基于随机森林算法的信道场景分类模型研究

Study of Model of Wireless Channel Classification Based on Algorithm of Random Forest

随着科技的进步移动通信产业已经与人类生活、社会发展产生紧密地结合在一起,对特定单一场景或一定区域内条件下的无线信道进行分类、识别,实现特定场景进行分析以及无线网络的优化,具有重要意义。针对这一背景,提出了一种结合随机森林算法的信道场景分类模型。对不同的信道场景物理特征进行提取与降维,基于多分类器集成的原理,结合多个决策树与随机向量决策树的构造原理,创建信道场景的随机森林分类模型。实验结果表明:所采用的分类方法的总模型准确率约为89.90%,能有效地实现信道场景的分类识别。

With the advances in technology, the mobile communications technology industry has beenclosely tied with human life, social development. Classifying and identifying the radio channel for asingle scene or under certain specific conditions of the region to achieve a specific scene analysis andoptimization of wireless networks is of great significance. Against this background this paper, weproposed a combination of random forests algorithm channel scene classification model. Firstly, thedifferent physical characteristics of the scene channel was extracted and dimensionality reduced, and based on the principle of integration of multiple classifiers, combining the principle of a plurality ofdecision trees constructed with random vector decision tree, we created a random forest classificationmodel of channel scene. Experimental results show that the overall accuracy of the model used in theclassification is about 89.90% , which is effective in achieving the classification channel scene.