Systematic Review of Task Scheduling in Fog–Cloud Computing: Machine Learning and Metaheuristic Approaches

Fog–cloud computing has emerged as a hybrid infrastructure designed to support low-latency, scalable, and intelligent processing for modern Internet of Things (IoT) applications. Efficient task scheduling within this environment remains a complex and NP-hard challenge due to heterogeneous distributed resources, fluctuating workloads, and diverse Quality of Service (QoS) requirements. This study presents a systematic review of state-of-the-art task scheduling techniques published between 2018 and 2025, with a focus on machine learning, heuristic-based, and metaheuristic-based approaches. The reviewed literature was classified into a taxonomy highlighting methodological characteristics, evaluation metrics, and simulation tools used in fog–cloud environments. Analysis of the selected studies indicates that energy consumption and makespan time are the most prioritized performance metrics, while Python is the most frequently used simulation environment. The findings reveal that although metaheuristic algorithms dominate the field due to their adaptability and effectiveness in large search spaces, hybrid approaches integrating artificial intelligence show increasing potential for improving scheduling performance. Finally, research gaps, limitations, and future directions are discussed, emphasizing the need for real-world implementations, improved QoS handling, and advanced hybrid optimization frameworks