Research On Resource Allocation For UAV-Enabled 5G Communication Scenarios

As the large-scale commercial deployment of the fifth generation mobile networks(5G)and the rapid development of industry 4.0,the application of the Internet of Things(IoT)has shifted from traditional machine-type com-munications to mission-critical communications.Diversified requirements in terms of reliability,latency,and the battery life of IoT devices have been put forward.As one of the three key scenarios of 5G,Ultra Reliable and Low La-tency Communications(URLLC)has attracted more and more attention.For some typical scenarios of URLLC such as autonomous driving,drone control,and augmented/virtual reality(AR/VR),several challenges are proposed to the existing cellular network architecture for achieving the extreme strict require-ments of delay and reliability.Unmanned Aerial Vehicles(UAV)have salient attributes,such as high maneuverability,flexible deployment and high probability of line-of-sight(LoS)links,and it can be used as flying base stations(BS)to provide commu-nication services for ground equipment.UAV-assisted intelligent heterogene-ous network architecture is considered to be an effective solution to support the three major application scenarios of 5G,which can effectively meet the technical requirements of 5G networks for capacity,delay,data rate,reliability,and scalability.For URLLC scenarios,UAVs have high probability of LoS link and flexible mobility,and they can ensure the requirements of latency and reliability by improving the channel quality of the ground users.This thesis is oriented to ultra-reliable and low-latency communications services in 5G networks.Considering the technical characteristics of URLLC services and salient attributes of UAV communications,a UAV-assisted URLLC network architecture has been proposed.This thesis focuses on the communication model and resource allocation issues in UAV-assisted URLLC networks.The main work of this thesis is as follows:Firstly,for a single UAV-assisted IoT network in URLLC services,this thesis proposes the LoS propagation model for air-to-ground channel and the data packet transmission error probability model with finite blocklength en-coding in URLLC networks.Then the problem of jointly optimizing the power control and deployment of UAV with the goal of minimizing total transmission power is proposed.Because the constraints of delay and reliability are com-plex and non-convex expressions,this resource allocation problem is difficult to solve directly.In this thesis,by deriving the mathematical properties of the constraint function,the delay and reliability constraints in the UAV-assisted URLLC networks are transformed into the power and signal-to-noise ratio constraints,and the UAV horizontal position and height are optimized alter-nately.Numerical simulation verifies the effectiveness of the proposed algo-rithm and proves the performance gain compared to other schemes.At the same time,it shows the variation of UAV minimum transmit power under dif-ferent delay and reliability constraints.Secondly,considering the insufficient coverage of a single UAV and the limit of UAV bandwidth resources,the joint optimization of power control and bandwidth resource allocation algorithms are studied in multiple UAVs as-sisted URLLC networks.Our goal is to minimize the average transmit power of the IoT devices.The resource allocation problem is a mixed integer optimi-zation problem,and it is difficult to find its optimal solution directly.This thesis divides it into three sub-problems based on Block Coordinate Descent(BCD)method.In the first sub-problem,given the UAV deployment scheme and bandwidth allocation policy,the optimal IoT device scheduling and asso-ciation are obtained by the MOSEK solver in CVX toolkit.In the second sub-problem,given the UAV deployment scheme and the optimal IoT device scheduling based on the first sub-problem,the optimal power control policy and bandwidth allocation are obtained by applying Lagrangian dual decom-position.In the third sub-problem,based on the optimal solutions obtained in the first two sub-problems,the optimal UAV deployment location are deter-mined.The BCD method is used to solve the sub-problems alternately to ob-tain the minimum average transmit power of IoT devices.Numerical simula-tions verify the convergence and effectiveness of the pro-posed resource allo-cation algorithm,and prove that the algorithm proposed in this thesus can achieve a performance gain of more than 25%compared with other schemes.Finally,in view of the limited coverage of fixed UAV deployment solu-tions and low economic benefits,the dynamic UAV-assisted URLLC network is considered.Meanwhile,for emerging URLLC services which requires not only the data transmission delay and reliability,but also data transmission rate,such as high-definition map downloading in autonomous driving and virtual environment loading in AR/VR,the resource allocation problem based on re-source blocks(RBs)is studied.The optimization of joint UAV location and RB allocation as well as the number of data retransmissions is a discrete NP-hard problem.The traditional convex optimization scheme is more compli-cated to solve and it is difficult to directly obtain the optimal solution.There-fore,this thesis first gives the communication delay,reliability model and sys-tem energy efficiency model,and then proposes a deep Q-learning network(DQN)-based multi-agent deep reinforcement learning resource management method to obtain the optimal resource allocation strategy.