融合型内容分发网络关键技术研究

发布时间：2018-04-01 15:21

本文选题：覆盖网　切入点：内容分发网络　出处：《哈尔滨工业大学》2014年博士论文

【摘要】：Internet流量指数性增长以及用户对网络速度的需求已促使网络资源分发的可用性、可靠性等成为互联网技术的关键问题。内容分发网络作为运行于整个互联网上的覆盖网，主要致力于互联网中的服务质量优化。随着云计算、移动互联网及动态网络内容技术的推进，内容分发技术逐步趋向于专业化、定制化，在内容路由、管理、推送以及安全性都面临新的挑战。本文从运营商角度出发，考虑流量结算成本、提升数据传输服务质量以及安全性三个方面对融合型CDN（Content Delivery Networks）进行研究。首先，考虑ISP（Internet Service Provider）的域内用户访问域外网络资源所产生的大量可存储数据由于被大量多次访问从而造成域间流量成本骤增，同时由于域内链路重复流量也导致传输服务的下降。针对该问题，本文首先提出针对ISP-CDN的内容节点两层部署机制，该部署机制分为两层，外层部署于域间出入口以降低域间流量，内层的部署策略采用基于贪心的网络流量代价算法对核心节点进行选取。并在此基础上设计基于RCBF（Record Counter Bloom Filter）的索引结构与同步机制，从而改进CDN的内容路由策略，提升融合型CDN的服务性能。第二，本文进一步研究内容数据的管理问题。由于内容节点位置固定后，如何对所存储内容进行管理是内容分发技术中的重要问题。内容管理主要分为两个方面：缓存管理与内容推送。本文首先提出基于访问间隔的替换策略，提高热度高的缓存对象在缓存空间中的生命周期；其次考虑到这些被访问的数据均依赖于用户自身的行为，在协议层对数据进行深入分析，提出多维度的缓存对象热度评价算法，进一步提升缓存命中率；最后提出一种基于价值度的热点数据推送策略，在所部署的节点间进行有效的内容预分发，降低用户的访问延时，，改善网络性能。第三，融合型CDN主要面对ISP域内用户，为了提升服务质量，降低CDN所面对的潜在网络威胁，如何结合ISP所拥有的拓扑和用户的数据优势，建立域内用户的可信机制是融合型CDN所面临的全新问题。本文提出基于信任的网络群体分类模型，结合对ISP所获取的用户行为数据，实现对用户的自动化分类，并采用传统检测方法设置用户类别可信度，以此构建可信融合型CDN。该模型首先使用网络交互拓扑信息建立网络节点间的信任矩阵；进而结合直接信任度和相关信任度计算网络节点间的相似度，进行用户分类；最后根据可信度对不同类别的用户进行差异化服务，提升低可信度的攻击代价，从而降低CDN所面临的网络风险。最后，本文提出了一种ISP融合的内容分发系统，该系统与传统的CDN系统有一定的区别：1、被动解析存储与层次化部署；2、安全预警。整个系统主要分为四个模块：信息采集、缓存管理、节点通信与离线审计。通过在实际系统中对本文提出的内容部署、管理及用户分类进行具体测试，实现了本文最初的研究目标。
[Abstract]:The availability of Internet flow index and user demand for network speed has prompted the cyber source distribution, reliability has become the key problem of Internet technology. The content distribution network as the overlay network running on the Internet, the Internet is committed to the quality of service optimization. With cloud computing, mobile Internet and dynamic web content the advance of technology, technology tends to specialization, content distribution customization, management, push content routing, and security are facing new challenges. This paper from the view of operators, considering the traffic cost, enhance the data transmission quality of service and security of the three aspects of the integration of CDN (Content Delivery Networks) to study.
First, consider the ISP (Internet Service Provider) a large number of data can be stored by a large number of visits resulting in inter domain traffic surge in the cost of intra domain users to access foreign cyber source generated by the intra domain link repeat traffic but also lead to a decline in transmission service. Aiming at this problem, this paper puts forward the content node ISP-CDN two layer deployment mechanism of the deployment mechanism is divided into two layers, the outer layer is deployed in the inter domain entrance to reduce inter domain traffic, inner deployment strategy based on greedy algorithm of network traffic cost for core node selection. And on the basis of this design based on RCBF (Record Counter Bloom Filter) index structure and synchronization mechanisms. In order to improve the content of CDN routing strategy, improve the service performance of the fusion of the CDN.
Second, this paper further studies the problem of management of content data. Due to the location of the content nodes is fixed, how to carry on the stored content management is an important problem in the content distribution. Content management is mainly divided into two aspects: buffer management and content push. This paper proposed replacement strategy based on access interval, improve cache object heat high in the cache space in the life cycle; secondly considering the accessed data are dependent on the user's behavior in the protocol layer for data analysis, the algorithm proposed hotnessevaluation cache object with multi dimensions, to further improve the cache hit rate; finally a hot data value based on the degree of push strategy the content, distribution in the pre effective deployment of nodes, reduce user access latency and improve network performance.
Third, the main face of fusion type CDN user ISP domain, in order to improve service quality, reduce the potential network threats facing CDN, how to combine ISP with topology and user data, establishing the mechanism of trusted domain user is a new problem faced by CDN fusion. This paper proposes classification of trust model of network group based on the combination of user behavior data obtained in the ISP, realize the automation of the customer classification, and the traditional detection method for setting user categories in order to build credibility, credible fusion CDN. the model first use of network interaction topology information to establish the trust matrix between nodes of the network; and then combined with the calculation of similarity between network nodes and direct trust trust degree of the user classification; according to the credibility of different categories of users differentiated services, enhance the credibility of the low price to attack. Reduce the network risk that CDN faces.
Finally, this paper proposes a combination of ISP content distribution system, the system with the traditional CDN system has certain difference: 1, analysis of passive storage and hierarchical deployment; 2, safety warning. The whole system consists of four modules: information collection, cache management, node communication and off-line through the proposed audit. The content in the actual system deployment, management and user classification of specific test, to achieve the goal of this research first.

【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】