Edge resource slicing approaches for latency optimization in AI-edge orchestration

摘要

Edge service computing is an emerging paradigm for computing, storage, and communication services to optimize edge framework latency and cost based on mobile edge computing (MEC) devices. The devices are battery-enabled and have limited communication and computation resources. X consolidation is a major issue in distributed heterogeneous MEC orchestrations, where X represents the task scheduling/device selection/channel selection/offloading strategy. The network entities need to enhance network performance under uncertain circumstances for such orchestrations. Haphazard X consolidation leads to abnormal resource and energy usage, quality of service (QoS) and latency of the edge framework. However, this study concentrates on analysing the impact of reinforcement learning-based edge resource consolidation models. The models are classified according to functionality, including device resource management, service request allocation, device selection, and offloading types. Finally, the article discusses and highlights some unresolved challenges for further study on MEC orchestration to enhance offloading strategy and resource management, as well as device and channel selection efficiency.