Report: Analytics Key to QoE for Complex Wireless Networks

Posted by Mae Kowalke on Tuesday, November 28, 2017 with No comments

Traditional wireless networks are not especially ‘smart’ or efficient, mostly serving to convey as much data as possible, without regard to importance of the service or app that data is tied to, noted Senza Fili in a recent report on analytics for big data and network complexity. But these networks and the traffic they carry are becoming more complex, so they must also become smarter and more efficient. Such a transformation is possible with analytics. 

“Network architectures continue to evolve, with the addition of Wi-Fi access, small cells and DAS, C-RAN, unlicensed access, carrier aggregation, VoLTE, virtualization, edge computing, network slicing, and eventually 5G. Managing networks that grow in size and complexity becomes difficult because there is a need to integrate new elements and technologies into the existing network in order to benefit from the technological advances,” explained Monica Paolini, founder and president of Senza Fili, and the report’s author in collaboration with RCR Wireless.

The solution is putting predictive analytics to work optimizing these networks, using automation paired with machine learning and artificial intelligence to extract and correlate valuable information from many data sources, generating insightful advice or predictions.

“The value analytics brings to optimization comes from expanding the range of data sources and taking a customer-centric, QoE-based approach to optimizing end-to-end network performance,” Paolini concluded. This give operators the ability to decide “which aspects of the QoE they want to give priority to, and surgically manage resources to do so,” rather than limiting optimization to throughput and selected KPIs like latency or dropped calls.

Focus on QoE
That ability to fine-tune traffic management is very valuable to operators, who are necessarily shifting to a quality of experience (QoE)-based model despite supply exceeding capacity in an environment with limited resources.

While operators may not be able to realistically give all users everything they want, Paolini said, they can still greatly improve the user experience using resources available, in a way that’s more fair and better aligned with what subscribers value most--for example, the quality of video calls taking higher priority than the ability to watch videos on YouTube and Netflix.

“Lowering latency across the board may be less effective in raising QoE than lowering it specifically for the applications that require low latency,” she explained. “The average latency may be the same in both cases, but the impact on QoE is different.”

Other advantages of this approach include the ability to:
  • Avoid over-provisioning parts of the network
  • Decide which KPIs carry the most weight for improving QoE
  • Determine the best way to allocate virtualized resources
  • Find root causes of network anomalies that result in QoE issues
  • Manage security threats
Toward Predictive Analytics
“The ultimate goal of analytics is to become able to predict the imminent emergence of an issue before it causes any disruption to the network,” Paolini stressed. Machine learning and artificial intelligence will make that possible, eventually.

For now, fitting analytics to each operator’s specific requirements involves making tradeoffs, most notably involving time (for how long, and at what time increments, data is collected) and depth (how macro or micro the data is).

“As operators move toward real time and closer to the subscriber, the volume of data that analytics tools have to crunch grows quickly, increasing the processing requirements, and hence the effort and cost,” Paolini pointed out. “But the reward is a more effective optimization.”

It’s more effective because it’s more targeted.

“Congestion or performance/coverage issues are likely to emerge at different places and times, but only in a small portion of the network…” and therefore “optimization has to selectively target these locations and not the entire network. And the lower the time resolution and the more precise the geolocation information, the more powerful the optimization can be,” Paolini concluded.

Adopting Analytics - Drivers
Operators are driven by several factors to adopt customer experience-focused analytics:
  • Cost and Services - Subscribers are more demanding and less willing to spend more. 
  • Usage - Subscribers use wireless networks more, and in new ways, resulting in a richer set of requirements.
  • Technology - 4G now and 5G in future benefit from more extensive and intensive use of analytics. 
A Cultural Shift
For operators, expanding the use of analytics is appealing but not without its challenges. The greatest of those “is likely to come from the cultural shift that analytics requires within the organization,” Paolini said in the report. “The combination of real-time operations and automation within an expanded analytics framework causes a loss of direct control over the network – the type of control that operators still have by manually optimizing the network. Giving up that level of control is necessary because the complexity of networks makes automation unavoidable.”

Yet, still, operators are increasingly committing to analytics because the benefits outweigh the challenges, enabling them to:

Improve support for existing services
  • Create new services
  • Customize service offerings
  • Optimize QoE for specific services and applications
  • Understand better what subscribers do, individually and within market segments
  • Implement network utilization and service management strategies that set them apart from competitors
Put another way, end-to-end network efficiency and service provisioning enabled by analytics result in significant financial benefits for an operator, by delivering:
  • Increased utilization of network resources
  • Lower per-valuable-bit cost
  • Lower operational costs
  • Better planning
  • Network slicing and edge computing
  • Better customer service and product offerings
  • Third-party revenues
Categories: , , ,