In this study, using the cognitive radio (CR)-based non-orthogonal multiple access (NOMA) technique, the total ergodic capacity value of the close user is estimated with high accuracy and fast training times through different training algorithms with the proposed feedforward backpropagation artificial neural network (ANN) and nonlinear autoregressive network with exogenous inputs (NARX) model. The data set used in the neural network was obtained from the CR-NOMA system model, which was modeled with the exponential fading channel characteristic. By training the input and output data to the ANN, which was designed using the supervised learning method, ergodic capacity estimates of the close user were made over the test data. While evaluating the performance of ANN and NARX neural networks, the training time, the number of iterations, and the conditions of the network not reaching saturation were taken into consideration. The actual ergodic capacity value of the close user and the predicted values of feedforward backpropagation ANN and NARX networks were compared. The performance analysis of the proposed neural networks under Levenberg-Marquardt, Bayesian and Scaled-Conjugate training algorithms has been examined in terms of epoch value graph, error histogram analysis, and training state analysis where the error reaches the minimum.
Bu çalışmada, bilişsel radyo (BR) tabanlı dikgen olmayan çoklu erişim tekniği (NOMA) kullanılarak, yakın kullanıcıya ait toplam ergodik kapasite değerinin, önerilen ileri beslemeli geri yayılımlı yapay sinir ağı (YSA) ve doğrusal olmayan dışsal girdili otoregresif ağ (Nonlinear Autoregressive Network with Exogenous Inputs, NARX) modeli ile farklı eğitim algoritmaları yoluyla yüksek doğruluk oranında ve hızlı eğitim sürelerinde tahmin edilmesi amaçlanmıştır. Sinir ağında kullanılan veri seti, üstel sönümleme kanalı karakteristiği ile modellenen BR-NOMA sistem modelinden elde edilmiştir. Denetimli öğrenme yöntemi kullanılarak tasarlanan YSA’ya girdi ve çıktı verileri öğretilerek yakın kullanıcıya ait ergodik kapasite tahmini yapılmıştır. YSA ve NARX sinir ağları başarımı değerlendirilirken eğitim süresi, iterasyon sayısı, ağın doygunluğa ulaşmaması durumları göz önünde bulundurulmuştur. Yakın kullanıcıya ait gerçek ergodik kapasite değeri ile ileri beslemeli geri yayılımlı YSA ve NARX ağlarının tahmin etmiş olduğu değerler karşılaştırılmıştır. Önerilen sinir ağlarının Levenberg-Marquardt, Bayesian ve Scaled-Conjugate eğitim algoritmaları altındaki performans analizi, hatanın minimuma ulaştığı epok değer grafiği, hata histogram analizi ve eğitim durum analizi açılarından incelenmiştir.
Primary Language | Turkish |
---|---|
Subjects | Communication and Media Studies, Electrical Engineering |
Journal Section | Research Articles |
Authors | |
Publication Date | April 30, 2023 |
Submission Date | October 25, 2022 |
Acceptance Date | January 24, 2023 |
Published in Issue | Year 2023 Volume: 28 Issue: 1 |
Announcements:
30.03.2021-Beginning with our April 2021 (26/1) issue, in accordance with the new criteria of TR-Dizin, the Declaration of Conflict of Interest and the Declaration of Author Contribution forms fulfilled and signed by all authors are required as well as the Copyright form during the initial submission of the manuscript. Furthermore two new sections, i.e. ‘Conflict of Interest’ and ‘Author Contribution’, should be added to the manuscript. Links of those forms that should be submitted with the initial manuscript can be found in our 'Author Guidelines' and 'Submission Procedure' pages. The manuscript template is also updated. For articles reviewed and accepted for publication in our 2021 and ongoing issues and for articles currently under review process, those forms should also be fulfilled, signed and uploaded to the system by authors.