Data scheme-based wireless channel modeling method: motivation, principle and performance

In recent years, data mining and machine learning technologies have made great progress driven by enormous volumes of data. Meanwhile, the wireless-channel measurement data appears large in volume because of the large-scale antenna numbers, increased bandwidth, and versatile application scenarios. With powerful data mining and machine learning methods and large volumes of data, we can extract valuable and hidden rules from the wireless channel. Motivated by this, we propose a channel-modeling method using PCA in this paper. Its principle is to utilize the features and structures extracted from the CIR data collected by measurements, and then model the wireless channel of the targeted measurement scenario. In addition, a noise removing method using a BP neural network is designed for the proposed model, which can recognize and remove the noise of the polluted CIR accurately. The performance of the proposed scheme is investigated with the actual measured CIR data, and its superiority is verified.