The Research On Intelligent Access Mechanism For Machine Type Communication

Internet of things(IoT)has been renewed dramatically as the widespread emerging diverse applications and the large-scale connection of user equipments in 5G-and-beyond systems,which graudually attracts lots of attention in both academia and industry where the researchers agree that emerging industries(i.e.,smart transportation,smart medical care)will set extremely high standards in different communication scenarios,i.e.,enhanced mobile broadband(eMBB),massive machine-type communications(mMTC),and ultra-reliable and low-latency communications(URLLC).Many studies show that the number of machine type communication devices(MTCDs)will reach about 50 billion by 2020.Specifically,low-power massive-connection has been identified as one of the key applications for 5G,with focusing on the service requirements of MTC,IoT and other vertically integrated industries.How to efficiently and intelligently access MTCDs in 5G scenarios to reduce network congestion is a hot issue that needs to be solved urgently.This article focuses on smart access mechanisms and resource allocation in MTC scenarios,which are summarized as follows:1)In the heterogeneous network scheme,the thesis proposes an intelligent access mechanism for MTC.Taking the number of devices connected to the network as the target and the system throughput and the average transmission rate of the devices as performance indicators,a theoretical optimization model of MTC intelligent access is established.Through the analysis of the model,the thesis uses genetic algorithm(GA)to solve the model and design the corresponding intelligent access mechanism.The simulation results show that compared with the traditional algorithm based on RSS algorithm and greedy algorithm,the proposed genetic algorithm based intelligent access mechanism(GAIAM)can achieve better performance in terms of access success rate,system throughput and average transmission rate.2)For the group paging network scenario,the thesis first considers the MTCD service type and mobility in the network,then group the devices.Using intelligent learning,the thesis observes the available resources and load conditions of the current base station,adjusts the access requests of devices dynamically,and obtains the better network performance.In this study,the thesis models the problem of devices access slots as a markov decision process(MDP),and based on multi-agent reinforcement learning(MARL),the thesis proposes a Q-learning based intelligent access mechanism.Through simulation experiments,the effectiveness of the algorithm is verified under the conditions of sufficient and insufficient base station resources.Numerical results show that the algorithm can effectively improve the MTCD access performance.3)The unique backoff mechanism of MTC can effectively reduce the collision probability of the network and avoid network congestion.Due to the dynamic nature of the network,the access selection of one device will affects the available resources in the network,and then affects the access selection of other devices.The thesis considers the combination of the backoff mechanism and the ACB mechanism,models the queuing model of the backoff mechanism as an MDP,and proposes a new backoff mechanism based on the AC algorithm.By optimizing the strategy parameters and estimating the value function,the optimal backoff access mechanism in the current state can be got.It can be obtained from the simulation results that the algorithm proposed in this study can effectively improve the access capability of MTC networks and reduce the access delay.According to the characteristics of the MTC communication network,the thesis analyzes the feasibility of machine learning in communication network scenarios,and designs an intelligent access mechanism for the actual network scenario,which improves network access capabilities,reduces network congestion,and provides a new idea for future AI-based MTC access control solutions.