A Post-sales hardware update platform for the internet-of-things

Abstract

In the software domain, updating applications after the product has been delivered to the customer is a common strategy to decrease the time-to-market and continuously improve the quality of the product. In the hardware sector, where time-to-market is generally longer, this strategy is not yet used due to technological barriers. This thesis presents a platform that allows post-sales hardware updates for Internet-of-Things (IoT) devices using dynamic partial reconfiguration. To demonstrate this platform in practice, we use an application using an Early Warning Score (EWS) system to determine a patient’s state of health as a showcase. The proposed platform not only allows updating the hardware design post-sales, but also includes dynamic partial reconfiguration capabilities that allows the updated designs, in this case Fourier Transform, used on the device to be changed during operations (at runtime). This feature is not only beneficial for hardware updates but it can improve runtime performance of the system too. A use case was shown in which a favourable hardware implementation is automatically configured depending on the battery status of the IoT device in order to adapt the performance of the unit to the runtime requirements (e.g., lower power profile when the battery is low). The performances of these implementations are compared in terms of: power consumption, accuracy, processing speed and utilisation.