Next-generation networks, such as millimeter-wave LAN, broadband wireless access systems, and 5th or 6th generation (5G/6G) networks, require enhanced security, diminished latency, and augmented reliability. Efficient congestion management is crucial for 5G/6G technologies, enabling operators to monitor many network instances on a unified infrastructure to provide enhanced quality of service (QoS). The increasing network traffic generated by these systems requires advanced methods for load balancing, preventing network slice failures, and offering alternatives when overloads or slice failures. This study introduces a reliable and efficient hybrid deep learning-based method for congestion reduction. The model combines Long Short-Term Memory (LSTM) and Support Vector Machine (SVM) techniques to improve traffic prediction and resource distribution. The model achieved an overall accuracy of 93.23% during a one-week simulation with unidentified gadgets and variable settings. Additional metrics, such as specificity, recall, time efficiency, and F-score, further demonstrate the model’s effectiveness in mitigating congestion and enhancing network performance.
Next-generation networks, such as millimeter-wave LAN, broadband wireless access systems, and 5th or 6th generation (5G/6G) networks, require enhanced security, diminished latency, and augmented reliability. Efficient congestion management is crucial for 5G/6G technologies, enabling operators to monitor many network instances on a unified infrastructure to provide enhanced quality of service (QoS). The increasing network traffic generated by these systems requires advanced methods for load balancing, preventing network slice failures, and offering alternatives when overloads or slice failures. This study introduces a reliable and efficient hybrid deep learning-based method for congestion reduction. The model combines Long Short-Term Memory (LSTM) and Support Vector Machine (SVM) techniques to improve traffic prediction and resource distribution. The model achieved an overall accuracy of 93.23% during a one-week simulation with unidentified gadgets and variable settings. Additional metrics, such as specificity, recall, time efficiency, and F-score, further demonstrate the model’s effectiveness in mitigating congestion and enhancing network performance. Read More
