With the rapid growth of wireless data demand and the shortage of global bandwidth,the use of millimeter-wave(mmWave)frequency band for wireless cellular networks has become the core content of the fifth generation ce...With the rapid growth of wireless data demand and the shortage of global bandwidth,the use of millimeter-wave(mmWave)frequency band for wireless cellular networks has become the core content of the fifth generation cellular network.Because mmWave communication has different characteristics from microwave communication,using traditional optimization techniques to manage the resource of mmWave communication networks is inappropriate.In this paper,we propose a neural network-based algorithm to solve the joint user association and resource allocation for mmWave communication system with multi-connectivity(MC)and integrated access backhaul(IAB).The resource allocation problem is formulated as a mixed-integer quadratically constrained quadratic programming(MIQCQP),which is very difficult to solve.First,we decompose the MIQCQP into two sub-problems,i.e.,binary associated matrix sub-problem and continuous IAB ratio sub-problem.Then we propose a neural network to solve the binary associated matrix inference problem and a resource allocation algorithm to find the sub-optimal IAB ratio.Simulation results show that the proposed algorithm can achieve good performance with a fast inference speed.展开更多
文摘With the rapid growth of wireless data demand and the shortage of global bandwidth,the use of millimeter-wave(mmWave)frequency band for wireless cellular networks has become the core content of the fifth generation cellular network.Because mmWave communication has different characteristics from microwave communication,using traditional optimization techniques to manage the resource of mmWave communication networks is inappropriate.In this paper,we propose a neural network-based algorithm to solve the joint user association and resource allocation for mmWave communication system with multi-connectivity(MC)and integrated access backhaul(IAB).The resource allocation problem is formulated as a mixed-integer quadratically constrained quadratic programming(MIQCQP),which is very difficult to solve.First,we decompose the MIQCQP into two sub-problems,i.e.,binary associated matrix sub-problem and continuous IAB ratio sub-problem.Then we propose a neural network to solve the binary associated matrix inference problem and a resource allocation algorithm to find the sub-optimal IAB ratio.Simulation results show that the proposed algorithm can achieve good performance with a fast inference speed.
