Steve Wong, vp of marketing at ClearSight Networks, explains to VoIP.biz-news.com some simple steps for remedying poor VoIP quality once it has been uncovered in a network.
As Voice over Internet Protocol has become a very common and inexpensive way to provide voice communication, it has expanded to include more forms of streaming audio and video.
However, there are many factors that can affect the quality of such a transmission, since it often has to compete for bandwidth over diverse networks that it shares with other traffic.
It is important that a network administrator or network provider know what Quality of Service (QoS) can be expected for VoIP communications on a given network.
In particular, it is desirable to have an easy to interpret way to measure that quality.
The Mean Opinion Score (MOS) has been developed to provide such a measure.
The original idea of MOS was developed by the ITU-T using human subjects to subjectively rate the quality of spoken sentences.
|
MOS
|
Quality
|
Impairment
|
|
5
|
Excellent |
Imperceptible |
|
4
|
Good |
Perceptible but not annoying |
|
3
|
Fair |
Slightly annoying |
|
2
|
Poor |
Annoying |
|
1
|
Bad |
Very annoying |
| |
|
TABLE 1.0 |
The result of these experiments was a quality scale of 1 to 5 (see table 1.0)
Factors Affecting Quality
Four of the most common factors that can degrade the quality of a VoIP transmission are:
– The compressor/decompressor (Codec) used
– Network latency
– Jitter
– Dropped packets
The choice of Codec establishes a maximum possible MOS score, irrespective of how well the network is working.
The situation in VoIP is different from other kinds of data compression.
For example, compressing fixed documents or graphics can often be done with lossless compression algorithms, meaning that they save bandwidth while still being able to reproduce the original data exactly.
VoIP compression is generally much more aggressive, and even the best Codecs are quite lossy (see Table 2.0).
|
Codec
|
Data Rate (kbps)
|
Max MOS Value
|
| G.711(ISDN) |
64
|
4.3
|
| iLBC |
15.2
|
4.14
|
| G.729 |
8
|
3.92
|
| G.729a |
8
|
3.7
|
| GSM FR |
12.2
|
3.5
|
| |
|
TABLE 2.0 |
R-Value
Another QoS metric in common use is R-value, which has a range of 1 (worst) to 100 (best).
In general terms it was designed to represent the percentage of users that might find the VoIP quality acceptable.
The relationship between R-value and MOS is not quite linear.
|
R-value
|
MOS Value
|
|
100
|
5.0
|
|
90
|
4.3
|
|
80
|
4.0
|
|
70
|
3.6
|
|
60
|
3.1
|
|
50
|
2.6
|
How Does CSA Calculates MOS and R-Value?
The ClearSight Analyzer (CSA) and the Network Time Machine (NTM) family of products includes an ability to calculate the expected quality of VoIP transmission, based on the Codec used, and on actual observed values of latency, jitter, and packet loss.
Basically it uses the formulas set forth by ITU-T Recommendations G.107 and G.113.
It derives measured statistics from the RTP packets in the audio/video stream, and applies coefficients that can be set by
the CSA or NTM user in an E-model configurati on screen.
The MOS values and R-values are calculated separately for each fl ow, and are displayed in stati sti cs tables and in VoIP reports.
CSA and NTM calculate and report Minimum, Mean, and Maximum values for MOS and R-value.
That way the user not only sees the overall average quality, but also gets an idea as to whether that quality varies much over time.
How to Remedy Poor VoIP Quality?
When CSA or NTM reports consistently low scores for MOS and R-value, there are oft en simple steps that can be taken to
improve things, even without deploying extra bandwidth.
When you have made these changes to your network, simply use CSA or NTM again to see what diff erence it makes to your
MOS and R-values.
So, for example:
– Load balancing: If some parts of the network are performing well and others are not, then it may be possible to re-route some of the traffic to better balance the load on selected network segments.
– Traffic Shaping: Often network providers contractually guarantee a certain level of performance in terms of committed bit rate, maximum burst length, average sustained bit rate, and other quantities characterizing available network resources.
This is known as provisioning a network segment. When traffic exceeds these guaranteed values, it may still be forwarded, but performance may deteriorate.
The process of traffic shaping controls the flow of traffic so that it does not exceed those contractually guaranteed performance
values.
If you suspect that you are overrunning some of these limits, you can check the flow of traffic over such provisioned segments, and change the traffic shaping algorithms to better match your traffic to the network’s capabilities.
– Traffic Prioritization: Situations may arise where VoIP quality deteriorates because of other traffic competing for bandwidth – for example, a large file transfer or the sudden occurrence of Windows updates.
A simple way to fix this is to program your switches and routers so that they use a higher priority for forwarding VoIP packets
than other data packets.
It doesn’t hurt a data packet that is part of a fi le transfer to be delayed for a few extra milliseconds, but delaying VoIP packets can seriously degrade the VoIP QoS.
We’d be interested to hear your feedback on Steve Wong’s advice. Please send us your comments or any questions.
