Delay-Oriented Offloading Mechanism For Computing-Intensive Tasks In Vehicular Networks

With the continuous progress of information technology,the Internet of Vehicles(IoV)has become one of the key technologies to realize the intelligent transportation network.In the process of vehicle driving,due to the complex real-time road conditions,mobile edge computing(MEC)is introduced to monitor the real-time information of vehicles and road conditions,plan the appropriate driving speed and route,and make reminders to drivers.At the same time,due to the threat of attack and tampering of vehicle information,it has caused great hidden dangers to road traffic safety.Using blockchain technology to ensure the safety of driving information has become a research hotspot.Both the above IoV services and blockchain node operation will produce many computingintensive tasks in the mining process,which cannot meet the requirements of low latency and high throughput of the IoV.In addition,the load of MEC servers varies greatly and the cluster computing resources are unevenly distributed.Therefore,combined with the characteristics of edge network and vehicular networks,this thesis studies the offloading mechanism of computing-intensive tasks,realizes the balanced allocation of network resources,reduces the system delay,which is of great significance to the high-quality operation of vehicular networks.At present,the application of blockchain technology can provide distributed storage scheme for data,and ensure the security and credibility of the system.However,this technology needs the support of a large amount of computing power,has the problem of high energy consumption of network resources.Moreover,the existing research methods rarely consider the mobility of terminal nodes,which cannot meet the needs of dynamic changes in services.On the other hand,MEC technology can provide powerful computing power for the network through edge server,but most of the existing MEC based computing offloading schemes adopt the binary offloading model and do not consider setting precise cooperative offloading scheme among edge nodes,resulting in low resource utility and unbalanced network load.In order to solve the problem that a large number of computingintensive tasks cannot be processed in time during the mining process of blockchain nodes,a delay-and-throughput-based edge collaboration offloading algorithm for computing tasks is proposed.Firstly,the vehicle sensor is used to obtain the vehicle dynamic and driving data,and the relevant parameters of the vehicle are stored in the blocks generated on the terminal layer.Then,the system model is established,including the offloading model,delay model,and block generation model.So as to solve the computing-intensive tasks,a multi-mode offloading algorithm is proposed.The deep reinforcement learning algorithm is used to jointly optimize the offloading decision,transmission power,block size and block interval.Simulation experiments show that this method can effectively integrate resource utilization,reduce total delay and optimize transaction throughput.In order to solve the problem of large load difference and uneven distribution of computing resources between cluster MEC servers in the IoV,a delay-oriented balanced offloading strategy of assisted driving service in IoV is proposed.Firstly,the mobility-aware task transmission model is constructed.After the service data of assisted driving is collected,comprehensive calculations and analysis are carried out,and the computing offloading method of edge collaboration is designed.At the same time,considering the usage of storage resources and computing resources of cluster MEC servers,this thesis constructs a load balancing model,designs the load balancing factor combined with the mean and standard deviation of load rate,designs a computing-intensive service balancing offloading algorithm based on deep reinforcement learning,and solves the accurate offloading rate.Simulation experiments show that the proposed strategy can effectively balance the load state of the cluster,reduce the cost and optimize the total delay of the system.