Research On Radio Resource Management For 5G Cellular Vehicle-to-Everything Communication

With the rapid development of the fifth generation(5G)and Internet of Things technology,there has been much research attention focused on the Internet of Vehicles.Cellular Vehicle-to-Everything(C-V2X)communication can operate in the commercial cellular spectrum and can be compatible with existing cellular networks.However,the rapid increase in the number of vehicles and the explosive growth of data traffic make the spectrum resources of the unlicensed band increasingly scarce.Nevertheless the corresponding unlicensed spectrum resources are quite abundant and the shortage of spectrum resources can be greatly alleviated by fully utilizing the unlicensed spectrum.Another,the high-speed development of the Internet of Vehicles also provides a variety of application services for vehicle terminals,whereas vehicle terminals usually have low computing ability and limited storage space.These applications with high computing requirements present a huge challenge to existing vehicle terminals,especially requiring more computing resources.The mobile edge computing(MEC)-based vehicle network which offloads computing tasks to the MEC servers not only solves the drawbacks of limited computing ability of the vehicle terminals,but also satisfies the requirements of the vehicle network for calculating delay.The main contents and innovations are briefly summarized as follows:Firstly,considering the shortage of licensed spectrum resources and sufficient unlicensed spectrum resources,this thesis proposes a coexistence scenario between C-V2X system and WiFi system.On the premise of ensuring C-V2X users’ quality of service(QoS)requirements,a portion of the C-V2X users are offloaded to the unlicensed band.In this thesis,the uplink transmission power of C-V2X users and the number of offloaded C-V2X users are jointly optimized to maximize the throughput of C-V2X systems,Long Term Evolution(LTE)systems and WiFi systems,and a multi-objective optimization problem is formulated.Different from the traditional methods of solving multi-objective optimization problems,the advanced nondominated sorting genetic algorithm III(NSGA-Ⅲ)is adopted to solve the original problem and the Pareto optimal solution set is obtained.The convergence of the algorithm is studied in simulations and the proposed algorithm outperforms the traditional single-objective optimization algorithm in terms of fairness and throughput.Secondly,for multiple C-V2X users offloaded to unlicensed spectrums,this thesis proposes a listen before talk(LBT)-based vehicle formation mechanism,by which a platoon of C-V2X users is formed to compete for the resources of unlicensed spectrums by the first vehicle in the platoon.The simulation verifies that the collision probability of users in unlicensed bands is effectively reduced and the communication delay of the C-V2X users is decreased significantly by using the proposed platoon formation mechanism.Finally,for the scenario with highly-efficient computation,this thesis proposes a C-V2X users’ tasks offloaded algorithm to achieve network load balancing in a multi-cell multi-MEC servers scenario.Aiming at minimizing the cost function,this thesis jointly optimizes the offloaded decision,the uplink transmission power of C-V2X users and computing resources of MEC servers.The mobility of the C-V2X users cause the state of the network to change dynamically,and the traditional optimization algorithm can not be applied in such a dynamic scenario.Therefore,double deep Q-learning network(Double DQN)framework in deep reinforcement learning is adopted to solve the problem.The convergence of the algorithm is analyzed by simulations,and it is verified that the proposed algorithm can obtain better performance in terms of the cost consumption than the traditional single-cell single-MEC server scene.