A vehicle-mounted edge computing task offloading method and system

Abstract

The application discloses a kind of vehicle-mounted edge computing task unloading method and system, using fuzzy comprehensive evaluation strategy to calculate the trust level of service vehicle in real situation, ensure the trust of unloading service provided by vehicle.In meeting the trust, the minimum expected delay service vehicle selection strategy is used, so as to select suitable vehicle under the constraints of vehicle trust, computing resource availability and distance accessibility, minimize task unloading delay, so as to ensure the safe driving of customer vehicle.The application uses fuzzy comprehensive strategy, and comprehensively judges the trust of vehicle from multiple attributes of vehicle;Effectively prevent malicious vehicle from participating in service attack customer vehicle, provide false results, so as to ensure the safety of customer vehicle.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of information security, and particularly relates to a vehicle-mounted edge computing task offloading method and system. BACKGROUND

[0002] With the rapid development of artificial intelligence and automatic driving technology, more and more computing-intensive and delay-sensitive applications appear in Internet of Vehicles (IoVs). However, improving the computing capacity of each vehicle will significantly increase the cost of the vehicle, and offloading the computing task of the vehicle to the remote cloud will bring great delay. In addition, in the Internet of Vehicles scenario, the computing resources and the number of roadside basic units (RSUs) are limited and cannot meet the needs of all vehicles. Therefore, as a supplement to the RSU in the Internet of Vehicles, many researchers propose to use the nearby VEC nodes to provide computing and storage services for customer vehicles.

[0003] However, not all vehicles are fully trusted. If a malicious vehicle participates in the service, it may return false results to the customer vehicle due to selfishness or cause task timeout, thereby causing traffic accidents. In addition, there are few studies on trust evaluation of vehicles in VEC networks at present, and little attention is paid to offloading delay.

[0004] Therefore, the joint optimization problem of trust evaluation and trusted and task delay of vehicles in VEC networks has gradually become an important problem to ensure the safety of vehicles with tasks and minimize offloading delay. SUMMARY

[0005] The technical problem to be solved by the application is to provide a vehicle-mounted edge computing task offloading method and system to solve the technical problem of low trustworthiness of the offloading service provided by the vehicle by using a fuzzy comprehensive evaluation (FCE) strategy to calculate the trust level of the service vehicle in a real situation.

[0006] The application adopts the following technical solutions:

[0007] A vehicle-mounted edge computing task offloading method, comprising the following steps:

[0008] S1, a roadside basic unit receiving a task offloading request and adjacent roadside basic units jointly form a roadside basic unit service vehicle management system RSMS;

[0009] S2, the roadside basic unit service vehicle management system RSMS obtained in step S1 extracts attribute information of vehicles in the optional vehicle set V from cloud and vehicle log information respectively;

[0010] S3. Using the vehicle attribute information obtained in step S2, calculate the trust evaluation index of each vehicle in the optional vehicle set V: the average distance AD ​​between each vehicle and the customer vehicle during the delay tolerance time, the available computing resources ratio between the customer vehicle and each vehicle ACR, the vehicle's transaction reputation value TR, and the vehicle's security score SR.

[0011] S4. Normalize the four indicators obtained in step S3 and input the normalized indicators into the trust assessment system.

[0012] S5. Based on the trust evaluation index obtained after normalization in step S4, the membership function is used to calculate the membership degree of each index value of each vehicle to the four evaluation results, and a trust evaluation matrix for each vehicle is formed.

[0013] S6. Select the weights of the four indicators based on the customer's vehicle preference, and determine whether to update the weights using the entropy method.

[0014] S7. Using the trust evaluation matrix obtained in step S5 and the index weights obtained in step S6, calculate the credibility and confidence score of each vehicle according to the weighted average fuzzy comprehensive evaluation model M(+,·).

[0015] S8. Based on the vehicle speed and location information in the attribute information obtained in step S2 and the customer vehicle task unloading request obtained in step S1, predict the unloading delay when the vehicle is successfully unloaded.

[0016] S9. Based on the vehicle attribute information obtained in step S2, the customer vehicle task unloading request obtained in step S1, and the vehicle credibility and confidence score obtained in step S7, filter out vehicles that meet the following conditions: computing resources > customer vehicle task requirements, the distance between the vehicle and the customer vehicle during the service period is less than 300m, and the credibility is above the passing grade, and add them to the candidate vehicle set.

[0017] S10. Based on the confidence score obtained in step S7 and the unloading delay obtained in step S8, calculate the expected delay of each vehicle in the candidate vehicle set obtained in step S9, select the vehicle with the smallest expected delay as the final service vehicle, and send the ID of the final service vehicle to the customer vehicle. The customer vehicle and the service vehicle establish secure communication and perform service unloading.

[0018] Specifically, in step S1, when the j-th roadside basic unit (RSU) j Received vehicle V within its coverage area c Task Unloading Request c ={Task size ,T max} and default weights for trust assessment Subsequently, the Roadside Basic Unit Service Vehicle Management System (RSMS) is represented as follows:

[0019] RSMS = {RSU j-1 , RSU j-1 , RSU j+1}.

[0020] Specifically, in step S2, the attribute information of the vehicle includes the historical transaction record HR i extracted from the remote end, the ID i , the security score SR i , the GPS position information (x i , y i ) extracted from the vehicle log information, the driving speed v i , and the CPU working frequency.

[0021] Specifically, in step S3, let i = 0 represent the ID value of the vehicle, i = i + 1, when the value i of the vehicle ID is greater than the number V of vehicles in the optional vehicle set V, step S10 is executed instead, otherwise the trust evaluation index of vehicle i is calculated; the trust evaluation index calculation method is as follows:

[0022] The average distance AD i between vehicle i and the client vehicle within the delay tolerance time is:

[0023]

[0024] Where T max is the delay tolerance time of the client vehicle, is the distance between vehicle i and the client vehicle at time t;

[0025] The available computing resource ratio ACR i between the client vehicle and vehicle i is:

[0026]

[0027] Where, is the available computing resource of vehicle i, is the computing resource requested by the client vehicle;

[0028] The transaction reputation value TR i of vehicle i is:

[0029]

[0030] Where, is the reward or punishment received by vehicle i for providing offloading service at time t; is the value weight of the transaction; is the service score of vehicle i at time t; T now is the current time; T(αt) is a time decay function.

[0031] Specifically, in step S5, the matrix composed of the single-factor evaluation sets of the four evaluation indexes of vehicle i is taken as the trust evaluation matrix

[0032]

[0033] wherein, is the membership degree of the jth evaluation index of vehicle i to the first evaluation result.

[0034] Specifically, in step S6, the updated weight w j is:

[0035]

[0036] wherein, is the initial default weight corresponding to the jth index, and ε′ j is the information entropy of the jth trust evaluation index, and m is the number of trust evaluation indexes.

[0037] Specifically, in step S7, the trust evaluation matrix of the ith vehicle is taken as the trust evaluation set TL The fuzzy vector on the trust evaluation set TL is calculated through fuzzy comprehensive transformation The evaluation result is calculated by using the weighted average model M(+,·) If The jth element in the trust evaluation set TL is the trust level of vehicle i.

[0038] Specifically, in step S8, the unloading delay of vehicle i as a service vehicle is is:

[0039]

[0040] wherein, is the time delay required by the client vehicle to send a task to vehicle i, is the time required by vehicle i to process the task, is the time required by vehicle i to return the task processing result to the client vehicle.

[0041] Specifically, in step S10, the unloading problem is defined as follows

[0042]

[0043] subject to

[0044]

[0045]

[0046] wherein, is the offloading latency, R max is the delay tolerance time of the client vehicle, is the computing resource requested by the client vehicle, is the available computing resource of the vehicle i, is the distance between the vehicle i and the client vehicle at time t, is the communication range of the vehicle, indicates that the trustworthiness of the vehicle i cannot be Bad.

[0047] In a second aspect, an embodiment of the present application provides a vehicle-mounted edge computing task offloading system, comprising:

[0048] an information module, configured to jointly form a roadside unit service vehicle management system RSMS with adjacent roadside units receiving the task offloading request, extract attribute information of vehicles in a selectable vehicle set V from the cloud and vehicle log information respectively using the roadside unit service vehicle management system RSMS, and calculate trust evaluation indexes of each vehicle in the selectable vehicle set V using the vehicle attribute information, wherein the trust evaluation indexes include an average distance AD between each vehicle and the client vehicle within a delay tolerance time, an available computing resource ratio ACR between the client vehicle and each vehicle, a transaction reputation value TR of the vehicle, and a security score SR of the vehicle;

[0049] a normalization module, configured to perform normalization processing on the average distance AD between each vehicle and the client vehicle within the delay tolerance time, the available computing resource ratio ACR between the client vehicle and each vehicle, the transaction reputation value TR of the vehicle, and the security score SR of the vehicle, and input the normalized indexes into a trust evaluation system;

[0050] a calculation module, configured to calculate, based on the trust evaluation indexes obtained after normalization by the normalization module, a membership degree of each index value of each vehicle to four evaluation results using a membership function, form a trust evaluation matrix of each vehicle, select weights of the four indexes according to the will of the client vehicle, judge whether to update the weights according to an entropy method, and calculate a trustworthiness and a confidence score of each vehicle according to a weighted average fuzzy comprehensive evaluation model M(+,·) using the trust evaluation matrix and the index weights;

[0051] a screening module, configured to predict an offloading latency when the vehicle offloads successfully according to speed and position information of the vehicle in the attribute information obtained by the information module and a task offloading request of the client vehicle, and select, according to the attribute information of the vehicle obtained by the information module, the task offloading request of the client vehicle, and the trustworthiness and the confidence score of the vehicle obtained by the calculation module, a vehicle that meets a condition that a computing resource > a task demand resource of the client vehicle, a distance to the client vehicle during service < 300 m, and the trustworthiness is above a passing grade, and add the vehicle to a candidate vehicle set.

[0052] The unloading module is used for calculating the expected delay of each vehicle in the candidate vehicle set according to the confidence score obtained by the screening module and the unloading delay, selecting the vehicle with the minimum expected delay as the final service vehicle, and sending the ID of the final service vehicle to the client vehicle.

[0053] Compared with the prior art, the present application has at least the following beneficial effects:

[0054] The vehicle-mounted edge computing task unloading method of the present application includes two parts of trust evaluation and service vehicle selection; the trustworthiness of the service vehicle is ensured to protect the unloading safety of the client vehicle, and the service quality of the unloading service is improved by considering the unloading delay of the vehicle; firstly, the trustworthiness of the vehicle is evaluated by using the fuzzy comprehensive strategy, which only needs to traverse the vehicle set and the trust evaluation matrix under the condition that the evaluation index number is determined. Therefore, the time complexity is O(n), and in addition, the minimum expected delay service vehicle selection strategy proposed by the present application only needs to traverse the vehicle set twice, once to calculate the expected delay of each vehicle, and once to select the vehicle with the minimum expected delay. Therefore, the present application has high time efficiency; secondly, in the actual Internet of Vehicles application scenario, the good or bad of a vehicle is not absolute, so the traditional trust evaluation method is difficult to achieve good results; the present application uses the fuzzy comprehensive strategy to divide the trust level of the vehicle into four levels of Excellent, Good, accepted and Bad, and evaluates the degree of good or bad of the vehicle, which is more in line with the actual situation. At the same time, the simulation results also prove that the trust evaluation result of the present scheme has high accuracy and can detect about 90% of malicious vehicles, thereby improving the accuracy of malicious vehicle detection; further, the traditional trust evaluation work only considers the trustworthiness of the node and does not consider the unloading delay and the possibility of unloading failure of the vehicle, so the selected vehicle may have a very high unloading delay or even cause the task to fail due to timeout. The minimum expected delay service vehicle selection strategy proposed by the present application jointly optimizes the trustworthiness and delay of the vehicle; the simulation results show that, compared with the traditional work, the present scheme can significantly reduce the task unloading delay; finally, when the RSU j receives the task request information of the vehicle V c , forms a service vehicle management system RSMS together with the two adjacent RSUs, can expand the selectable range of the service vehicle, fully utilizes the computing resources of all vehicles within the communication range, and avoids the large unloading delay or unloading failure caused by the small number of vehicles on the highway. The distance between vehicles on the highway is large, and the number of vehicles within the coverage range of one RSU is small, and the number of vehicles meeting the unloading condition is even smaller. In addition, the communication range of the vehicle and the coverage range of the RSU are both 300m, so the present application can expand the selectable range of the service vehicle, fully utilize the computing resources of all vehicles within the communication range, and avoid the large unloading delay or unloading failure caused by the small number of vehicles on the highway. cwhether in the center or at the edge of the RSU coverage range cannot fully utilize the computing resources of all optional vehicles within the communication range of the vehicle, and the success rate of offloading is improved.

[0055] Further, the historical transaction record HR of the vehicle is obtained i , ID i , security score SR i The requirement for latency is not high, and the update is not frequent. Storing and updating these attributes in the cloud can save storage and computing resources of the RSU and the vehicle. The GPS position information (x i , y i ), driving speed v i , CPU working frequency f i and other attributes have high timeliness and time effectiveness, and therefore need to be extracted from the vehicle log information. When the value i of the vehicle ID is greater than the number of optional vehicles, it indicates that the vehicle set V has been traversed, and therefore the next step of processing needs to be performed.

[0056] Further, the four indicators are formulated according to the actual situation of the Internet of Vehicles, which can more comprehensively consider the credibility of the vehicle and avoid unnecessary calculation consumption and evaluation interference caused by too many indicator items. The indicator AD considers the characteristics of fast movement and short communication distance of the vehicle. The service vehicle may exceed the communication range of the client vehicle V c during the service period, thereby causing service timeout or failure. Within the effective communication distance, the farther the distance between vehicles, the higher the communication delay, and the more unstable the communication connection, and the greater the possibility of service timeout. Therefore, the farther the distance between vehicles, the lower the service reliability. The range of available computing resources ACR limits the computing resources of the service vehicle to be greater than the resources required for task offloading. The greater the ACR, the smaller the offloading delay of the service vehicle, and the lower the possibility of service timeout due to accident. The vehicle transaction reputation value TR is an important indicator for evaluating whether the vehicle is selfish. If there is malicious behavior or low service quality in the service vehicle transaction record, the possibility of providing malicious service is higher. In addition, the use of value weight to measure the importance of each transaction can make the reputation value calculation more accurate and effective, and the use of time decay function to update the reputation value is more in line with the timeliness of the influence of transaction record on reputation value (recent transactions are more indicative of the current good or bad of the vehicle than distant transactions). The security score SR is an important indicator for evaluating the security situation of the vehicle. For example, a vehicle with an old system version has more vulnerabilities and is more likely to harm the safety of the client vehicle due to security attacks.

[0057] Further, the four evaluation indicators of the vehicle are normalized to facilitate the calculation of the subsequent indicator membership degrees.

[0058] Further, the single-factor evaluation set of the four evaluation indexes is combined to form a trust evaluation matrix In order to facilitate subsequent calculation of the trust level of the vehicle i, the model of the trust evaluation calculation is a weighted average model M(+,·), which is more convenient and efficient in calculation by using matrix multiplication.

[0059] Further, the entropy method is used to update the weight, which can reduce the influence of subjectivity of the initial weight, so that the calculation result is more objective and accurate.

[0060] Further, the weighted average model M(+,·) is used to calculate the evaluation result Compared with other models, all evaluation indexes can be considered according to the weight size, instead of only considering the indexes with greater influence, so that the role of each index can be fully played, and the accuracy and reliability of the evaluation result are improved.

[0061] Further, the unloading time delay of the vehicle i as a service vehicle is calculated In order to select a suitable vehicle as a service vehicle, the unloading service quality is further improved under the premise of ensuring the trustworthiness of the vehicle.

[0062] Further, the definition of the unloading problem is the joint optimization of the unloading time delay and the trustworthiness, which ensures the safety of the customer vehicle, reduces the service time delay, and improves the unloading efficiency.

[0063] It can be understood that the beneficial effects of the above-mentioned second aspect can be referred to the related description in the above-mentioned first aspect, which will not be repeated here.

[0064] In summary, the fuzzy comprehensive strategy is adopted to comprehensively judge the trustworthiness of the vehicle from multiple attributes of the vehicle, which effectively prevents malicious vehicles from attacking customer vehicles and providing false results, thereby ensuring the safety of the customer vehicle.

[0065] The technical solutions of the present application will be further described in detail below with reference to the drawings and examples. DESCRIPTION OF DRAWINGS

[0066] Figure 1 A highway Internet of Vehicles task offloading scene diagram;

[0067] Figure 2 A safe and efficient vehicle-mounted edge computing offloading system model diagram;

[0068] Figure 3 A safe and efficient vehicle-mounted edge computing offloading method flowchart;

[0069] Figure 4 membership function diagram of four trust evaluation indexes;

[0070] Figure 5 result comparison diagram of A-TVS, Max-TVS and Min-TVS from the aspect of unloading time delay;

[0071] Figure 6 result comparison diagram of RUSR, RUAR and RUVWTD from the aspect of transaction reputation value update;

[0072] Figure 7 result comparison diagram of Non-trusted scheme and the trust evaluation scheme of the present application from the aspect of malicious vehicle detection rate. DETAILED DESCRIPTION

[0073] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0074] In the description of the present application, it should be understood that the terms “include” and “contain” indicate the existence of described features, whole, steps, operations, elements and / or components, but do not exclude the existence or addition of one or more other features, whole, steps, operations, elements, components and / or sets thereof.

[0075] It should also be understood that the terms used in the present application specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, unless otherwise clear from the context, the singular forms “a”, “an” and “the” are intended to include plural forms.

[0076] It should be further understood that the term “and / or” used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations, for example, A and / or B can represent three cases of A alone, A and B together, and B alone. In addition, the character “ / ” herein generally represents an “or” relationship between the front and rear associated objects.

[0077] It should be understood that, although the terms first, second, third, etc. can be employed in describing the preset ranges, etc. in the embodiments of the present application, the preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from each other. For example, the first preset range can also be referred to as the second preset range, and similarly, the second preset range can also be referred to as the first preset range, without departing from the scope of the embodiments of the present application.

[0078] Depending on the context, the word "if" as used herein can be interpreted to mean "when" or "while" or "in response to determining" or "in response to detecting." Similarly, the phrase "if it is determined" or "if (a stated condition or event) is detected" can be interpreted to mean "when it is determined" or "in response to determining" or "when (the stated condition or event) is detected" or "in response to detecting (the stated condition or event)".

[0079] Various structural diagrams according to the disclosed embodiments of the present application are shown in the accompanying drawings. These diagrams are not drawn to scale, in which certain details are exaggerated for the purpose of clarity, and certain details can be omitted. The shapes of various regions, layers, and the relative size and positional relationship therebetween shown in the drawings are only exemplary, and in actuality, they can be deviated due to manufacturing tolerances or technical limitations, and a person skilled in the art can additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0080] The present application provides a vehicle-mounted edge computing task offloading method, which adopts a fuzzy comprehensive evaluation (FCE) strategy to calculate the trust level of a service vehicle in a real situation, to ensure the trustworthiness of the offloading service provided by the vehicle. While meeting the trustworthiness, a weighted expected delay service vehicle selection strategy is used to select a suitable vehicle under the constraints of vehicle trustworthiness, computing resource availability, and distance accessibility, to minimize the task offloading delay, thereby ensuring the safe driving of the client vehicle.

[0081] Please refer to Figure 1 and Figure 2 , a malicious vehicle participating in the service returns a false result to the client vehicle, resulting in a traffic accident. In this scenario, the vehicle V c wants to perform a task offloading service, and the task information is represented as a set of task size and maximum tolerated time delay, i.e. Task c ={Task size , T max}; the RSU on the roadside is represented as RSU={RSU1, RSU2,..., RSU 10}; the set of vehicles within the coverage of the RSMS system is represented as V={1, 2,..., V}, and the model diagram of the task offloading is shown in Figure 2 .

[0082] Please see Figure 3 The present invention provides a method for offloading vehicle-mounted edge computing tasks, comprising the following steps:

[0083] S1. When a Roadside Basic Unit (RSU) receives a task unloading request from a vehicle within its coverage area, the RSU and its neighboring RSUs together form a Roadside Basic Unit Service Vehicle Management System (RSMS).

[0084] When RSU j Received vehicle V within its coverage area c Task Unloading Request c ={Task size T max} and default weights for trust assessment Subsequently, the Roadside Basic Unit Service Vehicle Management System (RSMS) is represented as follows:

[0085] RSMS = {RSU} j-1 RSU j-1 RSU j+1}

[0086] The available vehicles within its coverage area are represented as a set, as follows:

[0087] V = {1, 2, ..., V}.

[0088] S2. Using the Roadside Basic Unit Service Vehicle Management System (RSMS) obtained in step S1, extract the attribute information of vehicles in the optional vehicle set V from the cloud and vehicle log information respectively.

[0089] The attribute information of vehicle i includes historical transaction records (HR) retrieved from a remote location. i ID i Safety rating SR i GPS location information (x) extracted from vehicle log information i ,y i ), driving speed v i CPU operating frequency f i .

[0090] S3. Let i = 0, i = i + 1. When the number of vehicles i is less than or equal to the number of available vehicles V, use the vehicle attribute information obtained in step S2 to calculate four credibility evaluation indicators: the average distance AD ​​between each vehicle and the customer vehicle within the delay tolerance time, the ratio of available computing resources between the customer vehicle and each vehicle ACR, the vehicle's transaction reputation value TR, and the vehicle's security score SR. Otherwise, proceed to step S10.

[0091] Credible evaluation metrics for vehicle i

[0092] S301, AD i is the distance between vehicle i and vehicle V c average distance between vehicle i and vehicle V c at time t The formula of AD

[0093]

[0094] where (x C , y C ), (x i , y i ) are the initial coordinates of vehicle V c and i, v C and v i are the average speed of vehicle V c and vehicle i, further, AD i is calculated as follows:

[0095]

[0096] S302, ACR i is the ratio of available CPU resources of vehicle i and the CPU resources requested by V c The formula of available CPU resources of vehicle i is as follows:

[0097]

[0098] where T max is the delay tolerance time; is the computing resource used by local tasks; M i can be approximated as a function of CPU frequency f i , i.e. and are estimation parameters;

[0099] The calculation method of CPU resources required by the tasks offloaded by vehicle V c is as follows:

[0100]

[0101] where Task size represents the size of the task to be offloaded; p c represents the computing density of the task, i.e. the number of CPU instructions required per bit of the task;

[0102] Further, the calculation method of ACR i is as follows:

[0103]

[0104] S303, Vehicle i's transaction reputation value TR i This is a method of judging its quality by using its historical transaction records. The calculation method is as follows:

[0105]

[0106] in, At time t, vehicle i provides unloading services and is rewarded or penalized. This indicates the value weight of the transaction; T represents the service score of vehicle i at time t; now This represents the current time; T(Δt) is the time decay function.

[0107] Due to the transaction The contribution to reputation score has a certain time sensitivity (the quality of service a year ago does not necessarily indicate whether a vehicle is trustworthy now). Therefore, this invention uses a time decay function T(Δt) = e {-Δt·ξ} To update the weight of each transaction. When T(Δt=T) now When -τ) < 0.01, the transaction has almost no impact on the calculation of reputation value, but these records will waste storage resources. Therefore, this system will periodically clean up transaction records before time τ.

[0108] S4. Normalize the four indicators obtained in step S3 and input the normalized indicators into the trust assessment system.

[0109] For the value of the j-th evaluation index of vehicle i The normalization formula is as follows:

[0110]

[0111] in, Represents the j-th evaluation indicator The lower bound; Represents the j-th evaluation indicator The upper bound; This represents the set of four evaluation index values ​​for vehicle i.

[0112] S5. Calculate the membership degree of each indicator value of each vehicle to the four evaluation results (Excellent, Good, accepted, Bad) using membership functions, and form a trust evaluation matrix.

[0113] Please see Figure 4 The membership function is:

[0114]

[0115] wherein, represents the degree to which x belongs to .

[0116] There are four evaluation factors, and through repeated experiments, the membership functions of the four evaluation indexes for each evaluation result are defined as triangular or trapezoidal formats, which are represented as:

[0117]

[0118]

[0119] For example, in the Figure 3 , for the membership function of TR, when the horizontal axis x = 0.45

[0120] Further, for the jth element of the ith vehicle, the single-factor evaluation set of the four evaluation results in the evaluation result set TL is represented as:

[0121]

[0122] wherein, represents the membership degree of the jth element of the ith vehicle to the first element (Excellent) in the evaluation result set TL.

[0123] The matrix composed of the single-factor evaluation sets of the four evaluation indexes of the ith vehicle is called a trust evaluation matrix:

[0124]

[0125] S6, according to the customer's vehicle intention, select the weight of the four indexes, and determine whether to update the weight according to the entropy method;

[0126] According to the default weight given by the customer and the customer's weight update intention to determine the weight value participating in the trust evaluation; if the customer is willing to update by entropy method, the method is as follows:

[0127] S601, according to each normalized evaluation index , the information entropy ε j of the jth evaluation factor is calculated.

[0128]

[0129] wherein, n represents the number of remaining vehicles in the set V.

[0130] S602, the entropy weight ε′ j is calculated.

[0131]

[0132] wherein m represents the number of evaluation indexes, m=4 in this example;

[0133] S603, in order to reduce the subjective influence of expert experience weight , the weight is modified.

[0134] The weight is modified as follows:

[0135]

[0136] S7, according to the weighted average fuzzy comprehensive evaluation model M(+,·) to calculate the credibility and confidence score of each vehicle;

[0137] According to the fuzzy comprehensive evaluation matrix of the ith vehicle The fuzzy vector on the trust evaluation set TL is calculated by fuzzy comprehensive transformation as follows:

[0138]

[0139] wherein, represents the comprehensive evaluation operator.

[0140] The weighted average model M(+,·) is used to calculate the evaluation result, so is the matrix multiplication operation, and the corresponding

[0141] Therefore, if , the jth element in TL is the credibility level of vehicle i.

[0142] S8, predict the unloading time delay when the vehicle is unloaded successfully;

[0143] Calculate the unloading time delay of vehicle i as a service vehicle composed of task transmission time, task processing time and task result return time:

[0144]

[0145] wherein, M i is the computer processing speed, unit: million instructions per second (MIPS); and The calculation of

[0146] wherein, the data transmission rate C i is calculated as follows:

[0147] ​

[0148] wherein, as the channel gain.

[0149] S9, screening vehicles meeting the following three conditions and adding to the candidate vehicle set: the computing resource > the customer vehicle task demand resource, the vehicle with the distance less than 300m during the service period, and the vehicle with the credibility above the passing mark;

[0150] deleting the vehicle in V not meeting any of the following conditions:

[0151]

[0152] that is, deleting the vehicle with the credibility of Bad. i the vehicle with the computing resource insufficient to complete the offloading task computation within the tolerance time T max

[0153]

[0154] wherein, denotes the communication range of the vehicle, in this case The condition indicates deleting the vehicle that may exceed the communication range of the customer vehicle during the service time.

[0155]

[0156] that is, deleting the vehicle with the credibility of Bad.

[0157] S10, calculating the expected delay of each vehicle in the candidate vehicle set, selecting the vehicle with the minimum expected delay as the final service vehicle, and sending the ID of the vehicle to the customer vehicle.

[0158] The offloading problem is defined as follows

[0159]

[0160] subject to

[0161]

[0162]

[0163] wherein, denotes that the credibility of the vehicle i cannot be Bad.

[0164] In another embodiment of the present application, a vehicle-mounted edge computing task offloading system is provided, which can be used to implement the vehicle-mounted edge computing task offloading method described above. Specifically, the vehicle-mounted edge computing task offloading system comprises an information module, a normalization module, a calculation module, a screening module and an offloading module.​

[0165] wherein the information module is configured to form the roadside unit receiving the task offloading request and adjacent roadside units into a roadside unit service vehicle management system RSMS, extract attribute information of vehicles in a selectable vehicle set V from the cloud and vehicle log information respectively using the roadside unit service vehicle management system RSMS, and calculate trust evaluation indexes of each vehicle in the selectable vehicle set V using the vehicle attribute information, wherein the trust evaluation indexes include an average distance AD of each vehicle from the client vehicle within a delay tolerance time, an available computing resource ratio ACR between the client vehicle and each vehicle, a transaction reputation value TR of the vehicle, and a security score SR of the vehicle.

[0166] a normalization module configured to normalize the average distance AD of each vehicle from the client vehicle within the delay tolerance time, the available computing resource ratio ACR between the client vehicle and each vehicle, the transaction reputation value TR of the vehicle, and the security score SR of the vehicle, and input the normalized indexes into a trust evaluation system;

[0167] a calculation module configured to calculate membership degrees of each index value of each vehicle to four evaluation results using a membership function based on the trust evaluation indexes obtained after normalization by the normalization module, form a trust evaluation matrix of each vehicle, select weights of the four indexes according to the will of the client vehicle, determine whether to update the weights according to an entropy method, and calculate a credibility and a confidence score of each vehicle according to a fuzzy comprehensive evaluation model M(+,·) of weighted average using the trust evaluation matrix and the index weights;

[0168] a screening module configured to predict an offloading time delay of each vehicle when offloading is successful according to speed and position information of the vehicles in the attribute information obtained by the information module and the task offloading request of the client vehicle, select vehicles that meet the conditions of computing resources > task demand resources of the client vehicle, a distance from the client vehicle < 300 m during service, and a credibility of passing the exam according to the attribute information of the vehicles obtained by the information module, the task offloading request of the client vehicle, and the credibility and confidence scores of the vehicles obtained by the calculation module, and add the vehicles to a candidate vehicle set;

[0169] an offloading module configured to calculate an expected time delay of each vehicle in the candidate vehicle set according to the confidence score and the offloading time delay obtained by the screening module, select a vehicle with the minimum expected time delay as a final service vehicle, send an ID of the final service vehicle to the client vehicle, and establish a secure communication between the client vehicle and the service vehicle to perform service offloading.

[0170] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0171] Referring to Figure 1 In actual application, a 7.5*2000m 2 double-lane road segment represents a highway scene. There are 10 RSUs with computing and storage resources evenly distributed on the highway segment, with a spacing of 200m between the RSUs, and the communication coverage of each RSU is 300m. There are 10-400 vehicles distributed on the highway, of which the proportion of malicious vehicles is 10%-40%. Due to selfishness, malicious vehicles often disturb or confuse client vehicles by participating in offloading services and initiating false result attacks, thereby causing collisions of client vehicles.

[0172] For example, after completing the offloading task, the malicious vehicle intentionally returns a false result and suggests the client vehicle to change the route to create a traffic-free road and travel quickly. The coordinates (x, y) of the vehicle are uniformly distributed within 7.5*2000m 2 , and the speed v is normally distributed within the interval [60, 90] km / h, N(75, 8.66) km / h. Considering the rapid iteration of the computing power of intelligent vehicles, the vehicles are equipped with different types of CPUs, such as ARM Cortex A57, Qualcomm Snapdragon SA8195P, etc. The CPU working frequency is extended to 700MHz-2700MHz, a total of 21 working frequency levels, and the vehicle communication adopts IEEE802.11p protocol with a communication bandwidth of 20MHz.

[0173] The size of each offloading task is 20MB, and 30 CPU instructions are required for each bit of data; the default weights of the trust evaluation indicators <AD, ACR, TR, SR> are <0.05, 0.05, 0.5, 0.4>, and the weight values can be set by the user as required or updated by using the entropy weight method.

[0174] Referring to Figure 4 and Figure 5According to the generated data, the membership function is defined, and the method result compares the unloading delay of the arbitrary trusted vehicle selection scheme (A-TVS), the maximum trustworthiness vehicle selection scheme (Max-TVS), and the minimum expected delay vehicle selection scheme (Min-TVS) proposed in the application.

[0175] Figure 5 The unloading delays of A-TVS, MAX-TVS and Min-EDVS vehicle selection strategy of the application are compared. It can be found that the performance of A-TVS is poor, and its delay even reaches the delay tolerance time T max . Similarly, the performance of MAX-TVS is also not high, and the delay reaches 150 ms. However, the performance of Min-EDVS of the application is very high, and the maximum delay is only 120 ms. With the increase of the number of vehicles, the performance of A-TVS is very unstable, and the delay fluctuates greatly. The delay of the MAX-TVS scheme decreases slowly and finally stabilizes at about 130 ms, while the delay of the Min-EDVS decreases fastest and finally stabilizes at about 100 ms. It shows that the scheme proposed in the application is far superior to the existing scheme in reducing delay, and the stability is significantly increased.

[0176] Please refer to Figure 6 and Figure 7 , the reputation update scheme based on service success rate (RUSR), the reputation update scheme based on average score (RUAR), and the reputation update scheme based on value weight and time decay function (RUVWTD) proposed in the application are compared from the change of the reputation value of the malicious service vehicle; the trust evaluation scheme proposed in the application is compared with the Non-trusted scheme from the detection rate of malicious vehicles.

[0177] Figure 6 The RUSR scheme, the RUAR scheme and the RUVWTD reputation update scheme proposed in the application are compared, and the process of updating the vehicle reputation value with the increase of the number of malicious services is shown. It is assumed that vehicle i trades once a day and gets high score for 50 times in a row However, due to some reasons such as ownership change, security attack, etc., vehicle i starts to provide five times of malicious services in a row and returns error results to the client vehicles. In order to observe the decline of the reputation value of the vehicle, it is assumed that the vehicle can still provide services when the reputation is low. It can be observed that the reputation value calculated by the scheme proposed in the application is more in line with the actual situation, and the distinction between good and bad services is higher. In addition, when the vehicle behaves maliciously, the reputation value of the scheme decreases faster, which can effectively prevent the malicious vehicle from providing malicious services again.

[0178] Figure 7The malicious vehicle detection rate of the trust evaluation scheme of the application and the Non-trusted scheme is compared. Obviously, the Non-trusted scheme only detects 40% to 50% of the actual proportion, and its detection rate fluctuates with the change of the number of vehicles. When the number of vehicles is small, its detection rate is low. However, the application can detect 90% of the malicious vehicles, which is very close to the true value, and the detection rate is very stable regardless of the number of vehicles. Therefore, the application is more accurate and stable in detecting malicious vehicles, and can effectively avoid malicious vehicles from participating in the service.

[0179] In summary, the vehicle-mounted edge computing task offloading method and system of the application, based on the fuzzy comprehensive evaluation (FCE) strategy, comprehensively considers the influence of the distance between vehicles on the vehicle communication connection and the communication transmission rate, the influence of the available computing resources of the vehicle on the task processing speed, the reputation value of the vehicle reflected by the vehicle transaction record, and the influence of the safety performance of the vehicle on the safety condition of the offloading service to calculate the credibility of the vehicle providing service. On this basis, the joint optimization of safety and efficiency of offloading is considered, and a service vehicle selection algorithm with minimum expected delay is designed. In addition, the concept of profit based on game theory is adopted to reward the vehicles with good service and punish the vehicles with malicious behavior during service, so as to encourage the good-willed vehicles with rich computing resources to provide service.

[0180] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems, or computer program products. Accordingly, the application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage media, etc.) having computer usable program code embodied therein.

[0181] The application is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0182] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 The functions of a flow or multiple flows and / or a block or multiple blocks in conjunction with the disclosed methods can be implemented on a computer or other programmable data processing apparatus. Figure 1

[0183] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions that are executed on the computer or other programmable apparatus provide steps for implementing the functions specified in the flow Figure 1 The functions of a flow or multiple flows and / or a block or multiple blocks in conjunction with the disclosed methods can be implemented on a computer or other programmable data processing apparatus. Figure 1

[0184] The above merely illustrates the technical idea of the present application, and cannot be used to limit the protection scope of the present application. Any modification made according to the technical idea of the present application, on the basis of the technical scheme, falls within the protection scope of the claims of the present application.​​

Claims

1. A vehicle-mounted edge computing task offloading method, characterized in that, The method comprises the following steps: S1, the roadside unit receiving the task offloading request and the adjacent roadside unit jointly constitute a roadside unit service vehicle management system RSMS; S2, using the roadside unit service vehicle management system RSMS obtained in step S1, attribute information of vehicles in the optional vehicle set V is extracted from cloud and vehicle log information respectively; S3, using the vehicle attribute information obtained in step S2, trust evaluation indexes of each vehicle in the optional vehicle set V are calculated: average distance AD of each vehicle and the client vehicle within the delay tolerance time, available computing resource ratio ACR between the client vehicle and each vehicle, transaction reputation value TR of the vehicle, and security score SR of the vehicle; S4, the four indexes obtained in step S3 are normalized, and the normalized indexes are input into a trust evaluation system; S5, based on the trust evaluation indexes obtained after normalization in step S4, the membership degree of each index value of each vehicle to the four evaluation results is calculated by using a membership function, and a trust evaluation matrix of each vehicle is formed; S6, the weights of the four indexes are selected according to the will of the client vehicle, and it is judged whether the weights are updated according to an entropy method; S7, using the trust evaluation matrix obtained in step S5 and the index weights obtained in step S6, the credibility and confidence score of each vehicle are calculated according to a weighted average fuzzy comprehensive evaluation model M(+,·); S8, according to the speed and position information of the vehicle in the attribute information obtained in step S2 and the task offloading request of the client vehicle in step S1, the offloading time delay when the vehicle offloading is successful is predicted; S9, according to the attribute information of the vehicle obtained in step S2, the task offloading request of the client vehicle obtained in step S1 and the credibility and confidence score of the vehicle obtained in step S7, vehicles meeting the conditions of computing resource> client vehicle task demand resource, distance from the client vehicle during service less than 300m and credibility above pass are selected and added to the selected vehicle set; S10, according to the confidence score obtained in step S7 and the offloading time delay obtained in step S8, the expected time delay of each vehicle in the selected vehicle set obtained in step S9 is calculated, the vehicle with the minimum expected time delay is selected as the final service vehicle, and the ID of the final service vehicle is sent to the client vehicle, the client vehicle and the service vehicle establish secure communication, and service offloading is performed. 2.The in-vehicle edge computing task offloading method according to claim 1, characterized in that, In step S1, when the j-th roadside basic unit (RSU) j Received vehicle V within its coverage area c Task Unloading Request c ={Task size T max } and default weights for trust assessment Subsequently, the Roadside Basic Unit Service Vehicle Management System (RSMS) is represented as follows: RSMS = {RSU j-1 , RSU j-1 , RSU j+1}. 3.The in-vehicle edge computing task offloading method of claim 1, wherein, In step S2, the attribute information of the vehicle includes the historical transaction record HR extracted from the remote end i , ID i , security score SR i , GPS position information (x i , y i ) extracted from the vehicle log information, driving speed v i , CPU working frequency.

4. The in-vehicle edge computing task offloading method according to claim 1, characterized in that, In step S3, i=0 represents the ID value of the vehicle, i=i+1, when the value i of the vehicle ID is greater than the number V of vehicles in the optional vehicle set V, step S10 is executed, otherwise the trust evaluation indexes of the vehicle i are calculated; The trust evaluation index calculation method is as follows: Average distance AD of vehicle i from the client vehicle within the delay tolerance time i is: wherein T max is the delay tolerance time of the customer vehicle, is the distance between vehicle i and the customer vehicle at time t. Available computing resources between the client vehicle and the vehicle i, ACR i is: wherein, is the available computing resource for vehicle i, is the computing resource requested by the customer vehicle; Transaction reputation value TR of vehicle i i is: wherein, is the reward or penalty for vehicle i to provide offloading service at time t; is the value weight for the transaction; is the service score of vehicle i for the transaction at time t; now is the current time; T(Δt) is a time decay function. 5.The in-vehicle edge computing task offloading method of claim 1, wherein, In step S5, the matrix of the single-factor evaluation sets of the 4 evaluation indexes of the vehicle i is taken as the trust evaluation matrix wherein, is the membership of the jth evaluation index of the vehicle i to the first evaluation result.

6. The in-vehicle edge computing task offloading method according to claim 1, characterized in that, In step S6, the updated weight w j is: wherein, is the initial default weight corresponding to the jth index, ε′ j is the information entropy of the jth trust evaluation index, and m is the number of indexes of the trust evaluation.

7. The in-vehicle edge computing task offloading method according to claim 1, characterized in that, In step S7, the trust evaluation matrix of the i-th vehicle is calculated according to the following formula The fuzzy vector on the trust evaluation set TL is calculated by fuzzy comprehensive transformation The evaluation result is calculated by using the weighted average model M(+·) If The j-th element in the trust evaluation set TL is the trust level of the i-th vehicle. 8.The in-vehicle edge computing task offloading method of claim 1, wherein, Step S8: Unloading delay of vehicle i as service vehicle is: wherein, T i is the time delay required by the client vehicle to send the task to vehicle i, T i is the time required by vehicle i to process the task, T i is the time required by vehicle i to return the result of the task processing to the client vehicle. 9.The in-vehicle edge computing task offloading method of claim 1, wherein, In step S10, the offloading problem is defined as follows subject to C1 : C2： C3： wherein, is the offloading latency, T max is the delay tolerance time of the client vehicle, is the computing resource requested by the client vehicle, is the available computing resource of vehicle i, is the distance between vehicle i and the client vehicle at time t, is the communication range of the vehicle, indicates that the trustworthiness of vehicle i cannot be Bad.

10. A vehicle-mounted edge computing task offloading system, characterized in that, It comprises: The information module is used for jointly forming a roadside unit service vehicle management system RSMS with adjacent roadside units after receiving a task offloading request of the roadside unit, extracting attribute information of vehicles in a selectable vehicle set V from the cloud and vehicle log information respectively using the roadside unit service vehicle management system RSMS, and calculating trust evaluation indexes of each vehicle in the selectable vehicle set V by using the vehicle attribute information, wherein the trust evaluation indexes include an average distance AD of each vehicle from the client vehicle within a delay tolerance time, an available computing resource ratio ACR between the client vehicle and each vehicle, a transaction reputation value TR of the vehicle, and a security score SR of the vehicle. The normalization module is used for normalizing the average distance AD of each vehicle from the client vehicle within the delay tolerance time, the available computing resource ratio ACR between the client vehicle and each vehicle, the transaction reputation value TR of the vehicle, and the security score SR of the vehicle, and inputting the normalized indexes into a trust evaluation system. The calculation module is used for calculating membership degrees of each index value of each vehicle to four evaluation results by using a membership function based on the trust evaluation indexes obtained after normalization by the normalization module, forming a trust evaluation matrix of each vehicle, selecting weights of the four indexes according to the will of the client vehicle, judging whether to update the weights according to an entropy method, and calculating a credibility and a confidence score of each vehicle according to a weighted average fuzzy comprehensive evaluation model M(+,·) by using the trust evaluation matrix and the index weights. The screening module is used for predicting an offloading time delay of each vehicle when offloading is successful according to speed and position information of the vehicles in the attribute information obtained by the information module and a task offloading request of the client vehicle, screening vehicles that meet the conditions of a computing resource > a task demand resource of the client vehicle, a distance from the client vehicle < 300 m during service, and a credibility of passing the test, and adding the vehicles into a candidate vehicle set according to the attribute information of the vehicles, the task offloading request of the client vehicle, and the credibility and confidence scores of the vehicles obtained by the calculation module. The offloading module is used for calculating an expected time delay of each vehicle in the candidate vehicle set according to the confidence score and the offloading time delay obtained by the screening module, selecting a vehicle with the minimum expected time delay as a final service vehicle, sending an ID of the final service vehicle to the client vehicle, establishing a secure communication between the client vehicle and the service vehicle, and performing service offloading.

Images