Мақола

Various Modulation Types Classification on the Air by Convolutional Neural Network

Yousif I. HammadiDep. of Medical Instruments Engineering Techniques, Bilad Alrafidain University College, IraqMokhalad Abdulameer Kadhim AlsaeediDepartment of Communications Engineering, University of Diyala, IraqOmar Abdulkareem MahmoodDepartment of Communications Engineering, University of Diyala, IraqAws Zuhair SameenDepartment of Medical Instrumentation Engineering Techniques, Al-Farahidi University, IraqMohammed Saleh Ali MuthannaInstitute of Computer Technologies and Information Security, Southern Federal University, RussiaAhmed AzizDepartment of computer science, Benha university, Egypt and International Business Management Department, Tashkent State University of Economics, UzbekistanAmmar MuthannaDepartment of Telecommunication Networks and Data Transmission, The Bonch Bruevich Saint Petersburg State University of Telecommunications, Russia and Department of Applied Probability and Informatics, Peoples Friendship University of Russia, Russia

2023en

ABI

Аннотация

Deep learning (DL), a relatively recent AI technique, has been successfully applied to the problem of automated modulation categorization (AMC), with promising results. An essential part of developing the spectrum-sensing capabilities needed by software-defined radio is carrying out radio signal classification jobs, which we did in this research by introducing a better deep neural architecture. Different detection techniques were utilized in the literature; however, DL neural networks are recently utilized in this field. Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM) which is one variation of the RNN network, and other techniques are some of the employed DL architectures in the field of AMC. In this paper, Convolutional Neural Network (CNN) model has been built to automatically classify ten-modulation formats correctly. The suggested CNN network is constructed mainly of six-Convolutional layers. The overall number of trainable parameters in the proposed network is 1,757,962. Thus, the network accuracy, which was achieved during the training reached more than 90%. Python programming packages have been used to implement the suggested approach using Kaggle.com platform, which provides wide-range of facilities.

Ҳали таржима қилинмаган

Мавзулар

Wireless Signal Modulation Classification Radar Systems and Signal Processing Advanced SAR Imaging Techniques

Идентификаторлар

DOI: 10.1145/3644713.3644855

Иқтибослар ва манбалар

0 та иқтибос25 та фойдаланилган манба

Кўрсаткичлар — AkademScholar