多覆盖网络共存环境下资源竞争理论与算法研究

发布时间：2018-10-22 08:29

【摘要】：随着互联网的不断发展和普及,越来越多的用户可以通过互联网享受到各种各样的服务。然而用户对于网络服务质量的要求越来越高,他们需求的服务种类也越来越丰富。传统的互联网面临着越来越严重的僵化问题,无法满足用户不断膨胀的服务需求。覆盖网络作为改善传统互联网的重要手段,具有在不改变原有网络的前提下提供新型网络功能的特性。因此,覆盖网络常常被部署在传统网络上,用来提供各种各样的服务,例如对等网络服务、组播服务、内容分发网络以及服务质量增强等。不仅如此,覆盖网络在新型网络上也有所应用,例如软件定义网络等。随着覆盖网络的大量部署,一个传统网络上构建着多个覆盖网络的情况越来越常见。这些覆盖网络往往是由不同服务提供商独立部署的。因此,它们不仅无法相互感知到对方的存在,而且只会自私地优化自己的表现。这种覆盖网络的自私性会导致共存的覆盖网络对底层的网络资源展开竞争,从而恶化网络环境,进而降低各自的表现。本文分析了共存覆盖网络资源竞争导致的性能损失问题。面对目前覆盖网络合作方案的诸多不足,以促进合作为指导思想,同时兼顾覆盖网络的个体利益以及覆盖网络合作的公平性,针对覆盖网络间合作关系,对同构/异构覆盖网的合作路由方案进行了研究,提出了改善覆盖网络间恶性竞争的解决方案。主要创新工作包括:(一)提出了覆盖网络自发合作激励机制。针对覆盖网络路由行为的自私性,该方案着重于避免覆盖网络在路由行为上的相互背叛。覆盖网络进行资源竞争时,各自路由的相互影响是重复性的。因此,我们基于重复博弈模型,将对背叛者的惩罚策略引入到覆盖网络路由策略中,使得产生背叛行为的覆盖网络长期利益受损。这样避免了覆盖网络的背叛,稳定了覆盖网络之间的合作关系。(二)提出了关注个体利益的合作路由算法。该算法关注于覆盖网络的个体利益,在不降低任何覆盖网络性能的前提下,达到改善网络整体环境的目的。通过对统一路由目标进行分布式的拆分,共存覆盖网络可以独立计算各自的路由。这样就减少了该合作方案实现的复杂度,提高了覆盖网络路由效率。该算法避免了传统的合作路由策略中导致部分覆盖网络表现下降的缺点,使参与合作的覆盖网络实现了双赢。(三)提出了基于联盟的覆盖网络合作方案。在多覆盖网络环境下,覆盖网络之间的利益关系十分复杂。部分覆盖网络可以通过合作获得双赢,而部分覆盖网络则具有不可调和的利益冲突。该合作方案使愿意合作的覆盖网络组成群体,群体内的覆盖网络进行局部合作。合作方案提供了针对群体内覆盖网络的利益分配算法,根据覆盖网络对群体的贡献度来分配它们的利益。我们也分析了群体组成的过程,并提供了快速收敛算法。该方案在保证覆盖网络个体间公平性的基础上,通过覆盖网络间的局部合作,使覆盖网络的收益大大增加。(四)提出了异构覆盖网络的合作方案。该方案引入了软件定义网络的概念来构建覆盖网络,解决了异构覆盖网络之间无法相互理解的问题。在此基础上,提出了异构覆盖网络路由算法。异构覆盖网络通过分布式的路由算法,可以协调彼此的流量安排,从而合理地利用底层网络资源。针对异构覆盖网络的表现需求不一致的问题,我们设计了一个权值平衡策略,使覆盖网络的个体利益得到保证。我们还提出了快速收敛算法,使该合作方案具有较高的效率。
[Abstract]:With the development and popularization of the Internet, more and more users can enjoy various kinds of services through the Internet. However, the user's demand for network quality of service is getting higher and higher, and the kind of service they need is becoming more and more abundant. The traditional Internet is faced with more and more serious rigidity problems, and can not meet the demand of continuous expansion of users. As an important means of improving the traditional Internet, the overlay network has the characteristics of providing new network functions without changing the original network. Thus, overlay networks are often deployed on conventional networks to provide a wide range of services, such as peer-to-peer networking services, multicast services, content distribution networks, and quality of service enhancements, among others. Furthermore, overlay networks are also applications on a new network, such as software-defined networks, etc. With the large deployment of overlay networks, multiple overlay networks are increasingly common across a traditional network. These overlay networks are often independently deployed by different service providers. Therefore, they can not only perceive the existence of each other, but also selfish to optimize their performance. The selfish nature of this overlay network can result in co-existing overlay networks competing against underlying network resources, thereby deteriorating the network environment, thereby reducing their performance. This paper analyzes the problem of performance loss caused by coexistence overlay network resource competition. In order to promote cooperation as the guiding ideology, we study the cooperative routing scheme of isomorphic/ heterogeneous overlay network in order to cover the individual interests of the network and to cover the fairness of network cooperation, and to cover the cooperative relation between networks. A solution to improve the vicious competition between overlay networks is proposed. The main innovation work includes: (1) putting forward the incentive mechanism to cover the spontaneous cooperation of the network. To cover the selfishness of network routing behavior, the scheme focuses on avoiding the mutual betrayal of overlay networks in routing behavior. When the overlay network performs resource contention, the mutual influence of the respective routes is repetitive. Therefore, based on the repeated game model, we introduce the penalty strategy of back-to-back to cover the network routing strategy, so that the long-term interests of the covering network that generate the betrayal behavior are damaged. This avoids the betrayal of the overlay network and stabilizes the cooperative relationship between overlay networks. (2) A cooperative routing algorithm is proposed to pay attention to individual interests. The algorithm focuses on covering the individual interests of the network, and achieves the aim of improving the whole network environment without reducing the performance of any overlay network. By carrying out distributed splitting on the unified routing target, the coexistence overlay network can independently compute the respective routes. Thus, the complexity of implementation of the cooperation scheme is reduced, and the routing efficiency of the overlay network is improved. The algorithm avoids the disadvantage of partial coverage network performance degradation in traditional cooperative routing strategy, and makes the overlay network participating in cooperation achieve win-win. (3) A coalition-based coverage network cooperation programme was proposed. In multi-coverage network environment, the relationship between overlay networks is very complicated. Some coverage networks can gain win-win through cooperation, while partial coverage networks have an irreconcilable conflict of interest. The cooperation programme makes local cooperation with coverage networks in groups and groups that are willing to cooperate. The cooperative scheme provides an allocation algorithm for the benefit of the coverage network within a group, and allocates their interests according to the contribution degree of the coverage network to the group. We also analyze the process of population composition and provide a fast convergence algorithm. By covering the local cooperation among networks, the scheme can greatly increase the revenue of covering the network. (4) A cooperative scheme of heterogeneous overlay networks is proposed. The scheme introduces the concept of the software-defined network to construct the overlay network, and solves the problem that the heterogeneous overlay network cannot understand each other. On the basis of this, a heterogeneous overlay network routing algorithm is proposed. Heterogeneous overlay networks can coordinate traffic arrangements with each other through distributed routing algorithms, thereby reasonably utilizing underlying network resources. Aiming at the problem that the performance demands of heterogeneous overlay networks are inconsistent, we have designed a right balance strategy to ensure the individual benefits of covering the network. We also propose a fast convergence algorithm which has higher efficiency.
【学位授予单位】：北京邮电大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP393.0

【相似文献】