边缘计算的物联网深度学习及任务卸载调度策略

为解决物联网深度学习模型的网络性能和隐私问题,提出一种边缘计算的物联网深度学习应用及任务卸载策略,以优化网络性能,保护数据上传中的用户隐私。深度学习的多层结构适用于边缘计算,边缘节点上传缩减的中间数据,因此减少了从物联网设备到云服务器的网络流量。考虑到边缘节点有限的服务能力,提出一种边缘计算环境中最大化任务数量的卸载调度策略,优化边缘计算的物联网深度应用性能。实验结果表明,该策略能够在边缘计算环境中执行多个深度学习任务,并且性能优于其他物联网深度学习优化解决方案。

DEEP LEARNING OF IOT AND TASK UNLOADING SCHEDULING STRATEGY FOR EDGE COMPUTING

To solve the network performance and privacy problems of IoT in deep learning model,we proposed an edge computing deep learning application and task uninstallation strategy of the IoT to optimize network performance and protect user privacy in data upload.The multi-layer structure of deep learning is suitable for edge computing,and edge nodes upload reduced intermediate data,thus reducing network traffic from IoT devices to cloud servers.Considering the limited service capability of edge nodes,we proposed an unloading scheduling strategy to maximize the number of tasks in edge computing environment to optimize the deep application performance of edge computing in the IoT.The experimental results show that the strategy can perform multiple deep learning tasks in the edge computing environment,and its performance is superior to other deep learning optimization solutions of the IoT.