Heterogeneous Wireless Network Selection Algorithms Based On Multiple Attribute Decision Making And Swarm Intelligence Optimization

With the rapid development of wireless network communication technology in recent years,more and more new network technologies have emerged.But for many reasons,the old network technology will not stop operation immediately,but will continue to operate for a long time.This universal fact in the world easily leads to a variety of networks around mobile terminals.Such a network environment is called heterogeneous networks.Furthermore,in different application scenarios,end users' demand for network quality of service also presents significant diversification characteristics.Handover of terminals between two disparate networks is called vertical handoff.Otherwise it means that the mobile terminal handovers from one base station in the current network to another base station,which is called horizontal handoff.This thesis studies the decision-making process and quality evaluation of vertical handoff algorithm in heterogeneous network scenarios without considering horizontal handoff.The problem model established in the research process is called heterogeneous network selection problem,which is called heterogeneous network selection for short.This thesis studies the decision-making process and quality evaluation of vertical handoff algorithm in heterogeneous network scenarios without considering horizontal handoff.The problem model established in the research process is called heterogeneous network selection problem,which is called heterogeneous network selection for short.According to whether the terminal can choose the network independently or not,it can be further divided into two types: the terminal self-selection network and the terminal passive access network.One thing they have in common is that they have strict requirements for timeliness.But the main differences are as follows:The main concern of the former is how to evaluate the comprehensive quality of each candidate network and rank them when the terminal encounters a decision point,and then select one as the target access network.It does not consider the impact on other entities before and after the selection,but simply considers how to obtain the maximum QoS by selecting the target network from the end-user's point of view.At the same time,it should avoid the ping-pong effect as much as possible,so as to obtain the most satisfactory QoE.Obviously,research related to this model is often built on the basis that the remaining available resources of the alternatives are abundant.There have been many algorithms to solve this model,but MADM(Multiple Attribute Decision Making)is often used for the issue,due to the fact that its principle is intuitive and simple,and it is easy to directly calculate the final results through corresponding formulas.While in the latter problem the function of the terminal is greatly limited.It is often necessary to give a reasonable network selection and allocation scheme from the perspective of network system resource administrators.Make each terminal get the best possible quality of service,while taking into account the impact on the network system,such as network load balancing,system throughput.Various heuristic algorithms,including swarm intelligence,are often used to solve such problems.According to the characteristics of the two models above,hybrid multi-attribute decision-making algorithm and hybrid group intelligent decision-making algorithm are proposed to solve the single-terminal/multi-terminal network selection problem.Therefore,the main contributions of this thesis are as follows:(1)The traditional Fuzzy Analytic Hierarchy Process(FAHP)algorithm based on fuzzy consistent matrix is improved,and then a network selection algorithm based on comprehensive utility value and ratio threshold is proposed.In order to control the running time of the whole algorithm,according to the characteristics of Entropy and TOPSIS,the proper formulas of assimilation and normalization are selected to complete the pre-processing process for Entropy and TOPSIS at one time.With different adjustment factors for each of four classical traffic classes defined by 3GPP,the three algorithms are flexibly integrated by linear combination which automatically matches the specific traffic class.After evaluating the comprehensive performance of each network,the ratio threshold is used to determine the target network.The simulation results show that the proposed algorithm is superior to the existing three algorithms in terms of controlling the number of vertical handoffs,restraining unnecessary vertical handoffs and improving the gains of vertical handoffs.(2)A network selection algorithm based on improved Grey Relational Analysis(GRA)and difference threshold is also proposed to solve the problem of terminal self-selection network.According to the characteristics of Standard Deviation(SD)and GRA,by choosing suitable formulas of assimilation and normalization,not only a normalization process is omitted,but also the subsequent calculation process of these two algorithms is simplified significantly,which helps to reduce the running time of the whole algorithm.The simulation results show that compared with the four existing algorithms,the proposed algorithm can better control the total vertical handover times,unnecessary vertical handovers and ping-pong effects for the four traffic classes considered.(3)A terminal passive access network algorithm based on QoS sequence evolutionary optimization is proposed.By setting the mapping function between the discrete multi-terminal network selection sequence and the continuous QoS sequence,the algorithm takes the continuous QoS sequence as the intermediary,and passes through the pre-discrete iteration of Genetic Algorithm(GA),the mid-term divergence test of Simulated Annealing(SA)and the late stage of Continuous Particle Swarm Optimization(CPSO).Continuous iteration is used to get the group optimal solution step by step.Through this optimization process,the algorithm not only has the advantages of GA,SA and CPSO,but also effectively controls the running time of the whole algorithm.The simulation results show that,compared with the four existing algorithms,the proposed algorithm can find high-quality optimal solution steadily and quickly,and significantly reduce the probability of falling into the trap of local optimum.