论文部分内容阅读
接收信号场强预测对无线通信网络的设计与规划非常重要.为此,提出了一种基于模块化神经网络的场强预测模型.对于给定的区域,选取一定数量的接收样本点,根据接收信号场强数据的分布特点,使用K均值(K-Means)聚类方法对全部样本点聚类,以实现对输入样本空间的分解,并建立对应的子神经网络模块.以某学校宿舍区域为例,选取了训练集和测试集样本点,通过对比单一神经网络模型和模块化神经网络模型的预测误差,发现模块化神经网络的预测结果优于单一神经网络,证明了所提出模型的有效性.
Field strength prediction of received signal is very important for wireless communication network design and planning.Therefore, a field strength prediction model based on modular neural network is proposed.For a given area, a certain number of received sample points are selected, Signal field strength data distribution characteristics, using K-Means clustering method for all sample points clustering in order to achieve the input sample space decomposition and the establishment of the corresponding sub-neural network module to a school dormitory area is For example, the training set and the sample points of the test set are selected. By comparing the prediction errors between the single neural network model and the modular neural network model, it is found that the prediction results of the modular neural network are better than the single neural network, which proves the validity of the proposed model .