Optimizing Server Load Distribution in Multimedia IoT Environments through LSTM-Based Predictive Algorithms
To address server overload and degraded Quality of Service (QoS)/Quality of Experience (QoE) in multimedia Internet of Medical Things (IoMT) networks—caused by explosive device proliferation and surging traffic—this paper proposes an SDN-driven dynamic load balancing framework. The method uniquely integrates a Long Short-Term Memory (LSTM) time-series prediction model with a fuzzy logic system within the SDN control plane, enabling fine-grained, adaptive scheduling of multimedia flows. LSTM accurately forecasts short-term server load trends, while the fuzzy logic system dynamically determines optimal flow-table installation policies in real time, enhancing both scheduling responsiveness and robustness. Simulation results demonstrate that, compared to baseline approaches, the proposed method reduces service response latency by 32%, improves server resource utilization by 27%, lowers operational costs by 19%, and significantly enhances end-user QoE.