异构无线网络软负载均衡算法研究

发布时间：2018-01-09 03:26

本文关键词：异构无线网络软负载均衡算法研究　出处：《华中科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：最近这些年，无线通信领域取得了空前的发展，人们不断提出了各种新颖的无线通信技术，一时间，出现了多种制式的异构无线网络。与此同时，随着移动互联网的高速发展和智能终端的不断普及，特别是移动端多媒体应用的迅猛发展，移动端的流量呈现爆炸性地增长。单一模式的通信网络由于端到端的吞吐量较低、延时较大，已无法满足现有需求。而未来将出现大量的异构无线网络重叠覆盖的场景，在该种场景下，利用异构无线网络间的优势互补，是解决上述问题的有效手段之一。因此，很有必要研究一种异构无线网络下的协作共享技术。从提升整个异构无线网络融合系统的资源利用率的角度来看，负载均衡显然是一种十分有效手段。我们研究的焦点是软负载均衡策略。该策略下，将用户的业务数据流分割成若干个子数据流，并将子数据流交由多个网络进行并行传输，，以实现异构无线网络融合系统的负载均衡。首先，我们提出了一种异构无线网络环境下的网络流量的最优分配机制。异构网络可以看作是多个可以同时为终端用户服务的同构网络的组合。为简单起见，每个网络都被建模成M/M/1队列。我们的目标是最小化每个队列中受约束的任务数。可以通过拉格朗日乘数解决最优解问题。根据各个不同网络的服务速率，将产生的数据速率最优地分配到各个网络中。仿真分析表明，该算法比平均分配算法和贪婪算法具有更高的吞吐量，提高了频谱效率。当并行链路传输多媒体业务和实时应用业务时，由于这两类业务对数据包时延、时延抖动和数据包重排序等因素十分敏感，因此，为了提高并行链路的利用率，研究一种有效的基于时延控制的负载分配算法显得至关重要。我们提出了一种基于链路时延控制的负载分配算法，称为DCSLB算法。DCSLB算法的目的是在异构无线网络融合场景下，考虑了多条链路的带宽和传播时延因素，通过动态调整分配到各条链路上的网络负载比例（分割比例）实现最大链路时延的最小化。一方面，减小了接收端数据包重排序的等待时延；另一方面，使得各条链路的端到端时延差最小，从而降低接收端数据包重排序的概率，进而使得系统成功传输一个数据包的时延减小。随后，我们搭建了基于LTE和WLAN融合的实物验证系统，以实现DCSLB算法，测试结果表明采用DCSLB算法的软负载均衡方案相对于传统的单模接入方案，在保证服务质量的前提下，实现了吞吐量的线性叠加。最后，我们对软负载均衡应用场景进行了拓展，针对未来蜂窝网络中宏小区和小小区协同覆盖以及WLAN AP组成mesh网络的场景，提出了一种分层半集中式网络架构，可以有效减小无线资源管理服务器的管理复杂度，并通过引入备用无线资源管理服务器，使得异构无线网络兼具容灾能力。
[Abstract]:In recent years, the field of wireless communication has made unprecedented development, people constantly put forward a variety of novel wireless communication technology, at a time, there are a variety of heterogeneous wireless networks. At the same time. With the rapid development of mobile Internet and the continuous popularity of intelligent terminals, especially the rapid development of mobile multimedia applications. Mobile traffic increases explosively. A single mode of communication network has a long delay due to the low end-to-end throughput. However, there will be a large number of overlapping scenarios in the future. In this scenario, the advantages of heterogeneous wireless networks will complement each other. Therefore, it is necessary to study a cooperation and sharing technology in heterogeneous wireless networks. From the point of view of improving the resource utilization of the whole heterogeneous wireless network fusion system. Load balancing is obviously a very effective method. The focus of our research is the soft load balancing strategy, in which the user's business data stream is divided into several sub-data streams. In order to realize the load balance of heterogeneous wireless network fusion system, the sub-data stream is transferred to multiple networks for parallel transmission. Firstly, we propose an optimal traffic allocation mechanism for heterogeneous wireless networks. Heterogeneous networks can be considered as a combination of multiple isomorphic networks that can serve end users at the same time. Each network is modeled as a M / M / 1 queue. Our goal is to minimize the number of constrained tasks in each queue. The optimal solution can be solved by Lagrange multipliers. . The resulting data rates are optimally allocated to each network. Simulation results show that the proposed algorithm has higher throughput and higher spectral efficiency than the average allocation algorithm and the greedy algorithm. When parallel links transmit multimedia services and real-time application services, these two services are sensitive to data packet delay, delay jitter and packet reordering. In order to improve the utilization of parallel links, it is very important to study an effective load allocation algorithm based on delay control. We propose a load allocation algorithm based on link delay control. The purpose of the algorithm called DCSLB algorithm. DCSLB is to consider the bandwidth and propagation delay factors of multiple links in heterogeneous wireless network fusion scenario. The maximum link delay is minimized by dynamically adjusting the network load ratio (partitioning ratio) allocated to each link. On the one hand, the waiting delay of packet reordering at the receiving end is reduced. On the other hand, the end-to-end delay difference of each link is minimized, thus reducing the probability of packet reordering at the receiving end, and then reducing the delay of successful transmission of a packet in the system. Then, we build a physical verification system based on the fusion of LTE and WLAN to implement the DCSLB algorithm. The test results show that compared with the traditional single-mode access scheme, the soft load balancing scheme based on DCSLB algorithm can achieve the linear superposition of throughput under the premise of guaranteeing the quality of service. Finally, we extend the application scenario of soft load balancing, aiming at the scenario of mesh network composed of macro cell and small cell co-coverage and WLAN AP in the future cellular network. A hierarchical and semi-centralized network architecture is proposed, which can effectively reduce the management complexity of the wireless resource management server. By introducing the backup radio resource management server, the heterogeneous wireless network has the capability of disaster tolerance.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.5

【相似文献】