Two-stage task cooperation unloading method in Internet of Vehicles

Abstract

The invention discloses a two-stage task cooperation unloading method in the Internet of Vehicles, and designs a task cooperation unloading framework fusing an LTE-V direct connection mode and an LTE-V cellular mode, and deploys a mobile edge computing server beside a roadside unit through wired connection. The mobile edge computing server at the roadside unit side cooperates with the idle service vehicle to provide computing service; meanwhile, a two-stage unloading decision scheme is provided, and the link quality is evaluated by a link quality evaluation model so as to improve the stability of data transmission between a prediction calculation node and a request vehicle. The response time is predicted through the unloading decision model, the unloading decision is made according to the predicted response time, and compared with an existing scheme, the proposed unloading decision scheme is more effective, the response time can be shortened, and energy consumption can be reduced.

Description

Technical Field

[0001] The present invention relates to a cooperative computing offloading method, and in particular to a two-stage task cooperative offloading method in a vehicle network. Background Art

[0002] With the popularization of the Internet of Things (IoT), the configuration of vehicles has become more complex, and the applications supported by vehicle terminals (VTs) have become more and more intensive, such as high-precision navigation, high-definition video streaming, and live broadcasting. However, resources and processing power are still limited and cannot meet the needs of computationally intensive and time-sensitive tasks.

[0003] The central cloud has abundant resources, and vehicle tasks can be offloaded to the cloud center, but long-distance transmission brings greater latency and higher transmission costs. In addition, the performance of cloud-based offloading technology is limited by the real-time changing wireless environment and communication requirements, and the quality of service (QoS) of non-line-of-sight (NLOS) is easily affected by external environmental factors (reference [1]). Vehicle cloud computing (VCC) integrates vehicle self-organizing networks and cloud computing (reference [2]) to alleviate these problems. VCC also realizes computing collaboration between the cloud center and VTs through vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) technologies, where VTs are used as computing carriers to improve resource utilization and computing efficiency. However, since the quality of service may be limited by the transmission rate and time fluctuations of the backhaul and backbone networks, VCC cannot meet the high requirements of time-sensitive and computing-intensive tasks (reference [3]). In recent years, mobile edge computing (MEC) has attracted widespread attention in the 5 / 6G field as a potential solution, which deploys communication, computing, and storage capabilities close to VTs. Distributed edge servers alleviate network congestion and reduce computing time (reference [4]). In vehicle edge computing (VEC), tasks are offloaded to the MEC server through the roadside unit (RSU), and the MEC server executes the task and returns the calculation results (reference [5]). Compared with VCC, VEC greatly reduces the transmission time. In addition, VEC expands the computing power of the vehicle and saves CPU cycles, further ensuring the quality of service requirements.

[0004] The Internet of Vehicles (IoV) is a key technology that can simultaneously meet the needs of comfortable driving experience and high-quality entertainment services. In recent years, the increase in the number of vehicles has brought challenges to the scarce wireless resources, limited computing resources, and the processing power of VTs, increasing the response time. Therefore, it is crucial to reasonably allocate MEC computing to reduce the response time. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a two-stage task collaborative unloading method in a vehicle network that can reduce the response time of task unloading and save energy consumption.

[0006] The technical solution adopted by the present invention to solve the above technical problems is: a two-stage task collaborative unloading method in the Internet of Vehicles, all vehicles are equipped with cellular interfaces, and each vehicle is provided with sensors for sensing position, speed and direction, and a task collaborative unloading framework integrating LTE-V direct connection mode and LTE-V cellular mode is established. In the task collaborative unloading framework, the computing node is composed of a mobile edge computing server on the roadside unit side and a service vehicle. When in the LTE-V direct connection mode, the computing node is the service vehicle, and when in the LTE-V cellular mode, the computing node is the mobile edge computing server on the roadside unit side. The specific unloading method is:

[0007] A link quality evaluation model for predicting the position of the vehicle in the next time slot and evaluating the link quality and an unloading decision model for predicting the response time are established. In the data request stage, the requesting vehicle sends a HELLO information packet to the adjacent computing node, and the link quality is obtained by the link quality evaluation model. When the task needs to be unloaded to the computing node, the unloading decision model obtains the computing node with the shortest response time based on the link quality and the task input data volume, the maximum tolerable time, the available CPU resources of the computing node and the wireless channel. The requesting vehicle sends the request information to the computing node with the shortest response time. When the computing node is the mobile edge computing server on the roadside unit side, the mobile edge computing server on the roadside unit side directly provides computing services to the requesting vehicle. When the computing node is the service vehicle, after obtaining the request information, the service vehicle determines whether it has sufficient computing resources and feeds back the resource status to the requesting vehicle. The requesting vehicle decides whether to unload the task.

[0008] Compared with the prior art, the advantages of the present invention are:

[0009] 1) A task collaborative offloading framework integrating LTE-V direct connection mode and LTE-V cellular mode is designed. The mobile edge computing server is deployed next to the roadside unit through a wired connection. The mobile edge computing server on the roadside unit side collaborates with idle service vehicles to provide computing services;

[0010] 2) A two-stage offloading decision scheme is proposed. The link quality assessment model evaluates the link quality to improve the stability of data transmission between the prediction computing node and the requesting vehicle. The offloading decision model then predicts the response time and makes an offloading decision based on the predicted response time.

[0011] 3) The simulation results show that compared with existing schemes such as linear regression prediction algorithm and combined prediction unloading algorithm, the proposed unloading decision scheme is more effective and can reduce response time and save energy consumption.

[0012] Preferably, at a turn or in a place blocked by obstacles, the requesting vehicle is in the LTE-V cellular mode, and the requesting vehicle sends a HELLO information packet to the roadside unit through the cellular network, and the mobile edge computing server on the roadside unit side directly provides computing services for the requesting vehicle.

[0013] Preferably, when the requesting vehicle is in LTE-V direct connection mode, information of neighboring service vehicles is obtained in an alternative network topology scenario.

[0014] Preferably, the content of the HELLO information packet includes a broadcast ID, location information, requested task information, and list information of neighboring computing nodes.

[0015] Preferably, the link quality assessment model is:

[0016] Among them, E ij Indicates link quality, T s represents a fixed time interval, T u Represents an update cycle, Indicates T s In the update cycle T u The proportion within, p M Indicates time T s The number of HELLO packets in R ij Indicates the distance between the requesting vehicle and the computing node

[0017]

[0018] Among them, (a i ,b i ) indicates requesting vehicle n i The longitude and latitude of the current location, (a′ i , b′ i ) indicates requesting vehicle n i the longitude and latitude of the predicted location, v i Indicates request vehicle n i The speed, v im Indicates request vehicle n i Maximum allowed speed, θ i Indicates request vehicle n i The angle (a j ,b j ) represents the computing node nj The longitude and latitude of the current location, (a' j ,b' j ) represents the computing node n j the longitude and latitude of the predicted location, v j Represents computing node n j The speed, θ j Represents computing node n j The angle, v jm Represents computing node n j Maximum speed allowed.

[0019] Preferably, the offloading decision model is a deep neural network consisting of an input layer, a hidden layer and an output layer, wherein the input layer is a network composed of link quality E ij , Task input data volume Maximum tolerable time Available CPU resources of the compute node and wireless channel b ij The task feature matrix The number of hidden layers is 10. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a task collaborative offloading framework diagram that integrates the LTE-V direct connection mode and the LTE-V cellular mode proposed in the present invention;

[0021] Figure 2 Schematic diagram of the two-stage offloading decision scheme proposed in the present invention, where (a) is a link quality assessment model and (b) is an offloading decision model;

[0022] Figure 3 The figure is a comparison diagram of the position prediction results of the present invention and the prior art at different iteration times, wherein (a) is the predicted longitude and (b) is the predicted latitude;

[0023] Figure 4 : is a comparison diagram of the root mean square error between the present invention and the prior art at different iteration times, where (a) is the predicted longitude and (b) is the predicted latitude;

[0024] Figure 5 A comparison chart of the response time of the present invention and the prior art at different iteration times;

[0025] Figure 6 This is a comparison diagram of average delays between the present invention and the prior art at different iteration times;

[0026] Figure 7 This is a comparison diagram of average delays between the present invention and the prior art under different numbers of vehicles;

[0027] Figure 8 This is a comparison chart of average delays between the present invention and the prior art under different task data volumes;

[0028] Fig. 9 The figure is a comparison of energy consumption between the present invention and the prior art under different task data amounts. DETAILED DESCRIPTION

[0029] The present invention is further described in detail below with reference to the accompanying drawings.

[0030] Embodiment: A two-stage task collaborative offloading method in a vehicle network, all vehicles are equipped with cellular interfaces, and each vehicle is provided with sensors for sensing position, speed and direction, and a task collaborative offloading framework integrating LTE-V direct connection mode and LTE-V cellular mode is established. In the task collaborative offloading framework, a computing node is composed of a mobile edge computing server on the roadside unit side and a service vehicle. When in the LTE-V direct connection mode, the computing node is a service vehicle, and the information of all nearby service vehicles is obtained in an alternative network topology scenario. At a turn or where there is an obstacle, the requesting vehicle is in the LTE-V cellular mode, and the requesting vehicle sends data information to the roadside unit through the cellular network. The mobile edge computing server on the roadside unit side acts as a computing node to directly provide computing services to the requesting vehicle. The specific offloading method is:

[0031] A link quality evaluation model for predicting the position of the vehicle in the next time slot and evaluating the link quality and an offloading decision model for predicting the response time are established. The link quality evaluation model is:

[0032] Among them, E ij Indicates link quality, T s represents a fixed time interval, T u Represents an update cycle, Indicates T s In the update cycle T u The proportion within, p M Indicates time T s The number of HELLO packets in R ij Indicates the distance between the requesting vehicle and the computing node

[0033]

[0034] Among them, (a i ,b i ) indicates requesting vehicle n i The longitude and latitude of the current location, (a′ i , b′ i ) indicates requesting vehicle n ithe longitude and latitude of the predicted location, v i Indicates request vehicle n i The speed, v im Indicates request vehicle n i Maximum allowed speed, θ i Indicates request vehicle n i The angle (a j ,b j ) represents the computing node n j The longitude and latitude of the current location, (a' j ,b' j ) represents the computing node n j the longitude and latitude of the predicted location, v j Represents computing node n j The speed, θ j Represents computing node n j The angle, v jm Represents computing node n j The maximum speed allowed, when in LTE-V cellular mode, since the position is fixed and the speed is zero, the distance between the requesting vehicle and the computational node is:

[0035] The longitude and latitude of the predicted position of the requesting vehicle and the computing node are obtained by the evolutionary neural network. In the evolutionary neural network, the particle swarm optimization algorithm based on weight adaptation is used to initialize the feedback neural network. The weight update formula is expressed as:

[0036]

[0037] Where w(t) is the inertia weight, which is between the maximum value w max With the minimum value w min d1, d2 are control factors, t is the current response time, t max is the maximum tolerable response time. The loss function of the particle swarm optimization algorithm based on weight adaptation is used as the training error, and the optimal weight and threshold of the feedback neural network are determined by searching. The loss function f i Defined as:

[0038]

[0039] Where D is the particle dimension, is the expected output value of the training data, x dis the actual output value of the model. In this way, the optimal weight and threshold that minimize the training error can be effectively found, thereby optimizing the performance of the feedback neural network. Therefore, by inputting the current position of the vehicle into the optimized feedback neural network, it is possible to predict the requesting vehicle and the computing node in the update period T u The position after (a′ i , b′ i ) and (a' j ,b' j ); The offloading decision model is a deep neural network consisting of an input layer, a hidden layer, and an output layer. The input layer is composed of the link quality E ij , Task input data volume Maximum tolerable time Available CPU resources of the compute node and wireless channel b ij The task feature matrix The number of hidden layers is 10.

[0040] In the data request phase, the requesting vehicle sends a HELLO packet to the adjacent computing node. The content of the HELLO packet includes the broadcast ID, location information, requested task information, and the list information of the adjacent computing nodes. The link quality evaluation model is used to obtain the link quality. When the task needs to be offloaded to the computing node, the offloading decision model determines the link quality E according to the link quality. ij And the amount of task input data Maximum tolerable time Available CPU resources of the compute node and wireless channel b ij The four prediction variables obtain the computing node with the shortest response time, and the requesting vehicle sends the request information to the computing node with the shortest response time. When the computing node with the shortest response time is the mobile edge computing server on the roadside unit side, the mobile edge computing server on the roadside unit side directly provides computing services to the requesting vehicle. When the computing node with the shortest response time is the service vehicle, after obtaining the request information, the service vehicle determines whether it has sufficient computing resources and feeds back the resource status to the requesting vehicle, and the requesting vehicle decides whether to unload the task.

[0041] The method of the present invention is simulated, and the results and performance analysis are as follows

[0042] A simulation environment was established on the MATLAB platform to evaluate the performance of the offloading mechanism. It is assumed that the configuration of the vehicle complies with the LTE-V standard, the parameters of the vehicle are set in accordance with the medium-sized SUV, and the computing power of each vehicle is from reference [6]. The environmental deployment of the MEC server complies with the provisions of the MEC white paper. The dataset of the vehicle path prediction model (references [7]-[8]) comes from the Microsoft T-Drive project. The dataset contains the trajectory data of 10,357 taxis in Beijing in 2008. The dataset contains 15 million coordinate points and the total trajectory distance exceeds 9 million kilometers. The task dataset referenced in the response time prediction is from reference [9]. The basic simulation parameters are listed in Table 1.

[0043] Table 1

[0044]

[0045]

[0046] 5.2 Prediction Performance Evaluation

[0047] The present invention proposes a two-stage task offloading decision algorithm, including an ENN-based link quality assessment model and an offloading decision model for predicting response time. In the first stage, the ENN predicts the future position of the vehicle, and the link quality assessment model calculates the link quality between the node and the vehicle after a certain period of time; in the second stage, the link quality and the available resource size of the candidate node are input into the deep neural network model to predict the response time. Figure 3 and Figure 4 The accuracy of the Evolved Neural Network (ENN) in predicting positions was evaluated. Figure 5 The accuracy of deep neural networks in predicting response times was evaluated.

[0048] Figure 3 The comparison of the location prediction performance of three models is shown, namely the ENN model, the classic BP network model and the AMRIA model (reference

[10] ). It can be seen that the ENN model has high accuracy in predicting longitude and latitude by adjusting the initial weights and thresholds, and has a good fitting effect. The classic BP neural network has better nonlinear processing capabilities, but due to the random initialization of values, the algorithm is prone to fall into local optimality.

[0049] Figure 4 The root mean square error (RMSE) of the three models is compared. RMSE is the inconsistency between the actual value and the predicted value. The closer the RMSE value is to zero, the better the performance is. It can be calculated by the following formula:

[0050]

[0051] Where N is the total number of sampled data, O i is the true value, P i is the predicted value.

[0052] from Figure 4 It can be seen that when the number of iterations increases, the RMSE values ​​of ENN, classic BP and AMRIA algorithms become smaller. Compared with other models, ENN has the best effect and faster convergence speed.

[0053] 2) Response time prediction

[0054] like Figure 5 As shown in Figure 2, we compare the prediction performance of the proposed two-stage prediction model with that of reference

[11] . Our solution has a better fit to the actual response time. The reason is that by fusing computing power and communication status, our solution adopts two-stage response time learning information. In addition, the nonlinear mapping characteristics of ENN are better than linear regression when dealing with multiple variables. Therefore, the two-stage prediction model can make offloading decisions more accurately.

[0055] 5.3 Offloading Decision Performance Analysis

[0056] Task offloading is an effective way to alleviate the limited computing power of vehicles. Time and energy consumption are key factors in evaluating offloading performance. In this section, the offloading performance of the proposed two-stage offloading decision algorithm (TSODA) is compared with the following schemes:

[0057] Linear regression prediction algorithm (LRP): Based on the relationship between historical status and response time, a linear regression equation is established for statistical analysis.

[0058] Combined Mode Offloading (CMR) algorithm [4]: ​​Adaptively offloads tasks to MEC servers through direct offloading in V2I mode or predictive relay transmission in V2V mode.

[0059] GPSR Algorithm (GPSR) Reference

[10] : Makes offloading decisions based only on the router's direct neighbor information.

[0060] exist Figure 6In , we show the average time when the task size is a certain value. It can be seen that, except for GPSR, the average time of task offloading is relatively stable. TSODA outperforms the other three algorithms because it first considers the quality of the communication link between nodes to ensure higher reliability, thereby effectively reducing the number of retransmissions and transmission time; then selects the optimal offloading node based on the response time, further reducing the computation time. GPSR extends an arbitrary routing mechanism based on the right-hand rule, resulting in more routing hops, resulting in a larger transmission time, so GPSR has the largest average time. CMR can find the next hop node to transmit the task forward, taking into account the available computing resources and the distance between adjacent vehicles. However, due to the limitations of environmental changes, the optimal relay node cannot be obtained, resulting in an increase in transmission time. In addition, the prediction accuracy of LRP is lower than that of ENN, so the average task execution time is higher.

[0061] exist Figure 7 In , it is shown that as the number of VTs increases, the average time also increases. The same trend verifies that the increase in the number of vehicles makes computing resources scarce and enhances the interference between vehicles. The curve of ENN is much lower than that of CMR, indicating the effectiveness of the two-stage response time prediction. When the candidate computing nodes are sparsely distributed, there are fewer available neighboring nodes and the movement of vehicles further aggravates the routing failure. Therefore, in the GPSR algorithm, the time is the longest because the vehicles are assigned to imperfect routing relays. The interaction effects between variables directly affect the prediction results of LRP. In addition, LRP cannot handle nonlinear causal relationships between variables.

[0062] Figure 8 and Fig. 9 The average time and energy consumption of VT when the task size is a certain value are shown respectively, and the performance improvement of the proposed ENN scheme is obvious. Figure 8 In,our mechanism adopts a cooperative offloading scheme, where fine-grained subtasks are executed in parallel on different computing nodes, which greatly reduces the transmission time and execution time.,GPSR performs the worst due to the increase in packet loss and timeout rates between routing packets.

[0063] The energy consumption calculation formula is as follows: Among them, z represents the energy consumption per unit CPU cycle, a j Represents the computing power of the vehicle. The energy consumption is related to the available resources of the computing node and the available resources required for the task. Through the response time prediction model, the computing node with the minimum energy consumption can be selected while meeting the time requirement.

[0064] exist Fig. 9In ,for the contrasted algorithms, the update of the available resources of the computing nodes will lead to the change of their computing capabilities, which indirectly leads to a larger prediction error than the proposed algorithm.,In addition, the energy consumption calculation formula reflects the instability of LRP when,processing nonlinear variables.

[0065] Summarize:

[0066] In this paper, the use of cooperative communication with task offloading in VEC systems is studied to minimize the response time and improve the vehicle quality of service (QoS). The proposed TSODA consists of two stages: First, a link quality model between vehicles is established to enhance the transmission stability. The ENN scheme is used to predict the location of the vehicle. Secondly, variables such as link quality, task size, and available resources of computing nodes are input into the DNN model to predict the response time and implement task offloading decisions. In the simulation results section, the accuracy of the two-stage prediction model is compared with the classic BP and AMRIA. It can be seen that the prediction results of the model have a strong fitting characteristic with the actual values. In addition, compared with the existing technologies, TSODA is able to reduce time and energy consumption.

[0067] References

[0068] [1] S. Olariu. A Survey of Vehicular Cloud Research: Trends, Applications and Challenges, IEEE Trans.Intell.Transp.Syst.21(6)(June 2020)2648-2663, https: / / doi.org / 10.1109 / TITS.2019.2959743.

[0069] [2] A.Alamer, Y.Deng, G.Wei,

[0070] [3]K.Zhang,Y.Mao,S.Leng,A.Vinel,Y.Zhang,Delay constrained offloadingfor Mobile Edge Computing in cloud-enabled vehicular networks,in:2016 8thInternational Workshop on Resilient Networks Design and Modeling(RNDM),2016,pp.288-294,https: / / doi.org / 10.1109 / RNDM.2016.7608300.

[0071] [4]K.Zhang,Y.Mao,S.Leng,Y.He,Y.ZHANG,Mobile-Edge Computing forVehicular Networks:A Promising Network Paradigm with Predictive Off-Loading,IEEE Veh.Technol.Mag,12(2)(June 2017)36-44,https: / / doi.org / 10.1109 / MVT.2017.2668838.

[0072] [5]H.Li,T.Zhang,J.Zhang,K.Shi,P.Zeng.Time deterministic routingalgorithm and protocol based on time-varying graph over the space-groundintegrated network.Journal on Communications,41(10)(2020)116-129.https: / / doi.org / 10.11959 / j.issn.1000-436x.2020188.

[0073] [6]W.Jung,J.Yim,Y.Ko,S.Singh,Acods:adaptive computation offloadingfor drone surveillance system,in:In 2017 16th Annual Mediterranean Ad HocNetworking Workshop(Med-Hoc-Net),2017,pp.1-6.https: / / doi.org / 10.1109 / MedHocNet.2017.8001647.

[0074] [7]J.Yuan,Y.Zheng,X.Xie,G.Sun.Driving with Knowledge from thePhysical World,(2011)316-324.https: / / doi.org / 10.1145 / 2020408.2020462.

[0075] [8]X.Hou,Y.Li,M.Chen,D.Wu,D.Jin,S.Chen,Vehcular Fog Computing:AViewpoint of Vehicles as the Infrastructures,IEEE Trans.Veh.Technol,65(6)(2016)3860-3873, https: / / doi.org / 10.1109 / TVT.2016.2532863 .

[0076] [9]H.Zhang,Q.Zhang,X.Du,Toward Vehicle-Assisted Cloud Computing forSmartphones,IEEE Trans.Veh.Technol,64(12)(2015)5610-5618, https: / / doi.org / 10.1109 / TVT.2015.2480004 .

[0077]

[10] B.Karp,H.Kung,GPSR:Greedy perimeter stateless routing forwireless networks,in:Proceedings of the 6th annual international conferenceon Mobile computing and networking,2000,pp.243-254,https: / / doi.org / 10.1145 / 345910.345953.

[0078]

[11] R.Valentino,W.S.Jung,Y.B.Ko,A design and simulation of theopportunistic computation offloading with learning-based prediction forunmanned aerial vehicle(uav)clustering networks.Sensors,18(11)(2018)3751.https: / / doi.org / 10.3390 / s18113751。

Claims

1. A two-stage task cooperative offloading method in a vehicle network, characterized in that: All vehicles are equipped with cellular interfaces, and each vehicle is provided with sensors for sensing position, speed and direction, and a task collaborative offloading framework integrating LTE-V direct connection mode and LTE-V cellular mode is established. In the task collaborative offloading framework, the computing node is composed of the mobile edge computing server on the roadside unit side and the service vehicle. When in the LTE-V direct connection mode, the computing node is the service vehicle, and when in the LTE-V cellular mode, the computing node is the mobile edge computing server on the roadside unit side. The specific offloading method is: A link quality evaluation model for predicting the position of the vehicle in the next time slot and evaluating the link quality and an unloading decision model for predicting the response time are established. In the data request stage, the requesting vehicle sends a HELLO information packet to the adjacent computing node, and the link quality is obtained by the link quality evaluation model. When the task needs to be unloaded to the computing node, the unloading decision model obtains the computing node with the shortest response time based on the link quality and the task input data volume, the maximum tolerable time, the available CPU resources of the computing node and the wireless channel. The requesting vehicle sends the request information to the computing node with the shortest response time. When the computing node is the mobile edge computing server on the roadside unit side, the mobile edge computing server on the roadside unit side directly provides computing services to the requesting vehicle. When the computing node is the service vehicle, after obtaining the request information, the service vehicle determines whether it has sufficient computing resources and feeds back the resource status to the requesting vehicle. The requesting vehicle decides whether to unload the task.

2. A two-stage task cooperative offloading method in a vehicle network as claimed in claim 1, characterized in that: At turns or places blocked by obstacles, the requesting vehicle is in the LTE-V cellular mode, and the requesting vehicle sends a HELLO information packet to the roadside unit through the cellular network. The mobile edge computing server on the roadside unit side directly provides computing services for the requesting vehicle.

3. A two-stage task cooperative offloading method in a vehicle network as claimed in claim 1, characterized in that: When the requesting vehicle is in LTE-V direct connection mode, information of neighboring service vehicles is obtained in an alternative network topology scenario.

4. A two-stage task cooperative offloading method in a vehicle network as claimed in claim 1, characterized in that: The content of the HELLO information packet includes broadcast ID, location information, requested task information and list information of neighboring computing nodes.

5. A two-stage task cooperative offloading method in a vehicle network as claimed in claim 1, characterized in that: The link quality assessment model is: Among them, E ij Indicates link quality, T s represents a fixed time interval, T u Represents an update cycle, Indicates T s In the update cycle T u The proportion within, p M Indicates time T s The number of HELLO packets in R ij Indicates the distance between the requesting vehicle and the computing node Among them, (a i ,b i ) indicates requesting vehicle n i The longitude and latitude of the current location, (a i ', b i ') indicates requesting vehicle n i the longitude and latitude of the predicted location, v i Indicates request vehicle n i The speed, v im Indicates request vehicle n i Maximum allowed speed, θ i Indicates request vehicle n i The angle (a j ,b j ) represents the computing node n j The longitude and latitude of the current location, (a' j ,b' j ) represents the computing node n j the longitude and latitude of the predicted location, v j Represents computing node n j The speed, θ j Represents computing node n j The angle, v jm Represents computing node n j Maximum speed allowed.

6. A two-stage task cooperative offloading method in a vehicle network as claimed in claim 1, characterized in that: The offloading decision model is a deep neural network consisting of an input layer, a hidden layer and an output layer. The input layer is composed of the link quality E ij , Task input data volume Maximum tolerable time Available CPU resources of the compute node and wireless channel b ij The task feature matrix The number of hidden layers is 10.

Images