Machine Learning-Based Methods for Enhancement of UAV-NOMA and D2D Cooperative Networks.

Area of Science:

• Wireless Communication Networks
• Machine Learning Applications
• Network Optimization

Background:

• Cooperative aerial and device-to-device (D2D) networks with non-orthogonal multiple access (NOMA) are crucial for future wireless systems.
• Artificial neural networks (ANNs) offer significant potential for enhancing performance in fifth-generation (5G) and beyond wireless networks.

Purpose of the Study:

• To develop and evaluate an ANN-based unmanned aerial vehicle (UAV) placement scheme for integrated UAV-D2D NOMA cooperative networks.
• To enhance overall communication quality and network efficiency through optimized UAV integration.

Main Methods:

• A novel placement scheme selection (PSS) method combining supervised and unsupervised machine learning (ML) techniques.
• Utilizing a two-hidden layered ANN with 63 neurons for supervised classification to select between k-means or k-medoids unsupervised learning.
• Implementing a cooperative scheme where UAVs serve NOMA pairs while D2D transmissions enhance communication quality.

Main Results:

• The proposed ANN model achieved a high accuracy of 94.12% for placement scheme predictions in urban environments.
• The integrated UAV-D2D NOMA cooperative network demonstrated significant gains in sum rate and spectral efficiency compared to conventional methods.
• The scheme effectively improves communication quality by enabling simultaneous NOMA service from the UAV and D2D cooperative transmissions.

Conclusions:

• The ANN-based UAV placement scheme is highly effective for optimizing integrated UAV-D2D NOMA cooperative networks.
• This approach offers substantial improvements in network performance metrics like sum rate and spectral efficiency.
• The proposed method represents a promising advancement for next-generation wireless network design and deployment.