MEC技术在视频监控领域的应用

随着移动通信技术和人工智能技术的共同发展,视频监控领域出现了越来越多的移动化巡查设备,由此产生的海量的监控数据,必然给移动网络带来巨大的传输压力,在移动网络边缘部署MEC服务器,将可能是解决这个问题的一种方案。介绍了视频监控智能化、移动化的进程,分析了视频监控"云边协同"的发展现状,重点研究了MEC技术以及MEC技术如何与视频监控业务相结合。     

移动边缘计算中基于联邦学习的视频请求预测和协作缓存策略

随着互联网社交平台的崛起和移动智能终端设备的普及,自媒体短视频、直播等视频业务蓬勃发展,人们对高质量视频服务的需求也急剧上升。与此同时,连接到核心网络的大量智能设备增加了回程链路的负载,传统的云计算难以满足用户对视频服务的低延迟要求。移动边缘计算(MEC)通过在网络边缘部署具有计算和存储能力的边缘节点,通过在更靠近用户的边缘侧提高计算和存储服务,降低了数据传输时延进而缓解了网络阻塞。因此,基于MEC架构,该文充分利用网络边缘资源,提出了基于联邦学习的视频请求预测和视频协作缓存策略。通过利用多个边缘节点对提出的深度请求预测模型(DRPN)视频请求预测模型进行联邦训练,预测视频未来的请求情况,然后量化缓存内容所带来的时延收益并协作地以最大化该时延收益为目的进行缓存决策。该文分析了真实数据集MovieLens,模拟了视频请求缓存场景并进行实验。仿真结果表明,相比于其他策略,所提策略不仅能有效降低用户等待时延,在有限的缓存空间中提高内容多样性,从而提高缓存命中率,降低缓存成本,还能降低整个系统的通信成本。     

基于云、网、边融合的边缘计算新方案:算力网络

边缘计算已经成为5G时代重要的创新型业务模式,尤其是其低时延特性,被认为是传统方案所不具备的,因此边缘计算能够提供更多的服务能力且具有更为广泛的应用场景。但边缘计算与处于中心位置的云计算之间的算力协同成为新的技术难题,即需要在边缘计算、云计算以及网络之间实现云网协同、云边协同,甚至边边协同,才能实现资源利用的最优化。在研究边缘计算算力分配和调度需求的基础上,提出了基于云、网、边深度融合的算力网络方案,并针对AI类应用给出了一个典型实施系统,该方案能够有效应对未来业务对计算、存储、网络甚至算法资源的多级部署以及在各级节点之间的灵活调度。     

Efficient Resource Management for Cloud-enabled Video Surveillance over Next Generation Network

The next generation video surveillance systems are expected to face challenges in providing computation support for an unprecedented amount of video streams from multiple video cameras in a timely and scalable fashion. Cloud computing offers huge computation resources for large-scale storage and processing on demand, which are deemed suitable for video surveillance tasks. Cloud also provides quality of service guaranteed hardware and software solutions with the virtual machine (VM) technology using a utility-like service costing model. In cloud-based video surveillance context, the resource requests to handle video surveillance tasks are translated in the form of VM resource requests, which in turn are mapped to VM resource allocation referring to physical server resources hosting the VMs. Due to the nature of video surveillance tasks, these requests are highly time-constrained, heterogeneous and dynamic in nature. Hence, it is very challenging to actually manage the cloud resources from the perspective of VM resource allocation given the stringent requirements of video surveillance tasks. This paper proposes a computation model to efficiently manage cloud resources for surveillance tasks allocation. The proposed model works on optimizing the trade-off between average service waiting time and long-term service cost, and shows that long-term service cost is inversely proportional to high and balanced utilization of cloud resources. Experiments show that our approach provides a near-optimal solution for cloud resource management when handling the heterogeneous and unpredictable video surveillance tasks dynamically over next generation network.     

边缘环境下的视频行为识别

边缘智能是一种新兴的智能计算模式,其将人工智能技术和边缘嵌入式设备结合,被广泛应用于物联网系统。智能摄像机是典型的边缘设备之一,它能提供低延迟的视频处理能力,适用于智能家居、智能交通、智能监控等领域。然而,由于摄像机的计算资源有限,传统的行为识别模型难以在本地完成计算任务。为解决这一问题,文中提出了一种基于边缘计算的架构,利用深度学习目标检测算法YOLO v3对视频行为进行识别。在该架构中,智能移动终端负责数据采集和压缩,边缘服务器承担大部分目标检测任务,而检测困难的目标和模型训练则由云服务器负责。为更好地适应边缘设备,本文采用轻量化的神经网络MobileNet替换YOLO v3模型的特征提取模块。经过测试,该架构能有效提取和识别视频中的静态和动态行为,为实现边缘计算环境下低成本、大规模的行为识别提供了有益的参考。     

基于5G边缘计算的深度学习架构与应用

针对深度神经模型在网络边缘难以训练的问题,构建了一种基于5G边缘计算的深度学习模型训练架构。架构利用5G边缘计算接入网打通边缘智能设备与边缘计算层的数据通信,模型训练过程采用各边缘计算节点利用本地数据进行全模型训练,再由中心服务器进行模型参数汇集和更新的分布式训练模式,既保证了模型训练的数据集多样性,又减少了网络压力和保障了本地数据隐私,是一种非常具有潜力的深度学习边缘计算架构。     

一种车载服务的快速深度Q学习网络边云迁移策略

智能网联交通系统中车载用户的高速移动，不可避免地造成了数据在边缘服务器之间频繁迁移，产生了额外的通信回传时延，对边缘服务器的实时计算服务带来了巨大的挑战。为此，该文提出一种基于车辆运动轨迹的快速深度Q学习网络(DQN-TP)边云迁移策略，实现数据迁移的离线评估和在线决策。车载决策神经网络实时获取接入的边缘服务器网络状态和通信回传时延，根据车辆的运动轨迹进行虚拟机或任务迁移的决策，同时将实时的决策信息和获取的边缘服务器网络状态信息发送到云端的经验回放池中；评估神经网络在云端读取经验回放池中的相关信息进行网络参数的优化训练，定时更新车载决策神经网络的权值，实现在线决策的优化。最后仿真验证了所提算法与虚拟机迁移算法和任务迁移算法相比能有效地降低时延。     

Energy Efficient Resource Scheduling Using Optimization Based Neural Network in Mobile Cloud Computing

The mobile cloud computing has become an emerging technology where the mobile computing is integrated with cloud computing to process the mobile data. Besides the advantages of mobile cloud computing, there are some issues which include power consumption, resource scarcity, quality of service, security and computational cost. In this paper, in order to minimize total power consumption with better performance, the neural network based optimization methods using artificial neural network and convolutional neural network models were implemented by varying variance and loudness. From the experimental results it is observed that, by using optimization in the neural network, the power consumption has been reduced by 53.68% and obtained improvement using convolutional neural network which further reduced the power consumption by 30.3% with minimum root mean square error compared with other algorithms.

     

基于云网边端协同计算的智能分发网络研究

智能服务机器人将成为社会发展的重要组成部分和人类工作与生活的重要助手。通过简要介绍由人工智能的云端大脑、基于5G构建的安全神经网络和多关节的机器人本体所组成的云端机器人,提出了基于云、网、边、端协同计算的智能分发网络(Intelligence Distribution Network,IDN)的概念,将IDN应用于云端机器人,以提升云端机器人的智能程度。详细阐述了IDN的架构,从算力、算法、通信、数据、安全等角度对IDN进行了研究,对比分析了智能分发网络与内容分发网络(Content Delivery Network,CDN)的异同点;IDN之于人工智能类似于CDN之于互联网内容,对IDN的典型应用场景进行了测试分析。通过使用IDN,提升了云端机器人智能程度和响应速度、降低了上行带宽和服务成本。尽管当前处于云端机器人的发展初期,但IDN将是实现云端智能机器人大规模商用的必经之路和关键技术,IDN将成为继CDN产业之后的又一个新领域。     

视频监控标准化最新进展

关注视频监控主流国际及国内标准化组织工作，分析了ITU-T、ONVIF、IEC、ISO、PSIA、HDcctv、IEEE、3GPP、ETSI、TC100、CCSA、AVS、全国信息技术标准化技术委员会等主要标准组织的视频监控相关标准化工作进展，介绍了视频监控系统、架构、设备、编解码、安全、智能应用、大数据应用、云存储、云计算、边缘存储、边缘计算等标准化研究最新进展。在此基础上，探讨了视频监控未来的产品研发方向和标准化内容，给出了视频监控标准化路线图，提出了未来的视频监控标准化及预研方向及建议。     

基于边缘云计算的光纤无线网络优化设计

为了提升网络边缘数据处理能力,满足终端大带宽和低时延的要求,构建了基于边缘基础设施的云计算平台,设计了具有动态带宽调整的光纤网络模型。提出了一种基于边缘云计算的时序优化算法,并将其应用于光纤无线网络。通过OPNET软件仿真分析了时序优化算法的传输时延均值,结果显示,优化后最大时延为43.1 ms,仅为传统方法的34.2%。实验对局域网内多个终端之间的数据通信进行分析,讨论了三种算法的传输能效、光纤信道利用率及传输能耗。实验结果显示,采用时序优化算法的测试结果具有明显改善,其传输能效提升了近1倍,边缘云数据传输时延均值信道利用率提升了约6.2%,网络传输能耗均值最优。该光纤无线网络模型及其优化算法在传输时延、信道利用率以及网络能耗方面具有明显提升。其在提升光纤通信链路选择及边缘端数据交互中具有一定的优势。     

Edge Computing Offloading Strategy Based on Dynamic Non-cooperative Games in D-IoT

With the vigorous development of Internet of Things technology, the current distribution network is developing towards the information-based and intelligent distribution Internet of Things (D-IoT). D-IoT adopts the mode of the cloud computing center and the edge cloud network working together. The edge cloud network has a large number of intelligent terminals, which can well adapt to the current sharply expanding power data scale. In order to further improve the ability of the edge network in D-IoT to process data in real time, and to maximize the quality of user experience (QoE) while minimizing energy consumption when performing computing offload, this paper proposes a dynamic non-cooperative game based edge Computing task offloading strategy, considering the dynamic nature of task generation, designed a distributed iterative optimization algorithm, which decomposes computing offloading into a series of sub-problems to solve. The results of simulation experiments prove that the calculation offloading mechanism proposed in this paper can greatly improve D -Compute efficiency of IoT system.

     

Survey on data security and privacy-preserving for the research of edge computing

With the rapid development and extensive application of the Internet of things (IoT),big data and 5G network architecture,the massive data generated by the edge equipment of the network and the real-time service requirements are far beyond the capacity if the traditional cloud computing.To solve such dilemma,the edge computing which deploys the cloud services in the edge network has envisioned to be the dominant cloud service paradigm in the era of IoT.Meanwhile,the unique features of edge computing,such as content perception,real-time computing,parallel processing and etc.,has also introduced new security problems especially the data security and privacy issues.Firstly,the background and challenges of data security and privacy-preserving in edge computing were described,and then the research architecture of data security and privacy-preserving was presented.Secondly,the key technologies of data security,access control,identity authentication and privacy-preserving were summarized.Thirdly,the recent research advancements on the data security and privacy issues that may be applied to edge computing were described in detail.Finally,some potential research points of edge computing data security and privacy-preserving were given,and the direction of future research work was pointed out.     

面向智能网联汽车边缘网络的分布式端-边协同算法

车联网高级安全服务中,智能网联车辆配备了摄像头,可以拍摄周围的视频,用于安全、交通监控和监视等目的。车辆将获取的视频上传到边缘计算节点后,可以对视频进行分析和备份,以满足不同的安全驾驶需求。然而,车辆连续直接向边缘计算节点上传生成的视频内容会非常消耗带宽,并消耗大量的能量。基于该问题,提出一种面向智能网联汽车边缘网络的分布式端-边协同算法。针对车联网高可靠低时延内容传输的特点,引入有限块长度编码机制。同时,引入车辆视频信息源的压缩编码功率消耗,建立车辆能耗模型。根据车辆视频信息源的视频质量要求,通过调整视频编码码率、信息源传输速率,以及车辆多路径路由的决策,提出一种完全分布式的优化算法,以提高网络资源利用率,并保证单个车辆的能耗公平性。     

基于广电5G的边缘云计算平台的建设实践

通过5G边缘云计算平台的部署以及云网边协同、边边协同等模式,创新地实现了在5G领域内分布式云计算架构的落地,既能发挥规模降本效应,又能提供规范化、标准化、高效化的技术服务,充分发挥边缘云计算平台快速部署、弹性扩容等核心技术特点,具备良好的技术创新和业务模式创新。     

基于通信云和承载网协同的算力网络编排技术

面向未来网络中计算与网络紧密结合、"算网一体"的技术发展趋势,提出了基于集中式和分布式两种控制方案的算力网络编排模型,并分别介绍了实现过程的关键技术。从方案与技术分析来看,基于电信运营商通信云和承载网协同的算力网络编排方案可以较好地适应未来移动边缘计算(MEC)站点成网后边边协同与云边协同的业务需求,增强了网络对业务的感知与调度能力,而集中式或分布式控制方案的具体选择与运营商通信云能力和承载网的演进阶段密切相关。     

Congestion‐aware multiaccess edge computing collaboration model for 5G

In 5G cloud computing, the most notable and considered design issues are the energy efficiency and delay. The majority of the recent studies were dedicated to optimizing the delay issue by leveraging the edge computing concept, while other studies directed its efforts towards realizing a green cloud by minimizing the energy consumption in the cloud. Active queue management‐based green cloud model (AGCM) as one of the recent green cloud models reduced the delay and energy consumption while maintaining a reliable throughput. Multiaccess edge computing (MEC) was established as a model for the edge computing concept and achieved remarkable enhancement to the delay issue. In this paper, we present a handoff scenario between the two cloud models, AGCM and MEC, to acquire the potential gain of such collaboration and investigate its impact on the cloud fundamental constraints; energy consumption, delay, and throughput. We examined our proposed model with simulation showing great enhancement for the delay, energy consumption, and throughput over either model when employed separately.     

基于IMS技术的三网融合网络多媒体业务部署

提出了基于云计算架构的三网融合网络多媒体视频业务部署技术,解决了三网融合网络海量终端用户访问及多媒体业务及用户的管理等问题.给出了三网融合网络的多媒体视频业务管理模型,并且详细说明了多媒体视频应用服务器及客户端的功能模块,以实例方式说明了QoS策略决策过程.为三网融合网络多媒体业务部署提供了指导性的技术方案.     

Research on cloud computing users’ public safety trust model based on scorecard-random forest

Traditional cloud computing trust models mainly focused on the calculation of the trust of users’ behavior.In the process of classification and evaluation,there were some problems such as ignorance of content security and lack of trust division verification.Aiming to solve these problems,cloud computing users’ public safety trust model based on scorecard-random forest was proposed.Firstly,the text was processed using Word2Vec in the data preprocessing stage.The convolution neural network (CNN) was used to extract the sentence features for user content tag classification.Then,scorecard method was used to filter the strong correlation index.Meanwhile,in order to establish the users’ public safety trust evaluation model in cloud computing,a random forest method was applied.Experimental results show that the proposed users’ public safety trust evaluation model outperforms the general trust evaluation model.The proposed model can effectively distinguish malicious users from normal users,and it can improve the efficiency of the cloud computing users management.     

Kubernetes容器云平台多租户方案研究与设计

在容器云平台中,租户共享底层的计算、存储、网络等资源,存在租户容器运行和数据安全问题。分析了 Kubernetes 访问控制和资源隔离实现方案基础上,提出了一种基于多租户访问控制模型的容器云平台多租户方案,涵盖多租户管理模型、多租户访问控制、计算资源隔离和网络资源隔离等,可切实提升基于Kubernetes的容器云平台的资源隔离能力,有效降低数据安全隐患。