It’s not about the network – it’s about the service users are experiencing

How many times have you phoned a company or colleague only to find that you cannot properly hear the person on the other end of the line, how many times have you tried to have a video conference to find that the picture quality is so bad, you revert to an audio only call?

Business users often complain of call distortion, echo, and noise when they experience voice quality problems with IP telephony systems, or picture blocking, distortion or loss of sync with video calls.  This is despite measures that are taken pre and post deployment to test the networks ability to carry this real time communications traffic. The issue is lack of visibility, the majority of IP telephony and network management software vendors (there are very few video conference management tools today) offer Quality of Service (QoS) based management and monitoring as their solution to manage real time VoIP, but in reality QoS based tools don’t cut it. For real time visibility of what the users actually experience and for operations to manage and diagnose these users call experience a totally new approach is needed.

With UC very much a user oriented set of tools and processes, traditional operating procedures and tools that focus inwardly on the network, devices, bandwidth, and traditional network ‘Quality of Service’ (QoS) measurements are just not up to the job of enabling the IT support organisation or provider to cope with the users needs.

Now, QoS must be implemented and measured, but QoS measures do what they say on their tin; control the way network traffic is carried and prioritized, enabling applications to work as effectively as possible when competing for network resources. Unfortunately, QoS may give the green light in terms of carrying and prioritizing different streams of network traffic, but this alone is not enough to guarantee users experience for real time communications applications such as voice and video. 

To put this in perspective, think about the process of obtaining a new car, lets compare this to the IP QoS mechanism. You place your order for the car, and agree to collect it at a certain time. The car manufacturer manages the production line and your car arrives at the dealer on time as agreed. However would you really be happy with your new car had it arrived looking exactly as expected but when you got in it – there was no steering wheel, and the gearbox was from a different model.

Applying this analogy to an IP network, QoS and QoS based management tools can make sure all voice and video data being carried across the network arrives at its designated end point (suitably prioritised because they are real time applications) and then report the fact that they did their job – effectively turning the management tools ‘green light’ on. What QoS cannot do, however, is ensure that the actual voice conversation, or the video signal carried within the data stream, was intelligible and undamaged, meaning that it cannot give any real indication of whether the end user experience was acceptable.

This is where the rubber hits the road with IP voice and video operations and support. Whereas existing network and VoIP management tools do their job of managing network performance and availability, and typically report an ‘IP MOS’ (mean opinion score) as their indicator of voice performance, these measures bear no relation to the actual users call experience, leaving operations and support effectively blind to visualising and diagnosing problems related to call quality.

The ability to measure actual user experience termed ‘Quality of Experience’ (QoE) – as well as network related QoS is fundamental to a consistent and acceptable IP telephony and video conferencing user experience, as well as enabling efficient (and profitable if you are a service provider) operations and support. It is the combination of QoS and QoE monitoring and reporting together that makes for a good IP voice and video applications performance management strategy.

If we look at the difference between QoS and QoE we can see why they are both needed and where they are significantly different: QoS enables you to provide better service to particular traffic flows by either raising the priority of a flow, or limiting the priority or throughput of other flows. When using QoS based congestion-management tools, they will raise the priority of the more ‘important’ flows using queuing mechanisms and servicing these queues with different priorities and traffic management mechanisms. Traffic policing and shaping effectively prioritise a flow by limiting the throughput of other flows, reducing the impact of jitter, packet loss and latency on the prioritised flow.

Now, QoE is significantly different and complementary to QoS. QoE is a purely subjective measure based on the user’s perspective of the overall value or quality of the service used. It is influenced by the type of device the user is using, and is typically represented by a different set of MOS scores, that accurately reflect the user experience.

Whereas QoE relies on the network’s QoS capabilities, it additionally adds a significantly increased degree of visibility. QoE monitoring does look at QoS characteristics, but on a per stream basis – e.g. for every phone call, and in both directions of that call simultaneously (network based QoS typically looks at the flows by traffic type and so will report all voice traffic as one flow, which is not much use if you are trying to diagnose problems with calls from a single user or location across a busy circuit). 

If we apply our car analogy here again, QoS ensures that the car ordered is delivered to the right place (the dealer) and on time, even when the manufacturers production line and the shipping carrier is busy. QoE on the other hand ensures that the customer gets what they ordered with the right fittings and accessories, all working, clean and shiny.

QoE monitoring, to be accurate, should also take into account any capabilities that the end device may have to mask network based problems from the user.  For example, if jitter is detected on a particular call, this may have an impact on the quality of the voice in the call.  But if the end device (handset, softclient, or video conferencing station) is recognised as having advanced jitter handling built into it, then there is no need to raise an alarm because this instance of jitter has not been able to impact the user. Using our car analogy again, if the car dealer knows that one of his new car customers is tolerant of a delivery delay he can push that car to the back of the pre-delivery workshop and concentrate on those customers that he knows will be impatient and get their card prepped first. Using this process all his customers remain happy even though there was a slight delivery problem.

Lastly, QoE based monitoring also looks at how the signal carried by the network is performing (the voice analogue signal carried inside the IP packets). This provides the ability to detect in real time whether the voice conversation has been corrupted by background noise, or has become distorted, or indeed whether there are other artefacts introduced such as echo that may contribute to a poor call experience. These QoE measurement techniques were pioneered by Psytechnics, and are now defined by the International Telecommunications Union and recognised throughout the industry.

So, user QoE monitoring provides more granular visibility for each and every call, and additionally QoE monitoring looks at the quality of the payload or voice signals being carried by the IP network to detect and report any factors that are known to impact users’ perception of quality. This additional level of visibility allows operations and support teams to not only work much more effectively, being able to fix problems right-first-time, e.g. by using the right resource at the right location, but also allows them to work proactively, detecting problems that affect call quality before users start to raise incident or problem requests.

Whilst the widespread use of IP telephony and eventually UC means that businesses can take advantage of advanced communications features and with application integration increasing productivity as well as reducing capital and call costs, unacceptable call quality can mean the loss of business, which can ultimately result in a substantial loss of customer confidence and brand reliability for the company as a whole.

The importance of call quality and user confidence cannot be underestimated. As illustrated above, customer-service focused companies can win and lose customers, revenue and staff through frustration with call quality. By taking a purely “network centric” view of a UC deployment and on going operations, using existing or traditional network and IP telephony management tools that focus on IP QoS and network and device availability, there is no ability to focus on the user or customer experience, and support and operational costs go unchecked. As UC, IP telephony and IP video conferencing roll out across organisations on a larger scale, it is the proactive monitoring of the system at both the network and the application level, i.e. the user’s perception of quality that becomes the key factor in controlling operating costs, and maintaining a consistent and acceptable user experience.

Stop trying to fix the network when it’s the service users are experiencing that’s broken!