Enhancing Next-Generation Wireless Communication Networks Through AI-Driven Resource Allocation and Intelligent Spectrum Management

Abstract

Next generation wireless communication networks require intelligent mechanisms to support high data rates, massive connectivity, low latency and efficient resource utilization. In dynamic 5G, beyond 5G and emerging 6G scenarios, traditional resource allocation and spectrum management techniques usually encounter challenges. In this paper, we propose an AI-based approach for QoS prediction and intelligent bandwidth/spectrum resource management based on 5G resource allocation dataset. The methodology consisted of data pre-processing, removal of duplicates, prevention of leakage, feature engineering, model training, validation and simulation-based resource-allocation analysis. We tried out different regression models like Linear Regression, Ridge, Lasso, Decision Tree, Random Forest, Extra Trees, Gradient Boosting and MLP Regressor. The tuned Random Forest had the best performance with R2, MAE and RMSE of 0.8954, 0.6204 and 0.9250 on the independent test dataset. The feature-importance analysis showed that resource_density was the most important predictor, i.e., the combination of CPU and available bandwidth per user is a strong determinant of QoS. Simulation results also indicated that QoS decreases with increasing user density, necessitating adaptive monitoring and dynamic resource management. The study concludes that QoS-aware resource planning in next-generation wireless networks can be aided by AI-based prediction.