【作者】 王昕；

【导师】 许胤龙；

【作者基本信息】 中国科学技术大学 ， 计算机软件与理论， 2017， 博士

【摘要】 随着通信技术的发展和移动设备的普及,因特网及其应用已经深入到了社会的各个领域并带来了巨大的经济效益。在因特网中,接入供应商提供的接入与传输服务实现了终端用户和内容供应商之间的数据通信。近年来,由于在线视频流等数据密集型应用的兴起,因特网中的流量呈现爆炸式增长并引起了严重的网络拥塞。为了保证服务质量和用户体验,接入供应商急需升级它们的网络设施和网络容量。然而,很多接入供应商没有足够的收益和动机升级网络。一个重要的原因是这些接入供应商对其接入与传输服务缺乏有效的定价策略,以至于因特网发展带来的红利大部分落入了内容供应商的口袋。因此,针对接入与传输服务制定有效的定价策略对于优化接入供应商的收益、帮助其改善服务质量以及维持因特网的快速发展非常重要。此外,不同于实物商品,接入与传输服务的质量——通常体现在网络的拥塞程度上——受到其价格的影响。具体地,价格会影响终端用户的数据流量以及接入供应商的网络负载,而网络负载又会影响网络的拥塞程度。这种价格对质量的作用导致接入与传输服务与实物商品往往有不同的定价策略,因而需要被专门研究。在本文中,我们考虑三种常用的接入与传输服务的计费模式,并研究了在这些计费模式下接入供应商的定价策略。第一种是两部计费模式,即接入供应商预设一个流量包。当终端用户的数据流量低于流量包时,接入供应商只收取固定的基本费用;而当终端用户的数据流量超过流量包时,对超过的部分按照每单位流量计费。第二种是双边计费模式,即接入供应商同时对终端用户和内容供应商这两边进行收费;第三种是差异化计费模式,即接入供应商向内容供应商提供不同质量和价格的多种传输服务。对于每种计费模式,我们研究了两类最优定价策略。第一类是最大化接入供应商利润(或收益)的定价策略;第二类是最大化社会福利的定价策略。而针对每一类最优定价策略,我们又从两个层面进行了分析。第一个层面是在因特网参与者的特征如接入供应商的网络容量、终端用户的流量需求和对网络拥塞的敏感度等给定时,我们刻画了最优定价策略的结构;第二个层面是在因特网参与者的特征改变时,我们描绘了最优定价策略的变化。我们使用建模分析的方法研究利润最优和福利最优的定价策略的结构及变化。首先,我们建立模型描述在不同的接入与传输服务的质量和定价下终端用户和内容供应商的行为。我们用网络的拥塞程度来反映服务的质量。由于网络的拥塞程度是市场的内生量(即不仅仅依赖于接入供应商的网络容量,还取决于终端用户的数据流量),我们引入“平衡拥塞”的概念来刻画内生的拥塞程度。然后,基于平衡拥塞模型,我们利用优化理论和敏感性分析等数学工具研究了两类最优定价策略的结构和变化。我们还使用数值模拟对最优定价策略进行了评估。我们的研究结果表明由于网络的拥塞程度具有内生性,接入与传输服务和实物商品往往有不同甚至截然相反的最优定价策略。例如,对于实物商品,当供应商的供应能力增强时,利润最优价格一定会降低;而对于接入与传输服务,当接入供应商的网络容量增大时,根据网络流量的类型(如在线视频流或文本内容)及它们的拥塞敏感度的不同,利润最优双边价格可能降低也可能升高。我们的研究结果还指出了接入与传输服务的定价策略随着因特网发展的变化趋势。例如,随着终端用户的流量需求和对网络拥塞的敏感度的不断增大,优化利润和社会福利的目标都会更强烈地驱使定价策略从固定费率向基于流量计费转变以及从单边计费向双边计费转变。我们相信,我们的研究结果可以帮助接入供应商设计有效的定价策略以增加利润并且指导监管部门制定监管政策以保护社会福利,进而促进因特网的持续健康发展。更多还原

【Abstract】 With the development of communication technology and the popularity of mo-bile devices, Internet and its applications have penetrated into all areas of society and brought great economic benefits. In the Internet, access and transit services offered by access providers (APs) enable end-users and content providers (CPs) to transmit data to each other. In recent years, due to the rise of data-intensive applications such as online video streaming, Internet traffic has been growing fast, causing serious network congestion. To guarantee the service quality and user experience, APs need to upgrade network infrastructures and capacities. However, many APs do not have enough rev-enues and incentives to upgrade their networks. Because the APs lack effective pricing strategies for the access and transit services, and most of the dividends brought by the development of the Internet go into the CPs’ pockets. Therefore, it is important to de-sign effective pricing strategies for the access and transit services to optimize the APs’revenues, help them improve the service quality, and keep the rapid development of the Internet. Besides, unlike physical commodities, the quality of access and transit service of an AP ——reflected in the congestion level of network ——is influenced by its price. In particular, the congestion level depends on the traffic usages of end-users and network load of the AP, which is impacted by the price of access and transit service. This price effect on quality makes the pricing strategies of access and transit services different from those of physical commodities, and therefore need to be studied exclusively.In this thesis, we consider three commonly used charging modes for access and transit services and study the APs’ pricing strategies under these modes. The first is the two-part charging mode, that is, an AP introduces a data cap. If a user’s traffic us-age is below the data cap, the AP only collects a fixed lump-sum fee; otherwise, extra charges are imposed on the usage above the data cap with a per-unit usage fee. The second is the two-sided charging mode. Under this mode, APs charge end-users and CPs simultaneously for data traffic between them. The third is the differentiated charg-ing mode, that is, APs provide CPs multiple transit services with different qualities and prices. For each charging mode, we study two types of optimal pricing strategies, which maximize the APs’ profits (or revenues) and social welfare, respectively. Furthermore,we analyze each type of optimal pricing strategy at two levels. The first level is when the characteristics of Internet participants, e.g., APs’ capacities and users’ traffic de-mand and congestion sensitivity, are fixed, we characterize the structures of the optimal pricing strategies. The second level is when the characteristics of Internet participants change, we describe the changes of the optimal pricing strategies.We study the structures and changes of the profit-optimal and welfare-optimal pric-ing strategies via modeling analysis. We first build a model to capture the behavior of end-users and CPs under varying quality and pricing parameters of access and transit services. We use the congestion levels of networks to measure the qualities of access and transit services. Since the congestion levels are endogenous variables of markets(i.e.,they depend not only on the APs’ network capacities but also on the end-users’traffic usages), we introduce an "equilibrium congestion" to characterize the endoge-nous congestion level. Based on the equilibrium model, we then use mathematical tools such as optimization theory and sensitivity analysis to study the structures and changes of the two types of the optimal pricing strategies. We also evaluate the pricing strategies by numerical simulations.Our results show that since the network congestion level has endogenous, access and transit services often have different or even opposite optimal pricing strategy from physical commodities. For example, for physical commodities, when providers increase their supply capacities, the profit-optimal prices must reduce. However, for access and transit services, when APs extend their network capacities, according to the types of net-work traffic (such as online video or text content) and their congestion sensitivities, the profit-optimal two-sided prices may either raise or reduce. Our results also point out the tendencies of the pricing strategies of access and transit services with the development of the Internet. For example, as end-users’ traffic demand and congestion sensitivity continue to increase, the targets of maximizing profit and social welfare will both drive the pricing strategies to shift from flat-rate to usage-based pricing and from one-sided to two-sided pricing more strongly. We believe that our results can help APs to design effective pricing strategies for optimizing their profits and help regulators to make reg-ulatory policies for protecting social welfare, which further promote the sustainable and healthy development of the Internet.更多还原