【作者】 张旭；

【导师】 李一兵；

【作者基本信息】 哈尔滨工程大学 ， 通信与信息系统， 2013， 博士

【摘要】 移动通信网络的飞速革新推动了多媒体技术的蓬勃发展，同时也使得有限的频域资源与用户不断增长的业务传输需求之间的矛盾日益突出。随着长期演进(Long TermEvolution，LTE)项目标准化工作的日臻完善，针对其无线资源管理模式的开发已成为业界关注的焦点。为此，动态子载波分配方案作为一种高效的频域资源调度手段，以其能够有效提升系统频谱利用率，增强多业务服务质量(Quality of Service，QoS)需求处理能力，而广泛研究应用于LTE系统上、下行链路进行频谱资源管理。本文致力于单载波频分多址(Single Carrier Frequency Division Multiple Access，SC-FDMA)系统跨层动态子载波分配技术研究，主要工作内容概括如下：首先，分析了SC-FDMA系统面向非实时业务的动态子载波分配问题。基于单一非实时业务模型提出一种最大平均容量动态子载波分配算法。该算法以最大化系统用户平均信道容量为分配目标，代替传统贪婪算法面向全体用户及资源块的寻优过程，从而以较低的计算复杂度实现高系统频率利用率。并针对具有变系统吞吐量或变分配公平度需求的单一非实时业务模型，提出一种可调公平度动态子载波分配算法。该算法通过可调性公平系数的引入，能够根据用户传输状态及业务需求合理判定适当的分配准则，从而实现系统吞吐量及分配公平性之间的可调性权衡。同时，针对多种非实时业务环境下系统频域资源的竞争问题，基于效用理论提出最大容量需求效用算法及其改进算法。所提出算法通过构建与用户信道传输能力及传输容量需求相关的效用评价体系，以实现系统吞吐量性能与各用户传输容量需求之间的有效权衡，从而有效提升多业务传输需求处理能力。然后，分析了SC-FDMA系统面向实时业务的动态子载波分配问题。针对有限队列缓存空间条件下单一实时业务模型的数据丢失问题，提出一种改进的最大时延效用算法。该算法在传统最大时延效用算法的基础上引入用户队列状态信息，通过提升长队列用户的分配优先级以有效控制各用户队列长度，从而抑制由于缓存溢出所造成的数据丢失。同时，为降低由用户队列长度信息引入所带来的信令开销，提出一种最大指数效用动态子载波分配算法。该算法采用与用户平均等待时间相关的指数效用函数以增大用户间分配优先级差异，充分保证高时延及长队列用户优先分得频域资源进行通信，在简化算法复杂度的同时兼顾系统数据丢失率及时延性能。此外，针对多种实时业务环境下系统频域资源的竞争问题，讨论了最大指数效用算法当根据不同业务传输特性设置分配权重从而实现业务间分配优先级划分时其QoS区分性能。并在此基础上引入分级分配理念，提出一种最大分级指数效用动态子载波分配算法。该算法为同种业务不同传输用户根据当前时延状态设定高、低两种不同的业务分配权重，在确保高分配优先级业务传输性能的同时，为时延状态较差且分配权重较低的用户提供接入系统的机会，从而提升算法多实时业务QoS需求处理能力。最后，分析了SC-FDMA系统面向混合业务的动态子载波分配问题。针对由于调度器分配的数据传输速率超出用户业务需求所导致的系统服务速率浪费问题，结合负载控制技术提出一种基于严格服务速率约束的最大分级指数效用算法。该算法通过在最大分级指数效用算法的基础上引入严格服务速率约束方案，能够在分配过程中严格限制调度器分配的数据传输速率，从而有效消除系统服务速率浪费。同时，针对服务速率控制所导致的频域资源不良竞争问题，结合接入控制技术提出一种基于自适应服务速率约束的最大分级指数效用算法。该算法根据用户传输状态及业务需求设立自适应系统接入阈值，通过在实时与非实时业务间实现分配优先级的严格划分及自适应调整，以保证混合业务间频域资源的合理分布。更多还原

【Abstract】 The rapid innovation of mobile communication network promotes the flourishingdevelopment of multimedia technology, and also causes a increasingly outstandingcontradiction between the limited frequency resources and the growing user service demands.With the gradual perfection of standardization work in Long Term Evolution (LTE), theexploitation of the radio resource management mode has become the focus of the informationindustry. Therefore, as an effective technique for frequency resources scheduling, DynamicSubcarrier Allocation (DSA) is now wildly studied and employed in both the uplink anddownlink LTE system due to its high system frequency efficiency performance and excellentquality of service (QoS) provisioning capability in multi-service environment.This paper focuses on cross-layer DSA of the Single Carrier Frequency DivisionMultiple Access (SC-FDMA) system, it mainly includes:Firstly, the non-real-time traffic oriented DSA problem for the SC-FDMA system isanalyzed. Based on the single non-real-time traffic model, maximum average capacity DSAalgorithm is proposed. The main idea of the proposed algorithm is to maximize the averagecapacity of system users without taking the searching process with respect to all the systemusers and resource blocks in traditional greedy algorithm, thereby achieving high systemfrequency efficiency performance with low computational complexity. And for meeting theservice requirements with changing system capacity and allocation fairness in singlenon-real-time traffic model, an adjustable fairness algorithm is developed. By introducing anadjustable fairness factor, the proposed algorithm can decide appropriate scheduling criterionaccording to the transmission state and service requirement for each allocation, so as tocontributing an adjustable tradeoff between the system capacity and allocation fairness.Furthermore, in order to solve the resource competition problem in multi non-real-time trafficenvironment, a max capacity demand utility algorithm and its improved algorithm areinvestigated. By establishing the utility-pricing framework which concerning both channelquality condition and traffic transmission demand, the proposed algorithms can achieve anattractive tradeoff between system rate-sum capacity and the user capacity demands, therebyefficiently improving the processing capability for multi-service requirements. Secondly, the real-time traffic oriented DSA problem for the SC-FDMA system isanalyzed. In order to avoid the data loss in single real-time traffic model with finite queuingspace assumption, an improved max delay utility algorithm is proposed. By introducing thequeue length information based on the traditional max delay utility algorithm, the proposedalgorithm can effectively eliminate the system data loss due to the buffer overflow.Meanwhile, a max exponential utility algorithm is developed for further signaling overheadreduction. By exploiting the exponential utility function with respect to user average delay,the proposed algorithm can broaden the priority variation between users, and fully guaranteethe allocation superiority of the user suffering from long waiting time and queue length. So asto reducing the computational complexity while keeping good data loss and delayperformance. Moreover, in order to solve the resource competition problem in multi real-timetraffic environment, the QoS differentiation performance of the max exponential utilityalgorithm is discussed when providing traffic priority division with appropriate schedulingweights setup among different services. Under this consideration, a max graded exponentialutility algorithm is presented by taking the concept of grading theory. By means of setting twokinds of traffic scheduling weights, the proposed algorithm can provide more accessopportunities for the users with poor delay performance and low scheduling weights, therebyimproving the QoS provision capability for multi real-time traffic.Finally, the mixed traffic oriented DSA problem for the SC-FDMA system is analyzed.In order to overcome the service rate waste caused by assigning more data rate than user’stransmission demand, a max graded exponential utility algorithm with strict service rateconstraint is proposed. By integrating the concept of load control, the proposed algorithm canstrictly restrict the data transmission rate assigned to each user, and completely eliminate theservice rate waste. Meanwhile, for solving the unjust competition in service rate controlmechanism, a max graded exponential utility algorithm with adaptive service rate constraint isdeveloped. By means of setting the access threshold which is aware of both user transmissionstate and service requirements, the proposed algorithm can provide a strict and adaptivescheduling priority division between real-time and non-teal-time traffics, and achieve areasonable frequency resources distribution among mixed traffic users.更多还原