Research On Handoff Algorithms Of Internet Of Vehicles Based On Mobile Edge Computing

With the rapid development of wireless communication technology,the deployment and application for Internet of Vehicles(IoVs)become possible.The high-speed mobile characteristics of vehicle terminals and the popularity of various intelligent applications make the time-delay requirement of IoVs higher and higher.Therefore,the low-latency IoVs based on mobile edge computing(MEC)is emerged.However,when the terminal is moving in the IoVs based on MEC,it will face the problem of handoff among different base stations(BSs)or cells.As a result,handoff as a key technology in the IoVs has becomes the research focus.In order to minimize the delay and reduce the ping-pong effect,decision tree handoff algorithm based on adaptive feedback and multi-attribute handoff based on ensemble learning are proposed in this dissertation,focusing on the handoff problem in the context of IoVs based on MEC.The main work of this dissertation is as follows.(1)First,the IoVs based on MEC is briefly introduced,including its architecture,characteristics and key technologies of resource management.Then the handoff technology which is one of the key technologies in the IoVs is introduced,mainly focusing on the theory and basic process of the handoff technology.(2)In the IoVs based on MEC,a self-adaptive feedback handoff(SAFH)algorithm is proposed to address the problem about dynamic handoffs for the IoVs,aiming at minimizing the handoff delay and reducing the ping-pong effect.In this algorithm,decision tree and feedback decision theory are introduced into the dynamic handoff problem of IoVs.Then,the main attributes and terminal movement trend,and the respective handoff probability distribution are given out.Based on handoff probability distributions,the structure of multi-attribute decision tree is determined.According to the dynamic change of vehicle terminal state,the terminal state at MEC server is updated by feedback mechanism,and the decision tree decision strategy is dynamically modified by adaptive feedback method based on incremental learning,so as to realize real-time and effective handoff of vehicle terminal.Simulation results show that the proposed algorithm is suitable for handoff decision-making of vehicle terminal with strong mobility and frequent business changes.SAFH algorithm can effectively reduce the pingpong effect and increase the effective time of network connection.In addition,handoff delay of SAFH algorithm is lower than some existing algorithms.(3)In the IoVs based on MEC,a multi-attribute handoff algorithm based on ensemble learning(ELMAH)is proposed.First of all,an algorithm to predict the link duration between vehicle and BS is proposed.After the vehicle is switched to the current BS,the MEC server would predict the link duration between vehicle and the current BS and all candidate BSs by using the link duration prediction method based on integrated learning.Then the vehicle terminal uses handoff decision-making method based on multi-attribute to decide the handoff result.The vehicle sends handoff request to the target BS in advance,so that the handoff operation can be performed after the link is interrupted,which can reduce the delay and improve handoff efficiency.Simulation results show that the proposed algorithm is suitable to the vehicles with high-speed.And the accuracy and effectiveness of handoff is improved,the number and delay of handoff are reduced,and the network performance is improved.(4)Finally,the current research work is summarized and the prospects of future research are given.