5G Network Slicing Using Deep Learning for Hospital of The Future

Effective health management is essential, yet hindered by challenges in traditional healthcare systems and an uneven physician-to-population ratio. The integration of 5G networks improves communication in healthcare. This paper delves into integrating deep learning (DL) within network slicing to provide tailored solutions for the Hospital of the Future (HoF). To the best of our knowledge, this paper presents the first instance of classification techniques being used in network slicing using DL demonstrated via OMNeT simulations. We evaluate three scenarios namely, network slicing using DL, network slicing without DL, and unsliced network in terms of throughput and delay. Throughput result for URLLC network slicing using DL shows approximately a 33.33 times improvement compared to network slicing without DL and unsliced network, while eMBB network slicing using DL exhibits approximately a 10 times improvement. Additionally, mMTC network slicing using DL demonstrates a 53% improvement. Regarding delay, URLLC network slicing using DL exhibits the lowest delay compared to network slicing without DL and unsliced network, while in eMBB, network slicing using DL shows the second lowest delay. In mMTC slice, network slicing using DL shows an overlapping performance with unsliced networks, and network slicing without DL exhibits the lowest delay. It’s noteworthy that the differences in delay among eMBB, mMTC, and URLLC slices compared network slicing without DL and unsliced network slices are minimal, approximately less than 1%. The intelligent distribution of resources by DL makes it ideal for critical healthcare applications, surpassing alternatives in heterogeneous networks.