Deadline-Aware Task Scheduling in Fog-Cloud Computing Using Multi-Agent Reinforcement Learning and Software-Defined Network Security

Task offloading and resource scheduling in fog-cloud Internet of Things environments face significant challenges, including high latency, constrained throughput, and unpredictable network conditions. These limitations hinder real-time responsiveness and efficient resource utilization, particularly in mission-critical Internet of Things applications. Moreover, ensuring robust data security under such dynamic and latency-sensitive scenarios is vital, as unsecured task execution and data exchange can lead to severe vulnerabilities. Therefore, optimizing both performance and security in low-latency conditions remains a crucial requirement for reliable and scalable fog-cloud computing infrastructures. Hence, this paper proposes a novel task scheduling framework such as Type−2 Fuzzy Multi-Agent Reinforcement Learning with Cauchy Mutation War Optimization algorithm within a secure Software-Defined Network architecture. The proposed model improves decision-making under uncertainty by analyzing the task scheduling process and optimizes resource allocation to strengthen network security against malicious attacks. The Cauchy mutation incorporates with war competition to explore the effectiveness of improving security and validates the control of dynamic functionality by estimating the routing process. The experimental results are analyzed by varied metrics and two benchmark datasets such as NASA Ames Research Center iPSC/860 and High Performance Computing Center North that demonstrate the superiority of the proposed model over state-of-the-art techniques. The results revealed that the latency is minimized for the proposed model by 43% and maximized throughput by 82.3% with better quality of service at 69%, and enhanced network security by 78.2%. Also, the proposed method diminishes response time by 37 s and optimizes resource utilization to conform to the robustness and efficiency in real-time Internet of Things applications. Thus, the results validate the capability of the proposed framework by improving offloading strategies with secure and scalable task scheduling.