【作者】 倪锐；

【导师】 卫国；

【作者基本信息】 中国科学技术大学 ， 通信与信息系统， 2011， 博士

【摘要】 本文以业务域(包括语音、视频、音频、文本、图像等)为研究对象,构筑了一套完整的业务理论框架体系。本文的研究既可以从理论上为无线网络的业务提供性能评价指标,又可以在工程实践中完整地给出端到端的业务服务质量和用户体验质量的分析方法。本文的研究成果和技术思路可以为未来的移动互联网提供来自业务域视角的指导性建议。本文首先对各种常见的业务类型进行枚举、归纳和分类,并根据业务模型的不同用途,分别有针对性地提出了网络规划业务模型、排队论业务模型,源发生器业务模型和一般意义上的业务模型。这些模型的表达形式、参数、含义、用途各不相同。由于不同领域和不同层面对业务模型的需求差异太大,本文未能提出一个统一的闭式表达的数学公式来刻画业务。以业务模型为基本颗粒,本文提出业务容量的概念,并将其作为无线网络对业务承载能力的评价指标。业务容量是用户体验加权的用户接收速率之和的上确界。它既具有类似于网络吞吐量的效果,又具有反映用户体验的能力,还可以刻画组播、广播等通信形式对网络性能的影响。本文的研究结果表明：无线业务的速率波动程度(突发性和自相似性)、到达离开特征、上下行的不对称性、业务用户间的距离(多跳)、组播系数这几个因素对无线网络的业务容量影响最大。为了提前预测业务在未来一段时间内的服务质量,进而提前计算出业务容量来作为网络管理和调度的指导信息,本文进一步展开了业务流的性能分析。以“点-线-面”的逻辑思路,本文分别介绍和提出了单一业务流一跳无线链路上的性能分析、单一业务流多跳串联链路上的性能分析,以及多条并行业务流竞争一个公用网络的性能分析。基于这三个分析,本文给出了一套完整的预测分析端到端业务流服务质量的方法。以业务模型来刻画无线业务，以业务容量来评价无线网络,以业务流分析来预测和计算业务的服务质量。这三者合在一起,就是本文研究的从业务域出发,指导下一代无线网络规划和优化的理论体系。此外,本文还分析了无线业务相对于固定业务的优势和特征,展望了未来移动互联网的杀手级业务,讨论了无线业务可能的商业模式和定价策略。本文认为,在技术性因素之外,价格机制不失为一种调整网络流量、引导用户行为的好方法。更多还原

【Abstract】 This paper took service domain as the research object, and built a theoretical framework for the communication services. This paper not only proposed a performance evaluation of wireless network in theory, but also gave an analytical method of end-to-end service quality and user experience in engineering practice. This paper’s technical ideas and results could be used to guide the development of future mobile Internet.Firstly, this paper put various types of services into enumeration, induction and classification. And then, this paper proposed four kinds of service model respectively target to network planning, queuing theory, source generator and general sense understanding. These service models have significant differences in expression, parameter, meaning and purpose. Because of the wide difference of service model in different areas and layers, this paper failed to find a unified mathematical closed expression to describe the service model.Taking service model as the basic particles, this paper proposed the concept of service capacity, and put it as wireless network’s performance evaluation of service supporting. Service capacity is the upper bound of user experience weighted receiver rate sum. It has a similar effect of network throughput, also has the ability to reflect the user’s experience, still can describe the multicast and broadcast influence to network performance. Our research shows that:the volatility rate of wireless service (burst and self-similarity), the service arrival and leave features, the asymmetry of the uplink and downlink, the distance between users (multi-hop), the multicast coefficient, and these five factors are the greatest impact for service capacity.In order to predict the quality of service (QoS) in the following period and put this predict to guide the network management and schedule, this paper further expanded the service flow performance analysis. Following the "point-line-surface" logic, this paper made a series of studies, including one-hop wireless link single service flow analysis, multi-hop single service flow analysis and parallel service flows analysis. Based on the three analyses, this paper proposed a complete methodology for end-to-end QoS prediction of service flow.The service model is used to describe wireless service. The service capacity is used to evaluate wireless network. The service flow analysis is used to predict and calculate the QoS. Integrating three tools together, it is the theoretical system to guide the next generation wireless network’s planning and optimization from service domain view.In addition, this paper also talked about wireless service’s features respect to fixed service, looked forward to the future mobile killer service, and discussed the possible commercial model and pricing strategy for wireless service. In addition to technical factors, the price mechanism is a good way to adjust network load and guide users’behavior.更多还原