elements. The details for many information elements can be found in Chapter 4. Several additional information elements were defined in 802.11h and are described in this section.
Value |
Type of spectrum management action frame |
---|---|
0 |
Measurement Request |
1 |
Measurement Report |
2 |
TPC Request |
3 |
TPC Report |
4 |
Channel Switch Announcement |
Measurement Request frame
The Measurement Request frame is used to request that a station make measurements and send the results to the sender. Its format is shown in Figure 8-23. The frame consists of a series of measurement request information elements. The number of potential measurements is limited by the size of the frame, rather than any aspect of the frame construction.
Periodic measurements are allowed by the standard. Periodic reports are enabled or disabled by sending a measurement request to a station with instructions to turn the periodic measurement on or off. Stations cannot disable measurements on access points in infrastructure networks.
Category
Set to zero to indicate spectrum management action frames.
Action
Set to zero to indicate that it is a measurement request.
Dialog Token
This field acts like a sequence number. It is set to a non-zero value to assist in matching measurement responses to outstanding requests.
A single Measurement Request frame may request multiple measurements by using multiple Measurement Request information elements within the body of the frame.
Figure 8-23. Measurement Request frame
A single information element is shown in exploded view, consisting of the following fields:
Element ID
Measurement Request elements are type 38.
Length
The length, in bytes, of the information element following this field.
Measurement Token
Each Measurement Request frame may make several requests by including several Measurement Request elements within the frame body. Each request is given its own value for the Measurement Token so that the different requests can be distinguished.
Measurement Request Mode bitmap
There are three bits in the Measurement Request Mode bitmap that are used to indicate what types of spectrum management frames are supported. Bit number 2 (numbered starting from 0) is the Request bit, and is set to one to signify that the transmitter will process incoming measurement requests. Bit number 3 is the Report bit, which is set to one to signify that the transmitter will accept unsolicited reports. The Enable bit is set to one when the other two bits are valid.
Measurement Type
The type of measurement being requested by the information element, as shown in Table 8-3.
Measurement Request
If a measurement is requested, there may be an additional field to give timing parameters. As it turns out, the three types of measurements that are currently standardized all have the same format, consisting of a channel number, the value of the timer function at which the measurement should start, and the duration of the measurement in time units. A timer start value of zero indicates the measurement should be taken immediately. This field is not present when the frame is used to turn measurements on or off.
Measurement Type value |
Name |
---|---|
0 |
Basic measurement |
1 |
Clear channel assessment |
2 |
Receive power indication (RPI) histogram |
Measurement Report
The Measurement Report frame is used to send the results of a measurement to the requester. Its format is shown in Figure 8-24. The frame consists of a series of measurement report information elements. The number of potential reports is limited by the size of the frame, rather than any aspect of the frame construction.
Category
Set to zero to indicate spectrum management action frames.
Action
Set to one to indicate that it is a measurement report.
Dialog Token
If the measurements were taken as the result of a measurement request from another station, the Dialog Token field from that request is copied into the response. If the frame is sent as an unsolicited report, the Dialog Token is zero.
A single Measurement Report frame may contain the results for several measurements, each transmitted in its own information element. For clarity, a single information element is shown in the information element header, and the three possible report elements are shown below.
Element ID
Measurement Request elements are type 39.
Length
The length, in bytes, of the information element following this field.
Measurement Token
Each Measurement Request frame may make several requests by including several Measurement Request elements within the frame body. Each request is
Figure 8-24. Measurement Report frame
given its own value for the Measurement Token so that the different requests can be distinguished.
Measurement Report Mode bitmap
There are three bits in the Measurement Report Mode bitmap that are used to indicate why a measurement is being refused, if the report frame is being used to deny a measurement. The Late bit is set to one when the measurement request arrives after the start time specified in the measurement. The Incapable bit is set to one when a station is unable to perform the requested measurement. The Refused bit is set to one when the station is capable of performing the measurement, but does not wish to do so.
Measurement Type
The type of measurement being requested by the information element, as shown in Table 8-3.
Measurement Report
The Measurement Report frame contains the requested measurement data. Unlike the Measurement Request, the contents are different for each type of report. All three reports are shown underneath the information element header. All share a common header that reports the channel number that the request is for, the time the measurement started, and the duration of the measurement. However, the data is reported in different ways for each type of measurement.
In a basic report, the data reported are a series of bit flags for the channel:
BSS (1 bit)
This bit will be set if frames from another network are detected during a measurement period.
OFDM Preamble (1 bit)
This bit is set if the 802.11a short training sequence is detected, but without being followed by the rest of the frame. HIPERLAN/2 networks use the same preamble, but not the same frame construction.
Unidentified Signal (1 bit)
This bit is set when the received power is high, but the signal cannot be classified as either another 802.11 network (and hence, set the BSS bit), another OFDM network (and hence, set the OFDM Preamble bit), or a radar signal (and hence, set the Radar bit). The standard does not specify what power level is high enough to trigger this bit being set.
Radar (1 bit)
If a radar signal is detected during a measurement period, this bit will be set. Radar systems that must be detected are defined by regulators, not the 802.11 task group.
Unmeasured (1 bit)
If the channel was not measured, this bit will be set. Naturally, if there was no measurement taken, nothing can be detected in the band and the previous four bits will be set to zero.
In a CCA report, the main field is the CCA Busy Fraction, which describes the fraction of time on which the clear channel assessment function was set to busy. It is a single byte, so the fraction is multiplied by 255 to convert to an integer ranging from 0 to 255, where higher values indicate the channel was busy more often.
RPI histogram reports are used to report the spread of received power on an interface. Stations can request an RPI histogram to determine how well another station is able to see signals from the current network, or it may use the report to scout out other channels when it is time to change the operating channel. An RPI histogram report contains information on the strength of received signals. However, unlike a one-frame measurement, it can report the spread of power received over the measurement duration to give the receiver an indication of the overall level of transmissions. The histogram contains eight bytes, each of which represents a range of received power, as shown in Table 8-4. Each byte has a value representing the fraction of power signals that fall into its range. The time fraction over which the received signal falls into the power range for the byte is scaled so that each byte ranges from 0 to 255, with its value depending on the fraction of time signals were received at that power level.[*]
[*] Although the values of all 8 bytes theoretically add to 255, the standard notes that they may add up to 262 due to rounding effects.
RPI number |
Corresponding power (dBm) |
---|---|
0 |
Less than -87 |
1 |
-87 to -82 |
2 |
-81 to -77 |
3 |
-78 to -72 |
4 |
-71 to -67 |
5 |
-66 to -62 |
6 |
-61 to -57 |
7 |
-56 and higher |
TPC Request and Report
TPC Request and TPC Report frames are both shown in Figure 8-25. They are both straightforward, consisting of Action frames with the spectrum management type. Each frame contains its corresponding information element, as described in Chapter 4. As with other frames, the Dialog Token field is used to match up requests with responses.
Figure 8-25. TPC Request and Report frames
Channel Switch Announcement
When the channel must be changed, stations that are part of the network must be informed of the impending change so that they may prepare to switch to the specified new channel. Channel Switch Announcement frames, shown in Figure 8-26, are essentially Action frame wrappers around the Channel Switch Announcement element that was described in Chapter 4. As such, they have all the functions of a channel switch announcement element, and are used to specify the time at which the network will switch to a new channel.
Figure 8-26. Channel Switch Announcement frame