期刊
IEEE INTERNET OF THINGS JOURNAL
卷 7, 期 7, 页码 5976-5985出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JIOT.2019.2953108
关键词
Task analysis; Resource management; Games; Cloud computing; Heuristic algorithms; Reinforcement learning; Double auction game; experience-weighted attraction (EWA); latency-sensitive businesses; mobile edge computing (MEC)
资金
- National Key Research and Development Plan [2018YFB1800805]
- National Engineering Laboratory for Public Safety Risk Perception and Control by Big Data (PSRPC)
In recent years, we have witnessed the compelling application of the Internet of Things (IoT) in our daily life, ranging from daily living to industrial production. On account of the computation and power constraints, the IoT devices have to offload their tasks to the remote cloud services. However, the long-distance transmission poses significant challenges for latency-sensitive businesses, such as autonomous driving and industrial control. As a remedy, mobile edge computing (MEC) is deployed at the edge of the network to reduce the transmission delay. With the MEC joining in, how to allocate the limited computing resource of MEC is a critical problem to guarantee efficient working of the whole IoT system. In this article, we formulate the resource management among MEC and IoT devices as a double auction game. Also, for searching the Nash equilibrium, we introduce the experience-weighted attraction (EWA) algorithm performing behind each participant. With this AI method, auction participants acquire and accumulate experience by observing others' behavior and doing introspection, which accelerates the trading policy's learning process of each agent in such an opaque environment. Some simulation results are presented to evaluate the convergence and correctness of our architecture and algorithm.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据