面向医疗大数据的云雾网络及其分布式计算方案
发布时间:2018-01-31 08:51
本文关键词: 医疗大数据 云计算 云/雾混合网络 负载均衡 出处:《西安交通大学学报》2016年10期 论文类型:期刊论文
【摘要】:针对云计算应用于医疗大数据场景时存在业务处理时延较高的问题,提出了一种基于边缘计算的新型云/雾混合网络架构,该架构利用医院中的路由器或交换机等边缘设备,在云服务器与医疗检测设备之间构建一个雾计算层,通过将云服务器中的医学影像等医疗大数据分析结果主动缓存至雾计算设备,并与雾设备上来自医疗检测终端的数据进行对比计算,得出诊断结果,达到降低业务处理时延的目的。考虑到边缘设备的计算能力较弱,进一步提出了一种多设备分布式计算方案,利用带约束的粒子群优化负载均衡(CPSO-LB)算法,达到任务处理时延最小的目标。仿真结果表明:基于CPSO-LB算法的云/雾混合网络能有效地降低医疗数据处理时延;当采用10个雾计算设备,处理的医疗数据量在6~10Gb时,与云计算网络相比时延性能提升了50.95%~37.37%。
[Abstract]:A new cloud / fog hybrid network architecture based on edge computing is proposed to solve the problem of high service delay when cloud computing is applied to the medical big data scenario. The architecture uses edge devices such as routers or switches in a hospital to build a fog computing layer between a cloud server and a medical detection device. The results of medical big data analysis such as medical images in the cloud server are actively cached to the fog computing equipment, and compared with the data from the medical detection terminal on the fog equipment, the diagnosis results are obtained. In order to reduce the delay of service processing, a multi-device distributed computing scheme is proposed considering the weak computing power of edge devices. The constrained particle swarm optimization (PSO) load balancing algorithm (CPSO-LB) is used. The simulation results show that the cloud / fog hybrid network based on CPSO-LB algorithm can effectively reduce the delay of medical data processing. When 10 fog computing devices are used and the amount of medical data processed is 610Gb, the delay performance of cloud computing network is improved by 50.95g / 37.37g / h compared with cloud computing network.
【作者单位】: 西安电子科技大学ISN国家重点实验室;国网吉林省电力有限公司信息通信公司;
【基金】:国家自然科学基金资助项目(61401331) 港澳台科技合作专项资金资助项目(2015DFT10160) 中央高校基本科研业务费专项资金资助项目(20101155739)
【分类号】:TP311.13;TP338.8
【正文快照】: 在医患比例严重失调的当今社会,挖掘分析医疗大数据,利用获得的有价值的信息辅助医生诊断治疗疾病,提高医生诊断效率,成为了迫切需要解决的问题。云计算作为目前医疗大数据分析处理的支撑平台,为医院信息化建设提供了强大的技术支持[1]。近年来,专家学者们基于云计算提出了医,
本文编号:1478730
本文链接:https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/1478730.html