Towards In-Fiber Computation and Networked Fabric Systems

There has been growing interest in the development of wearable and mobile computing systems for not only collecting physiological data, but also contextualizing this data into meaningful health insights. However, conventional wearables often rely on rigid devices with small areas of contact, presenting both a mechanical mismatch with the human body and limiting the depth of physiological insights these systems can acquire. In contrast, fibers and fabrics are lightweight, flexible, and ubiquitous in daily life, offering a unique opportunity for physiological monitoring based on their large, conformal contact with the human body. Despite advancements in fiber electronics, challenges remain in embedding fibers and fabrics with the computational capabilities necessary for fully integrated health monitoring while maintaining the properties of traditional textiles, including flexibility, elasticity, and environmental resilience.

Here, we present our progress towards the development of a soft, flexible fiber that combines sensing, storage, processing, and communication capabilities. To fabricate these fibers, we investigate the use of a flow-based thermal draw process that integrates multiple functionally distinct microelectronic devices into a single fiber. For robust fabric integration and durability in daily use, we explore material combinations and processing techniques that provide both fiber flexibility and elasticity. Using the fiber-based computing architecture, we work towards implementing protocols and customized algorithms that leverage the fiber’s edge-processing capabilities to contextualize physiological data into meaningful health metrics. Finally, to create networked fabric systems that avoid failure modes associated with rigid wire interconnects, we explore various wireless communication modalities between individual fibers. To demonstrate the potential of such a networked system for whole-body health monitoring, we develop federated machine-learning algorithms that aim to capitalize on the advantages granted by this distributed fabric approach as compared to the traditional single-device approach.