A Layered Framework for Energy-Efficient Edge Computing in Sustainable IoT Systems
TL;DR
The proposed design uses adaptive sampling, data aggregation, compression, a workload-aware offloading mechanism, dynamic voltage/frequency scaling where available, and container consolidation and renewable-energy awareness, and carbon-aware workload placement, all of which are aligned with the direction of current Multi-access Edge Computing standardization.
Abstract
The proliferation of Internet of Things (IoT) installations has resulted in an increased number of sensing, communication, storage, and computation operations performed by devices with limited resources. A typical cloud-centric approach results in unnecessary communication overhead, latency, and additional energy expenditure to transport large volumes of raw sensor data to remote datacenters. The proposed design uses adaptive sampling, data aggregation, compression, a workload-aware offloading mechanism, dynamic voltage/frequency scaling where available, and container consolidation and renewable-energy awareness, and carbon-aware workload placement, all of which are aligned with the direction of current Multi-access Edge Computing (MEC) standardization, and are now considering energy management, power information, and application-specific energy policies as first-class concerns.
