Dynamic Bandwidth Allocation Technology Of Internet Of Vehicles Based On Proximal Policy Optimization Algorithm

The rapid increase in the number of vehicles has overloaded the traffic system,and the limited functions of the vehicle warning system and the fatigued driving of the driver have led to an increasing number of traffic accidents.In order to ensure the safety of vehicles during driving,the potential of vehicle communication in improving the intelligence of traffic systems has attracted great attention in recent years.Due to the high mobility of Internet of vehicles communication terminals and the limitation of bandwidth and computing resources,bandwidth resources must be used reasonably and effectively.However,in practice,the high mobility of the vehicle makes the channel change time-varying and the actual vehicle computing resources are limited.The traditional algorithm cannot make real-time decision-making,and the computational complexity is high.The traditional deep reinforcement learning algorithm is limited by the state and action space.Dimensional training is inefficient.In this context,how to perform dynamic bandwidth resource allocation for each link to improve system resource utilization is particularly important.Based on 3GPP urban vehicle communication scenarios,this thesis proposes a deep reinforcement learning algorithm under random strategy actor-critic mode to solve the problem of dynamic bandwidth resource allocation for vehicle communication.In a specific vehicle-to-vehicle communication scenario,considering the strong time-varying characteristics of the inter-vehicle channel under large-scale and smallscale fading,this thesis firstly constructs the dynamic bandwidth resource optimization problem under the vehicle mobility communication model,and based on the vehicle networking scenario A multi-objective optimization problem of bandwidth allocation for vehicle-to-vehicle communication is established to build a model and a simulation platform.Then,according to the characteristics of vehicle-to-vehicle communication scenarios,a state,action and revenue transformation mechanism suitable for vehicle-to-vehicle communication is designed,and a near-end strategy is proposed.The optimization algorithm for the dynamic bandwidth allocation method of the Internet of Vehicles realizes the real-time dynamic bandwidth resource allocation under the continuous high-dimensional state action space of the Internet of Vehicles scene and the strong time-varying channel state.The simulation results show that the method proposed in this thesis not only ensures the high capacity of the vehicle-to-infrastructure(V2I)link,but also improves the successful communication of the vehicle-tovehicle(V2V)link as much as possible.Compared with the traditional random allocation algorithm and traditional deep reinforcement learning algorithm,the proposed algorithm can effectively improve the system on the premise of realizing the large capacity of V2I link communication and successful V2V communication,makes the bandwidth utilization rate of ascension and algorithm complexity,high real-time and high training efficiency,satisfy the vehicle under the high speed motion state of dynamic bandwidth allocation of resources,which can meet the dynamic bandwidth resource allocation under the high-speed motion state of the vehicle.