Moving target defense in 5G and beyond networks: A comprehensive survey and research directions

5G and beyond networks rely on SDN/NFV, network slicing, and multi-access edge computing (MEC) to support highly heterogeneous and mission-critical services; however, the same capabilities enlarge the attack surface and make static protection policies increasingly brittle under fast reconfiguration and adversarial adaptation. Moving Target Defense (MTD) is therefore particularly significant in this setting because it can continuously reshape exposed assets, communication paths, and virtualized functions to invalidate reconnaissance, reduce attacker dwell time, and raise the cost of exploitation. This paper presents a deployment-oriented survey of MTD for 5G, with special emphasis on learning-assisted designs, especially reinforcement learning (RL), that optimize what to mutate, when to mutate, and how to enforce mutation under slice-isolation, URLLC, and orchestration constraints. The reviewed literature is organized by research category and by 5G operational domain, and then linked to mutation targets, timing policies, SDN/NFV implementation components, and the attacks they are designed to mitigate. Beyond cataloguing prior work, the survey clarifies the trade-offs among security gain, reconfiguration overhead, policy stability, scalability, and reproducibility, and highlights benchmark, secure-learning, and deployment-realism gaps that currently limit operational adoption. The resulting taxonomy and synthesis provide a clearer foundation for designing robust, low-overhead, and AI-assisted MTD mechanisms for 5G and emerging 6G networks.