In PRTG, QoS sensors have a channel that shows a so-called MOS value. What does this number mean and how does PRTG calculate it? Which factors are considered to get this score?
This article applies as of PRTG 22
MOS calculation in PRTG
MOS (Mean Opinion Score) is an established metric to obtain the quality of VoIP. PRTG uses this value to indicate an overall benchmark for Quality of Service (QoS) in your network.
The MOS channel is available in both the QoS (Quality of Service) Round Trip sensor and the Cisco IP SLA sensor. For the QoS sensor, PRTG must calculate MOS itself, while the Cisco IP SLA sensor monitors this value directly on the device.
However, MOS is not an absolute metric, and many parameters influence this value. PRTG does its own measurements via the QoS Round Trip sensor and calculates the MOS score based on these parameters. For this purpose, PRTG considers packet loss, the average latency, and jitter.
Furthermore, an R-value that expresses the subjective quality of a connection is needed. This R-value is the standard for MOS calculation and ranges from 0 to 100, where higher means better. The R-value is defined for MOS calculation in advance.
PRTG starts with 93, the number that indicates maximum user satisfaction and equals the MOS score 4.4. Based on the network conditions (that means packet loss, latency, and jitter), PRTG deducts from there to map this R-value to an MOS score.
Mapping R-value to MOS score
- Latency and jitter are summed up and a defined value for computation time is added (for example, 10ms). For the estimation of MOS, PRTG increases the impact of latency by a certain factor (for example, doubled). The resulting number is called effective latency.
effectiveLatency = latency + jitter * latencyImpact + compTime
- The next step is to subtract the effective latency from the defined R. The higher the effective latency, the lower R gets. Also, R gets a much more aggressive deduction if the latency exceeds a certain amount of time. The lower the latency, the more is deducted from the effective latency, and the higher stays R.
R = 93 – (effectiveLatency / factorLatencyBased)
- Afterward, PRTG deducts the percentage of packet loss from R. Depending on what impact packet loss should have, it is multiplied with a certain factor, for example, 2.5 times packet loss.
R = R – (lostPackets * impact)
- Finally, the reduced R is converted into an MOS value by applying a widely-used formula for this purpose.
MOS = ( (R - 60) * (100 – R) * 0.000007R) + 0.035R + 1)
I have two questions from your good-self i.e.
1). Can I use the above PRTG based R-value and MOS Calculation formulas for evaluating the Network MOS of any network protocol...???
2). Also please can you explain 'latencyImpact', 'factorLatencyBased' and 'impact' in the above mentioned formula #1, #2 and #3 respectively? If I want to use them, then, how can I assign them the appropriate values for measuring QoE of a system???
Thank you in advance.
I notified our QoS developer of your question, note that it might take a while until he has time to answer your question, depending on his schedule.
Hi rehan, 1.) It depends .... The MOS value and the specified formula is targeting VoIP Traffic. The different factors count different for other protocols. So if the protocol you have in mind has the same properties like VoIP Traffic (stateless connection, no resending, etc.) then you can use such a formula.
2.) These are constants we evaluated. The value of them influences the curve used to calculate the MOS and their concrete values are internal. There some sources on the internet which share sample Values