Research On Downlink Reliable Transmission Technology Of Proactive Access Network Under Open-Loop Transmission Framework

With the vigorous development of novel mobile communication technologies,Ultra-Reliable and Low Latency Communications,as one of the three major scenarios of 5th Generation Mobile Networks,effectively supports the rapid implementation and continuous updating of technologies such as autonomous driving,power automation,and virtual reality.However,with the exponential growth of equipment in industrial networks and the popularization of micro base stations and small cells,the frequent signaling interactions caused by the access of massive mobile devices in wireless networks will lead to signaling storms and seriously affect the transmission latency performance.The proactive network,which is based on the open-loop transmission framework,greatly reduces the communication latency by discarding the feedback mechanism and control signaling,and becomes a promising nextgeneration ultra-low-latency network architecture.Considering the characteristics of non-feedback transmission in the proactive network,the reliable transmission of downlink communication still faces the following two challenges:(1)Aiming at the problem that channel state information(CSI)cannot be obtained by using the openloop transmission,there are no effective downlink resource allocation schemes.(2)For the problem that the single proactive network cannot provide high-quality communication services in extreme scenarios such as ultra-high traffic,there is a lack of technologies for coexistence with the traditional closed-loop network in downlink communication.Based on reinforcement learning theory,this paper studies the reliable downlink transmission technology of the proactive vehicular network using openloop transmission.The following are the main research work and innovation points of the paper:1.Aiming at the problem of downlink communication multiple access interferences caused by the lack of CSI in the proactive vehicular network,this paper constructs a resource management framework based on a "generalized closed-loop",and guides the network side anchor node(AN)through the high-quality radio resource information provided by the vehicle in the immediate past uplink communication.At the same time,the long-term data transmission success rate of the system is used as the reliability index,and a downlink radio resource allocation model based on the bidirectional optimization of the vehicle and the AN is established.On the vehicle side,this paper proposes two strategies for generating high-quality radio resource information based on local experience and global experience.On the network side,an intelligent radio resource allocation algorithm based on deep reinforcement learning is proposed.Simulation results verify that under the resource load rate of 40%,the radio resource allocation algorithm based on the cooperation between the vehicle and the network side can obtain a data transmission success rate of more than 98%.2.Aiming at the problem that the proactive network cannot meet the downlink reliable communication in ultra-high-traffic urban centers and remote mountainous areas lacking access points(AP)deployment,this paper constructs a heterogeneous proactive vehicular network composed of AP network and high-power node network.Simultaneously,a fallback mechanism is proposed.When the current AP network cannot provide reliable transmission,a dedicated channel is used to perform closed-loop transmission through high-power nodes by fallback.Under the constraint of latency,the optimization problem of minimizing the network’s longterm packet loss is established to realize downlink adaptive transmission,and the selection process of network-side transmission mode is constructed as a Markov decision process.This paper proposes an adaptive transmission algorithm based on the Dueling deep Q-network.Simulation results show that the proposed algorithm can reduce the network packet loss rate to 10%while ensuring low-latency transmission.