Efficient resource allocation through CNN-game theory based network slicing recognition for next-generation networks

Abstract

Fifth generation (5G) and sixth generation (6G) networks are examples of next-generation networks that need higher levels of safety, lower latency, and more capacity and dependability. Reconfigurable wireless connection slicing becomes essential for satisfying these sophisticated networks' requirements, enabling many network instances on the same hardware to improve Quality of Service (QoS). Nonetheless, the centrally managed resource allocation for network slicers presents difficulties, particularly as the quantity of User Equipment (UEs) increases. This puts pressure on Radio Resource Management (RRM) and makes slice customization more difficult. In order to address these issues, this study presents an organizational radio resource distribution architecture in which the neighborhood radio resource managers (LRRMs) receive sub channel allocations from the RRM in slices, and the LRRMs then distribute the assigned capabilities to the corresponding UEs. The suggested model, which runs in MATLAB, uses an original method called CNN-Game Theory to achieve an exceptional 98 % accuracy, outperforming CNN-LSTM, RNN, DeepCog, and DHOA by 29.27 %. This method combines ideas from game theory with neural network weight optimization to produce an improved model with increased efficiency and accuracy. Many experiments illustrate how effective this method is and how it can be used to improve different machine learning applications. Metrics like slice type utilization, average packet delay for each LTE/5G category, and others are used to assess game optimization for resource allocation