基于遗传和人工鱼群融合算法的频谱分配优化

发布时间：2018-06-24 13:50

本文选题：图论模型频谱分配 + 网络效益　；参考：《吉林大学》2017年硕士论文

【摘要】：总体频谱资源的紧缺性俨然成为了阻碍当今社会无线通信发展的一个绊脚石。因此现在各级频管部门也在大力呼吁和倡导认知无线电技术,旨在发现闲置信道资源,或者利用合法指定用户不使用信道的间隙,把基本闲置和暂时闲置的信道分配给更多有需求的合法接入用户,进而解决诸如由于定频专向分配造成的频谱资源浪费现象。在技术层面,它可以类似移动通信中CDMA复用技术思想一样,充分利用各种复用技术,例如OFDM等等,达到同频间的资源共用。还可以积极寻找出闲置频点,或者法定占用信道的用户不使用该信道时,把这种情况下的频道动态的分配给有需求的合法用户。这样既保障了合法授权用户的权益,也最大限度的利用了频谱资源。只是在这样的利用频谱资源的过程中,还需要考虑干扰、吞吐量等技术问题,还要准确判断想要接入的用户是否合法等法律问题。总之高效率的分配已发现的闲置频谱资源,积极利用好合法用户不占用信道的时间段以及如何高效安全的实现频谱复用技术是当下重要的科研攻关方向,同时也是必然趋势。本文将首先介绍我国当前频谱资源使用所面临的一些问题,并简要讨论解决办法和研究重点。然后系统介绍和梳理该领域电磁频谱分配的主要技术和经典模型,进而从众多模型中最终选择图论模型作为讨论重点,详细介绍该模型下闲置矩阵、干扰矩阵、效益矩阵和分配矩阵数学描述。这些经典的基础的知识储备,为遗传、人工鱼群算法和融合后的改进算法的优化仿真提供了理论基础。在算法讨论方面,在第三、第四章将系统介绍两种基于仿生特点的优化算法,遗传算法和人工鱼群算法。并利用图论模型,利用算法本身的特点,对频谱分配的实际问题进行优化。通过仿真实验,变化三个变量参数,得到的仿真结果进行合理化分析,验证算法在该实践领域的有效性。在实际工作中,我利用研一暑假和研二大约一年的时间对比了多种优化算法,最终选定对于遗传和人工鱼群算法进行研究,因为他们仿真结果符合定性的分析结论,与认知是吻合的。但是在常用的粒子群优化算法中,仿真的结果与认知吻合度不高,或者说还有些现象与认知还有差距,因此没有选用。在核心创新点方面,是基于频谱分配的图论模型,依托融合算法对其进行优化分配。融合算法是根据遗传和人工鱼群各自算法特点,做到优势互补。基本思想是避免鱼群算法在寻优过程中方向性差和容易陷入僵局的问题产生。此时引入了遗传算法的相关操作,来克服这一缺陷,求得真正的全局最值,而不是局部极值。通过这一理念,使得融合后的算法在频谱分配网络效益优化方面要优于单独使用遗传和人工鱼群算法的网络效益,可以验证该融合算法在这一分配模式下的效果还是较为显著的。
[Abstract]:The scarcity of overall spectrum resources has become a stumbling block to the development of wireless communication in today's society. Therefore, at present, the frequency management departments at all levels are also vigorously calling for and advocating cognitive radio technology, with the aim of finding idle channel resources or taking advantage of the gap between legally designated users who do not use the channel. The basic idle channel and the temporarily idle channel are allocated to more and more legitimate access users, thus solving the waste of spectrum resources caused by the special allocation of fixed frequency. At the technical level, it can make full use of all kinds of multiplexing technologies, such as OFDM, to achieve the resource sharing between the same frequency, just like CDMA multiplexing technology in mobile communication. We can also find out the idle frequency points actively, or when the users who occupy the channel do not use the channel, we can dynamically assign the channel to the legitimate users who need it. This not only protects the legitimate rights and interests of authorized users, but also maximizes the use of spectrum resources. However, in the process of using spectrum resources, we still need to consider the technical problems such as interference, throughput and so on, and determine the legal issues such as whether the users who want to access are legal or not. In a word, it is an important research direction and inevitable trend to allocate the idle spectrum resources that have been found efficiently, to make good use of the time period that the legitimate users do not occupy the channel and how to realize the spectrum multiplexing technology efficiently and safely. This paper will first introduce some problems in the use of spectrum resources in China, and briefly discuss the solutions and research emphasis. Then the main techniques and classical models of electromagnetic spectrum allocation in this field are introduced and combed systematically, and then the graph theory model is chosen as the focus of discussion from many models, and the idle matrix and interference matrix under the model are introduced in detail. Mathematical description of benefit matrix and distribution matrix. These classical basic knowledge reserves provide a theoretical basis for the optimization simulation of genetic artificial fish swarm algorithm and fusion improved algorithm. In the aspect of algorithm discussion, in the third chapter, in chapter 4, we introduce two optimization algorithms based on bionic characteristics, genetic algorithm and artificial fish swarm algorithm. The practical problem of spectrum allocation is optimized by using graph theory model and the characteristics of the algorithm itself. Through the simulation experiment, the simulation results are analyzed by changing the parameters of three variables, and the validity of the algorithm in the practical field is verified. In the actual work, I compared a variety of optimization algorithms with one summer vacation and two years of research, and finally selected genetic and artificial fish swarm algorithms to study, because their simulation results were consistent with qualitative analysis conclusions. It is consistent with cognition. But in the commonly used particle swarm optimization algorithm, the simulation results are not consistent with cognition, or there are still some phenomena and cognitive gaps, so there is no choice. In the core innovation, it is a graph theory model based on spectrum allocation, which is optimized by fusion algorithm. The fusion algorithm is based on the characteristics of genetic algorithm and artificial fish swarm algorithm to achieve complementary advantages. The basic idea is to avoid the problem of poor directionality and easy deadlock in the optimization process of fish swarm algorithm. In order to overcome this defect, the real global maximum value, rather than the local extremum value, is obtained by introducing the relevant operations of genetic algorithm. Through this idea, the fusion algorithm is superior to the genetic algorithm and artificial fish swarm algorithm in the optimization of the efficiency of the spectrum allocation network. It can be verified that the fusion algorithm is more effective in this allocation mode.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN92;TP18

【相似文献】