Wi-Fi MultiMedia (WMM), formerly known as Wireless Multimedia Extensions, refers to QoS over Wi-Fi. QoS enables Wi-Fi access points to prioritize traffic and optimize the way shared network resources are allocated among different applications.
This section describes the following three considerations for WMM implementation:
WMM is a Wi-Fi Alliance certification of support for a set of features from an 802.11e draft. This certification is for both clients and APs, and certifies the operation of WMM. WMM is primarily the implementation of enhanced distributed coordination function (EDCF) component of 802.11e. Additional Wi-Fi certifications are planned to address other components of the 802.11e.
WMM uses the 802.1P classification scheme developed by the IEEE (which is now a part of the 802.1D specification).
This classification scheme has eight priorities, which WMM maps to four access categories: AC_BK, AC_BE, AC_VI, and AC_VO. These access categories map to the four queues that are required by a WMM device, as shown in the following table.
Table 2 802.1P and WMM Classification
The following figure shows the WMM data frame format.
Figure 3. WMM Frame Format
Even though WMM maps the eight 802.1P classifications to four access categories, the 802.1D classification is sent in the frame.
The WMM and IEEE 802.11e classifications are different from the classifications that are recommended and used in the Cisco network, which are based on IETF recommendations. The primary difference in classification is the change of voice and video traffic to 5 and 4, respectively. This allows the 6 classification to be used for Layer 3 network control. To be compliant with both standards, the Cisco Unified Wireless solution performs a conversion between the various classification standards when the traffic crosses the wireless-wired boundary.
The following figure shows the queuing that is performed on a WMM client or AP.
There are four separate queues, one for each of the access categories. Each of these queues compete for the wireless channel, with each of the queues using different interframe space, contention window (CW) minumum (CWmin) and contention window maximum (CWmax) values as defined by EDCF. If more than one frame from different access categories collide internally, the frame with the higher priority is sent, and the lower priority frame adjusts its backoff parameters as though it had collided with a frame external to the queuing mechanism.
The following figure shows the principle behind EDCF where different interframe spacing and CWmin and CwMax values (for clarity, CwMax is not shown) are applied per traffic classification.
Figure 5. Access Category (AC) Timing
Different traffic types can wait different interface spaces before counting down their random backoff, and the CW value used to generate the random backoff number also depends on the traffic classification. For example, the CWmin for voice traffic is 23-1, and CWmin for best effort traffic is 25-1. High priority traffic has a small interframe space and a small CWmin value, giving a short random backoff, whereas best effort traffic has a longer interframe space and large CWmin value, that, on average, gives a large random backoff number.
The following figure shows the WMM information in a probe response.
Figure 6. Probe Response WMM Element Information
The elements on the client not only contain WMM AC information, but also define which WMM categories require admission control. For example, in the preceding figure, the admission control for voice AC is set to Mandatory. Therefore, the client is required to send the request to the AP, and have the request accepted, before it can use this AC.
Unscheduled automatic power-save delivery
Unscheduled automatic power-save delivery (U-APSD), a WMM feature of Wi-fi devices, provides two key benefits:
Allows the voice client to synchronize the transmission and reception of voice frames with the AP, allowing the client to transition into power-save mode between the transmission or reception of each voice frame tuple. The WLAN client frame transmission in the access categories supporting U-APSD triggers the AP to send any data frames that are queued for that WLAN client in that AC. A U-APSD client remains listening to the AP until it receives a frame from the AP with an end-of-service period (EOSP) bit set. Once the client receives a frame with the EOSP bit set which indicates there are no other frames, the client goes back into power-save mode. This triggering mechanism is a more efficient use of client power than the regular listening for beacons method, at a period controlled by the delivery traffic indication map (DTIM) interval. This is because the latency and jitter requirements of voice and video are such that a voice and video over IP (VVoIP) client would either not be in power-save mode during a call, resulting in reduced talk times, or would use a short DTIM interval, resulting in reduced standby times. U-APSD allows the use of long DTIM intervals to maximize standby time without sacrificing call quality. You can apply this feature individually across access categories; however, only voice ACs in the AP use U-APSD and other ACs still use the standard power-save feature.
Increases call capacity The coupling of transmission buffered data frames from the AP with the triggering data frame from the WLAN client allows the frames from the AP to be sent without the accompanying interframe spacing and random backoff, thereby reducing the network contention.
The following figure shows an example of traffic flow with U-APSD.
Figure 7. U-APSD Traffic Flow
In this example, the trigger for retrieving traffic is the client sending traffic to the AP. When the AP acknowledges the frame, it indicates the client that data is in queue and must wait. The AP then sends data to the client typically as a transmit opportunity (TXOP) burst where only the first frame has the EDCF access delay. All subsequent frames are then sent directly after the acknowledgment frame. In a Real-Time Traffic over WLAN implementation, only one frame is queued at the AP, and the real-time-capable WLAN client becomes idle after receiving that frame from the AP.
The U-APSD approach overcomes both the disadvantages of the previous scheme, thus making it efficient. The timing of the polling is controlled through the client traffic, which in the case of voice and video is symmetric. If the client is sending a frame every 20 ms, it waits to receive a frame at each 20 ms time interval. This introduces a maximum jitter of 20 ms, rather than n * 100 ms jitter.
The following figure shows an example frame exchange for the standard 802.11 power-save delivery process.
Figure 8. Standard Client Power-Save
The client in power-save mode first detects that there is data waiting for it at the AP from the traffic indicator map (TIM) in the AP beacon. The client must power-save poll (PS-Poll) the AP to retrieve that data. If the data that is sent to the client requires more than one frame to be sent, the AP indicates this in the sent data frame. This process requires the client to continue sending power-save polls to the AP until all the buffered data is retrieved by the client.
The standard client power-save has two disadvantages.
It is inefficient for the PS polls and the normal data exchange to go through the standard access delays associated with distributed coordination function (DCF).
Retrieving the buffered data is dependent on the DTIM, which is an integer multiple of the beacon interval. Standard beacon intervals are 100 ms. This introduces a level of jitter that is unacceptable for voice and video calls, and voice and video capable wireless endpoints handsets switch from power-save mode to full transmit and receive operation when a call is in progress.
This standard client power-save mode gives acceptable voice and video quality but reduces battery life. The Cisco Unified Wireless IP Phones address this issue by providing a PS-Poll feature that allows the phone to generate PS-Poll requests without waiting for a beacon TIM. This allows the device to poll for frames when it has sent a frame, and then go back to power-save mode. This feature does not provide the same efficiency as U-APSD, but improves battery life for Cisco Unified Wireless IP Phones on WLANs without U-APSD.
Traffic Specification Admission Control
Traffic Specification (TSPEC) allows an 802.11e client to signal its traffic requirements to the AP. In the 802.11e media access control (MAC) definition, the following two mechanisms provide prioritized access, both provided by the transmit opportunity (TXOP):
Contention-based EDCF option
Controlled access option
With the TSPEC features, a client can specify its traffic characteristics, which automatically results in the use of controlled access mechanism. The controlled access mechanism enables the client to grant a specific TXOP to match the TSPEC request. However, the reverse mechanism is also possible; that is, a TSPEC request can be used to control the use of various ACs in EDCF. In a TSPEC mechanism, a client must send the TSPEC request before it sends any priority-type traffic.
For example, a WLAN client device that requires to use the voice AC must first make a request for use of that AC. You can configure the use of voice and video ACs by TSPEC requests but the use of best effort and background ACs can happen without TSPEC requests.
The use of EDCF ACs, rather than the 802.11e hybrid coordinated channel access (HCCA), to meet TSPEC requests is possible because the traffic parameters are simple to allow them to be met by allocating capacity, rather than creating a specific TXOP to meet the application requirements.
Add traffic stream
The Add traffic stream (ADDTS) function is how a WLAN client performs an admission request to an AP. Signaling its TSPEC request to the AP, the admission request can be in two forms:
ADDTS action frame: Used when a voice or video call is originated or terminated by a client associated to the AP. The ADDTS contains TSPEC and might contain a traffic stream rate set (TSRS) information element (IE) (Cisco Compatible Extensions Version 4 clients).
Re-association message: Uses the re-association message when the re-association message contains one or more TSPEC and one TSRS IE if an STA roams to another AP.
The TSPEC element in ADDTS describes the traffic request. Apart from data rates and frame sizes, the TSPEC element also tells the AP the minimum physical rate that the client device will use. This helps to determine the time that the station consumes to send and receive in this TSPEC, therefore allowing the AP to calculate whether it has the resources to meet the TSPEC. The WLAN client (VoIP handsets) uses TSPEC admission control during a call initiation and roaming request. While the WLAN client is roaming, the TSPEC request is appended to the reassociation request.