Intelligent resource allocation scheme for cloud-edge-end framework aided multi-source data stream

Abstract To support multi-source data stream generated from Internet of Things devices, edge computing emerges as a promising pattern with low latency and high bandwidth compared to cloud computing. enhance the performance within limited communication computation resources, we study cloud-edge-end architecture, where one server multiple computational access points can collaboratively process compute-intensive streams that come sources. Moreover, environment is considered, in which wireless channel characteristic are time-varying. adapt dynamic network environment, first formulate optimization problem markov decision then decompose it into offloading ratio assignment sub-problem resource allocation sub-problem. Meanwhile, order reduce action space, further design novel approach combines proximal policy (PPO) scheme convex optimization, PPO used for assignment, while employed allocation. The simulated outcomes this work help development application stream.