基于深度强化学习的移动群智感知调度研究

Mobile Crowd Sensing Scheduling Based on Deep Reinforcement Learning

为了提高基于移动群智感知技术应用的性能,提出基于深度学习模型的层次化调度策略,利用边缘计算技术实现移动群智感知的任务调度和资源分配。首先设计了具有层次化结构群智感知框架以实现高效的资源分配。然后,考虑物联网节点位置和性能对服务质量的影响,提出了基于深度强化学习的策略来进行任务调度。最后,采用仿真实验评估本方法的有效性。由结果可知,本方法能够获得较高的奖励,其性能要优于现有的调度方法,说明本方法适用于动态的网络环境。

To improve the performance of the mobile-crowd-sensing-based application,this paper proposes a hierarchical scheduling strategy based on a deep learning model,leveraging edge computing to implement task scheduling and resource allocation for mobile crowd sensing.First,we introduce the hierarchical mobile crowd sensing framework to realize resource allocation.Then,considering the location and performance of Internet of Thing’s node will affect the quality of service,we design the deep reinforcement learning-based strategy to schedule the tasks.Finally,simulation experiments are used to evaluate the effectiveness of this method.Experimental results show that our proposal can achieve high reward,indicating our proposal is superior to the existing scheduling methods,which means our method is suitable for dynamic network environment.