Why is MOS so easily dismissed?

If the mere mention of MOS (Mean Opinion Score) conjures up some high level sales terminology that’s meant to gloss over the details and provides little impact to troubleshooting VoIP QoS, then don’t worry because you are not alone.  Over the past few years, I’ve noticed a growing trend amongst our technology peers who have similar feelings when discussing MOS as a factor of visibility to VoIP networks. Most anecdotal responses to my queries of dismissal end in “…I can’t fix my network using MOS.” So I ponder the dismissal; Has the audience expertise surpassed the oversimplified value of MOS? Has MOS outgrown it’s value in how we operate our VoIP networks? or is there something more subtle creating the perception of dismissal?

Background

MOS originated from a need to demystify the complex impairments that affect conversations and provide a user perception for voice quality. Measuring analog voice conversations include impairments such as noise, echo, distortion, signal-to-noise ratio, etc… Simplifying the voice quality to a scale between 1 and 5 allowed providers and consumers to identify if an impairment exceeded thresholds that could manifest in an acoustic perceivable impairment.

As network migration towards Voice over IP was introduced, MOS played/plays another critical role in determining if the VoIP Quality of Service is similar or better than the analog voice quality. Certainly VoIP QoS measurements account for packetized impairments such as packet loss, jitter and delays, however the acoustic perception of a MOS=3.5 (for example) voice quality should be similar between analog and VoIP.

Today, we see the proliferation of VoIP increase to a dominant role within the enterprise network and comparison to analog circuits is no longer a primary requirement. Instead, we now prefer access to all the raw impairments when managing the VoIP network and too-often focus on the peaks/valleys of impairments as the key driver in managing VoIP QoS.

As a fellow technology enthusiast, I understand the need to analyze and operate the VoIP network using the variances in the individual impairments, however I also see merit in utilizing MOS scores.

  • When attempting to quantify the overall status of user experiences within the VoIP network, utilizing MOS provides a quick management performance indicator.
  • When determining the expected VoIP QoS between codec variances, MOS once again creates the necessary differentiation to determine the acceptable impairments for each codec utilized.
  • Lastly, MOS helps to establish a baseline of acceptable VoIP QoS within the network and aids to discover any variances when network conditions change such as load or device firmware.

Summary: MOS still has a place within our operational processes of the VoIP network, however we must adapt our usage when maintaining VoIP services. Since we have become more educated in real-world VoIP experiences, the need to simplify the impairments and identify if the PSTN is better/worse than VoIP is not greatly utilized. However, there are times where even I look at call with packet loss >1.3% on a G.711 codec and wonder if the conversation quality was actually impacted. MOS has not outgrown it’s value as a key performance indicator, instead the usage of MOS needs to change to a more macroscopic indicator of the VoIP network instead of the metric we use day-to-day to troubleshoot call quality.