| With the official commercialization of 5G,the era of "Internet of Everything" is coming.In order to cope with the continuous explosive growth of mobile terminals and exponentially multiplying data traffic,the ultra-dense network(UDN)with intensively deployed access nodes is one of the key technologies of future mobile communication networks.UDN usually consists of a large number of low-power and low-cost small base stations,and their deployment density far exceeds that of current mobile communication networks.Meanwhile,UDN has the advantages over improving spectrum efficiency,increasing cell user coverage and deploying freely.However,the shortage of spectrum resources and the densification of base stations will inevitably bring about serious interference problems,which will not only affect the user's communication quality of service and reduce network throughput,but also increase additional power consumption.This consequence seriously violates the development of green communication.Therefore,this paper optimizes the energy efficiency of the communication system by studying the resource allocation problem of UDN.The main innovative research work is as follows:(1)Considering that the resource allocation problem in the UDN scenario is an NP-hard problem,this paper transforms the original problem into two subproblems of resource block allocation and power allocation to solve separately.Firstly,a frequency resource allocation sub-problem model is established and an improved K-means algorithm to cluster micro base stations in the communication network is proposed.Then,the concept of local base station distribution density is defined to greatly realize the initial aggregation of the dynamic allocation K-means algorithm cluster center,which reduces inter-cluster interference effectively.Last,the user grouping strategy in base station clusters based on spectral clustering algorithm is proposed by combining with channel loss to construct a user interference relationship weight graph.It allocates orthogonal resources to different user groups,which minimizes inter-cluster interference greatly and improves network throughput significantly.Meanwhile,in the centralized scenarios,the mixed integer nonlinear optimization sub-problem model of power resource allocation is established to optimize energy efficiency.Then,the fractional objective function form is converted into a non-fractional programming problem by using Dinkelbach method,and the Lagrangian multiplier algorithm is used to solve the optimal power on each resource block iteratively,which allocates the power in each base station cluster reasonably.Compared with KMUG and RARB algorithms,the simulation results show that the algorithm proposed has good convergence and gets a great improve on energy efficiency of UDN.(2)To further cope with the rapidly increasing number of base stations and terminals,a power adaptation problem model for a larger number of base stations in a distributed scenario is constructed.As for distributed scenarios,the distributed ultradense heterogeneous cellular network power allocation method based on ADMM is proposed by trading every base station access point as a center,which split the energy efficiency optimization goal of the entire system into optimizing the energy efficiency of each base station.For transforming the nonlinear fractional optimization problem into a subtractive optimization problem,the augmented Lagrangian function iterative optimization combined with the Dinkelbach method is constructed to solve the optimal power distribution scheme.The simulation results show that the proposed algorithm has good convergence,and in a higher-density heterogeneous cellular network scenario,compared to the centralized algorithm,it is more excellent to improve the calculation efficiency. |
PDF Full Text Request