Reinforcement Learning-Based Resource Sharing in Hierarchical Architecture for Fog Computing Environment with Stackelberg Game Approach

• Shifa Manihar,
• Ravindra Patel,
• Sanjay Agrawal,

Vol. 19, No. 10, October 31, 2025
10.3837/tiis.2025.10.010, Download Paper (Free):

• Reinforcement Learning
• Stackelberg Game
• Fog Computing
• Resource Allocation
• game theory

Abstract

Machine learning's newest hot commodity, reinforcement learning, allows agents to make smart decisions in response to the ever-changing dynamics of their environments. Fog nodes distribute processing tasks requested by Internet of Things (IoT) devices, thereby reducing response times in IoT-based systems, rather than relying on centralized cloud servers. However, this benefit is challenging to achieve in systems with high request rates due to the delayed and presumably ineffective offloading caused by the long task queues in the fog nodes. Hence, this paper, Reinforcement Learning-Based Resource Sharing in Hierarchical Architecture (RL-RSHA), has proposed improving performance in a fog computing environment by offloading and executing tasks. The suggested method highlights the demand-supply mismatch using a Stackelberg game-based resource-sharing method. It encourages all computing providers (CPs) to trade in a way that accurately reflects their resource capacity and needs. Users' service selection among CPs is also analyzed using an evolutionary game-based replicator dynamics. Using the Limit Searching Algorithm (LSA) to determine the ceiling usage, the Greedy Energy-aware Algorithm (GEA) is used to choose functional servers. Simulated results demonstrate that the proposed method for sharing resources operates as expected, revealing where user selection and allocation converge and reach equilibrium.

Statistics

Cite this article