LTE中广播多播服务的资源分配算法研究

发布时间：2018-08-14 13:17

【摘要】：随着移动互联网的蓬勃发展,多媒体业务在移动通信业务中所占比重越来越大,高效使用多媒体广播多播技术(MBMS)传输多媒体业务方面的研究成为目前的研究热点,MBMS技术通过无线资源共享可以极大缓解日益紧张的资源需求。对于MBMS资源分配算法的研究可以进一步提高多播系统资源利用率并提高多播用户的用户体验。LTE的MBMS包括单频网(MBSFN)和单小区两种传输模式,这两种传输模式有不同的应用场景,本文基于单小区模式进行资源分配算法研究。本文对LTE下行链路物理资源和资源分配框架进行了研究,剖析了LTE中MBMS的网络架构、信道支持、协议栈、传输模式和业务流程。从LTE经典的单播资源分配算法入手,对比单播资源分配算法深入分析了经典的多播资源分配算法,其中包括单速率算法和多速率算法。本文针对目前单速率资源分配算法的缺点设计了动态调制编码策略(MCS)选择方案和基于Kuhn-Munkres算法的资源分配算法;针对目前多速率分配算法的缺点改进设计了基于优先级的分层多速率MBMS资源分配算法PLRA。由于现有仿真平台无法满足仿真需求,因此设计实现了基于Python的MBMS系统级仿真平台。动态MCS选择方案对每一个多播组内的用户的平均丢包率进行策略调整,可以在保证给定丢包率阈值情况下有效提高系统吞吐量。在动态MCS选择方案基础上设计了三种单速率资源分配算法,包括D-BKM、D-IKM和D-MaxKM。D-BKM算法先采用Kuhn-Munkres算法进行最低速率保证的资源块分配,再采用最大吞吐量原则进行第二次分配;D-IKM算法利用资源块数量需求估计,在进行最低速率保证的分配中提高算法收敛速度;D-MaxKM算法是D-BKM算法的逆向,首先进行最大吞吐量原则资源块分配,再从已分配的资源块中寻找富余资源块,在富余资源块和未满足最低速率要求的多播组间使用Kuhn-Munkres算法进行分配。基于优先级的分层多速率MBMS资源分配算法PLRA分为基层分配和扩展层分配两个阶段。在基层分配时首先满足资源块需求较少的多播组,并且在选择多播组时引入公平性因子改善在系统容量不足时的速率公平性;在扩展层分配时设计一种优先级计算方式,优先将资源块分配给优先级高的多播组扩展层,优先级计算时考虑了频谱效率、吞吐量以及公平性等多种因素。基于Python语言设计和实现的MBMS系统级仿真平台可以在多种场景下验证了本文设计的算法在系统吞吐量、丢包率、速率满意度、速率公平性和频谱效率等方面的性能。仿真结果表明,第三章中设计的三种算法可以实现在保证业务最低速率的情况下使系统吞吐量有所提升,其中D-MaxKM算法在三种算法中吞吐量最大,而D-IKM算法通过资源块数量估计有效提高D-BKM算法运行效率并且D-IKM算法性能与D-BKM算法基本相当;第四章改进设计的PLRA算法在牺牲少量的频谱效率的情况下,可以有效提高分层多速率多播系统的吞吐量和公平性。
[Abstract]:With the vigorous development of mobile internet, the proportion of multimedia services in mobile communication services is increasing. Efficient use of multimedia broadcast multicast technology (MBMS) to transmit multimedia services has become a research hotspot. MBMS technology can greatly alleviate the growing demand for resources through wireless resource sharing. The research of S resource allocation algorithm can further improve the resource utilization of multicast system and the user experience of multicast users. LTE MBMS includes single frequency network (MBSFN) and single cell transmission mode, which have different application scenarios. This paper studies the resource allocation algorithm based on single cell mode. Link physical resources and resource allocation framework are studied. The network architecture, channel support, protocol stack, transmission mode and service flow of MBMS in LTE are analyzed. The classical unicast resource allocation algorithms, including single-rate algorithm and multi-rate algorithm, are compared with the classical unicast resource allocation algorithms in LTE. This paper designs a dynamic modulation coding strategy (MCS) selection scheme and a resource allocation algorithm based on Kuhn-Munkres algorithm to overcome the shortcomings of the current single-rate resource allocation algorithm, and designs a priority-based hierarchical multi-rate MBMS resource allocation algorithm PLRA to overcome the shortcomings of the current multi-rate allocation algorithm. The real platform can not meet the simulation requirements, so a system-level simulation platform for MBMS based on Python is designed and implemented. Dynamic MCS selection scheme adjusts the average packet loss rate of each multicast group, which can effectively improve the system throughput under the given packet loss rate threshold. Three single-rate resource allocation algorithms are proposed, including D-BKM, D-IKM and D-MaxKM.D-BKM. Kuhn-Munkres algorithm is used to allocate resource blocks with minimum rate guarantees, and then the maximum throughput principle is used for the second allocation. D-MaxKM algorithm is the inverse of D-BKM algorithm. Firstly, the maximum throughput principle is used to allocate resource blocks, then the redundant resource blocks are searched from the allocated resource blocks, and the Kuhn-Munkres algorithm is used to allocate the redundant resource blocks and the multicast groups which do not meet the minimum rate requirements. The algorithm PLRA is divided into two stages: primary allocation and extended layer allocation. In the primary allocation, multicast groups with fewer resource blocks are satisfied, and fairness factor is introduced to improve the rate fairness when the system capacity is insufficient. In the extended layer allocation, a priority calculation method is designed to give priority to resource blocks allocation. The system-level simulation platform of MBMS based on Python language can be designed and implemented to verify the proposed algorithm in a variety of scenarios, such as system throughput, packet loss rate, rate satisfaction, rate fairness and frequency. Simulation results show that the three algorithms designed in Chapter 3 can improve the system throughput under the condition of guaranteeing the lowest traffic rate. Among the three algorithms, D-MaxKM algorithm has the largest throughput, while D-IKM algorithm can effectively improve the efficiency of D-BKM algorithm by estimating the number of resource blocks and D-MaxKM algorithm has the highest throughput. The performance of IKM algorithm is almost the same as that of D-BKM algorithm. In the fourth chapter, the improved PLRA algorithm can effectively improve the throughput and fairness of Hierarchical Multicast System at the expense of a small amount of spectral efficiency.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN929.5

【相似文献】