Telecommunication Systems Electrical and Electronic Engineering
Smart resource sharing mechanisms in device-to-device communication for future cellular networks: a review
Abstract
Device-to-Device (D2D) communication has emerged as a pivotal component in modern wireless networks, enabling enhanced spectral efficiency, reduced latency, and direct peer-to-peer connectivity. However, most existing survey articles focus narrowly on conventional resource management or fail to capture recent advances driven by artificial intelligence, edge computing, and programmable radio environments. In this review, we address these gaps by providing a comprehensive and up-to-date analysis of smart resource allocation strategies in D2D-enabled cellular networks. We categorize and compare traditional centralized techniques with modern machine learning-based approaches, including Q-learning, deep reinforcement learning (DRL), and actor-critic methods. Furthermore, this survey explores emerging paradigms such as Reconfigurable Intelligent Surfaces (RIS), Federated Learning (FL) for privacy-preserving collaboration, cognitive D2D communication for dynamic spectrum access, and edge intelligence for real-time optimization in beyond-5G and 6G wireless networks. Based on the limitations and research gaps identified in existing reinforcement learning-based D2D communication studies, the survey further discusses a conceptual DRL-EKF-assisted future research direction for adaptive and intelligent resource management under dynamic wireless environments. Finally, the paper highlights key open challenges and outlines future research opportunities for developing intelligent, scalable, energy-efficient, and secure D2D communication systems for next-generation wireless networks.