基于SDN的边缘缓存技术研究与原型验证

发布时间：2018-04-23 09:10

  本文选题：SDN + 边缘网络　； 参考：《电子科技大学》2017年硕士论文

【摘要】：随着车联网、智慧城市、虚拟现实等新型网络应用的兴起,边缘无线网络中的服务类型不断增多。不同服务对时延的要求各不相同,从而给边缘缓存技术带来了新的挑战。本论文基于软件定义网络(SDN)来构建一种面向低时延的边缘无线网络缓存系统,旨在减少用户的访问时延和下载时延。其中,访问时延是指用户访问存储有所需内容的节点需要花费的时间,该时延主要由内容的存储位置来决定;下载时延则表示内容从缓存节点传输回用户需要的时间,该时延主要由与用户关联的基站所能提供的最大传输速率(即吞吐量)来决定。因此,本文主要围绕用户关联和缓存决策两个方面进行研究。(1)基于负载均衡的关联决策模型。传统的基于RSSI的关联决策,用户仅从自身角度出发选择基站进行关联,并未考虑基站的负载以及网络的整体性能,这无疑会降低网络的系统吞吐量,增加用户的平均下载时延。因此,关联决策不仅需要保证基站负载的相对均衡,还需要从无线网络的整体出发,最大化系统吞吐量。基于该目标,本文提出了基于负载均衡的关联决策模型。然而,随着网络规模的扩大,关联决策模型的求解难度较大。为了降低求解难度,本文提出了基于KLA的关联算法,即通过引入图划分来创建虚拟基站。因此,用户首先需要接入虚拟基站,随后在各个虚拟基站内与具体的物理基站相关联。(2)基于网络切片的缓存决策模型。边缘无线网络中通常存在多个服务提供商,不同服务提供商对访问时延的要求各不相同。因此,缓存决策需要在保证不同服务提供商对访问时延要求的同时,减少网络内用户的平均访问时延,从而提高用户体验。基于该目标,本文通过引入网络切片对缓存资源进行虚拟化,来满足不同服务的访问时延要求,并提出了基于网络切片的缓存决策模型。同时,本文证明了缓存决策模型具备NP完全性,并针对缓存决策模型的特点设计出了一种启发式算法来求解。基于上述的研究内容,本文还进行了大量的仿真来验证评估基于负载均衡的关联决策模型和基于网络切片的缓存决策模型各方面的性能。最后,本文使用车载无线服务器对整个缓存系统进行了原型验证。
[Abstract]:With the rise of new network applications, such as car networking, intelligent city, virtual reality, the types of services in edge wireless networks are increasing. Different services have different requirements for delay, which brings new challenges to edge caching technology. Based on the software definition of SDN, this paper constructs a low delay oriented edge wireless network cache system, which aims to reduce the user's access delay and download delay. Among them, the access delay refers to the time spent by the user to access the node with the required content, which is mainly determined by the storage location of the content, while the download delay indicates the time that the content is transferred back to the user from the cached node. The delay is mainly determined by the maximum transmission rate (i.e. throughput) that the base station associated with the user can provide. Therefore, this paper focuses on two aspects: user association and cache decision. In the traditional association decision based on RSSI, the user chooses the base station to associate only from its own point of view, without considering the load of the base station and the overall performance of the network, which will undoubtedly reduce the system throughput of the network and increase the average download delay of the user. Therefore, the association decision not only needs to ensure the relative balance of the base station load, but also needs to maximize the system throughput from the wireless network as a whole. Based on this goal, this paper presents a load-balancing-based decision-making model. However, with the expansion of network scale, it is difficult to solve the associated decision model. In order to reduce the difficulty of solving the problem, this paper proposes an association algorithm based on KLA, which is to create virtual base station by introducing graph partition. Therefore, the user needs to access the virtual base station first, and then associate with the physical base station in each virtual base station. There are usually multiple service providers in edge wireless networks. Different service providers have different requirements for access delay. Therefore, cache decision-making needs to reduce the average access delay of users in the network while ensuring that different service providers require access delay, so as to improve the user experience. Based on this goal, this paper introduces network slicing to virtualize cache resources to meet the access delay requirements of different services, and proposes a cache decision model based on network slicing. At the same time, this paper proves that the cache decision model has NP completeness, and designs a heuristic algorithm to solve the problem according to the characteristics of the cache decision model. Based on the above research, a large number of simulations are carried out to verify and evaluate the performance of the load balancing based decision model and the cache decision model based on network slicing. Finally, this paper uses the vehicle wireless server to verify the whole buffer system prototype.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN915.0

【参考文献】

中国期刊全文数据库 前1条

1 许阳;高功应;王磊;;5G移动网络切片技术浅析[J];邮电设计技术;2016年07期

，

本文编号：1791322