Empowering Intermittent IoT Networks

Internet of Things (IoT) forms a network of interconnected devices that generate data and process them on cloud. From about 8.6 billion active IoT devices in 2019, a recent survey shows that, at present there are over 15.1 billion IoT devices connected worldwide. The number of active IoT devices is expected to double by 2030. The primary aim of such devices is to automate various processes and to reduce human labor. For generating legitimate data, any IoT device requires sufficient power and uninterrupted communication. Some of these devices may also be placed at remote unserviceable locations. Hence, these devices are gradually adopting battery-less, energy harvesting solutions leading to Transiently Powered embedded systems. These represent a novel category of embedded systems relying exclusively on energy harvested from external sources to perform long-running computations. Performing complex computations require sufficient memory and a stable power source. Enabling long-running computations on these systems is a major challenge due to highly intermittent nature of the power supply, resulting in frequent system reboots. While distributed task concurrency holds promise for intermittent networks, data unavailability due to frequent failures poses a challenge. The maj or challenge faced by energy harvesting devices is that due to the fluctuating nature of the power supply, the communication between interconnected IoT devices are interrupted, leading to loss of data and degrades the computation progress. The proposed work introduces a novel approach to improve the forward computation progress and ensures that no data loss occurs while performing large computations in an intermittent network. The future scope is to distribute a long running process among various devices in an intermittent network by also ensuring that there is no loss or duplication of data.