Deep Learning and Optimization Approaches in Joint Power and Delay Optimization Based Resource Allocation in MIMO-OFDM System- Deep Convolutional Red Piranha Pyramid-Dilated Neural Network: A Review

Abstract

The increasing demand for high data rates, low latency, and efficient spectrum utilization in 5G and emerging 6G networks has made resource allocation in MIMO-OFDM systems a critical research area. Joint optimization of power and delay is essential for achieving high spectral efficiency while satisfying Quality of Service (QoS) requirements. Traditional optimization techniques such as convex optimization and heuristic algorithms face limitations due to high computational complexity and inability to adapt to dynamic wireless environments. Recently, deep learning-based approaches have emerged as promising solutions for resource allocation problems. These methods leverage neural networks to learn complex relationships between channel conditions and resource allocation decisions. Advanced architectures such as convolutional neural networks (CNNs), reinforcement learning (RL), and hybrid models have demonstrated superior performance in optimizing power and delay simultaneously. Furthermore, pyramid and dilated convolution structures enable efficient multi-scale feature extraction, improving optimization accuracy.  This paper provides a comprehensive review of deep learning and optimization approaches for joint power and delay optimization in MIMO-OFDM systems. It analyses recent advancements, identifies key challenges, and highlights future research directions. The study shows that deep learning-based methods significantly outperform conventional techniques in terms of efficiency, adaptability, and scalability, making them suitable for next-generation wireless communication systems.