5G Signal Identification Using Deep Learning

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Spectrum awareness, including identifying different types of signals, is critical in a cellular system environment. The Fifth Generation Mobile System (5G) achieves a considerable promise in terms of high data rate, low latency, and low power consumption. This work explores a neural network to identify 5G signals, among other cellular communications signals, including Long-Term Evolution (LTE) and Universal Mobile Telecommunication Service (UMTS). We investigate the use of deep learning in wireless communications systems. The signals of different cellular systems, including 5G are generated to train different conventional neural networks. We consider the effects of training dataset size, features extracted, and channel fading in our study. Besides, we studied the impact of several signal-to-noise ratios. Experiment results demonstrate the effectiveness of deep learning neural networks in identifying cellular system signals, including UMTS, LTE, and 5G. This dissertation focuses on 5G signal classification using machine learning and deep learning algorithms. For deep learning algorithms approaches, convolutional neural networks are utilized. Various cellular signals were generated by using the MATLAB toolbox for UMTS, LTE, and 5G NR signals. Noise environment and interference are considered in the classification task. The actual data has been covered to test the model, including 3G, 4G, and 5G, by using Huawei’s GENEX Probe. A network optimization and drive test data collection system is an air interface test tool for WCDMA/HSDPA/HSUPA/GSM/GPRS networks. Our research demonstrates the effectiveness of deep learning algorithms to identify 5G signals among various cellular signals. This work explores the features of deep learning in cellular signals identification. The cellular signals data are used to train Convolutional network (CNN) LeNet-5 Based and then test those networks with various signals. We Investigate the CNN under different scenarios. Besides, the Signal-to-Noise Ratio is considered. The performance of the CNN is substantially improving when we increase the dataset size.