Reliability and Trust Aware Task Scheduler for Cloud-Fog Computing Using Advantage Actor Critic (A2C) Algorithm

Abstract

Because of the wide variety of heterogeneous jobs and the internet’s inherent complexity in relation to runtime capabilities, task scheduling in cloud computing poses a paradigm issue. Some tasks are very computationally intensive and sensitive to delays, which is a problem. This research uses fog computing to streamline computational resources for scheduling tasks that are computationally intensive and sensitive to delays in order to address these challenges. For cloud and fog computing, RTATSA2C, a trust-and reliability-aware task scheduler is proposed that optimizes the scheduling process for scalability. Among its features, RTATSA2C is a method for task segmentation that improves dependability, scalability, trust, and makespan efficiency. The number of Virtual Machines (VMs) is automatically adjusted by the scheduler to match the demands of the workload, allowing for dynamic resource scaling. The integration of Advantage Actor-Critic (A2C) model with RTATSA2C reinforcement learning is performed for improved decision making in task scheduling. Simulation results showed a significant reduction of makespan while improving reliability, fault tolerance, trust, and scalability. The proposed RTATSA2C task scheduling dynamically determines whether to schedule tasks on fog nodes or cloud nodes by assigning actor-critic policies to tasks and virtual machines. SimPy is used to implement the algorithm and used real-time datasets from Google Cloud Jobs for simulation. RTATSA2C is evaluated against baseline algorithms, including Long Short-Term Memory (LSTM) and Deep Q-learning Network (DQN).

Keywords

• Fog computing
• reinforcement learning
• makespan
• scalability
• trust
• reliability