FR-CNN: The optimal method for slicing fifth-generation networks through the application of deep learning

Abstract

The 5G network is expected to accommodate numerous novel use cases originating from vertical businesses in mobile broadband communication service. Higher standards of execution, affordability, security, and board-level adaptability are only a few of the difficult needs brought on by these recently changed conditions. The current organizational strategy of using a one-size-fits-all blueprint is not practicable. An emerging strategy for sustainably meeting these diverse criteria is to split a single physical network into multiple logical networks, each tailored to a unique set of requirements. The authors of this work created a hybrid learning approach to network slicing. Improving weighted feature extraction (OWFE), data collection, and slicing classification are the three processes recommended for this work. A dataset of 5G network slices is used as an initial input. This dataset contains metrics such as bandwidth, duration, modulation type, delay rate, jitter, speed, user device type, packet loss ratio, and packet delay budget. The last step is to use the Faster R-CNN model, which includes the RPN model, to classify the values provided. From this model, one can generate precise network slices like URLLC, mMTC, and eMBB. A change in the configuration of accurate 5G organization slicing would be brought about by the suggested approach, according to the findings of the study.