Trusted task unloading method based on autonomous collaboration of vehicle side cloud

Abstract

The invention discloses a trusted task unloading method based on autonomous collaboration of vehicle side cloud. The method can be applied to the dynamic heterogeneous Internet of Vehicles to solve the low-efficiency problems of task accumulation, resource preemption, poor cooperation and the like caused by false tasks and malicious nodes. Firstly, an unloading task and cooperative node trust evaluation model is provided, and false tasks and malicious nodes are accurately identified and filtered by analyzing data sensitivity, resource preemption, historical unloading success rate and cooperative feedback evaluation; then, a discrete particle swarm algorithm is utilized to iteratively obtain a vehicle edge cloud collaborative unloading decision, particle codes, a fitness function and a correction mechanism are redesigned, and random quantities, greedy and mapping functions are introduced to enhance the diversity of unloading decisions. Compared with a traditional task unloading method, the method shows better energy consumption and delay performance in a scene with compact and uniformly distributed tasks, and the unloading efficiency and safety are effectively improved.

Description

Technical Field

[0001] The present invention belongs to the field of Internet of Vehicles covered by the new generation of information technology, and relates to a trusted task offloading method based on autonomous collaboration between vehicle-side and cloud. Background Art

[0002] The Internet of Vehicles uses roadside perception, edge computing and cloud fusion technologies to achieve information exchange and sharing between people, vehicles and roads, and provide collaborative perception, decision-making and control of complex traffic environments. With the development of Internet of Vehicles technology, emerging applications such as real-time navigation, collaborative lane change, and collision warning are installed in the vehicle environment, which puts higher requirements on the vehicle's computing, storage and response speed. For example, in real-time navigation, collision warning and other services, it is necessary to recommend multiple driving routes in real time based on dynamic road conditions and vehicle positions, and the task computing density is high; in collision warning services, in order to solve problems such as conflict avoidance and lane change, the computing delay is generally required to be within tens of milliseconds; in addition, in intelligent assisted driving scenarios such as unmanned driving, task computing usually has multiple characteristics such as computing intensive, delay sensitive, and resource consumption. Therefore, in the face of a large number of computing intensive and delay sensitive computing tasks generated by applications, traditional traffic intelligent terminals such as vehicles have limited computing power and severe battery loss, and it is difficult to meet the system task processing requirements. Especially when the vehicle density is high, there may be an imbalance between supply and demand with limited available resources but too high demand resources.

[0003] Task offloading is a key technology to improve the efficiency of task processing in the Internet of Vehicles. It offloads tasks that are difficult for vehicles to handle to the edge or cloud for execution, and makes use of the powerful computing performance of the cloud and the latency advantage of the edge to make up for the shortcomings of vehicles in computing, storage, and energy efficiency. In the Internet of Vehicles system with vehicle-side cloud collaboration, there are multiple intelligent nodes such as vehicle terminals, edge servers, and cloud servers. They have heterogeneous computing and storage capabilities. The autonomous collaborative offloading of multiple intelligent agents in the vehicle-side cloud can achieve resource integration and computing collaboration, effectively improving the efficiency of task processing and the level of transportation services.

[0004] However, although a large number of task offloading methods have been proposed in the current academic and industrial circles, most of these methods aim to reduce latency and energy consumption, and rarely analyze the security and trust issues that may be faced during task offloading. In fact, due to the open, collaborative, and dynamic characteristics of the Internet of Vehicles environment, and the large number of vehicles and service nodes involved, the security and reliability issues faced are more complicated than those of ordinary networks. First, due to the existence of malicious request nodes such as free-riding and selfish nodes, edge and cloud service node resources are seized and consumed, resulting in the death of service nodes due to excessive resource consumption, causing problems such as accumulation of offloaded tasks and high execution failure rate. Secondly, task offloading is essentially the outsourcing of computing. If it is offloaded to malicious service nodes, it will seriously threaten privacy and data security, especially in the computing mode of vehicle-side cloud collaboration, conflicts of interest between service nodes lead to frequent situations such as bad collaboration and resource preemption. Finally, the Internet of Vehicles has the characteristics of heterogeneous resources, dynamic access, and vehicle mobility, and the offloading process is affected by channel quality, propagation loss, server availability, etc., making it difficult for a small number of existing task offloading methods to be transplanted and applied to real Internet of Vehicles scenarios.

[0005] In response to the above problems, the present invention provides a trusted task offloading method based on autonomous vehicle-side cloud collaboration to meet the needs of trusted and efficient task offloading. First, a trust mechanism is used to identify and filter false tasks and malicious nodes, and then a discrete particle swarm algorithm is used to iteratively obtain offloading decisions. Finally, trust perception and collaborative offloading are mutually enhanced to improve offloading efficiency and security. Summary of the invention

[0006] The present invention provides a trusted task offloading method based on autonomous vehicle-side cloud collaboration, which can be applied to dynamic heterogeneous Internet of Vehicles to solve inefficiencies such as task accumulation, resource preemption, and poor collaboration caused by false tasks and malicious nodes.

[0007] The technical solution adopted by the present invention is: first, by analyzing data sensitivity, resource preemption, historical unloading success rate and collaborative feedback evaluation, false tasks and malicious nodes are accurately identified and filtered; then, the discrete particle swarm algorithm is used to iteratively obtain the vehicle-side cloud collaborative unloading decision, redesign the particle encoding, fitness function and correction mechanism, and introduce random quantity, greed and mapping function to enhance the diversity of unloading decisions, which mainly includes the following steps:

[0008] Step 1: Perform authenticity assessment on pending tasks generated in the current cycle, based on the data sensitivity of the task itself and the task request frequency of the source node generating the task. i , based on its task sensitivity and source node resource preemptibility, a authenticity evaluation function is constructed, which is specifically expressed as:

[0009]

[0010] Among them, α and β are weight adjustment factors, and It is task T i The amount of sensitive data and total data involved, K is the total number of tasks in the task generation cycle, is the number of task processing requests initiated by the task source node in the kth observation period. Then, based on the authenticity of the task evaluation Classification is performed, if the authenticity is lower than the set threshold Then the task is judged to have an attack risk and removed from the task set If the authenticity is higher than the threshold, Keep it in the collection and enter the subsequent uninstallation decision stage.

[0011] Step 2: Conduct trust evaluation on the vehicles and edge nodes involved in the task calculation, based on the node's historical task processing success rate and the feedback evaluation based on the collaborative interaction nodes. i , analyze its historical task success rate and interactive collaboration reputation, introduce the weight adjustment factor ω, and construct its trust evaluation function in the kth observation period:

[0012]

[0013] in, and is node N i The number of successful and failed tasks processed during the k-th observation period, is the information provided by the jth interactor about node N i Feedback evaluation, R j is the reliability of the jth interactor, and M is the total number of interacting nodes that provide feedback evaluation. Then, the behavior trend and time decay characteristics of the nodes in multiple observation periods are integrated to build a comprehensive evaluation and filtering model. Each period is divided into M trust observation periods. When considering time decay, the node N i The comprehensive trust in the kth observation period can be expressed as:

[0014]

[0015] in,

[0016] in, is the decay function of the node trust memory retention with respect to time t, η1 = 1.84, η2 = 1.25, t represents the difference between the corresponding trust relationship and the current time, in minutes. Then, classification is performed based on the node trust. If the node credibility is lower than the trust threshold Then the node is determined to be a malicious node and removed from the node set If the node credibility is higher than the threshold It is then allowed to enter the subsequent offloading decision phase and serve as a candidate node for task offloading and calculation.

[0017] Step 3: Taking real tasks and trusted nodes as objects, improve and redesign the population initialization, fitness evaluation, speed and position update, and validity correction steps of the discrete particle swarm algorithm to iteratively obtain the best task offloading decision. First, determine the particle position and speed coding, fix the coding length to n, n is the number of tasks, and each position code value is null or n i , Represents the task processing node, and each bit of the particle speed encoding takes a value of 0 or 1. 0 means adjusting the corresponding position encoding segment in the next iteration, and 1 means not adjusting the position segment in the next iteration. Then, the local priority-random adjustment algorithm is used to generate the initial particle population; then, the fitness function of particle evaluation is constructed based on the weighted and optimized objective of minimizing energy consumption and delay, which is specifically expressed as:

[0018]

[0019] in, It is task T i About Node n j If the processing decision This means that task T i Assign to n j , at this time the task processing cost is λ is the weight adjustment factor of energy consumption and delay, and the individual optimal particle and the global optimal particle are found based on the evaluated particle fitness; then, random quantities, mapping functions, etc. are introduced to update the velocity encoding of the particle in the next iteration, and the position encoding in the next iteration is updated based on the velocity vector and combined with the individual and optimal particles; finally, the particle validity is corrected against the node resource constraints and task delay constraints; it is repeated until a converged offloading decision is obtained or the maximum number of iterations is reached. After obtaining the best offloading decision, the task offloading service is executed, and the node trust evaluation model is reversely updated based on the task completion quality, node collaboration, etc.

[0020] The method of the present invention has the following beneficial effects: first, by evaluating the authenticity and trust of the unloading tasks and the collaborative nodes, false tasks and malicious nodes can be identified and filtered before unloading, the decision space can be reduced, and the unloading cost can be saved; then, by using the discrete particle swarm algorithm to obtain the unloading decision and introducing random quantity, greed and mapping functions, the global optimization ability of the unloading decision can be improved, and the efficiency of task allocation and unloading can be improved. In general, the present invention has good energy consumption and delay performance in the Internet of Vehicles scenario with compact or uniform task distribution, and simultaneously improves the unloading efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a network architecture for application of the present invention;

[0022] Figure 2 The overall framework diagram for the implementation of the method of the present invention;

[0023] Figure 3 A schematic diagram of particle coding involved in implementing unloading decision-making in the present invention;

[0024] Figure 4 A schematic diagram of particle speed and position update involved in implementing unloading decision-making in the present invention;

[0025] Figure 5 A schematic diagram showing the comparison of node recognition rates between the method of the present invention and the comparative method;

[0026] Figure 6 A schematic diagram showing a comparison of task recognition rates between the method of the present invention and the comparative method;

[0027] Figure 7 A schematic diagram showing a comparison of energy consumption between the method of the present invention and a comparative method;

[0028] Figure 8 It is a schematic diagram of delay comparison between the method of the present invention and the comparative method;

[0029] Fig. 9 A schematic diagram showing a comparison of the task benefits of the method of the present invention and the comparative method;

[0030] Fig.10 It is a schematic diagram comparing the task completion rates of the method of the present invention and the comparative method. DETAILED DESCRIPTION

[0031] In order to facilitate the understanding of the essential content of the present invention, the main technical parameters involved in the present invention are first defined and explained.

[0032] Table 1 Symbols, definitions and descriptions of main technical parameters

[0033]

[0034]

[0035] Next, network scenarios to which the present invention can be applied are described.

[0036] Figure 1 The figure shows a schematic diagram of the network architecture of the present invention. The entire system consists of m vehicle nodes, n edge nodes and 1 cloud node. Multiple roadside units and base stations are also deployed to assist in communication and data processing. The vehicle set is denoted as The edge server set is denoted as Represents all task computing nodes n i The task offloading of the present invention adopts a binary complete offloading mode. Each task can only be executed as an independent whole in the vehicle or offloaded to the edge server or cloud. The task set is recorded as For each task T i , whose features are available in tuples Indicates that T ID is the task number, is the source node identity of the task, c i Calculate the required CPU cycles for the task, d i is the amount of input data, t i is the maximum tolerable delay, represents the task processing decision, Indicates that task T i Assigned to computing node n j , This means that the task is not assigned to n j .

[0037] The optimization goal of the method of the present invention is to minimize the average task processing energy consumption and the weighted sum of delays. i , assuming it consists of node n j The energy consumption during the processing is Delay is The objective optimization function is expressed as follows:

[0038] P1:

[0039]

[0040] in, is the number of tasks, is the number of nodes, is the number of successfully executed tasks, is the task processing decision, λ is a weight parameter, λ∈[0,1], which is used to achieve a dynamic trade-off between task delay and system energy consumption. C1-C4 are constraints, where C1 and C2 indicate that a task can only be executed completely and independently by one node at most, C3 indicates that the resources required for task execution cannot be greater than the remaining available resources of the computing node, and C4 indicates that the task T i The execution time must be within its tolerance delay.

[0041] The specific implementation process of the present invention is described below.

[0042] Figure 2 The overall framework diagram of the implementation of the method of the present invention is shown, and the complete implementation steps included are as follows

[0043] Step 1: Evaluate the authenticity of pending tasks generated in the current cycle, based on the data sensitivity of the task itself and the frequency of unloading requests from the source node where the task is generated. If the authenticity of the task is lower than the set threshold, Then the task is judged to have an attack risk and removed from the task set The remaining real tasks enter the offloading decision stage;

[0044] Step 2: Conduct trust evaluation on vehicles and edge nodes involved in task calculations. The first step is based on the node's historical task processing success rate, and the second step is based on the feedback evaluation of collaborative interaction nodes. The node's comprehensive credibility is obtained by integrating the node's behavioral trends and time decay characteristics over multiple observation periods. If the node's credibility is lower than the set threshold, Then the node is determined to be a malicious node and removed from the node set The remaining trusted nodes enter the uninstallation decision phase;

[0045] Step 3: Taking real tasks and trusted nodes as the objects, improve and redesign the discrete particle swarm algorithm's population initialization, fitness evaluation, speed and position update, and validity correction steps, iteratively obtain the best task offloading decision, execute the offloading service, and reversely update the node trust evaluation model based on the task completion quality, node collaboration status, etc.

[0046] It can be seen that the implementation of the method of the present invention mainly includes three key steps: offloading task authenticity evaluation, cooperative node trust evaluation and offloading decision based on discrete particle swarm. The specific implementation plans of these three steps are described in detail below:

[0047] ① Perform authenticity assessment on pending tasks generated in the current cycle. To prevent the current task from excessively requesting relevant data from non-self nodes, thus stealing data and leaking privacy, the present invention first analyzes the sensitivity of the task's own data. For any task T i , assuming that the amount of sensitive data and the total amount of data involved are and Then the data sensitivity characteristics of this task can be expressed as:

[0048]

[0049] Among them, α is the weight adjustment factor, 0≤α≤1, the first added item is the proportion of sensitive data in the task itself, the second added item is the proportion of sensitive data in the task among all tasks, and K is the total number of tasks generated in the cycle of the task.

[0050] In addition, the present invention will also analyze the resource preemptibility of the source node that generates the task, mainly based on the frequency of task processing requests by the source node in the current time window. i The corresponding source node is In the kth observation period, the source node The number of task processing requests initiated is Then the resource preemptibility of the source node that generates the task can be expressed as:

[0051]

[0052] Then, considering the task sensitivity and the resource preemptibility of the source node, the weight adjustment factor β is introduced to establish the task evaluation authenticity expression as follows:

[0053]

[0054] Finally, according to the authenticity of the task Classify the tasks and if the authenticity is lower than the set threshold Then the task is judged to have an attack risk and removed from the task set If the authenticity is higher than the threshold, Keep it in the collection and enter the subsequent uninstallation decision stage.

[0055] ② Carry out trust evaluation on vehicles and edge nodes involved in task calculation. To ensure the comprehensiveness and objectivity of trust evaluation, the present invention uses the node's historical behavior as the basis for trust evaluation, and adopts two methods to obtain it directly or indirectly: one is based on the node's previous task processing success rate, and the other is based on the feedback evaluation of collaborative interaction nodes. Therefore, node N i The trust feature in the kth observation period can be expressed as

[0056] For features Assume that in the kth observation period, node N i The number of successful and failed tasks processed is and Then we have:

[0057]

[0058] For features Provided by other nodes that have had direct interactions with the evaluated node. Since this method relies on the reliability of the provider, it is necessary to ensure that the node providing the trust evidence is trustworthy. Suppose that in the kth observation period, there are M nodes providing information about node N. i Feedback The jth node has a trust level of R j , then:

[0059]

[0060] Finally, considering the node task completion quality and interactive collaboration reputation, and introducing the weight adjustment factor ω, the basic trust of the node in the kth observation period can be expressed as:

[0061]

[0062] The present invention takes into account that node trust is a property that decays dynamically over time, and behaviors that are more recent can better reflect the current behavior trend of the node. Therefore, based on Ebbinghaus's theory of human brain memory forgetting, a decay function of the node trust memory retention with respect to time t is constructed. The specific description is as follows:

[0063]

[0064] in, The retention rate of trust information that decays over time is expressed as a percentage. It starts from the current time. t represents the difference between the generation of the corresponding trust relationship and the current time, in minutes. In experiments, η1=1.84 and η2=1.25 are often taken.

[0065] Each cycle is divided into M trust observation periods. Taking time decay into account, the trust evaluation of nodes in M ​​observation periods is integrated to obtain the trust value of node N. i The comprehensive trust in the kth observation period is:

[0066]

[0067] Finally, classification is performed based on node trust. If the node trustworthiness is lower than the trust threshold, Remove it from the node collection Delete; if the node credibility is higher than the threshold It is then allowed to enter the subsequent offloading decision phase and serve as a candidate node for task offloading and calculation.

[0068] ③ Use discrete particle swarm algorithm to iteratively obtain the vehicle-side cloud collaborative unloading decision. After completing the unloading task and collaborative node evaluation and filtering, the real task set is obtained. and a set of trusted nodes The discrete particle swarm algorithm is used for task allocation and offloading decision. The present invention maps the discrete problem of offloading decision to the continuous particle motion space. The specific steps are as follows: first, particle encoding and initialization; then, particle fitness evaluation; then, particle speed and position update; finally, particle validity correction. Repeat the second to fourth steps until convergence or the number of iterations is reached.

[0069] Figure 3 The figure shows a schematic diagram of particle encoding involved in the present invention. In the present invention, each particle represents an allocation scheme, including two parameters: position and speed. Assuming there are n tasks in total, the position parameters are encoded in sequence according to the task encoding, which can be expressed as The position fragment represents the allocation plan for task T7. A valid position code should meet the following conditions: (1) the code length is fixed to n, where n is the number of tasks; (2) each bit can be null or n. i , represents the task processing node, including vehicle, edge and cloud nodes. The particle speed can be expressed as An effective speed code should meet the following conditions: (1) the code length is fixed to n; (2) each bit has a value of 0 or 1. Indicates that the position coding segment corresponding to T7 is adjusted in the next iteration; if The allocation status of task T7 remains unchanged.

[0070] Then, the local priority-random adjustment algorithm is used to initialize particles. Create a candidate set of tasks to be assigned Node Candidate Set Node traversal set The initialization process is as follows: (1) From the set Randomly select a task T i By default, its processing node is to generate T i The source node vehicle v SR , check whether the C1-C4 constraints are met. If it is a valid allocation that satisfies the constraints, T i The corresponding position fragment is encoded as the vehicle number; if it does not meet the requirements, Randomly find a node n that satisfies the constraint k , encode the corresponding position fragment as node n k Number; Finally, if you traverse the entire If no node that meets the requirements is found, it will be encoded as null; (2) Update Ti Process the remaining resource parameters of the node. If it is found that the node has no available resources, remove it from the collection. Delete and set the next round of node traversal collection (3) Randomly generate 0 or 1 as T i The initial speed code is T i from Delete from; (4) Randomly reselect a task and repeat the above process until all tasks are traversed; (5) Sort the tasks in order according to their numbers to obtain a legal task assignment particle.

[0071] Next, the particle fitness is evaluated. The present invention constructs a fitness function based on the goal of minimizing the weighted sum of energy consumption and delay, and designs the fitness evaluation function as follows:

[0072]

[0073] in, It is task T i About Node n j If the processing decision This means that task T i Assign to n j Calculate, the task processing cost is λ is the adjustment weight factor of energy consumption and delay. When the task processing cost is higher, the fitness is smaller, indicating that the particle is inferior. Then, find the individual optimal particle and the global optimal particle to guide the particle to the optimal particle. Assume that the particle swarm Including μ particles, After I rounds of update iterations, the individual optimal particle and the global optimal particle are defined as follows:

[0074] represents the individual optimal particle, for each particle For , it refers to the position value with the highest fitness in this round;

[0075] represents the global optimal particle, which refers to all particles The position value of the particle with the highest fitness in these I rounds.

[0076] Then, update the particle velocity and position. The speed is expressed as For any velocity segment v i , if v i = 0, then keep task T in the next iteration i Location fragment unchanged, otherwise, adjust T iAssume that the particle The position at the kth iteration is represented by The individual optimal particle is The global optimal particle is Then its speed in the k+1th iteration is as follows:

[0077]

[0078] in, represents an XOR operation, r1 and r2 are two random numbers between 0 and 1, c1 and c2 represent acceleration, which are used to adjust the maximum learning step size, usually set to 1.494. The particle speed update mainly considers three factors: the particle's own speed inertia Second, the particle’s own historical experience affects When c1 is 0, it means that its own experience is not taken into account, and the group diversity will be lost at this time; the third is affected by the historical experience of other particles When c2 is 0, it means that the experience of others is not taken into account. At this time, the convergence will be slow due to the lack of information sharing.

[0079] To ensure is a binary vector to represent the 0 and 1 states of the particle velocity. Let Y = {y1, y2, ...y n}, For each task T i Speed ​​fragment The present invention defines its calculation as:

[0080]

[0081] in, is a random number between 0 and 1, which can balance the convergence speed and avoid premature particle maturation and falling into local optimum. It is symmetrical around 0.5 and can be used without being restricted by the domain. Mapped to the 0-1 numerical range, thus achieving The binary value of .

[0082] After obtaining the particle velocity of the next iteration, the present invention uses To update the particle position, for particles Any position fragment Its position at the k+1th iteration is as follows:

[0083]

[0084] In the above formula, if We adjust the task T in the particle i, so that it is adjusted to The position fragment with the highest fitness.

[0085] Figure 4 This is a schematic diagram of particle speed and position update involved in the present invention. The particle encoding and fitness in the figure are randomly set and are only used for example explanation. First, obtain the particle As well as individual optimal particles and global optimal particles, randomly generated A decimal between 0 and 1, updated to get the speed. Particle speed Will follow Different results are presented, which helps to improve particle diversity. Then, according to the updated particle speed, the particle position is adjusted. Assuming that the particles in the figure The fitness of its individual optimal particles and global optimal particles are {0.1, 0.2, 0.1, 0, 0.1, 0.2, 0.1}, {0.1, 0.2, 0.1, 0.3, 0.4, 0.2, 0.1}, {0.1, 0.1, 0.1, 0.3, 0.4, 0.4, 0.1} respectively. For the position segment to be adjusted, the method of the present invention selects the encoding scheme with the highest fitness and updates it to obtain Figure 5 The position coding shown shows that the particle fitness after adjustment is {0.1, 0.2, 0.1, 0.3, 0.4, 0.4, 0.1}, which indicates that the present invention can retain high fitness fragments into the next round, thereby guiding the particles to continuously approach the optimal value.

[0086] Finally, the particle validity is corrected. Since the particle fragments are all obtained based on the previous legal particle adjustment, the constraints C1, C2, and C4 are all satisfied. The main purpose is to solve the node resource allocation conflict problem of C3. The specific correction process is as follows: (1) Traverse the set Each node n in k If n is found k If the required resources for task calculation exceed its available resources, the valid task subset with the maximum fitness is retained from the original task allocation, and the others are allocated to n k The position code corresponding to the remaining tasks is adjusted to null. (2) Repeat the above process until the task resource allocation of all nodes is no longer conflicting. (3) Reuse the remaining resources and traverse the unassigned tasks in turn to see if a node with remaining resources and meeting the constraints is found, then assign the task to the node and repeat the above process until no further allocation is possible.

[0087] After obtaining the offloading decision, the task-related data is uploaded to the corresponding computing node, the task offloading service is executed, and the result is returned to the vehicle node after the calculation is completed. Through vehicle-side cloud collaborative computing, the task processing efficiency is improved.

[0088] The beneficial effects of the method of the present invention are introduced below, and experimental comparison data of the method of the present invention and a comparative method are provided.

[0089] The experimental scenario of the present invention is set up as follows: an experimental scenario consisting of 200 intelligent vehicles, 50 edge servers and 1 cloud server is built, which are distributed within 25 kilometers centered on the longitude and latitude (116.41667, 39.91667). In order to simulate the communication and cooperation process between vehicles and between vehicles and servers more realistically, the present invention uses the data set "parking places of Beijing" to build a compact distribution scenario of vehicle nodes and tasks, and uses the data set "T-drive" to build a compact distribution scenario of edge servers. At the same time, a uniform distribution scenario is manually created to avoid the conclusion being affected by the characteristics of the scenario. The present invention involves other experimental parameters as follows: (1) 10% of the 200 vehicles contain malicious nodes, which will initiate resource grabbing behavior from time to time. The available computing resources of each vehicle are 1-2GHz, the task processing rate is 1GHz / s, the transmission power is 0.2W, and the computing power is 0.5W; (2) 10% of the 50 edge servers contain malicious nodes, which will initiate task abandonment behavior from time to time. The available computing resources of each edge server are 3-5GHz, the edge server task processing rate is 2.6GHz / s, and the cloud server task processing rate is 4.2GHz / s; (3) Ordinary vehicle nodes randomly generate 1-3 tasks in each cycle, and resource grabbing nodes generate 3-6 tasks, of which 5%-10% are randomly generated false tasks, the CPU cycle required for each task is 0.5-1GHz, the data volume is 0.5-1.5MB, the maximum tolerable delay is 0.5-2s, and the task benefit is 2-5 currency units. (4) In the trust evaluation of tasks and nodes, the node's most recent 5 cycle information is always retained for comprehensive trust evaluation. (5) When making unloading decisions, the number of particles per iteration is set to 10 and the maximum number of iterations is set to 50.

[0090] In order to objectively evaluate the effectiveness of the proposed method, this paper selects three comparative methods. The first is the comparative method RG, which aims to minimize the cost of task offloading and makes task offloading decisions based on random and greedy ideas. Under the condition of satisfying the task allocation constraints, the node with the minimum task execution cost among the ε candidate nodes is selected as the task calculation node each time. The second is the comparative method TENG, which first identifies and filters the node trust, and then performs task offloading through non-cooperative games. The third is the comparative method MOPS, which iteratively obtains the optimal offloading decision between multiple tasks and multiple nodes based on the traditional particle swarm optimization method.

[0091] Figure 5 and Figure 6The figure is a schematic diagram comparing the node and task recognition rates of the method of the present invention and the comparative method. Combining the trust evaluation values ​​of the tasks in the compact and uniform distribution scenarios, it can be seen that compared with the comparative method TENG, the method of the present invention improves the edge node recognition accuracy by 1.69%-22.59%, improves the vehicle node recognition rate in the uniform distribution scenario by 8.49%-21.28%, and improves the task recognition rate by 26.69%-29.24%.

[0092] Figure 7 The figure is a schematic diagram comparing the energy consumption of the method of the present invention and the comparative method. Since the MOPS algorithm of comparative method 4 is more complex, its energy consumption is significantly higher than that of other methods. In the first few rounds, the energy consumption of the method of the present invention is not much different from that of other methods, but as the network runs, the energy consumption of the method of the present invention gradually decreases. This is because the method of the present invention uses a trust evaluation mechanism to evaluate node trust, filters out malicious nodes and false tasks, and avoids additional energy consumption and waste. Compared with comparative method RG, the method of the present invention reduces energy consumption by 1.30%-3.84%, and compared with comparative method MOPS, the method of the present invention reduces energy consumption by about 15%.

[0093] Figure 8 The figure is a schematic diagram of the delay comparison between the method of the present invention and the comparative method. The method of the present invention fully considers the delay characteristic constraints of the task when making the unloading decision, and introduces random quantities, optimization functions, etc. in the process of particle swarm optimization to solve the unloading strategy, so as to avoid particles falling into the local optimum. The experimental results also prove that the final decision-making performance of the method of the present invention is relatively optimized. In general, the method of the present invention has obvious delay advantages, and compared with the three comparative methods, it reduces the average task delay by 6 to 21ms.

[0094] Fig. 9 The figure is a comparison diagram of the task benefits of the method of the present invention and the comparative method. The benefit here refers to the profit generated by the successfully executed task. Since the method of the present invention filters out false tasks and malicious nodes based on the task and node trust evaluation, the task execution success rate is higher, so the generated task benefits are also significantly higher than other methods. Overall, in scenario 1 where tasks and nodes are compactly distributed, the method of the present invention improves the benefits by 3.68%-21.07%; in scenario 2 where tasks and nodes are evenly distributed, the method of the present invention improves the task benefits by 10.03%-19.12%.

[0095] Fig.10 The task completion rate comparison diagram of the method of the present invention and the comparative method is shown in FIG. The task completion rate refers to the ratio of successfully executed tasks to the total number of tasks. Since the method of the present invention randomly generates 5% to 10% malicious tasks, the task success rate of the present invention is still very high at 89%, which increases the task success rate by about 5%.

Claims

1. A trusted task offloading method based on vehicle-side cloud autonomous collaboration, characterized in that: By analyzing data sensitivity, resource preemption, historical unloading success rate, and collaborative feedback evaluation, false tasks and malicious nodes can be accurately identified, and discrete particle swarm algorithm can be used to iteratively obtain vehicle-side cloud collaborative unloading decisions, effectively improving unloading efficiency and security. The main steps include: Step 1: Evaluate the authenticity of pending tasks generated in the current cycle, based on the data sensitivity of the task itself and the frequency of unloading requests from the source node where the task is generated. If the authenticity of the task is lower than the set threshold, Then the task is judged to have an attack risk and removed from the task set The remaining real tasks enter the offloading decision stage; Step 2: Conduct trust evaluation on vehicles and edge nodes involved in task calculations. The first step is based on the node's historical task processing success rate, and the second step is based on the feedback evaluation of collaborative interaction nodes. The node's comprehensive credibility is obtained by integrating the node's behavioral trends and time decay characteristics over multiple observation periods. If the node's credibility is lower than the set threshold, Then the node is determined to be a malicious node and removed from the node set The remaining trusted nodes enter the uninstallation decision phase; Step 3: Taking real tasks and trusted nodes as the objects, improve and redesign the discrete particle swarm algorithm's population initialization, fitness evaluation, speed and position update, and validity correction steps, iteratively obtain the best task offloading decision, execute the offloading service, and reversely update the node trust evaluation model based on the task completion quality, node collaboration status, etc.

2. The trusted task offloading method based on vehicle-side cloud autonomous collaboration according to claim 1 is characterized in that: The specific operation of step 1 is: for any task T i , based on its task sensitivity and source node resource preemptibility, a trust evaluation function is constructed, which is specifically expressed as: Among them, α and β are weight adjustment factors, and It is task T i The amount of sensitive data and total data involved, K is the total number of tasks in the task generation cycle; is the number of task processing requests initiated by the task source node in the kth observation period; then, based on the task trust Perform classification processing. If the trust level is lower than the threshold Then remove it from the task set If the trust level is higher than the threshold, Keep it in the collection and make subsequent uninstall decisions.

3. The trusted task offloading method based on vehicle-side cloud autonomous collaboration according to claim 1 is characterized in that: The specific operation of step 2 is: for any node N i , analyze its historical task success rate and interactive collaboration reputation, introduce the weight adjustment factor ω, and construct its trust evaluation function in the kth observation period, which is specifically expressed as: in, and is node N i The number of successful and failed tasks processed during the k-th observation period, is the information provided by the jth interactor about node N i Feedback evaluation, R j is the reliability of the jth interactor, M is the total number of interacting nodes that provide feedback evaluation; each cycle is divided into M trust observation periods, and considering time decay, the trust evaluation of nodes in M ​​observation periods is integrated to obtain the node N i The comprehensive trust in the kth observation period is: in, in, is the decay function of the node trust memory retention with respect to time t, η1 = 1.84, η2 = 1.25, t represents the difference between the corresponding trust relationship and the current time, in minutes; then, classification is performed based on the node trust. If the node credibility is lower than the trust threshold Remove it from the node collection Delete; if the node credibility is higher than the threshold It is then allowed to enter the subsequent offloading decision phase and serve as a candidate node for task offloading and calculation.

4. The trusted task offloading method based on vehicle-side cloud autonomous collaboration according to claim 1 is characterized in that: The specific operation of step three is as follows: First, determine the particle position and velocity code, fix the code length to n, n is the number of tasks, and the position code value is null or n. i , represents the task processing node, and each bit of the particle speed encoding takes a value of 0 or 1. 0 means adjusting the corresponding position encoding segment in the next iteration, and 1 means not adjusting the position segment in the next iteration. Then, the local priority-random adjustment algorithm is used to generate the initial particle population. Next, the fitness function of particle evaluation is constructed based on the weighted and optimized objective of minimizing energy consumption and delay, which is specifically expressed as: in, It is task T i About Node n j If the processing decision This means that task T i Assign to n j , at this time the task processing cost is λ is the weight adjustment factor of energy consumption and delay, and the individual optimal particle and the global optimal particle are found based on the evaluated particle fitness; then, random quantities, mapping functions, etc. are introduced to update the velocity encoding of the particle in the next iteration, and the position encoding in the next iteration is updated based on the velocity vector and combined with the individual and optimal particles; finally, the particle validity is corrected according to the node resource constraints and task delay constraints; the iteration is repeated until a converged offloading decision is obtained or the maximum number of iterations is reached.

Images