Digiot: Proactive and Adaptive Network Parameter Configuration for Sustainable IoT Devices

The sustainability of beyond 5G networks is of paramount importance, driven by the increasing number of supported applications, end-user devices, and their dynamic requirements. As the connectivity of end-user devices experiences exponential growth, the focus shifts toward reducing the energy consumption of each individual device while ensuring a satisfactory end-user quality of experience. To address the aforementioned challenge, this paper presents Digiot, a novel framework that leverages contextual information from devices and their application demands to dynamically configure network parameters controlling device operation states. Specifically, Digiot accurately models per packet energy consumption and transmission latency of a device through a detailed Markov Chain. From this model information is derived to efficiently adapt network parameters. To choose network parameter configuration that reduces the device energy consumption while fulfilling its demands Digiot relies on an optimization problem, which is solved via heuristic and meta-heuristic techniques. We thoroughly evaluate Digiot through extensive simulations on real traces. We demonstrate that Digiot reduces the per packet energy consumption by an average of 20%, while maintaining all delay requirements.