Voice over Wireless LAN Solution Guide v1.0 December 2005
______________________________________________________________________________________________________
Page 38
2.5.1.2 WMM description
To accelerate adoption of 802.11e (even before finalization of the standard) the Wi-Fi Alliance
defined two tiers of QoS capabilities, each being a subset of 802.11e: WMM and Wi-Fi
Multimedia Scheduled Access (WMM-SA). The Wi-Fi Alliance has also started certifying WMM
compatibility.
The primary difference between the two WMM and WMM-SA is that WMM is based on 802.11e
EDCA, which is backwards compatible with the distributed coordination function (DCF) (that is,
the original 802.11 MAC), and WMM-SA is based off of the HCCA portions of 802.11e, which are
not compatible with DCF. WMM-SA can support legacy devices but only by allowing for periods of
legacy MAC operation.
WMM specifies four classes of traffic: Voice, Video, Best Effort, and Background (listed here in
order of priority). Each class has a statistical advantage over lower classes. Under old DCF rules,
all devices had the same transmit opportunity, but with WMM the old DCF is modified to give
statistical advantages to each of four tiers of transmit opportunities, that is, each of the four
classes. The Background class specifies contention window sizes and DCF interframe space
(DIFS) that are identical to DCF; therefore, non-WMM devices fit very nicely into WMM as the
lowest class of device. Specifically, legacy data-only devices do not need to support WMM to fit
into the WMM framework.
There are a few advantages of WMM over SVP. First, WMM allows for several classes of traffic.
Under WMM, video can be prioritized above other data while not sacrificing the supreme priority
of voice. Only with WMM can true multimedia capabilities be supported over WLAN. Second,
WMM allows an abstraction of devices, so that handset voice and non-handset voice can both be
given voice precedence. Third, WMM allows for admission control to be implemented, regardless
of device, on a per-service-class basis. This means that handset voice (SVP) and non-handset
voice can potentially be treated as equals. Lastly, WMM allows for individual devices to have a
mix of voice, video, and data, each of which is given the proper prioritization. As an example, with
WMM, a laptop could potentially have an MCS call, webcast, and e-mail running simultaneously
over the WLAN while providing proper classification to each of the applications.
2.5.1.3 Prioritization capabilities
The WSS 2300 contains the classification and prioritization capabilities; the AP 2330 contains the
queuing and scheduling capabilities. The WSS 2300 uses ACLs to prioritize on a per-packet
basis. An ACL usually has an action to permit or deny packets, but with the WSS 2300, an
additional action can be to set the class of service (CoS) value. This CoS value is used to mark
the type of service (ToS) field of the IP packet as well as determine the transmit queue. After the
ToS field is marked, this value is copied to the outer encapsulating packets between WSS and
AP so that QoS can be maintained on an end-to-end basis.
Because of the use of ACLs, the flexibility exists for a number of VoWLAN deployment scenarios:
1. You can implement a converged voice/data SSID, including voice applications on a
laptop, where each application’s traffic is given separate QoS treatment.
2. You can implement QoS at the device level through an ACL bound to the device, which
marks all traffic with a single CoS value. The binding of ACL to device is controlled by an
AAA server or the WSS 2300 local database.
3. Through Location Policies, you can give an entire voice SSID a specified CoS value. This
is similar in function to the way the WSS 2270 implements QoS.
2.5.1.4 Queuing with WMM disabled—Native SVP mode of operation
The major difference between WMM being enabled and disabled is the queuing behavior in the
AP 2330. That is to say that the classification and prioritization steps remain the same on the