Intelligent Wireless Resource Management Technologies In Ultra-Dense Networks

The future mobile network will enter the era of "Internet of everything(IoE)",and the access of massive mobile devices will bring unprecedented challenges to network capacity.The increasingly large and complex wireless network and various emerging mobile applica-tions make the network planning,optimization and maintenance more and more complex.Among many wireless network technologies,the ultra-dense network,which could improve the network capacity significantly by importing near end-to-end transmission and spatial reuse,is a promissing technology for the 5th generation(SG)and future wireless networks to achieve the explosively increasing requirement of network capacity.The intelligentization will greatly enhance the situational awareness of network operators and make the network self-planning,self-optimization and self-maintenance possible.Intelligentization will pro-mote the evolution of mobile network from simple "network connection" to a new era of"intelligent interconnection".This paper focuses on the wireless resource management in ultra-dense networks.It solves the problems of network interference modeling and spectrum resource allocation by utilizing intelligent resource management technologies such as game theory and machine learning algorithms under the framework of wireless cloud network(C-RAN).The main work and innovation of this paper are as follows:Firstly,this paper studies the intelligent resource management algorithm based on game theory under the condition that the real-time and accurate geographical location of network nodes could be known.A user-centric coalition formation game is proposed to find a optimal or sub-optimal network partition under a given resource allocation algorithm.In this method,the whole network is divided into several coalitions by merging and splitting operations.An unweighted conflict graph,which could reflect the interference relations,is constructed ac-cording to the network partition.The user allocation order is determined in a distance-aware method according to the geographical location information of the network nodes.In order to eliminate the strong interference between users,the orthogonal subchannels are allocated to the users who are connected in the conflict graph.The subchannel with the largest net profit will be allocated when the idle subchannel is unavailable.In addition,to overcome the limitation of "each user can only be allocated with one subchannel" in some previous work,this paper proposes a low-complexity supplementary allocation algorithm to allocate the remaining subchannels,so as to improve the spectral efficiency of the system.Finally,simulation results show that this method could effectively suppress network interference and improve the network capacity.Secondly,in the case that the real-time and accurate geographical location information of network nodes cannot be obtained,this paper studies the interference modeling problem for uplink communications by utilizing the large amount of up-link signal to interference and noise ratio(SINR)data,resource block(RB)allocation data collected from the network based on the artificial neural network.The proposed method could effectively separate the interference from different interfering sources without adding additional data transmission overhead to the network.Then construct weighted conflict graph based on the interference modeling results,the throughput maximization problem is approximately decomposed into a user clustering subproblem and a subchannel allocation subproblem.The former is solved by proposing a low complexity user clustering algorithm with modified balanced Min k-Cut,which identifies low-interference entities(i.e.,clusters)for spectrum reuse;and the latter is solved by presenting a subchannel allocation algorithm with accumulative inter-cluster interference considered,which could further reduce the interference caused by spectrum reuse.Finally,the simulation results show that the proposed interference modeling method could achieve a high accuracy and the corresponding resource allocation algorithm could effectively improve the network capacity.Finally,the relative interference modeling problem for downlink communications,which mines the large-scale RB allocation data,new data indicator(NDI)data,acknoledgement(ACK)and negative acknolesgement(NACK)data collected from the network and utilizes the association rule algorithm,under the condition that the real-time and accurate geographi-cal location information of network nodes is unavailable is studied in this paper.In addition,this paper proposes a load-aware resource allocation approach which calculates each user’s boundary of reusing the common RB s and allocating the orthogonal RBs with its interfering sources based on the relative interference intensities modeled above and the time-varying network load.The orthogonal interfering source set of each user is generated based on its time-varying boundary.Then allocating the spectrum resources according to each user’s orthogonal interfering source set.Simulation results show that the proposed relative inter-ference modeling method is very accurate,and the proposed load-aware resource allocation approach could achieve superior performance under most network densities and network loads especially when the network load is heavy and the network is ultra-dense.