Fuzzy-DRL: Cost and Energy Efficient Task Scheduling in Cloud-Fog Computing Environment

Task Scheduling (TS) is a key feature of Cloud-Fog Computing (CFC) that ensures the efficient execution of latency-restrictive and computation-intensive applications and ensures the efficient utilization of the system and its sustainability. In this paper, a hybrid Fuzzy Logic and Deep Reinforcement Learning (FL-DRL)-based task scheduling model that co-optimizes makespan, energy cost, and execution cost will be proposed to operate in heterogeneous cloud-fog systems. The utilization of fuzzy logic is used to deal with uncertainty and dynamically prioritize tasks depending on the nature of workload and resource conditions and a deep reinforcement learning agent is used to learn the best policies of task-to-resource allocation through ongoing interaction with the environment. The suggested framework is tested on Alibaba GPUs workload traces and analysed in the SimPy simulation environment and evaluated in terms of its performance in comparison with ACO, pure DRL, and EEOA algorithms. Experimental outcomes show that FL-DRL can achieve as much as 60 % of makespan reduction, 57 % energy savings, and 45 % baseline method execution costs reduction with different task loads. These significant improvements confirm that integrating uncertainty-aware fuzzy reasoning with learning-driven decision-making enables robust and scalable multi-objective scheduling. The suggested FL-DRL model is an efficient framework that can be applied to energy-efficient, low-cost, and high-performance tasks scheduling in the next-generation cloud-fog computing systems.