QoE Oriented Network Selection For Indoor Hybrid VLC And Radio Frequency Network Framework

The rapidly increasing number of wireless devices like smart phones and tablets,the emering virtual reality,online live broadcast and other network services,and the evolving application background of smart homes and the Internet of Things are all posing a great challenge to present radio frequency(RF)wireless network who is already facing spectrum deficit.Thus,wireless heterogeneous network will play an important role toward the goal of using a diverse spectrum to provide high quality-of-service,especially in indoor environments where most data are consumed.Researchers gradually realize that the future wireless communication network will be a variety of wireless access technologies coexist and complement each other,providing users with secure and seamless connections to meet diverse communication needs.The new indoor wireless communication technology,Visible Light Communication(VLC),is considered to be a powerful complement to existing RF communication technologies because of its various technological advantages.In VLC/RF hetergeneous network,the key to effective use of the potential performance advantages of each network lies in network selection.The premise of network selection is to determine the network performance evaluation criteria.However,the current research mainly based on the objective evaluation index of physical service parameters,lacking consideration to end users.The study focuses on the characteristics of VLC/RF heterogeneous wireless networks,introduces the concept of Quality of Experience(QoE),establishes an evaluation mechanism for combining subjective and objective of network performance,and then studies network selection problem.The main work and research results are summarized as follows:(1)Network selection for single-user,single-service type.In the VLC/RF heterogeneous wireless network system,the impact of other users is regarded as a dynamic external environment,and network selection decisions are optimized from a single user perspective.When a user performs a certain service,the position is moved or the network environment changes,so that the user continuously performs network switching during network selection,increases network switching overhead(delay,energy consumption,etc.),and thus reduces user QoE.This chapter first establishes a discrete QoE level model based on the average subjective scoring mechanism,and divides the user-side network experience into five levels.If and only if the network improves the QoE experience level of the user,it will switch to a better network to ensure the user service experience.At the same time reduce the number of network switching and increase system robustness.Finally,the performance of the proposed mechanism was analyzed through simulation experiments.(2)For single user multi-service type network selection.Study in VLC/RF heterogeneous networks,users perform multiple types of network communication services.Because different network service capabilities are different,different services have different network service parameter requirements.Users need to select suitable access in dynamic wireless networks.This chapter uses Q-learning in reinforcement learning to overcome environmental dynamic uncertainty,optimizes QoE cumulative expectations as the maximum goal.Through repeatedly interacts with the controlled environment through perception,choice of actions,and rewards,and finally the user optimizes it current decision making through historical experience.In order to overcome the slow convergence problem of classical Q-learning algorithm in the actual scenario,using three kinds of observed context information as prior knowledge to construct Q-learning algorithm with knowledge transfer to improve the convergence speed of the learning algorithm.Finally,the algorithm simulation result and discussion are given.(3)For system-level network selection.In VLC/RF heterogeneous networks,attention is paid to the multi-user network selection behavior to optimize the global performance optimization as the goal.Because in multi-user decision-making system,the user's decision-making is coupled,the complex role of the multi-user relationship is the biggest challenge that network to face.Since game theory provides a complete theoretical framework for multi-user distributed intelligent decision making,matching game algorithm is used to model the network selection as a many-to-one bilateral matching problem.Through the QoS parameter mapping method in the objective QoE quantization method,the matching parties are ranked on each other.To reduce the network congestion rate and improve the user experience as the goal,establish an optimization model.By solving the optimization model,the user side and the network side can be selected bidirectionally to improve the overall performance of the heterogeneous network.Finally,a simulation experiment was conducted to verify the analysis.