Research On User Scheduling And Resource Allocation In The 5G Hybrid Service Scenario

With the continuous development of communication technology,the number of intelligent terminals is increasing,and the terminal data flow is growing exponentially,which brings great pressure to the communication system.The new radio standard of stand-alone（SA）architecture for the 5^thGeneration（5G）mobile communication system was frozen in June,2018,which marks a new era for the communication industry.On June 6,2019,China’s Ministry of industry and information technology announced the issuance of 5G commercial license,and 5G was officially commercial in China.In the early stage of 5G construction,the enhanced mobile bcroadband（e MBB）scenario will be deployed first,which has faster network communication rate.At present,the 3rd Generation Partnership Project（3GPP）is developing the release 16 specification,which is expected to freeze by the second quarter of 2020.The release16 specification is mainly applied to intelligent transportation and other applications that have strict requirements on delay and reliability,so how to improve the reliability of ultra reliable low latency communications（URLLC）service is a worthy research direction.Aiming at the low-latency and high-reliability characteristics of URLLC services,this thesis studied how to improve the reliability of URLLC services in terms of scheduling and resource allocation.The main contents and innovations of this thesis are as follows:（1）Considering that it will be possible to deploy e MBB and URLLC services on the same frequency band in the future,the first part of the work focused on how to schedule users in a hybrid scenario and how to allocate frequency resources to users.In the coexistence scenario,delay parameters and priority adjustment factors were added for each user.By adjusting the priority factors,URLLC users could be scheduled preferentially.In the process of frequency allocation,a part of the bandwidth was reserved from the system and allocated to URLLC users as additional bandwidth to reduce the encoding rate of URLLC users and improve their transmission reliability.But it would have effect on throughput of e MBB.The simulation results show that it can greatly improve the reliability of URLLC services at the cost of sacrificing a little throughput by reserving appropriate proportion of system bandwidth.（2）The scenario considered in the second part of the work was URLLC business scenario only.Using reinforcement learning,a how to improve the reliability of URLLC services was studied from the perspective of power allocation.By setting the action space,the state space,and the reward function,the agent could adaptively adjust the transmission power to ensure the reliability of the user,and at the same time tried to save the transmission power on the codition of a certain retransmission probability.Link-level simulation was performed and the results show that as the number of learning times increases,the agent’s knowledge is gradually saturated,the reliability of URLLC users is improved,and the probability of air interface delays over 1ms gradually decreases.