<p>The Cloud-Based (CB) Internet of Things (IoT) ecosystem is commonly used to run latency-sensitive applications such as surveillance, healthcare monitoring, and online gaming. However, the exponential growth in IoT devices leads to an overwhelming amount of data and service requests. Due to the centralized nature of cloud data centers, which are typically located far from edge devices, CB architectures often struggle to meet such applications’ stringent real-time processing requirements. This results in degraded Quality of Service (QoS) experienced by end users. Fog computing has emerged as a promising solution to address these limitations by positioning computational and storage resources closer to the data source, often within a single hop. It extends cloud computing capabilities to the edge of the network, thereby reducing latency and bandwidth usage and enabling faster response times for critical IoT applications. This study proposes and implements an efficient Electroencephalography (EEG)-driven Emotion Recognition System (ERS) deployed on a Fog computing-based infrastructure. The system extracts features from 105 EEG-derived signals and employs Support Vector Machines (SVMs) to classify emotions according to the three affective dimensions of the Valence-Arousal-Dominance (VAD) model. Additionally, the deployment strategy is QoS-aware and tailored for Fog computing environments. Experiments were conducted on the AMIGOS dataset, and the proposed ERS model demonstrated improved classification accuracy for valence and arousal, along with significant recognition capability for the dominance dimension, which was previously underrepresented in the baseline. The implementation was executed in a simulated Fog computing environment using the iFogSim toolkit. The results indicate that the proposed approach substantially improves QoS metrics, underscoring the benefits of Fog computing for real-time, resource-aware IoT applications like ERS.</p>