Research On Wireless Resource Management Of MmWave MEC System

Mobile edge computing(MEC)technology will be an important direction for future mobile network development in order to support a wider and more diverse range of highcomputing terminal services.Millimeter wave(mmWave)communication has better spectrum resources,which can provide higher transmission rate and throughput.In addition,mmWave communication greatly reduces the impact of multipath propagation effect by using beamforming technology for directional transmission,which improves the reliability and spectral efficiency of communication.As a result,mmWave communication technology can provide MEC systems with more offload transmission resources to support higher user density and more computational offload tasks.These advantages make mmWave MEC technology promising for a wide range of applications in future mobile communication and computational offloading scenarios,enabling more efficient data transmission and computational resource management to support the demand for wireless services in more vertical industries.However,mmWave MEC technology also has some disadvantages,such as susceptibility to factors such as co-channel interference and blockages,which may lead to link interruptions and transmission failures.It also has limited performance in heterogeneous scenarios,and difficulty in resource management to bring out its advantages.Therefore,in order to push mmWave MEC technology to practical applications,these factors need to be fully considered,especially for the MEC offloading process,and wireless resource management techniques need to be studied to ensure the efficiency and reliability of the system.In this thesis,highly reliable and targeted resource management techniques are investigated for the two challenges of frequent interference blocking and scenario service heterogeneity in future mmWave MEC systems.The details of the research and innovative results are described as follows:1.For the current situation of frequent co-channel interference in mmWave MEC systems under the demand of dense access of massive users,this thesis investigates the resource management scheme of mmWave ultra-dense networks.To realize the joint allocation of multi-dimensional resources with low complexity,firstly,the beam,time and frequency are modeled as three-dimensional resource blocks in their feasible domains.Second,to maximize the sum rate of the ultra-dense mmWave MEC network,the system capacity maximization problem is formulated under the resource constraint,and a low-complexity resource allocation scheme is proposed to solve it.The proposed solution scheme includes best option first(BOF)beam scheduling algorithm and M20-ME subcarrier allocation algorithm.Finally,the simulation results show that the proposed low-complexity resource allocation scheme can improve the total rate of the ultra-dense access mmWave MEC system by about 3 times compared to the conventional greedy algorithm and random resource allocation strategy.2.To address the problem that the uplink of mmWave MEC system is susceptible to blocking and generating link interruptions,this thesis investigates the dynamic offloading resource planning and blocking link rebuilding strategy.First,a mmWave MEC model with dynamic offloading function is modeled,which serves users with mobility and no complete information about burst computing tasks,and a multibeam transmission scheme is introduced to overcome the blocking problem.Second,the matching-aided-learning(MaL)solution architecture is proposed and a learnable weight adjustment mechanism(LW-AM)based on an attention mechanism is introduced to adaptively and dynamically adjust the weights of different targets.Finally,the characteristics and simulation results of the proposed MaL resource allocation scheme are comprehensively analyzed.The proposed scheme can approach the performance of the benchmark scheme with perfect offloading task information.Compared with the conventional deep deterministic policy gradient(DDPG)algorithm,the proposed MaL scheme can speed up the training process by nearly 90%.3.A joint multidimensional resource allocation technique for mmWave MEC systems is studied for computing communication coexistence requirements.First,computing users scheduling,beamwidth allocation,and power allocation are jointly considered,and the original multi-objective optimization problem is converted into a singleobjective problem by the ε-constraint method,and a very low-complexity three-stage iterative optimization algorithm is proposed.The results show that the difference between the Pareto fronts generated by the proposed algorithm and the true fronts is less than 0.16%.4.For the problem of coexistence of MEC service and cache service,a joint scheduling and time-domain resource sharing strategy for MEC and cache tasks is investigated in this thesis.The optimization of task scheduling variables,beamwidth variables,and power allocation variables are jointly considered in the mmWave MEC system with computing caching coexistence to maximize the resource sharing capability of this system.A MODRL/HA algorithm is proposed to solve the multi-objective Markov decision process(MOMDP)problem for a mixed action space with both discrete and continuous actions.Simulation results show that the proposed resource sharing strategy can improve the delay and energy performance by about 15%compared to the scheme without task scheduling.In addition,the proposed MODRL/HA algorithm can achieve about 22%performance improvement compared to the original DDPG and MODDPG algorithms.