Research On Key Technologies Of D2D And M2M Terminal Access

The evolution of mobile network is a process of constantly improving the utilization of limited storage,computing and transmission resources to meet the constantly critical demands of users,especially in the case of terminal access technologies.In particular,the Device-to-Device(D2D)and Machine-to-Machine(M2M)technologies have been extensively investigated for their potential to address many technical challenges in future wireless networks.In this paper,on the one hand,we present an indepth study of social-aware D2MD based content distribution and D2Dbased computing offloading.On the other hand,we also make an in-depth study on random access control of mass M2M communication.Specifically,the first part of the paper includes the study of D2Doriented terminal access technology under two typical scenarios:Mobile Content Distribution(MCD)based on D2MD and Multi-access Edge Computing(MEC)based on D2D.Study Point 1:User Grouping for Social-Aware Device-to-MultiDevice(SA-D2MD)Video Distribution.Offloading BS traffic in case of asynchronous user requests while considering their time-varying characteristics and the Quality of Experience(QoE)of Video Request Users(VRUs)is a pressing problem.We exploit communication stability and Video Loading Duration(VLD)-tolerant property to group VRUs and Seed Users(SUs)to offload BS traffic.Further,a social-aware D2MD user selection algorithm based on finite horizon optimal stopping theory and a social-aware D2MD user matching algorithm based on heuristic ideas are proposed to solve the aforementioned problem.The simulation experimental results show that our algorithms outperform prevalent algorithms.Study Point 2:Reinforcement Learning-Based User Grouping for SA-D2MD Content Distribution.In order to address the problem that spatial-temporal model and social-aware model of Content Request User(CRU)are complex and difficult to describe,and maximizing BS offloading in the scenario of selfish SU,the author models the problem as a game and long-term optimization joint problem,which can be solved based on a matching-Stackelberg hierarchical game and multi-agent deep Q-learning algorithm.The results of simulation experiments show that compared with existing algorithms,the proposed algorithm improve the BS offloading capacity.Study Point 3:Reinforcement Learning-Based D2D Computing Offloading for Scalable High-efficiency Video Coding(SHVC).To address the problem of minimizing the completion time and energy consumption of SHVC computing task in D2D-based MEC computing offloading scenarios where the computing amount of the SHVC computing task is uncertain and the computing process has multiple execution steps,the author models the SHVC computing tasks by ten atomic operations,and make it has multiple decision-making attributes,and employing a double-agent deep Q-learning algorithm to solve the problem.The simulation results show that the algorithm makes the requestor spend less completion time and energy consumption while maintaining the high video quality.In the second part of the paper,the author studies M2M-oriented terminal access technology under the co-existence of Delay-Tolerance Device(DTD)and Delay-Sensitive Device(DSD)scenario.Study Point 4:Random-Access Control Scheme for DSD Access Success Rate in Massive M2M Communication.When massive M2M devices access the network,they quickly scramble preambles,and induce significant network congestion.In particular,the access success rate of DSD will decrease sharply when DTDs and DSDs coexist.Therefore,we propose a Markov chain-based Access Class Barring(M-ACB)scheme,which can guarantee the random-access success rate of a large number of M2M devices when DTDs and DSDs coexist,while ensuring the efficient use of network resources.Simulation results show that the M-ACB scheme improves the access success rate of M2M devices while reducing the conflict rate.