一种移动边缘计算下基于高斯隐藏贝叶斯的服务质量监控方法

QoS Monitoring Approach Based on Gaussian Hidden Bayes in Mobile Edge Computing

移动边缘计算是指在网络边缘执行计算的一种新型计算模型,其通过部署在靠近移动设备端的边缘服务器为用户提供服务,面对这种计算模型下时效短,变化快的数据,如何及时有效的监控出服务的运行状态显得尤为重要,在服务监控系统中,QoS(服务质量)通常作为是否满足用户调用服务时的需求的重要指标.在移动边缘环境下,用户的移动性和QoS属性值之间的依赖性往往会导致监控结果偏离真实结果.然而现有的QoS监控方法都未考虑以上问题.本文提出了一种移动边缘计算下基于高斯隐藏贝叶斯的QoS监控方法,该方法假设边缘服务器的QoS属性值服从高斯分布,为每个属性构造一个父属性,从而减少属性间取值的依赖性,在训练阶段为每个边缘服务器构造对应的高斯隐藏贝叶斯分类器,在监控过程中基于用户的移动性动态切换分类器,并结合KNN算法实现边缘计算下的服务质量监控.实验结果表明了本文所提方法的有效性.

Mobile edge computing is a new computing model that performs computing on the edge of the network.It provides services for users by deploying edge servers near mobile devices.Faced with such time-consuming and fast-changing data under this computing model,it is particularly important to monitor the running status of services timely and effectively.In the system of service monitoring,the quality of service(QoS)is usually used as an important indicator whether the needs of users to be satisfied when invoking a service.In mobile edge environment,user mobility and the dependence between the value of QoS attributes often cause the monitoring results to deviate from the real results.However,the existing QoS monitoring methods don’t take these problems into account.In this paper,a method of QoS monitoring based on Gaussian Hidden Bayes for mobile edge computing is proposed.This approach assumes that the value of QoS attributes of edge servers obeys Gaussian distribution,then constructs a parent property for each property,thus reducing the dependence between attributes.At the training stage,Gaussian Hidden Bayesian classifier is constructed for each edge server.During monitoring,changing classifier dynamically based on user mobility and combining KNN algorithm to realize quality of service monitor-ing under edge computing.The experimental results show the effectiveness of the proposed method.