6.4.e [ii] Voice optimization
Voice packets traveling through an IP network are no different from data packets. In the plain old telephone system (POTS), voice traffic travels over circuit-switched networks with predetermined paths and each phone call is given a dedicated connection for the duration of the call. Voice traffic using POTS has no resource contention issues, but voice traffic over an IP network has to contend with factors such as delay, jitter, and packet loss, which can affect the quality of the phone call.
Delay (also referred as latency ) for voice packets is defined as the delay between when the packet was sent from the source device and when it arrived at a destination device. Delay can be measured as one -way delay or round-trip delay. The largest contributor to latency is caused by network transmission delay. Round-trip delay affects the dynamics of conversation and is used in Mean Opinion Score (MOS) calculations. One-way delay is used for diagnosing network problems . A caller may notice a delay of 200 milliseconds and try to speak just as the other person is replying because of packet delay. The telephone industry standard specified in ITU-T G. 114 recommends the maximum desired one-way delay be no more than 150 milliseconds. Beyond a one-way delay of 150 milliseconds , voice quality is affected. With a round-trip delay of 300 milliseconds or more, users may experience annoying talk-over effects.
Jitter means inter-packet delay variance. When multiple packets are sent consecutively from source to destination, for example , 10 ms apart, and if the network is behaving ideally , the destination should be receiving them 10 ms apart. But if there are delays in the network (like queuing, arriving through alternate routes , and so on) the arrival delay between packets might be greater than or less than 10 ms. Using this example, a positive jitter value indicates that the packets arrived more than 10 ms apart. If the packets arrive 12 ms apart, then positive jitter is 2 ms; if the packets arrive 8 ms apart, then negative jitter is 2 ms. For delay-sensitive networks like VoIP, positive jitter values are undesirable, and a jitter value of 0 is ideal.
Packet loss can occur due an interface failing, a packet being routed to the wrong destination, or congestion in the network. Packet loss for voice traffic leads to the degradation of service in which a caller hears the voice sound with breaks. Although average packet loss is low, voice quality may be affected by a short series of lost packets.
Mean Opinion Score (MOS)
With all the factors affecting voice quality, many people ask how voice quality can be measured. Standards bodies like the ITU have derived two important recommendations: P. 800 (MOS) and P. 861 (Perceptual Speech Quality Measurement [PSQM]). P. 800 is concerned with defining a method to derive a Mean Opinion Score of voice quality. MOS scores range between 1 representing the worst voice quality, and 5 representing the best voice quality. A MOS of 4 is considered “toll-quality” voice.
PfR voice traffic optimization provides support for outbound optimization of voice traffic on the basis of the voice performance metrics, delay, packet loss, jitter, and MOS. Delay, packet loss, jitter and MOS are important quantitative quality metrics for voice traffic, and these voice metrics are measured using PfR active probes. The IP SLA jitter probe is integrated with PfR to measure jitter (source to destination) and the MOS score in addition to measuring delay and packet loss. The jitter probe requires a responder on the remote side just like the UDP Echo probe. Integration of the IP SLA jitter probe type in PfR enhances the ability of PfR to optimize voice traffic. PfR policies can be configured to set the threshold and priority values for the voice performance metrics: delay, packet loss, jitter, and MOS.
Configuring a PfR policy to measure jitter involves configuring only the threshold value and not relative changes (used by other PfR features) because for voice traffic, relative jitter changes have no meaning. For example, jitter changes from 5 milliseconds to 25 milliseconds are just as bad in terms of voice quality as jitter changes from 15 milliseconds to 25 milliseconds. If the short-term average (measuring the last 5 probes) jitter is higher than the jitter threshold, the prefix is considered out-of-policy due to jitter. PfR then probes all exits, and the exit with the least jitter is selected as the best exit. MOS policy works in a different way. There is no meaning to average MOS values, but there is meaning to the number of times that the MOS value is below the MOS threshold. For example, if the MOS threshold is set to 3.85 and if 3 out of 10 MOS measurements are below the 3.85 MOS threshold, the MOS-low-count is 30 percent. In the output of the show commands the field, ActPMOS, shows the number of actively monitored MOS packets with a percentage below threshold. If some of the MOS measurements are only slightly below the threshold, with percentage rounding, an ActPMOS value of zero may be displayed. When PfR runs a policy configured to measure MOS, both the MOS threshold value and the MOS-low-count percentage are considered. A prefix is considered out-of-policy if the short term (average over the last 5 probes ) MOS-low-count percentage is greater than the configured MOS-low-count percentage . PfR then probes all exits, and the exit with the highest MOS value is selected as the best exit.
Adam, Paul (2014-07-12). All-in-One CCIE V5 Written Exam Guide (Kindle Locations 6310-6315). . Kindle Edition.