Deep Learning based Adaptive Congestion Control in 5G Wireless Networks

Abstract

The rapid expansion of data-intensive applications such as IoT, ultra-high-definition video streaming, augmented reality, and mission-critical communications has significantly increased traffic demands in 5G wireless networks. This surge in heterogeneous traffic leads to severe congestion, packet loss, increased latency, and degradation of Quality of Service (QoS). Traditional congestion control mechanisms, including rule-based and TCP-variant approaches, rely on static threshold parameters and lack the capability to adapt dynamically to highly fluctuating 5G network environments. To address these limitations, this paper proposes a Deep Learning-based Adaptive Congestion Control framework designed specifically for 5G wireless networks. The proposed model employs a Long Short-Term Memory (LSTM) network to learn temporal traffic patterns using real-time network metrics such as throughput, packet loss rate, round-trip time (RTT), and buffer occupancy. Based on predicted congestion states, the system dynamically adjusts transmission rates and congestion window parameters to optimize network performance. Simulation results demonstrate that the proposed approach significantly improves throughput, reduces end-to-end delay, and minimizes packet loss compared to conventional congestion control algorithms. The adaptive learning capability of the model enhances network stability and ensures efficient resource utilization in dense 5G environments. The proposed framework provides a scalable and intelligent solution for next-generation wireless communication systems.