Home Wi-Fi QoE: Post-AI Hype Connected Home Management

Artificial Intelligence (AI) receives a lot of hype these days. The tech industry uses it extensively, in ranges of use from home digital assistants, to autonomous vehicles, to predictive analytics. In our industry, AI can be extremely helpful for improving subscriber Wi-Fi quality-of-experience.

We have been using AI technologies in our broadband product line, Expresse, for many years. Our AI algorithms diagnose broadband issues, automatically optimize performance, and recommend next-best service steps to call center and field technicians. For our CloudCheck product, we have drawn upon these years of AI experience and expertise to manage home Wi-Fi for our service provider customers.

In this post, and its supporting video presentation, we share our AI learnings and cover:

  • The cost of poor Wi-Fi QoE to service providers
  • The role AI plays in-home Wi-Fi QoE
  • The challenges the dynamic Wi-Fi ecosystem makes for managing home Wi-Fi
  • The decision carriers must make regarding their role in managing home Wi-Fi
  • The practical application of AI to improve home Wi-Fi
  • Some requirements for a managed home Wi-Fi solution

The Cost of Poor Wi-Fi QoE

Poor Wi-Fi performance negatively impacts a service provider’s operating expenses. The typical subscriber does not distinguish between Wi-Fi and broadband issues. When subscribers experience poor home Wi-Fi, they call their service provider assuming something is wrong with their internet connection. Often their problem is a Wi-Fi issue.

This table details the cost of service calls. Our customer data, derived from over 100 million customer accounts shows that poor QoE increases the number of these calls by a factor of 4.8.

Support Level Cost Per Ticket
Vendor $471
Field Support $196
Level 3: IT (apps, networking, NOC, etc.) $85
Level 2: Desktop Support $62
Level 1: Service Desk $22

AI and Home Wi-Fi QoE

The key to a good Wi-Fi QoE for home users is a sensible QoE score defined by operational, rather than marketing metrics. The most important factors that deliver a high Wi-Fi QoE are:

  • Picking the channel correctly
  • Picking the band correctly
  • Steering devices to the best access point for them
  • Balancing the load when necessary

Mesh networks add a little more complexity due to their need for topology management.

But the scale and dynamic nature of home Wi-Fi environments create a challenge to manage home Wi-Fi QoE. Artificial intelligence helps by:

  • Monitoring all the existing conditions in real-time
  • Detecting any changes
  • Processing and analyzing the data coming from the devices and CPE
  • Learning from what has worked well in similar environments
  • Applying those best-practice learnings to the current home environment

Customers using our AI technology to manage QoE:

  • Improve customer retention 15-20%
  • Improve network QoE 25-35%
  • Improve home Wi-Fi speeds
  • Reduce customer service calls 30-50%
  • Reduce access points with interference 40%
  • Reduce new installation costs 15-20%
  • Reduce field dispatches 44%.

Home Wi-Fi’s Dynamic Ecosystem

As mentioned above, the Wi-Fi ecosystem constantly changes, which makes managing home Wi-Fi QoE a challenge for service providers. Each new technology comes with high expectations for improving home Wi-Fi but creates new issues. The 5GHz band, 11ax, 11ay, and mesh all help in their way, yet home Wi-Fi performance issues persist. For example:

  • The move from 2.4 to 5GHz replaced the noise issues in 2.4 with coverage issues in 5 GHz.
  • The 5GHz band has marginal impact on throughputs—44% of homes still see less than 30 Mbps, which is not enough to handle 4K video streaming, gaming, and multiple IoT and mobile devices.
  • Adding access points, if unmanaged, only improve 2.4GHz by 2.5% and do not impact 5Ghz at a cost of $50-100 per access point.

Going forward, multi-dwelling units, 5G, and the convergence of 5G and Wi-Fi add new challenges.

The Carrier Decision Point

The issues with home Wi-Fi QoE and the challenges created by the dynamic ecosystem force service providers into difficult business decisions that impact their service and brand. They must decide where they focus their resources, do they:

  • Optimize for bandwidth-to-the-device or settle with optimizing to the premises
  • Optimize for dynamic QoE or settle for optimizing for average QoE

The Practical Application of AI

AI technology, when practically applied, makes this decision easier and balances the polarized nature of the decision. AI bridges the gap between the business priorities and the complex physical assets that impact home Wi-Fi QoE. AI technology relieves service providers from making an either/or decision. They can phase capabilities in overtime, based on their business priorities.

As the figure below shows, by adding virtual probes to elements in the physical layer on the left, AI and Data Science optimize home Wi-Fi QoE based on the business priorities of the service provider on the right. Ergodic Spectrum Management is implemented for wireless networks, and Dynamic Line Management for fixed-line networks. Everything is phased in accordance with the business priorities of the carrier.

Home Wi-Fi QoE AI

Home Wi-Fi QoE Solution Requirements

To successfully apply AI to home Wi-Fi QoE in this phased manner, some core technology capabilities are required. These include:

  • Vendor-neutral network element interface across the environment, radio, and spectrum
  • Predictive analytics to dynamically learn and predict network failure and performance patterns, and their impact on customer churn and service requests
  • Real-time diagnostics for all the connections in the network
  • Extended data input from systems beyond network elements such as customer care requests, service offers, etc.
  • Prescriptive analytics to dynamically learn and recommend the best connection profiles for target stability and QoE for proactive care, maintenance, and network design
  • Dynamic layered optimization to dynamically learn, fix, and recommend best connection profiles for target stability and QoE
  • Communication and collection so that all connections are monitored in real-time and detailed diagnostics are triggered based on pre-set QoE events
  • Commitment to standards so that it will work well with new technology advancements in the future

Home Wi-Fi QoE Recommendation

Service providers do not have to settle for rudimentary solutions configured to solve problems based on the lowest common denominator or averages. With the practical application of the right AI solution, they can implement intelligent solutions that learn and adapt to solve a wide range of home Wi-Fi QoE problems in a variety of environments.

To learn more: