Research On Intelligent Wireless Access Mechanisim In Heterogeneous Networks

Novel heterogeneous wireless networks are evolving into complex systems.Wireless access will operate in a highly heterogeneous environment,so it requires new intelligent methods to break through the performance range.At present,research on wireless network access optimization at home and abroad generally considers different scenarios such as multiple radio access technology,cell switching,cell selection,user association and resource allocation,as well as various algorithms such as game theory,convex optimization and reinforcement learning.But there are still three problems:First of all,the access method cannot adapt to environmental changes.The distributed access algorithms are lack of cooperation among users,which affects the user experience and network efficiency.The parameters of model-based access algorithms are difficult to obtain,and it is difficult to model when the environment is complex,and the problems that can be modeled are difficult to solve due to high-dimensional non-convex.The convex optimization algorithm needs to be solved again after each environmental parameter change,and cannot adaptto environmental changes.Secondly,the inadequate allocation of network resources for multiple types of services cannot meet the rate requirements and delay requirements of services with different cost-effectiveness,which limits the flexibility of wireless networks.Finally,the rigid mapping of resource allocation and access can not only satisfy the flexible coexistence of multiple types of services but also cannot guarantee the efficient use of network resources.In the scenario where heterogeneous networks and multiple types of services coexist,an intelligent wireless access mechanism is implemented from access to resource allocation,making it adaptable to changes in the environment.Firstly,the user-adaptive access mechanism is studied in heterogeneous networks.In the scenario of the coexistence of macro and micro base stations,a distributed access strategy combining information sharing is designed to solve the incomplete information interaction in the network and the access competition between users,and an adaptive intelligent access strategy is proposed.The problem is modeled as a Markov process.An appropriate reward function is designed,and the reward weight is added to the action set so that the user can adaptively select the base station to access to.The simulation compares the effect of the proposed method with the access mechanism of fixed weight and multiple attribute decision making(MADM)algorithm on user-side Quality of Experience(QoE)and network-side load balancing.And the effectiveness and rationality of the proposed mechanism are proved.Secondly,the intelligent resource allocation mechanism for multiple types of services is studied.The thesis studies the characteristics and traffic models of three services based on Voice over IP(VoIP),Video Streaming and HyperText Transfer Protocol(HTTP).It explores requirements such as delay and transmission rate.In the case of the coexistence of multiple types of service scenarios,an intelligent resource allocation mechanism based on reinforcement learning(RARL)is designed in detail to simultaneously meet the service rate and delay requirements of users from different network service providers.The problem is modeled as a Markov process and a suitable reward optimization function is designed in the thesis.It enables the base station to allocate resources adaptively based on the current situation of network and feedback from the current user access.The simulation compares the performance of RARL with resource allocation algorithms such as predicted resource allocation and proportional resource allocation in the user-side QoE and network-side spectral efficiency(SE)performance.At the same time,the convergence of the RARL and the impact of minimum delay and transmission rate requirements on the performance of the algorithm are verified.Finally,the intelligent access optimization mechanism of heterogeneous networks is studied.The intelligent resource allocation mechanism under multiple types of services cannot completely solve the problem of uneven use of base station resources,and may not meet the delay and rate requirements of all access users.The base station resources are not fully utilized.An intelligent access optimization mechanism is proposed in this thesis based on the greedy algorithm and recommendation algorithm,which establishes the flexible mapping relationship between wireless resource allocation and wireless access.It minimizes the number of users that do not meet the delay and rate requirements in the network.Furthermore,it improves the performance of the user-side QoE and network-side SE.The simulation verifies the convergence and superiority of the intelligent access optimization mechanism.At the same time,the effects of optimization indicators of the greedy algorithm and recommendation algorithm on simulation performance are explored.