IRAM5G : Intelligent Resource Allocation in 5G Networks for Enhanced Mobile Computing

Abstract

The implementation of 5G networks necessitates novel resource allocation strategies to satisfy requirements for elevated throughput, minimal latency, and optimal bandwidth utilization. Conventional approaches frequently squander resources by emphasizing equal data rates across varied applications. IRAM5G, a machine learning-driven system, tackles these difficulties with real-time, predictive resource management, using technologies such as Network Function Virtualization (NFV) and Software-Defined Networking (SDN). This method improves network adaptability, speed, and fairness, providing excellent support for applications such as enhanced mobile broadband (eMBB) and ultrareliable low-latency communication (URLLC). It consists of 16,539 training samples and 345 testing samples. The dataset includes an unbalanced class distribution with 5,757 unique classes in the training set and 220 unique classes in the test set. Preprocessing steps included data normalization, handling missing values, and feature selection to enhance model performance.