How to solve Wi-Fi problems using AI

The general approach is quite traditional once we have established Wi-Fi Bandwidth as the key metric for Quality-of-Service. The system will routinely identify patterns of issues that need diagnostics, analysis, and solution. When a solution is found it is matched to pattern and added to toolbox — ready to be deployed on live Wi-Fi networks.

The real-time engine is triggered when observing unusual or unacceptable QoS metrics — i.e. abnormal values for latency or bandwidth. Then there is a process of diagnosing for cause, following a tree-structured data and context analysis as shown in the flowchart below.

Once the system has successfully identified what is causing the problem, it will trigger action in form of optimizing router configuration or engaging the user in an automated dialogue to help resolve the issue. A typical dialogue exchange looks like the example below:

We have found that data and especially the objective QoS metric is key to successfully applying the traditional AI methodologies. Bandwidth and latency to internet and over Wi-Fi have proven to be the required objective QoS metric.