5G Tactile Internet Modeling And Performance Research For Mission-critical Internet Of Things Service

Mission-Critical Internet-of-Things(MC-Io T)will play a vital role in remote healthcare,haptic interaction,and industrial automation.On the one hand,in such application fields,haptic applications have become more critical.Benefit from the development of the fifth generation(5G)wireless communication networks and the technological advances of Internet of Things(Io T),the Tactile Internet has been envisioned as a promising enabler of MC-Io T services.On the other hand,different MC-Io T services have diverse requirements.This requires a flexible network architecture for enabling different MC-Io T services.With the surge in computing needs from different MC-Io T services,multi-access edge computing(MEC)can further meet MC-Io T service requirements at the network edge.To provide MC-Io T services flexibly,a 5G network architecture based on the network function virtualization(NFV)technology is designed to support the implementation of the Tactile Internet.Moreover,a utility function model is proposed to evaluate the performance of the 5G NFV-based Tactile Internet.Considering the just-noticeable difference(JND)in human perception and the corresponding network costs on providing MC-Io T services,a hum An per Ception-based Tactile Internet utility Optimizatio N(ACTION)algorithm is developed to optimize the 5G NFV-based Tactile Internet utility.Simulation results indicate that the maximum utility achieved under the proposed ACTION algorithm is improved by 35.4% to the network slice requests(NSRs)implementation algorithm.Further,considering the computational delay and the computational energy consumption required to complete the MC-Io T services,a cost function model has been constructed.To minimize the system cost,a task offloading and resource optimization method based on reinforcement learning is developed.Simulation results show that the maximum system cost under the Q-learning based reinforcement learning method is reduced by 86.6%,89.7%,and 92.1% compared to the local computing mode,AP computing mode,and cloud computing mode,respectively.