Optimizing 5G Sustainable Networks Performance in Urban Areas Using Machine Learning Techniques

ABSTRACT The rapid expansion of 5G networks offers ultra‐reliable low‐latency communication (URLLC), enhanced mobile broadband (eMBB), and massive machine type communication (mMTC) demands for intelligent and adaptive optimization strategies. This is particularly critical in ultra‐dense urban environments, where interference, traffic variability, and energy inefficiencies pose major challenges. This study offers a next‐generation optimization framework driven by machine learning. It combines random forest classifiers, long short‐term memory (LSTM) networks, and gradient boosting regressors. Together, these models improve key performance indicators, such as the signal‐to‐interference‐plus‐noise ratio (SINR), reference signal received power (RSRP), latency, throughput, and energy efficiency. The proposed system demonstrated a strong predictive performance. It achieved an R 2 score of 0.89, mean squared error (MSE) of 0.25, and mean absolute percentage error (MAPE) of 5.20%. Under high load and dense interference conditions, the framework delivered a 22.4% increase in average throughput, a 17.8% reduction in packet loss, and a 19.5% drop in end‐toto‐end latency. The LSTM model effectively responded to changes in signal strength caused by user mobility. The random forest classifier detects interference with 93.4% accuracy, enabling quick mitigating action. Moreover, the energy efficiency of the proposed system is improved by 18.7%, making it both optimized and cost‐effective for better performance evaluation. These outcomes show that future 5G infrastructure depends on artificial intelligence‐driven self‐organizing, self‐optimizing networks. This framework provides a reliable path toward scalable, resilient, and adaptive service delivery in smart cities.