Research On User-centric Network Management Schemes In Software Defined WLANs

Recent years,with the rapid development of mobile Internet,various new types of mobile services are emerging.As one of the most important ways of wireless access,wireless local area networks(WLANs)are carrying rapidly growing wireless data traffic,and have become increasingly important in wireless networks.Although the technology and standard of WLAN have made great progress,they still face two problems.First,in order to meet people's growing demand for wireless services,more and more access points(APs)are deployed.In some hotspot areas,WLANs are becoming increasingly dense.However,the existing WLAN management mode is fixed and rigid,and the management efficiency is low,which make it difficult to perform flexible and effective management of dense WLANs.Second,new mobile applications and services represented by real-time voice and video often have more stringent requirements for network performance indicators including handover delay and throughput.But the existing WLAN solutions mainly implement network-centric management strategies,focus on the configuration and management of network devices,and lack the network management mechanism that guarantees the quality of service experience of users.In order to address the problems of inefficient network management and lack of user experience quality assurance mechanisms in the existing WLAN architecture,a new WLAN architecture and management model need to be designed.Therefore,this thesis will focus on building an user-centric WLAN architecture and management strategies.First,a software-defined WLAN architecture is designed based on the concept of Software Defined Network(SDN),and then the performance analyses of dense WLANs are performed.Finally,based on the obtained analytical results,a handover management strategy and a load balancing strategy based on the proposed WLAN architecture are designed to realize an user-centric management mode and provide a seamless and consistent service experience for users.The main research work and innovative achievements are as follows:1.The design and implementation of software-defined WLAN architectureThe emergence of SDN technology provides a solution for WLAN management and user experience quality assurance issues.This thesis designs a new Software-Defined WLAN architecture,named SWAN,which can achieve an user-centric,flexible and open WLAN management mode.According to the concept of SDN,this thesis specifically designs the system composition and function division of the control plane and data plane,and the southbound protocol between them.In addition,in order to flexibly control the connection of Stations(STAs)to the network,we use virtualization technology to abstract the connection status between APs and STAs.On this basis,this thesis studies and implements the seamless handover and load balancing mechanisms in the proposed SWAN architecture.The experimental results prove that the proposed SWAN architecture has a flexible network management model,and can provide users with a seamless and consistent connection experience.2.The construction of performance analysis models for dense WLANsPerformance analysis of dense WLANs is the foundation for designing user-centric network management strategies.This thesis constructs practical models to analyze the performance of dense WLANs.The interested performance indicators include coverage probability and throughput.Firstly,this thesis uses stochastic geometric theory to analyze the coverage probability performance of dense WLANs,and an improved Hard Core Point Process(HCPP)model called TM-HCPP is proposed.Then we analyze and compare the coverage probability performance by computer simulation and theoretical calculation.The results show that TM-HCPP can mitigate the intensity underestimation problem of traditional HCPP.Secondly,we use the Continuous Time Markov Chain(CTMC)model to analyze the throughput performance of dense WLANs with arbitrary topologies.We propose a feasible state search algorithm to obtain all CTMC state information in the network.Considering that interference will affect the accuracy of throughput analysis,we introduce coverage probability model into the CTMC model,thus capturing the effects of interference on the transmission process.3.The user-centric WLAN handover decision strategyThe handover management in WLANs involves two aspects,that is,handover mechanisms and handover decision algorithms.The proposed SWAN architecture has the mechanism of ensuring seamless handover of STAs in the network.In order to achieve user-centric handover management,it is necessary to improve the data rate of STAs in handover processes.Therefore,this thesis designs a Deep Q-Network(DQN)architecture for SWAN architecture,and proposes a DQN based handover decision algorithm.In this algorithm,we use the real-time throughput of STAs as the reward value of DQN,use the SINR received by APs to characterize the state of STAs in the network,and use convolution neural network and recurrent neural network to extract the spatial characteristics and temporal characteristics of STAs' status respectively.By training the algorithm in the actual network environment or the simulation environment,a handover decision strategy which is suitable for the corresponding network scenario can finally be obtained.Simulation analyses show that the proposed DQN based handover decision algorithm can significantly improve STAs' data rate in the handover process compared with the traditional handover decision algorithm.4.The user-centric WLAN load balancing strategyThe load balancing management in WLANs involves the definition of load metrics and the design of load adjustment mechanisms.This thesis studies the load balancing strategy in the proposed SWAN architecture.Firstly,in order to accurately measure the load status of WLANs,we innovatively propose the concept of Channel Busy Time Ratio(CBTR),and use it as an indicator of WLANs' load.Secondly,in order to embody the user-centric design principles,the load balancing strategy not only needs to reassign the unbalanced distributed load,but also needs to improve the overall throughput of the network,and ensure the service experience of users.Therefore,the connection assignment problem of STAs to APs in a WLAN is modeled as the problem of finding the maximum weighted semi-matching in a bipartite graph.The weight of each edge in the bipartite graph is represented by the SINR value received by the AP from the corresponding STA.To solve this problem,we first transform the weight matrix of the bipartite graph according to the load status of the APs in the network,then use the new weight matrix as the input of the classical K-M algorithm.The output of the K-M algorithm is the maximum weighted matching scheme.Simulation results show that the proposed load balancing strategy can effectively improve the load balancing level and the long-term throughput of the entire network.