Improving IoT Module Testability with Test-Driven Development and Machine Learning

Critical applications supported by the Internet of Things (IoT) such as health, finance and safety require robust, fault tolerant and available systems. Besides the large amount of real time data generated to extrinsic systems (considering Big Data, AI, Smart Cities and Industry 4.0 applications), there is a plethora of data that could be used intrinsically to diagnose an IoT system. A Test Driven Development (TDD) approach for automatic module assessment based on ESP32 IoT Module and unsupervised machine learning is proposed to classify an IoT module status as bad or good based on memory, temperature and WiFi level metrics. An automatic test case using the trained classifier is presented as a result of the proposed approach validation with real testbed data. The IoT module scripts for ESP32, and Python scripts for data collection, model training and inference with KNN algorithm are available under a Creative Commons license for replicability.