Large-scale, distributed sensor networks are the projected weapon of choice for future pervasive computing applications such as, for example, environment monitoring, surveillance, (big) data mining and patient monitoring. However, state-of-the-art approaches face major challenges: specialized sensors are expensive and require careful calibration. Hardware sensors operating in uncertain, harsh environments eventually suffer from stress, ageing and physical damage, which leads to unforeseen effects that can render the device and the data recorded useless. Highly-tuned data processing algorithms are often not scalable and are not robust against faulty sensors delivering wrong data. Generally, systems can only adapt, if at all, in some predefined limited ways and are not capable of autonomously “inventing” new ways of adapting to unexpected changes in their internal and external environment.

This project follows a different approach to sensor network development drawing inspiration from nature, where biological systems self-organize, selfheal, sense, process information and react, emerging from evolutionary and developmental processes. Exploiting the advantages of a distributed structure of cellular systems this project perceives sensor networks as spatially distributed multicellular organisms where each sensor node represents a cell colony or a DNA bank. It is hypothesized that incorporating artificial cell development mechanisms and gene regulatory networks within each sensor node enable emerging beneficial properties of fault-tolerance, adaptivity and scalability in a sensor network. In order to investigate if this hypothesis holds in the context of a constrained distributed embedded system, the model is optimized for the Arduino microcontroller.

External link to the publication >>