Linkage Transformer: an Attention Based Neural Network for Multi-Cell Traffic Prediction

Zhenqian Wang, Meixia Fu, Qu Wang, Yifan Lu, Jiansheng Wu, Le Chen, Wanqing Guan, Wei Li, Jianquan Wang

Abstract: With the growing complexity of base station network, the research of traffic prediction has attracted increasingly attention in the world. Especially, artificial intelligence technology is applied to traffic prediction, which plays important role in communication system. In this article, we propose a Linkage Transformer model for multi-cell traffic prediction, which contains three main procedures. First, the traffic data of multi-cell with multi properties is sent to the encoder module for extracting data characteristics. Second, an attention module is designed in the middle part of the Linkage Transformer to learn spatial-temporal relation among cells’ traffic data. Finally, decoder module is utilized to predict traffic value of multi-cells. In the experiment, we use China Union traffic data of Haidian district to demonstrate the performance of Linkage Transformer in multi-cell traffic prediction task.