Why Quality Attenuation (∆Q) is the key to measuring better broadband

Why Quality Attenuation (∆Q) is the key to measuring better broadband

Lucy Hazell, Packet Flow Performance Analyst, PNSol

For service providers to deliver better broadband at lower costs, and by extension have happier customers when using demanding applications, accurately measuring a network’s performance is imperative to guaranteeing solid service.

Focusing on average network characteristics like latency, jitter, and packet loss in isolation doesn’t consider the complexity caused by next-generation services. Such an approach is unable to capture the relationship between the end-user and the technology: it shows us aspects of technical performance but cannot give the full picture of how functional a network is.

Customer demand for applications like VR headsets and online gaming to work seamlessly together has spawned QoE, or ‘Quality of Experience’, to subjectively measure how the customer experiences the network service. Its objective complement, UX, measures how applications respond to network performance. In conjunction these offer both a qualitative and quantitative insight and can show us the application impact of data sent over the network, not just the efficiency of data transport across the network itself.

A joint Broadband Forum and Omdia report revealed that 72% of surveyed telcos planned to differentiate customer QoE on a per application basis at some point in the future.

A perfect network?

New applications place new demands on networks and require the use of a combination of active and passive network traffic monitoring mixed with automation and real-time analysis.

Combining these principles lets us look beyond conventional averaged measurements to allow more meaningful analysis and improvement of the overall broadband experience. Core to this approach is Quality Attenuation (∆Q).

A perfect network would let the IP packets carrying the end user’s application data – for example during a video conference – across the network with zero delay and zero packet loss. In other words, the customer would experience zero defects. Of course, a perfect network is intrinsically impossible – in real networks, all packets take time to travel (i.e. are delayed), and some may get lost, even at low loads, hence quality is ‘attenuated’ compared to the perfect network outcome. This is the basis for Quality Attenuation measurement and analysis. It lets us statistically characterize the real-world degradation of quality and hence quantify the ‘delta’ from a perfect network, which we denote as ∆Q. This ∆Q is a mappable proxy to UX, and thus to user experience and satisfaction.

The beauty of Quality Attenuation performance analysis is that it can specifically isolate the component and degree of quality degradation due to any inadequate scheduling operation when the network is under load. This essentially means that network operators can optimize broadband performance more cost-effectively via configuration changes rather than just increasing link speeds, which is costly and can be counter-productive.

Broadband Forum’s Quality Experience Delivered (QED) project uses ∆Q methods to deliver greater insight into QoE and application outcomes, by measuring the statistical distribution of delay and loss.

From increasing throughput to improving quality

Another key concept of QoE and the use of ∆Q is the move away from throughput as the key measurement of network performance in the age of gigabit broadband.

In the past, bandwidth quantity and speed has been used as a measure of how good a broadband network is. Bandwidth, however, is an increasingly weak proxy to measure the application outcomes that customers value. For example, when intending to use a connection for small cell/mobile backhaul timing for synchronisation is crucial, and almost completely independent of bandwidth.

Bandwidth or capacity is necessary, but it is not sufficient. The phrase ‘insufficient bandwidth’ really means that the packet loss/delay exceeds the acceptable performance bounds of the application. Quality Attenuation unifies packet delay and loss (capturing them as statistical distributions) which enables it to reflect structural/environmental attenuation as well as that imposed by load on the available capacity.

Mainstream application of ∆Q via the BBF

Measuring ∆Q and understanding its mapping as a proxy for application QoE metrics can give a far better picture of how network performance translates into customer satisfaction. It is providing operators with the tools they need to meet current user requirements.

Led by Vodafone and Predictable Network Solutions (PNSol), Broadband Forum’s Broadband Quality Experience Delivered (Broadband QED) initiative brought together one of  the organization’s oldest members with one of its newest to create “invisible” networks that will greatly enhance the QoE of today’s broadband networks.

For more information about Broadband Forum’s work, read the Marketing Report (MR) 452.4 ‘QED Uses in Lab Evaluation & Network Design’ here https://www.broadband-forum.org/pdfs/mr-452.4-1-0-0.pdf.