Programming Wearable IoT Apps

Purpose and goal

The project aims to design a programming system that simplifies the orchestration of multiple machine-learning (ML) algorithms, particularly neural networks, along with sensors and actuators on or in the body. These algorithms process physiological data and enable wearable IoT applications. The system will help non-experts program and integrate these components more easily, allowing them to develop ML applications for real-time physiological data processing and interaction.

Expected effects and result

The expected result is to design a programming system that adopts a “building-block” approach, similar to Blocky or Node-RED, that allows developers to compose wearable IoT applications that span data acquisition and machine learning on resource-constrained sensor nodes, the edge, and cloud devices. Towards the end of the thesis, the student will then realize a prototype implementation to demonstrate the system in action.

Planned approach and implementation

The study will begin with a literature review on machine learning applications in in- and on-body sensors. This will be combined with an analysis of the programming process of IoT devices. Based on these insights, an orchestrating mechanism will be designed and implemented, with a potential prototype for real-world evaluation. Finally, the findings will be documented in the thesis, which will be written and discussed.