Publication Details
Abstract
Transitioning to dedicated resources for smart treatment using industrial technology is not just an improvement, but a necessity for all the risks facing networks in the unlikely future of 6G. We are working to create self-organizing networks that dynamically optimize resources to provide unparalleled service and build trust. This is achieved through results of e-learning, particularly in deep reinforcement learning. In this paper, a deep learning approach is used to study intelligent resource allocation in wireless communication networks. First, the concepts related to CSCN architecture are discussed. The throughput of small base stations (SBS) in CSCN architecture is analyzed. Next, the long short-term memory network (LSTM) model is used to predict users’ mobile locations. Transmission conditions of users are scored based on two factors: multiple network environments, and their real-time adaptability and computational efficiency are verified. Compared with the SOTA method, TRDM shows better real-time adaptability, especially under low latency and dynamic load conditions. In addition, TRDM maintains high computational efficiency in complex network environments. It is suitable for practical applications such as large-scale IoT or urban cellular networks. The experimental settings include hyperparameters such as a 6-layer self-attention network of the Transformer architecture, a learning rate of 0.001, and a batch size of 128. Through comparative experiments, TRDM outperforms existing methods in multiple key performance indicators, with statistically significant differences. This study offers an effective solution for real-time network scheduling and provides important reference for future deployment in IoT and remote urban networks.