Research On Resource Allocation Method Based On Interference Control In MMTC Scenario

As one of the three application scenarios of the Fifth Generation Mobile Communication System(5G),massive Machine-Type Communication(m MTC)has obtained extensive attention from researchers.In this scenario,the number of Machine to Machine(M2M)communication terminals is growing rapidly,which not only causes the lack of spectrum resources in the network,but also causes network congestion.This thesis focuses on wireless resource management and allocation in the m MTC scenario.Firstly,a resource allocation strategy with the help of interference graph is proposed to solve the problem of resource block and power allocation between cellular user(CUE)and data aggregator(DA).In this algorithm,an interference graph based on graph theory is constructed with the purpose of revealing the interference relationship between communication entities in the network.In the interference graph,CUE and DA are the vertices,and the interference relationship between CUE and DA is expressed as the edge of the interference graph.Through the interference graph,the interference relationship between the communication entities in the system can be clearly obtained,which in turn can guide resource allocation more effectively and reduce system interference.The simulation results show that the algorithm has certain benefits in improving system throughput and reducing interference.Secondly,in terms of the resource allocation problem among MTCDs in the m MTC network,this thesis divides the research into two stages: MTCD clustering and resource allocation within the MTC network,and proposes corresponding algorithm to solve them.In the first stage,a new clustering algorithm based on Cosine Similarity(CS)is proposed to cluster the MTCD in the system into several small MTC networks.In the second stage,for the purpose of solving the joint allocation of resource blocks and power within the MTC network,two Q-learning algorithms are proposed,namely,centralized Q-learning(Team Q-learning,Team-Q)and distributed Q-learning(Distributed Q-learning,Dis-Q).From the simulation results,a conclusion can be obtained that Team-Q learning algorithm and Dis-Q learning algorithm can both improve the system throughput when compared with the traditional algorithm.At the same time,the Dis-Q algorithm has a lower complexity,so its convergence speed is significantly faster than that of Team-Q algorithm.Finally,to solve the problem of power allocation among MTCDs and DAs,and under the constraints of the transmission delay of both MTCD and DA,the power allocation problem is transformed into a convex optimization problem.A sub-optimal power control strategy that can minimize the energy consumption in the process of MTCD data transmission process while meeting the MTCD transmission delay restriction is proposed..