There only is one reason for 1st class voice quality monitoring

sprechblasen

QIQO – Why Data Quality is the Base for Good Analytics

Voipfuture’s “Quality In, Quality Out” Principle

Data quality isn’t just nice to have. If you input data that is junk, you will get junk as a result. The outcome depends entirely on the quality of the beginnings.

 

No doubt: We have to take every care to ensure that data will be grouped, aggregated, and analyzed thoroughly. And when nicely presented, results become convincing and believable.

 

On the outcome level, low quality input is often hardly noticeable. Nevertheless, the quality of decisions relying on them suffers considerably.

 

All the intelligence we invest relies on the data quality of the very first chunk. This quality is not about a MOS value. Or about the slimness of time slices.

Of much more importance is the fundamental quality of one single time slice itself.

We deliver hundreds of metrics per slice – voice quality metrics, impairment pattern indicators and so forth. Together they shape a picture of quality:

 

  • Per 5-second time slice
  • Per stream
  • Per call
  • Per network element
  • Per route
  • Per destination
  • Per customer

 

This is how we can contribute to premium voice quality – it simply is premium voice quality monitoring.

Just to point out what we mean by monitoring: all calls in the network, both call directions, 24×7. A voice quality monitoring solution can only be considered good in light of measurement and processing intelligence.

HD Voice – A New Experience and Some Old Truths

Paradise ahead

High-definition (HD) voice delivers a richer, better calling experience. Gone are the days of dull tones and dampened voice. Making a call will become a pleasure.

The problem is, the more (bandwidth) CSPs deliver, the more customers expect – causing quality standards to rise. The different conversion modes (R-factor to MOS value) for narrow band and wide band calls reflect this.

The equation “more bandwidth = less problems” has never proven to be true. All involved parties face many challenges in the period of transformation from 4G to 5G ahead. Voice prioritization will compete against other services like video or IoT.

The truth is, voice quality is by no means a given success. Packet loss and jitter will remain. And High-definition voice remains a task to which we must devote our full attention.

Voice Quality and Mean Opinion Score

Why it makes more sense than ever before: assessing Voice over IP quality by monitoring the Real-Time Transfer Protocol (RTP)

The increasing need for a clear understanding of voice quality

In telephony, voice quality guides whether the experience is a good or bad one. It is the customer’s view on the value of the service provided. The underlying questions are: does voice quality match the users’ expectations, and does it justify the costs of service?

When network resources are congested the quality of Voice over IP (VoIP) traffic will suffer – causing a poor user experience for the subscribers. This results in increased numbers of trouble tickets for the Voice Service Provider. The impact: decreasing revenue due to increasing customer churn. Consequently, the need to monitor Quality of Service (QoS) for VoIP traffic grows.

What are your organization’s most important strategic priorities over the next three years? Customer Experience Management clearly stands out as No 1

82%Customer experience management
50%Cost control and business efficiencies
35%Network upgrades and modernization
68

Operators that view customer experience as their number one strategic priority

Understanding MOS

What is Mean Opinion Score?

 

Put simply, the Mean Opinion Score (MOS) is a way of ranking the voice quality of a call. MOS has been used for decades in telco networks, and traditionally has been a subjective measurement. Listeners sit in a “quiet room” and score call quality on a scale from 1 (worst) to 5 (best). The average or the arithmetic mean of all the individual scores is called MOS.

The days of solely focusing on the control plane are gone. Properly establishing and closing a call remains important. But nowadays, voice travels packet-by-packet through “bursty” IP environments – with congestion, delay and packet loss on the agenda. What is happening on the media plane becomes more and more important.

Consequently, the need has risen to evaluate not only one call by one person, but to automatically measure real-time traffic through entire networks with millions of customers.

In 2007, Voipfuture was the first to introduce a technology to directly monitor the Real-time Transfer Protocol (RTP) which is used to transport voice in IP networks.

Graphic_1b

Why one MOS value for one call is not enough?

 

In the good old network days things were easy: Circuit-switched networks were used for phone calls and packet-switched networks handled data. Once a call was set up one did not need to worry about voice quality.

 

Due to the migration of telecommunications towards IP, voice packets now have to travel on bumpy roads. They have to wait for other packets. In the worst case, packets get lost.

 

The effect: Voice quality isn’t stable during a call anymore. It changes over time. If we want to fully understand speech quality, we can’t check RTP streams once and think the job is done. Voice quality measurements, over time, at different points in the network tell very different stories. Two calls with the same MOS value can show distinct voice quality patterns with a diametrically opposed Quality of Experience.

Easily slice and dice network quality

 

A couple of years ago, Voipfuture established time slicing in voice quality monitoring. Monitoring a network’s full traffic requires us to break the full amount of data down into smaller parts – to better examine and understand it.

For this purpose, every stream of a call is cut into slices. 5 seconds by 5 seconds by 5 seconds. For each of these time slices a highly detailed Quality Data Record (we call it simply QDR) is produced.

These atomic voice quality data records can be clustered – aggregated, correlated or grouped. For example by

 

  • Destination
  • Route
  • eNodeB or eNodeB Group
  • Call Direction
  • Network Segment

 

The nice thing about a highly detailed input is, one can look both downward and upward. Downward to the voice quality of a specific call at a specific monitoring point in the network. Upward to check quality KPI and SLA input.

The latest correlation: QCDR

RTP streams and call signaling data belonging to the same call are merged into one Qrystal Call Data Record (QCDR). A QCDR provides 100% call visibility.

It is an all-in-one representation of the recorded data including:

 

  • Media summary quality
  • Aggregated call quality
  • Root cause analysis
  • Definition of the worst stream in the call
  • Recognition of differences between multiple streams and multiple xDRs of one call
  • De-duplication of streams (if necessary)

This approach brings a couple of highly interesting additional use cases and benefits

 

  • Not just typical records, but analysis of call data
  • Media end-to-end quality summary of each call
  • Only the metrics you need (QCDR visualization adjusted to OSS needs)
  • Settling of billing disputes – calculation of call minutes allows you to have a good overview of your expenses
  • LTE/VoLTE (cell quality and call quality)
  • Reports on the basis of normalized call numbers
Contact Us

captcha

Please type in:

* mandatory

Voipfuture Whitepaper MWC 2017
Will VoLTE really benefit from EVS?
Your Information will never be shared with any third party.