【作者】 袁俊；

【导师】 张平； 王莹；

【作者基本信息】 北京邮电大学 ， 电路与系统， 2011， 博士

【摘要】 未来的B3G通信网络将是基于IP的异构融合网络。在异构融合网络中,智能化的多模终端能够自由地选择利用一种或同时利用多种不同的网络技术,实现多样化的业务服务,以获得更好的用户体验。这种应用场景下的网络设备、通信技术以及智能化的多模终端等所有元素构成了异构多连接系统。随着网络技术和业务需求的迅速发展,网络服务质量和用户体验的保障技术成为约束异构多连接系统向个性化、智能化方向发展的瓶颈之一。异构多连接系统中基于QoS保障的关键技术已经引起了广泛的关注和研究。选择多样性、连接差异性以及移动广泛性这三方面的特征,给异构多连接系统中的QoS保障都带来了巨大的挑战。本文分别就来自这三方面的挑战提出了面向QoS保障的相关算法和解决方案,包括：单业务单连接承载方式下的连接选择技术、单业务多连接承载下的QoS技术以及多连接承载方式下的移动性管理技术。具体地,本文的研究内容主要包括以下几个方面：首先,对异构多连接系统进行了简要介绍,随后归纳了在异构多连接系统中实现QoS保障所面临的三方面挑战。接着详细列举介绍了国际国内学术界和相关研究组织在应对这三方面挑战的研究现状。基于深入的比较和分析,本文从异构多连接系统的三方面特征出发,提出了相应的改进和完善面向QoS保障的相关技术的解决方案。针对异构多连接系统选择多样性特征：本文基于虚拟目标网络思想,利用MADM理论提出了一种异构多连接系统环境下的连接选择算法,该算法在保证QoS的前提下,选择最适合的连接作为目标接入连接,实现了以尽可能小的资源代价换取业务的连续性以及QOS保证,同时提高了网络资源的利用效率。进一步地,基于信息熵理论提出了能根据实时环境进行动态更新权重计算方法,使该算法更具备实用性。另一方面,当网络信息不完全或者测量不准确或者无法测量部分参数信息时,提出了两种基于加权马尔可夫链的算法对排序融合算法加以改进,从而解决了连接选择时出现参数值缺失和不完全信息的问题。针对异构多连接系统连接差异性特征,本文分别从用户侧和网络侧提出了面向QoS保障的相关解决方案。在用户侧,设计了一种适用于降低端到端多连接并行传输的重排序时延的负载分配算法,通过对负载分配理论值和实际传输值之间的误差和误差门限值之间的对比,根据不同的对比结果采用不同的业务分割方法,以获取QoS保障性能的提升；在网络侧,设计了一种适用于降低端到端多连接并行传输的重排序时延的网络控制方法,通过将决策功能置于网络侧,同时根据包的到达状态,决策并控制发送端数据包的发送,既实现降低重排序时延,同时还为终端节省能耗。针对异构多连接系统移动广泛性特征,本文提出了一种新型的异构网络移动性管理架构,设计了主要由触发模块、多连接管理模块、切换模块等三个逻辑功能模块组成的异构网络下移动性管理架构,并实现了与实际3GPP演进网络架构与协议栈的映射。同时,本部分还提出了一种在多业务场景中使用扩展MIH业务的高效切换方案。所提方案允许即将要切换的接口以与相同MN的其他激活链路协同工作的方式来管理切换问题。论文最后对全文进行了总结,并指出了今后的研究方向。更多还原

【Abstract】 The main structure of future communication systems will be the IP-based heterogeneous network, in which intelligent multi-mode terminals can connect multiple different networks simultaneously to obtain better user experience for diverse services. This kind of heterogeneous multi-link systems is composed of network equipments, communication technologies and intelligent multi-mode terminals. With increasing diverse service requirements, quality of service (QoS) becomes one of the bottlenecks restricting the development of hetergeneours multi-link systems in the direction of personality and intelligenty. Therefore the research on QoS guarantee has received extensive attention.There are three main characteristics in the heterogeneous multi-link system:multiple options of target access network, diversity of link characteristics and mobility. These characteristics have brought significant challenges to QoS guarantee in the heterogeneous multi-link system. This thesis proposes algorithms and solutions for QoS guarantee to tackle these challenges. The research contents include:link selection technology in the scenario of single service on one bearer, QoS technology in the scenario of single service on multiple bearers, and the mobility management technology in the scenario of multiple bearers scenario. Specifically, this study covers the following aspects:Firstly, the heterogeneous multi-link system is introduced briefly. The challenges of achieving QoS guarantee in the heterogeneous multi-link system are summarized. Then the states of the arts of tackling the technology challenge mentioned above are reviewed. Detailed analysis produces the research emphases of this thesis. As for the selection of the target access network, a link selection algorithm based on virtual target network (VTN) and multiple attribute decision making (MADM) theory is proposed. The network which is similar with VTN is selected as the access target network, so that the continuity of service and QoS are guaranteed with the minimum cost and the network resource is fully utilized. Furthermore, an algorithm of weight allocation to different link attributes is also proposed. This algorithm is based on information entropy theory and it is practical. Two novel weighted Markov chain (WMC) approaches based on rank aggregation are proposed as for the situation that the network information is not complete or inaccurate measurement happens. By using the proposed approaches, partial knowledge due to incomplete measurement can be easily used.As for the diversity of link characteristics, the solutions in the user side and the network side for QoS guarantee are proposed respectively. On the user side, a load distribution algorithm is proposed. It optimizes the end-to-end delay by decreasing packet reordering delay. The pivotal idea of this algorithm is as follows:when the error between theoretical distribution and practical transmission capabilities of the links is under the predefined threshold, the traffic is split proportionally for all available links; when the error between theoretical distribution and practical transmission capabilities of the links is over the predefined threshold, the traffic is split proportionally for selected links. On the network side, a network control method is proposed. It also optimizes the end-to-end delay by decreasing packet reordering delay. In this method, decision-making function is placed on network side. The packet transmission is controlled by this function based on the state information of packet arrival. This method can optimize the reordering delay and save power consumption for the terminal.As for the mobility aspect, a novel type of mobility management architecture is proposed for heterogeneous multi-link system. This architecture is composed of trigger module, multi-link management module and handover module. These modules are logical function modules, which achieve the mapping to 3GPP evolved network architecture and the protocol stack. An efficient handover scheme based on MIH services is also proposed. It performs the opration by allowing the cooperation between the handover initiated interface and other activated interfaces on the same terminal.A summary is given at the end, where the future research directions related to this thesis are also pointed out.更多还原