Its opening night for a new play. The crowd settles in its seats. The orchestra tunes in the pit. Finally in costume and makeup, each actor knows his lines. However, theres a catch: the actors have rehearsed separately, not together. Not one actor has yet heard the others speak their lines.
One can bet on miscues, confusion, dropped lines, and a weak performance at best. The audience will snicker, then remember the ticket price and walk out ready to spread the bad word. The critics will not be kind.
Testing the readiness of a group of actors separately rather than together may seem the height of foolhardy behavior. To truly view, troubleshoot, and refine what the audience will experience, requires viewing and listening to all the actors, lighting, props, and the rest operating together.
Yet a similar scenario awaits any Internet telephony provider who fails to analyze end-to-end performance of VoIP networks. Testing each piece of network equipment in a lab, while necessary, will reveal little about how that equipment will interoperate with the rest of the network in delivering a quality VoIP performance when the curtains are raised to a paying audience.
To ensure a flawless opening night, VoIP providers must supplement their lab testing and conduct pre-deployment VoIP readiness assessment on the actual network. Paying subscribers have proved themselves willing to accept some quality tradeoffs in exchange for valuable benefits such as mobility. However, for lifeline voice and premium IP video and multimedia services, providers should expect little subscriber tolerance for quality that fails to meet or exceed expectations set up by years of user experience.
In short, voice and video make measuring quality of experience (QoE) more critical then ever.
Distributing Testing For Distributed Networks
Assessment of network readiness during pre-deployment should be conducted under a variety of scenarios, evaluating each network link under stress testing not only of peak capacity, but also utilization curves over time. Common VoIP scenarios also will include interconnections of on-net VoIP with the public switched telephone network (PSTN). Consequently, testing should cover not only IP, but also hybrid IP/PSTN links and protocol conversions.
Readiness assessment also requires distributed, link-by-link testing, if it is to reflect the end users actual experience of VoIP and IP multimedia applications. For example, if the operator wishes to validate VoIP readiness in a four-city network, then tests must be applied to each city and every link. Testing each link produces an end-to-end picture of QoE, protocol conformance, and other factors from City A to Cities B, C and D; from City B to Cities A, C and D; and so on.
Through distributed testing the operator can then isolate specific problems. For example, the cause of dropped packets in the B to C link might be traced to a concurrent rush of multicast IP video for training sent across the entire network. Indeed, the mix of voice, video, and data at any given time will affect performance of real time applications making it imperative that validation testing takes this larger picture into account.
Distributed test systems that place test platforms across the network can yield substantial cost reductions and time savings, compared to moving test gear and repeating tests in serial fashion at each location. The distributed approach also yields integrated reporting and analysis that covers the entire network, rather than requiring rationalization of multiple reports.
Real Time QoE: A Multidimensional Challenge
Network readiness assessment for real time voice and video applications requires a wide range of testing capabilities. Among requirements that test system suppliers should meet are the abilities to:
Emulate real-world traffic from multiple points on the network;
Test the full range of IP communications and collaboration applications likely to travel the network; and
Test the full range of network statistics and media quality, including packet loss, jitter, delay, voice and video MOS, and echo.
Key call statistics will include such metrics as call setup time, percent of call completions and call termination reason. Key QoE analysis may be derived from metrics such as Mean Opinion Score (MOS) and Perceptual Evaluation of Speech Quality (PESQ).
The QoE issues examined below represent some of the most common problems encountered in readiness assessment. These issues are best tested and evaluated during pre-deployment using a distributed network assessment solution.
Data Packet Loss
Dropped packets that produce no hiccups for non-real-time applications like e-mail and file transfers can create unacceptable degradation of voice and video traffic quality.
That degradation may appear as garbled voices on a phone call to graininess and dropouts in a video signal. Enough packets lost, and entire spoken phrases or frames of video will not reach their destination. Severe packet loss can cause a break in the entire voice or video session.
Testing of packet loss must take into account not only the voice and video bearer traffic, but also signaling information (such as Session Initiation Protocol, or SIP, and H.323 signaling) required to set up and tear down connections.
TCP/IP routing technologies such as Differentiated Services (DiffServ) and IP Precedence packet markings are now being used to instruct routers and gateways to give real-time applications priority treatment in a congested network.
Latency, or delay, causes two problems echo and talker overlap. Echo is caused by the signal reflections of the speakers voice. Echo becomes a significant problem when delay is greater than 50 milliseconds. Talker overlap becomes significant if one-way delay is greater than 250 milliseconds.
Exceeding these acceptable delay/ latency levels can make for a voice call in which parties are forced to pause, waiting for the other party to hear what theyve said. Video impairments can include frozen frames, tiled images or, in videoconferencing, sound out-of-sync with lip movements. These problems can be solved by QoS mechanisms that monitor and buffer packets as they travel across the network, ensuring that they are transmitted with acceptable amounts of delay and in the proper order.
Because delay can be caused by constantly changing factors like traffic volume, testing should be conducted under varying traffic scenarios to accurately determine maximum latency that may occur, and to properly configure network capacity and QoS solutions to accommodate that level.
In a hybrid network, analog phones can be the source of echo, but an IP telephony configuration is also susceptible, since some calls will inevitably interface with TDM network elements at some point when they go off net. Factors such as the volume and delay of the echo and the impact of hardware, such as handsets and routers, can contribute to the echo level.
Echo cancellation and suppression processors, generally used in media gateways, audio coder/decoders and other equipment, are the most common fixes for unacceptable echo. Consequently, providers must test both echo itself and the performance of echo cancellers.
Voice And Video Jitter
The human ear can detect too much variation in delay (jitter) as inconsistency in quality or simply something unnatural about the voice. Jitter is caused by actions like queuing and routing that affect the path of packets through the network. Network congestion generally produces higher levels of jitter, but quality of service controls like queuing and bandwidth allocation can control the problem.
Video tolerances for jitter are even lower, given the demand for smooth motion in film, sports or other TV applications. While great advances in video compression continue to squeeze more and more video information into less bandwidth, compression does not address issues that generate delay or jitter, such as congested router ports or malfunctioning buffers at any given point the end-to-end path. Most QoS solutions also include a jitter buffer, which adds small amounts of delay to packets received so they all appear to have equal and acceptable amounts of latency. This also ensures that packets are transmitted in the correct order.
Testing of the performance of these QoS mechanisms, as well as testing of application quality from the end-users point of view (i.e., from the origination and destination end-points), is essential to determining how much jitter a network will experience and how it should be handled.
Clipping within VoIP calls occurs when either the beginning or end of words, or whole words, seem to be cut off during a conversation. This can occur when voice activity detectors and other solutions that work in tandem with echo cancellers are thrown out of sync.
Echo cancellers deal with background noise and double talk in addition to echo. For instance, an absence of background noise during a call can confuse users into thinking a call has been dropped. Yet improper levels of background noise can result in annoying voice clipping.
Winning The Customer On Opening Night
By addressing each of the QoS issues detailed above, service providers can substantially increase their chances to win the hearts and minds of paying subscribers. Failing to resolve these issues effectively prior to service turn up will result in subscriber unrest and revolt.
Network operators who undertake distributed pre-deployment testing on VoIP services stand the best chance to pull off VoIP opening day to rousing reviews.
As an added bonus, operators employing these tools to ensure the performance of demanding, mission-critical VoIP services can enjoy maximum confidence that their networks will support high quality delivery of all other triple play real time and non-real-time applications for some time to come. IT
Bahaa Moukadam is vice president, IP Telephony at Spirent Communications, a provider of integrated performance analysis and service assurance systems that enable the development and deployment of next-generation networking technology. For more information, please visit www.spirentcom.com.
If you are interested in purchasing reprints of this article (in either print or PDF format), please visit Reprint Management Services online at www.reprintbuyer.com or contact a representative via e-mail at [email protected] or by phone at 800-290-5460.