IDEAL: Reducing Carbon Footprints of IoT Devices through Extension of Active Lifespans (EP/Z533749/1)
About this project
This project is supported by EPSRC standard (EP/Z533749/1) in partner with the University of Oxford and the University of Glasgow.
The CO2 emissions from manufacture (the so-called embodied carbon) of end-user ICT devices makes up the majority of their carbon footprint over the typical current useful life of such devices. To achieve sustainable ICT, in addition to reducing run time energy consumption, it is therefore essential to extend the active life, ideally to several decades. The largest growth in end-user devices is in the Internet of Things (IoT) market, projected to rise to 30 billion devices by 2030.
Our vision with the IDEAL project is to demonstrate the possibility of extending the lifetime of IoT devices from several years to several decades through a combination of novel hardware design technologies, hardware/software co-design techniques, formal methods, machine learning and circular economy. As current business models assume short replacement cycles for end-user devices (typically a few years), in addition to the technologies to extend the useful life, we will also demonstrate the viability of alternative business models based on very long-lived devices, co-created with our partners.
Our technology will allow not only to extensively prolong the useful life, but also to accurately assess the degradation of any given device, enabling the repurposing of devices with reduced capabilities for new tasks that match these capabilities. In this way, our proposal fits both with Sustainable ICT, as it will enable a drastic reduction in the embodied carbon of IoT devices, and the Circular Economy, as our technology will allow devices to be repurposed repeatedly throughout their useful life.
Our proposed approach is to instrument the integrated circuits at the lowest level with a novel, ultra-low power, unobtrusive monitoring and data aggregation technology and additional self-healing capabilities. The generated data will be analysed using low power machine learning nodes executing close to the actual IoT devices rather than in the cloud for reliable and early detection of anomalies in the system operation, indicative of early system degradation, and produce the optimal strategy for addressing each anomaly before it can affect the lifetime of the system. Because this is a close-loop networked system with a considerable degree of control over the IoT devices, it is essential to have guarantees of correctness and security by design in the communication between the analysis and decision making nodes and the IoT devices, which will be enabled by the use of a formal mechanism know as session types.
The system we propose to develop is a subsystem of any IoT system: an IoT system consists of end-user devices located at the so-called “edge”, connected to a cloud data centre, with increasingly, part of the processing performed in the “fog” between the edge and the cloud. The aim is that the introduction and operation of our subsystem will substantially reduce the full-lifecycle carbon footprint of the overall system. In business terms, our system fits within an emerging environment of IoT-as-a-service and circular economy.
Large-scale introduction of our technology and associated business models can result in the emergence of entirely new types of businesses taking care of the re-purposing of devices or offering monitoring and adaptation services.