无线移动自组网下主动性流媒体认知服务研究

发布时间：2018-05-09 21:42

本文选题：视频点播 + 移动对等网络　；参考：《北京邮电大学》2014年博士论文

【摘要】：在无线移动自组网(Mobile ad hoc Networks, MANETs)中提供基于对等网络(Peer-to-Peer, P2P)的具有主动性流媒体特性的视频点播(Video-on-Demand, VoD)服务,以支持大规模移动用户主动和交互式的访问视频资源,已经成为未来移动流媒体服务的关键技术之一。然而,在MANETs中部署P2P VoD系统却面临着巨大的挑战：(1)VoD服务允许用户异步接入和主动变更当前播放内容,造成P2P网络中节点间逻辑连接抖动程度高,严重影响用户播放的连续性。(2)现有P2P技术存在组网策略僵化、网络维护负载高和查询效率低下的问题,难以满足移动环境下对资源高效管理与分发的要求。(3)现有移动网络环境变化感知策略严重匮乏,移动节点不具备自适应动态网络环境的能力,节点间协作机制相对静态,使得网络中可用视频资源短缺、资源获取成本高和分享效率低下。针对以上问题,本文对用户播放行为感知、视频资源分发体系和节点移动性感知等三个关键技术进行了深入研究,取得以下成果： (1)设计了面向可靠性的用户播放行为感知和预测模型。感知具有主动性和交互式等特性且能够反映当前和未来用户对视频内容兴趣变化、资源请求和使用抖动程度的用户播放行为,是主动性流媒体服务研究的认知驱动源的基础。本文首先提出了资源需求预测算法,精确预测用户未来播放内容,并预先下载到本地,确保用户播放连续性；其次,提出了用户播放状态评估模型,基于历史播放内容评估用户播放行为的可靠性,确保资源提供者与接收者间逻辑连接稳定性。 (2)提出了基于移动对等网络(Mobile P2P, MP2P)的视频资源管理和分发机制。由于移动设备存储能力相对有限和动态变化的资源需求对MP2P构建和资源分发策略产生严重影响。本文首先提出了MP2P的组网策略和资源搜索算法,根据用户历史与当前播放内容建立基于微社区的MP2P结构,利用用户播放行为预测模型,动态建立微社区间逻辑关联,实现资源快速查找;其次,提出了MP2P的维护策略,通过维护具有可靠播放行为的节点,以降低维护成本。 (3)设计了基于网络环境动态变化感知的资源分享和协作获取机制。移动网络动态的拓扑结构造成连接不可靠、数据传输性能低下,影响用户平滑的播放体验。本文首先设计了基于跨层感知和邻居协作的视频资源分享机制,通过感知一跳邻居通信质量变化,设计了资源获取代价评估模型,建立以资源为中心的簇结构,提高移动环境下资源分享效率和传输性能；其次,设计了一跳邻居移动稳定性和通信质量评估与预测模型,利用移动稳定性高且通信质量好的节点协助下载预测的播放内容,减少节点移动性和通信质量变化对视频数据传输的影响。
[Abstract]:In wireless mobile ad hoc Networks, MANETs), video-on-demand (VOD) services with active streaming characteristics based on peer-to-peer (P2P) network are provided to support active and interactive access to video resources by large-scale mobile users. It has become one of the key technologies of mobile streaming media service in the future. However, the deployment of P2P VoD system in MANETs is facing a great challenge. The service allows users to access and actively change the current playback content asynchronously, resulting in a high degree of jitter of logical connections between nodes in P2P networks. The existing P2P technology has some problems such as rigid networking strategy, high network maintenance load and low query efficiency. It is difficult to meet the requirements of efficient resource management and distribution in mobile environment. (3) the existing mobile network environment awareness strategy is seriously lacking, mobile nodes do not have the ability of adaptive dynamic network environment, and the cooperation mechanism between nodes is relatively static. It causes the shortage of available video resources, the high cost of resource acquisition and the low efficiency of sharing. In order to solve the above problems, three key technologies of user playing behavior perception, video resource distribution system and node mobility perception are studied in this paper. The following results are achieved: 1) A reliability-oriented model for user playback behavior perception and prediction is designed. Perception has the characteristics of initiative and interaction, and it can reflect the changes of current and future users' interest in video content, resource request and user playing behavior with jitter degree. It is the basis of cognitive driving source of active streaming media service research. In this paper, we first propose a resource demand prediction algorithm to accurately predict the user's future playback content, and download it to the local area in advance to ensure the continuity of user playback. Secondly, a user playback status evaluation model is proposed. The reliability of user playback behavior is evaluated based on historical playback content to ensure the stability of logical connection between the resource provider and the receiver. A video resource management and distribution mechanism based on Mobile P2P (MP2P) in mobile peer-to-peer networks is proposed. Due to the relatively limited storage capacity of mobile devices and the dynamic changes in resource requirements, MP2P construction and resource distribution strategies are seriously affected. In this paper, we first put forward the strategy of MP2P networking and resource search algorithm. According to the user history and current content, we build a MP2P structure based on microcommunity, and use user playback behavior prediction model to dynamically establish logical association between microcommunities. Secondly, the maintenance strategy of MP2P is proposed to reduce the maintenance cost by maintaining nodes with reliable playback behavior. 3) the mechanism of resource sharing and collaborative acquisition based on dynamic change perception of network environment is designed. The dynamic topology of mobile network results in unreliable connection and low performance of data transmission, which affects the user's smooth playing experience. This paper first designs a mechanism of video resource sharing based on cross-layer awareness and neighbor cooperation. By sensing the quality change of one-hop neighbor communication, a cost evaluation model of resource acquisition is designed, and a resource-centric cluster structure is established. Secondly, a one-hop neighbor mobile stability and communication quality evaluation and prediction model is designed to use the nodes with high mobile stability and good communication quality to help download the predicted playback content. Reduce the influence of node mobility and communication quality change on video data transmission.
【学位授予单位】：北京邮电大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN929.5

【参考文献】