频谱数据挖掘算法研究与应用

发布时间：2018-07-28 19:42

【摘要】：无线通信技术发展迅速,频谱资源也日益紧缺。运用数据挖掘技术对频谱感知数据进行分析,在大量数据中找出潜在的频谱使用规律,并预测未来一段时间内的频谱使用情况,从而为认知用户的动态频谱接入提供可靠依据。认真研究了数据挖掘算法理论及其无线电频谱感知技术,其中深入研究了最小二乘支持向量机算法(LSSVM)的原理和其模型的寻优方式。本文主要工作和创新点如下:1.提出了基于免疫算法的特征加权LSSVM算法。在传统的最小二乘支持向量机算法的基础上,引入了特征加权向量,与原模型的超参数形成新的超参数。以分步寻优的方式,利用免疫算法得到最优的特征加权向量,从而提高了模型的预测精度。2.提出了基于耦合模拟退火的特征加权LSSVM算法。以改进原模型的超参数方式,实现对不同特征维度数据加权,并采用联合寻优的方式,利用耦合模拟退火算法得到新的超参数,提高了算法的性能。另外针对频谱数据挖掘场景,在离线预测的基础上提出来在线预测算法。3.设计了完整的模拟卫星通信系统的频谱数据采集与预测仿真系统,系统包括地理运动模块、用户接入模块和信号产生模块等;并用仿真得到频谱数据验证了本文提出的两种算法的都能提高模型的预测精度,同时使用真实的电力负荷数据验证得到的结论。本文通过对频谱数据挖掘技术的研究,改进了最小二乘支持向量机算法,同时将预测结果推荐给动态频谱接入用户作为参考,具有理论和应用的双重价值。
[Abstract]:With the rapid development of wireless communication technology, spectrum resources are increasingly scarce. Using data mining technology to analyze spectrum sensing data, to find out the potential spectrum usage rule in a large number of data, and to predict the spectrum usage in a period of time in the future, so as to provide a reliable basis for the dynamic spectrum access of cognitive users. The theory of data mining algorithm and its radio spectrum sensing technology are studied in detail. Among them, the principle of least squares support vector machine (LSSVM) algorithm and its model optimization are studied. The main work and innovation of this paper are as follows: 1. A feature weighted LSSVM algorithm based on immune algorithm is proposed. Based on the traditional least square support vector machine (LS-SVM) algorithm, the feature weighted vector is introduced to form a new superparameter with the superparameter of the original model. In order to improve the prediction accuracy of the model, the optimal feature weighting vector is obtained by using the immune algorithm. A feature weighted LSSVM algorithm based on coupled simulated annealing is proposed. By improving the hyperparametric method of the original model, the data of different feature dimensions are weighted, and the new superparameters are obtained by using the coupled simulated annealing algorithm, which improves the performance of the algorithm. In addition, an online prediction algorithm. 3. 3 is proposed on the basis of offline prediction for spectrum data mining scene. The spectrum data acquisition and prediction simulation system of analog satellite communication system is designed. The system includes geographic motion module, user access module and signal generation module. The two algorithms proposed in this paper can improve the prediction accuracy of the model and verify the conclusion by using the real power load data. In this paper, the least-squares support vector machine (LS-SVM) algorithm is improved through the research of spectrum data mining technology. At the same time, the prediction results are recommended to the dynamic spectrum access users as a reference, which has both theoretical and application value.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.13;TN925

【相似文献】