Traffic-Aware User Association Technique In The Ultra-dense Networks

In the fifth-generation(5G)mobile communication system,base stations are deployed in a dense and heterogeneous way,and user equipment becomes more powerful.As the key technology of 5G,the UDNs bring opportunities to enhance the users' data rate and the spectrum efficiency via dense deployment.For one thing,the users may be within the coverage of multiple SBSs,which means that the users suffer a complex interference environment introduced from dense SBSs.For another,the enormous traffic may cause load imbalance among SBSs and MBSs,which may cause network congestion and a waste of resources.Through adjusting the user association,i.e.,network selection,we can improve throughput,enhance user experience,and balance the network load.To overcome and exploit the challenge and gain of the UDN,we study the multiple association technique in homogeneous networks and the intelligent user association technique in heterogeneous networks based on load sensing.Due to that the users in the UDNs can receive more than one powerful signals,multiple association is suitable for the UDNs.In the multiple association,the users can access multiple SBSs and MBSs and distribute their traffic to different BSs to improve the data rate and mitigate the load imbalance.However,existing multiple association schemes are centralized and they depend on reference signal received power(RSRP),which are high signal overhead,and can't adapt to the time-varying traffic.Therefore,it calls for a new distributed multiple association scheme.We model the multiple association problem as a state-based potential game(SPG).With the aid of SPG,the association policy that we designed could drive the ultra-dense networks(UDNs)to evolve towards the global optimization in the flow level performance.Finally,the performance of our proposed TAMA algorithm is investigated through a practical and discrete simulation method.However,the dense and heterogeneous environment of multi-RATs challenges the user association because of the more frequent and complex decision process along with increased complexity.Most of traditional user association approaches are executed based on air interference parameters,like RSRP.The oscillation of air interference parameters exacerbates the frequent switching problem.Introducing artificial intelligence to ultra-dense heterogeneous networks can improve the way we address user association today,and can execute efficient and intelligent user association.Whereas,there still exist difficulties to be noted.On one hand,the contradiction between real-time communications and time-consuming learning exacerbates the difficulty of convergence.On the other hand,the black-box learning mode suffers from oscillation due to the diversity of multi-RATs,which can result in arbitrary convergence.In this paper,we propose a model-driven framework with a joint off-line and on-line way,which is composed of feature learning and strategy learning,and able to achieve fast and optimal user association through an alliance of machine learning and game theory.The feature learning is designed for the analysis of link quality,and the strategy learning is responsible for decision.Further,we implement a distributed algorithm at the user side based on the proposed framework,which can reduce the number of frequent switching,increase the possibility of gainful switching,and provide the individual service.The simulation results confirm the performance of algorithm in accelerating convergence rate,boosting user experience,and improving resource utilization.