In our previous tutorial, we looked at the concepts of the Mean Opinion Score (MOS) and subjective voice quality testing that is defined in the International Telecommunications Union—Telecommunication Standardization Sector’s (ITU-T’s) P.800 Recommendation titled Methods for Subjective Determination of Transmission Quality (http://www.itu.int/rec/T-REC-P.800/en).
The ITU-T, being an agency of the United Nations, is quite instrumental in developing methods of testing global communication circuits. Much of the work of interest to VoIP net managers falls under the guidance of ITU-T Study Group 12,the lead study group on quality of service and performance. This group is responsible for four key areas of recommendations:
G-Series:Transmission systems and media, digital systems and networks
I-Series:Integrated Services Digital Network
P-Series:Telephone transmission quality, telephone installations, local line networks
Y-series: Global information infrastructure, Internet protocol aspects and next generation networks
Under the G-series we find ITU-T Recommendation G.107, titled <.i>The E-model, a Computational Model for use in Transmission Planning(http://www.itu.int/rec/T-REC-G.107/en). The E-model is an algorithm originally developed in 1998 and updated annually, which is designed as a transmission planning tool, that can evaluate the effects of transmission degradations that come from several sources. The output of the algorithm is a quality rating value, called R, which varies directly with the overall quality of the conversation. The general premise is that the various transmission factors are additive, and that a composite result will more accurately represent the quality of the communication circuit. The transmission rating factor Ris derived as follows:
R = Ro—Is—Id—Ie—eff + A
Rorepresents the signal-to-noise ratio, including the circuit noise and room noise
Isrepresents a combination of impairments which occur simultaneously with the voice signal
Idrepresents the impairments caused by delay
Ie-eff, the effective equipment factor, represents impairments caused by low bit-rate codecs
A, the advantage factor, compensates for impairment factors when access advantages (such as cellular or satellite circuits) are available to the end user.
While the standard acknowledges some limitations of the E-model, nevertheless, it provides a systematic approach that (one hopes) exceeds the “expert, informed guessing” (as the standard says) that goes into many transmission planning exercises.
In addition to the MOS and E-model tests, the ITU-T has also developed some objective tests that were originally devised for codec testing. These include: ITU-T Recommendation P.861, Objective Quality Measurement of Telephone-band (300-3400 Hz) Speech Codecs (http://www.itu.int/rec/T-REC-P.861/en), and the more recent ITU-T Recommendation P.862, Perceptual Evaluation of Speech Quality (PESQ): An Objective Method for End-to-End Speech Quality Assessment of Narrow-Band Telephone Networks and Speech Codecs,published in 2001, and updated in 2007. These methods were originally developed for the lab testing of codecs, and determine the distortion introduced into the system by comparing the original input signal with an impaired signal at the output.
Other recent work includes ITU-T Recommendation P.561, In-service Non-intrusive Measurement Device—Voice Service Measurements, (http://www.itu.int/rec/T-REC-P.561/en), which defines devices that can be used in-line to measure various voice-grade parameters including speech and noise levels, echo loss, and so on.
ITU-T P.562, Analysis and Interpretation of INMD Voice Service Measurements (http://www.itu.int/rec/T-REC-P.562/en) describes methods to analyze the individual measurement parameters over single and multiple calls, and how those measurements can be applied to network planning and operations activities.
A third document, ITU-T Recommendation P.563, Single-ended Method for Objective Speech Quality Assessment in Narrow-band Telephony Applications (http://www.itu.int/rec/T-REC-P.563/en) describes an objective method for predicting the subjective quality of narrow-band telephony applications. According to the standard, this method is recommended for non-intrusive speech quality assessment and live network monitoring, and is able to predict the speech quality on a perception based Mean Opinion Score—Listening Quality Objective (MOS-LQO) scale as defined in ITU-T Recommendation P.800.1, thus providing some degree of correlation between subjective and objective measurements.
Our next tutorial will continue our examination of this real time performance challenge, by looking at the latest network performance metric: Quality of Experience.