Test parameters for VoWLAN.
VoIP IP traffic is split into control plane and data plane traffic. Control plane traffic is characterized by the VoIP protocol that the VoIP application or WLAN IP phone is using, such as session initiation protocol, H.323, media gateway control protocol or Cisco's SCCP (Skinny) protocol. These protocols execute the "handshaking" between the VoIP client and, for example, the media gateway/IP PBX.
For the vast majority of enterprise networks, stress testing data plane traffic to determine voice quality and related measurements, such as jitter and latency, is usually sufficient, since the initial session establishment and taardown only generate a minimal amount of traffic on the network. Also, few bidirectional voice conversations have a 100% activity rate. Most conversations are characterized by a series of activity and inactivity conditions, (i.e., speech and silence), hence VoWLAN test scenarios should maintain this authenticity.
Quality of service (QoS) is most useful in VoIP deployments to help achieve a consistent level of performance. Typical QoS techniques used in enterprise networks include DiffServ, GqoS and IP TOS.
From a performance-measurement perspective, creating test scenarios that generate VoIP streams both with and without QoS will help IT managers to:
* understand the need to implement QoS throughout the whole network prior to VoWLAN rollout;
* measure the impact VoIP with QoS will have on other mission-critical application traffic running across the 802.11 network; and
* decide on a potential upgrade from lower-speed 802.11h to 802.11a or 802.11g networks.
The 802.11 standard that the VoWLAN phone or station is using can greatly influence the coverage area, and therefore the available bandwidth for VoWLAN. Even though most VoIP codecs rely on efficient compression mechanisms that require relatively little bandwidth, any coverage test should always assume a worst-case scenario with regard to the number of concurrent calls, as well as standard distance attenuation tests.
Moreover, a range of WLAN switch solutions include dynamic RF functionality that automatically provides dynamic transmit power control to reduce interference and congestion with neighboring radios, thereby allowing a device to always he connected at the best data rate possible. To accommodate this new functionality, IT managers should also consider VoWLAN test cases with varying access point (AP) transmit power levels.
Advanced security mechanisms such as Wi-Fi protected access (WPA) can introduce significant delay and packet loss, especially when roaming between APs. These performance drawbacks are caused by the WPA requirement to rotate encryption keys, and can create unacceptable delay in the VoIP stream when re-establishing sessions while the WLAN stations and/or VoIP handsets roam across the 802.11 network.
In contrast, using less-secure encryption mechanisms such as the static wireless equivalency protocol (WEP) generally creates results well within the acceptable range for critical VoIP performance measurements, such as jitter, delay and consecutive lost datagrams. If WEP alone is deter mined to be insecure for VoIP traffic. IT managers also have the choice of simply using MAC address authorization to create access right policies for the VoWLAN stations. VoWLAN tests should therefore include performance measurements for all security mechanisms currently deployed in the specific 802.11 network.
There are essentially two types of 802.11 stations that need to he considered for VoWLAN testing. The first is a range of dedicated VoWLAN phones that usually operates on the 802.11b standard. The second are general-purpose WLAN stations, such as laptops or PDAs that may run VoIP applications locally. In the latter case, standard performance end points (or lightweight software-based traffic generators) are available that can be loaded on to the station to generate the required VoIP test traffic.
Running these types of software agents on dedicated VoWLAN phones is usually not possible due to the typical proprietary nature of the operating systems running on these devices. From a traffic-generation and performance-measurement perspective, however, the only area where this may cause issues is if proprietary voice-priority protocols are being tested. For the majority of WLAN enterprise networks, emulating VoIP traffic with standard PC/PDA platforms can he regarded as sufficient.
There are a number of recognized VoIP quality-measurement approaches that also apply to the 802.11 space, including perceptual speech quality measure; perceptual evaluation of speech quality; perceptual analysis measurement system; and the E-model.
All of these algorithms map their scores to a mean opinion score (MOS), which identifies how two people perceive the quality of their conversation. The E-model is often regarded as the most suitable way to assess call quality on an IP network in an enterprise, due to its focus on network issues such as delay, jitter and datagram loss. In addition to M0S scores of 4 and above, jitter values of less than 10 ms, end-to-end delay of less than 50 ms, and a single digit number of consecutive lost datagrams can be regarded as benchmarks for high voice quality in WLAN networks.
For more information from Ixia: www.rsleads.com/408cn-258
This article was provided by Chris Buerger, a product manager at Ixia, Calabasas, Calif.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Voice Networks|
|Date:||Aug 1, 2004|
|Previous Article:||Prepare your WLAN for voice: new standards will address quality-of-service and management requirements.|
|Next Article:||Secure your mobile road warriors: VPN choices or managed security providers are options to ensure network integrity.|