| With the rapid development of intelligent transportation,various applications such as mobile ultra-clear TV,security warning,etc.,which have millisecond requirements for delay,appear in the vehicular heterogeneous networks.Since the multipath transmission control protocol has low latency characteristics and it is especially suitable for high speed mobile wireless environments,it is gradually being applied to vehicular heterogeneous networks to meet the needs of delay sensitive applications.In addition,based on the use of traditional cloud computing,offloading these delay-sensitive applications to edge nodes becomes a trend of development.The conventional edge node workload distribution method is to distribute the workload to the edge node closest to the local device.However,with the increasing number of vehicular applications,the traditional workload distribution method will make the workload distribution unbalanced,causing some edge nodes to be overloaded,resulting in corresponding application response time being too long,which cannot meet low latency requirements.In this case,how to design an effective workload distribution method to enable workload balancing between edge nodes becomes an urgent problem to be solved.This paper focuses on the multi-path transmission workload balancing optimization technology based on mobile edge computing for the vehicular heterogeneous network environment.The main research contents can be generalized as follows:Firstly,this paper proposes a packet scheduling optimization algorithm based on probability prediction for the problem of network throughput degradation caused by the receive buffer blocking of multipath transmission control protocol in vehicular heterogeneous network.The algorithm introduces the throughput prediction model,and defines the similarity parameters between each path and the maximum throughput path.The parameter is compared with the established threshold to judge the performance of the paths.The algorithm predicts the buffer blocking problem by judging the number of out-of-order packets,and stops the data transmission on the bad path to improve the performance of the multi-path transmission control protocol.The simulation results show that compared with MPTCP-ML algorithm,the average network throughput of the packet scheduling optimization algorithm based on probability prediction is increased by 0.18 megabits per second.Secondly,this paper proposes a multipath transmission workload balancing optimization algorithm for the problem that the edge node overload causes a large application delay.The algorithm first allocates delay-sensitive applications,determines the allocation priority of applications according to their relative network delay differences,and then allocates delay-relatively insensitive applications,assigning the vehicular applications as far as possible to the nearest edge of each vehicle.Then,when the workload of the edge node exceeds its capacity,the algorithm will iteratively select the vehicular application with the longest response time,and if its response time exceeds the response time processed by the cloud computing center,it will be reassigned to the cloud computing center,otherwise it is reassigned to the alternate edge node to minimize its response time until no better edge nodes are found for each vehicular application.Finally,the algorithm allocates different sizes of computing resources for different types of virtual machines in each edge node,thereby realizing dynamic on-demand allocation of computing resource.We simulate the proposed algorithm,and the response time is compared under different application types,different workloads and different edge node capacities.The simulation results show that the multipath transmission workload balancing algorithm can effectively reduce the computing delay of the applications and reduce the average response time of the applications by 88 milliseconds compared with the traditional workload distribution algorithm. |
PDF Full Text Request