Multi‐Objective Deep Q‐Network for Efficient Application Placement in Fog Computing

ABSTRACT From smart sensors to wearable monitors, the increasing usage of IoT devices in healthcare has elevated effective resource management to a high priority. These devices generate massive volumes of sensitive real‐time data that require immediate processing, strict privacy safeguards, and reliable system performance. By bringing processing power closer to patients, healthcare professionals, and medical equipment, fog computing overcomes these issues and offers better data security and faster reaction times than conventional cloud‐based systems. However, the dynamic and resource‐constrained nature of fog environments presents significant challenges for application placement and load balancing. In this study, we propose a novel deep‐reinforcement‐learning‐based method identified as dueling double deep Q‐networks (D3QN) which integrates capability‐aware node modeling and a comprehensive reward function to ensure intelligent, efficient, and balanced application placement across fog computing nodes. To guarantee experimental consistency and realism, we employ iFogSim to generate heterogeneous and dynamic fog computing environments that accurately emulate healthcare scenarios. These environments are directly integrated into our Python‐based D3QN framework, enabling comprehensive system performance. The experimental results demonstrate that the proposed strategy delivers significant improvements in placement time, latency, energy efficiency, and balanced resource utilization within a fog computing environment.