IoT: Wi-Fi Sensing and Diagnostic for Customer Premises Equipment using Transfer Learning and USP

The new routers are connected through USP over MQTT and allow geographically near-real-time monitoring. The aim of the thesis is to analyze and build a predictive algorithm based on a model trained using the information collected from the routers. A new controller/agent will be developed for diagnostic elements gathering (software, hardware, telemetry, environment). The new agent can send the information to backend for a distributed computation but also act locally through the BLE, when Internet is not available.