This systematic literature review analyzes machine learning (ML)-based techniques for resource management in fog computing. Utilizing the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol, this paper focuses on ML and deep learning (DL) solutions. Resource management in the fog computing domain was thoroughly analyzed by identifying the key factors and constraints. A total of 68 research papers of extended versions were finally selected and included in this study. The findings highlight a strong preference for DL in addressing resource management challenges within a fog computing paradigm, i.e., 66% of the reviewed articles leveraged DL techniques, while 34% utilized ML. Key factors such as latency, energy consumption, task scheduling, and QoS are interconnected and critical for resource management optimization. The analysis reveals that latency, energy consumption, and QoS are the prime factors addressed in the literature on ML-based fog computing resource management. Latency is the most frequently addressed parameter, investigated in 77% of the articles, followed by energy consumption and task scheduling at 44% and 33%, respectively. Furthermore, according to our evaluation, an extensive range of challenges, i.e., computational resource and latency, scalability and management, data availability and quality, and model complexity and interpretability, are addressed by employing 73, 53, 45, and 46 ML/DL techniques, respectively.

UAV-assisted fog computing places drone-mounted edge nodes between IoT devices and the cloud, enabling ultra-low-latency services where fixed infrastructure is unavailable. However, battery constraints, node heterogeneity, and bursty workloads make container scheduling particularly difficult: unnecessary migrations deplete drone battery and violate SLA deadlines. We propose MORA (Multi-Objective Resource Allocation), a stability-oriented deep reinforcement learning scheduler based on Twin Delayed Deep Deterministic Policy Gradient (TD3), with a normalized multi-objective reward that jointly optimizes energy, latency, SLA adherence, and proactively minimizes container migrations. Evaluation via COSCO under AzureFog traces over 200 intervals demonstrates that MORA achieves 79.6% task completion, reduces migrations by 94% versus DRLMOTS (27 vs. 182), cuts energy by 21–27%, and limits SLA violations to 47 over 200 intervals with a stable per-interval range of 9–14, outperforming TD3, DDPG, and DRLMOTS across all metrics.