Active and trace back based trust management method for crowd sensing

Abstract

The present application relates to the field of big data, and relates to an active and trace back based trust management method for crowd sensing. The method comprises: acquiring task submission data of all participants, and performing standardization processing on data in each dimension; sending an unmanned aerial vehicle to an area with participant reliability uncertainty to collect data and record the data as real data; calculating the sum of difference values of data in each dimension to determine the trustworthiness of the participants; calculating the ratio of the difference value of each participant to a real value, defining a cut-off point, and determining a trust value trend of the data submitted by the participant; setting a trust value update formula; comparing traced historical data with the real data, retroactively comparing historical behaviors of the participant, and analyzing and determining the trustworthiness of the participant; calculating the amount of increase or decrease in reputation value within each timestamp; and calculating the total amount of increase or decrease in the reputation value of the participant, and determining a trust trend of the participant. The method of the present application can be used for identifying malicious participants in a system, and improving the quality of crowd sensing data.

Description

A trust management method for crowd-sensing based on active and retrospective

[0001] This application claims priority to a Chinese patent application filed with the Patent Office of China on December 13, 2023, with application number 202311715072X and application name “A method for trust management based on active and retrospective crowd intelligence perception”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] The present application belongs to the field of big data technology, and in particular relates to a crowd-sensing trust management method based on active and backtracking. Background Art

[0003] With the advancement of wireless communication and microprocessor technologies, mobile devices have increasingly powerful computing, storage, and communication capabilities. Mobile crowdsensing is becoming a widely applicable and affordable method for acquiring data. Mobile crowdsensing involves forming an interactive, participatory perception network using existing mobile devices and distributing perception tasks to individuals or groups within the network.

[0004] However, a key challenge currently lies in ensuring the authenticity of the acquired perception data. In real-world scenarios, malicious, fraudulent, and dishonest actors often provide erroneous information, potentially leading to significant deviations from the true value of the collected data, thus impairing the quality of crowdsensing services.

[0005] There are two main problems with trust detection:

[0006] (1) It is difficult for the platform to obtain the actual results of the task. Such tasks are usually closely related to time and location. When the time or location changes, it is difficult to restore the original state to obtain the actual task data. For example, if you want to obtain the traffic volume at a certain time in a certain place, it is difficult for the platform to know the actual traffic volume afterwards.

[0007] (2) The cost of obtaining real results is too high. If a comparison is performed on each participant, the system will actually collect all the real data. There is no need to obtain the data submitted by the participants, and the cost is also very high.

[0008] Patent CN113642978A discloses a crowd evacuation method and system based on a crowd intelligence trust management mechanism. This method uses the perception data acquired by preset personnel within a sensing unit as the actual task data to calculate each person's trustworthiness. If a malicious person intentionally submits falsified data, this can lead to inaccurate trustworthiness judgments for other personnel. Furthermore, the new credibility value calculated for an individual based on previous credibility scores and current information is not objective and accurate, and does not take time into account. Summary of the Invention

[0009] In view of the above deficiencies in the prior art, the purpose of the invention is to provide a crowd-sensing trust management method based on active and retrospective analysis, which can be used to identify malicious participants in the system and improve the quality of crowd-sensing data.

[0010] This application proposes a crowd-sensing trust management method based on active and retrospective intelligence, including:

[0011] S1, obtain the task submission data D of all participants all , the edge nodes normalize each dimension of data so that its value range is between [0,1];

[0012] S2: For areas with uncertain participant reliability, send a drone to the area with uncertain participant reliability to collect data, and record the collected data as real data D real ; and the real data D real Each dimension of data is standardized; according to the data D submitted by the participants all Real data collected by drones real The sum of the difference values ​​of each dimension of the standardized data is used to judge the credibility of the participants;

[0013] S3, calculate the ratio of each participant’s difference value to the true value δ i , define the dividing point θ; set δ i Compare it with θ and determine the trust value trend of the data submitted by the participant based on the comparison result;

[0014] S4, according to the trust value trends of the participants’ reliable and dishonest behaviors in the actual scenario, set the trust value update formula;

[0015] S5: Submit the participant's previously unsubmitted data to the upper level. By tracing back the data from various historical periods and obtaining the real data at a specific time and place, we can compare the historical data with the real data, retrospectively compare the participant's past behavior, and analyze and judge the participant's credibility.

[0016] S6: Find the cutoff value θ from the database for calculating the reputation value of other participants in the same period when collecting data in the area at each timestamp. tj , and then calculate the increase or decrease in the reputation value at each timestamp; where θ tj is the cutoff value between credible and uncredible for the difference ratio of all other participants who participated in the same task at time stamp tj;

[0017] S7, calculating the total increase or decrease in the participant's reputation value based on the time decay factor and the increase or decrease in the reputation value within each timestamp, and judging the trust trend of the participant based on the increase or decrease in the total reputation value.

[0018] Furthermore, in S1, the following formula is used to standardize each dimension of data of each participant:

[0019] in, is the normalized value of the j-dimension data of the i-th participant; D i,j The j-th dimension data submitted by the i-th participant; u j is the maximum value of the j-th dimension data; l j is the minimum value of the j-th dimension data; i is the i-th participant; j is the j-th dimension data.

[0020] Furthermore, in S2, the data D submitted by the participant is calculated using the following formula: all Real data collected by drones real The sum of the difference values ​​of each dimension of the standardized data,

[0021] Among them, Δ i is the sum of the difference values; The normalized value of the j-th dimension data submitted by the i-th participant; is the normalized value of the j-th dimension data of the true value; k is the current dimension, and m is the total dimension;

[0022] Δ i The larger the value, the lower the credibility of the data submitted by the participant, and the greater the possibility that the participant is a malicious participant; Δ i The smaller the value, the closer the data submitted by the participant is to the actual value collected by the drone, and the higher the reliability.

[0023] Further, in said S3,

[0024] When δ i When ≤θ, the difference is judged to be within the acceptable range, the participant is trustworthy, and its trust value should be increased;

[0025] When δ i When θ > θ, it is judged that the data provided by the participant is false and the trust value should be reduced.

[0026] Furthermore, in S4, the trust value update formula is:

[0027] in, is the updated trust value; r iis the trust value of the i-th participant, θ is the median selected after sorting the difference values ​​of all participants, δ i is the ratio of the difference value to the true value; the function S(x) is the Sigmoid function, and the graph of this function conforms to the changing law of the trust value; α1 and β1 are used to control the rate of increase and convergence boundary of the credibility value; α2 and β2 are used to control the rate of decrease and convergence boundary of the credibility value.

[0028] Furthermore, when the data submitted by a participant in the past is judged to be credible multiple times, the participant is judged to be a credible participant, and the average of the data collected by all credible participants in the same area is taken as the real data D for this task. real .

[0029] Furthermore, in said S5, a time decay function μ(h) is added to reasonably weight the data submitted at different times.

[0030] Among them, h is the current h-th timestamp, and v is the maximum timestamp.

[0031] Furthermore, in S6, the following formula is used to calculate the increase or decrease in the reputation value within each timestamp:

[0032] in, is the increase or decrease in reputation value; i is the i-th participant, tj is the tj-th timestamp, θ tj is the cutoff value between credible and uncredible for the difference ratio of all other participants who participated in the same task at time stamp tj, ρ i,tj For participant P at timestamp tj i The percentage of difference between the submitted data and the determined true data.

[0033] Furthermore, in said S7, the following formula is used to calculate the participant P i The total credit value increase or decrease,

[0034] in, is the total credit value increase or decrease; μ(t) is the time decay function value in the t-th timestamp; tv is the maximum value of the timestamp, The increase or decrease in the reputation value of the i-th participant at the tj-th timestamp;

[0035] Further, in said S7, if It indicates that the participant may be credible and the probability of being credible is high; if This indicates that the probability that the participant is a malicious participant is greater.

[0036] The beneficial effects of this application are as follows:

[0037] (1) Traditional trust detection methods in crowd intelligence perception mostly use the aggregated results of the data submitted by all participants as the real results for comparison, such as the average, median, etc. This application adopts an active trust evaluation method, using drones to fly to the area to be collected to collect real data, and using the collected data as a comparison benchmark for the real data, thereby verifying the credibility of the data submitted by the participants, solving the problem of difficulty in obtaining real data in traditional methods. The comparison benchmark obtained by this application is more objective, true and accurate. In traditional methods, if there are a majority of malicious participants, the aggregated results will deviate from the true value, and it will lose its meaning as reference data. The active trust evaluation method adopted in this application can resist the joint attack of malicious participants, is not affected by the existence of malicious participants, and the judgment results are more accurate.

[0038] (2) Compared with the traditional method of sending many participants to collect task data for each task, the method of this application can only send one or two trusted participants to the area to collect data after obtaining enough trusted participants, which greatly saves costs and resources.

[0039] (3) Compared with the traditional method of conducting trust tests on a group of participants, the retrospective trust evaluation method proposed in this application solves the problem that the traditional method is difficult to quickly obtain the credibility of a specific participant. This application recalculates the results of the participant's previous tasks and assigns a higher weight to the newly participated tasks, making the trust judgment more efficient, fast and accurate. This application can judge the credibility of a participant very efficiently and quickly. This is very useful when the system wants to quickly know whether a specific participant is trustworthy.

[0040] (4) Compared with the traditional method that does not make full use of the trusted participants identified in the system, resulting in low verification efficiency, this application adopts the method of moving the trusted participants with shortcomings in long-term cooperation to the area to be detected, and using the data collected by the trusted participants as the real data for comparison. On the one hand, it saves system resources, and on the other hand, it can flexibly expand the detection range.

[0041] (5) The trust detection inference model and algorithm proposed in this application are consistent with real-world scenarios: a participant's reputation tends to grow slowly after some reliable behavior, but should drop sharply after some dishonest behavior. By continuously providing reliable data, a participant can gain a high reputation, while if they continue to provide false data, they will receive a very low reputation.

[0042] (6) This application improves the recognition rate of malicious participants in the crowd-sensing system. Through two trust detection methods, active and retrospective, the true results of the task can be obtained. A series of trust derivation and evolution formulas are proposed, which can quickly reduce the trust value of malicious participants to 0. In addition, the system will not be affected by the joint attack of malicious participants, thereby improving the recognition rate of malicious participants.

[0043] (7) This application reduces resource consumption in the crowd-sensing system. When the system obtains enough trusted participants in the later stage, the trusted participants move into the area to collect data, reducing the use of drones. At the same time, the system can distribute tasks to a small number of participants, greatly reducing the system's resource consumption.

[0044] (8) This application improves the accuracy of task results in the crowd-sensing system: by distributing tasks to participants with high trust values, the task submission results can be made closer to the true value, thereby improving the accuracy of the task results. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are only for the purpose of illustrating specific embodiments and are not to be considered as limiting the present application. Throughout the drawings, the same reference numerals represent the same components. Obviously, the drawings described below are only some of the embodiments described in the present application. Those skilled in the art can also obtain other drawings based on these drawings.

[0046] FIG1 is a flow chart of a method for trust management based on proactive and retrospective crowdsensing according to an embodiment of the present application;

[0047] FIG2 is a schematic diagram of a collective intelligence perception system framework model according to an embodiment of the present application;

[0048] FIG3 is a schematic diagram of a perception task release process according to an embodiment of the present application;

[0049] FIG4 is a schematic diagram of the active data collection process of a drone according to an embodiment of the present application;

[0050] FIG5 is a schematic diagram of a trusted participant moving and collecting data according to an embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions of the present application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without making creative work should fall within the scope of protection of this application.

[0052] Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts disclosed in this application.

[0053] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application. In addition, the terms "first", "second" and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance. The terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the internal parts of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0054] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of methods and systems consistent with certain aspects of the present application, as detailed in the appended claims.

[0055] The following is an explanation of the technical terms involved in the active and retrospective crowd-sensing trust management method of this application:

[0056] Mobile Crowd Sensing (MCs) refers to a new data acquisition model that combines crowdsourcing and the sensing capabilities of mobile devices. Specifically, mobile crowd sensing involves a large number of ordinary users using their smart mobile devices to collect perception data and upload it to a server. Service providers then record and process this data, ultimately completing the perception task and using the collected data to provide users with daily services.

[0057] The proposed active and retrospective crowdsensing trust management method uses both active and retrospective trust detection methods to effectively identify the trustworthiness of participants in crowdsensing tasks. Figure 2 is a schematic diagram of the crowdsensing system framework model; Figure 3 is a schematic diagram of the perception task release process.

[0058] (1) Active trust evaluation

[0059] In the proactive trust evaluation method, drones are dispatched to areas where suspicious participants exist to collect relevant data as a baseline value to verify the authenticity of the data submitted by the participants. This allows the trustworthiness of participants to be acquired proactively. Data reported by highly credible participants can also be used as a secondary baseline value to verify the trustworthiness of certain participants. This application addresses the dilemma of previously reported data being unable to be compared with real data by using drones to obtain baseline data. Furthermore, this application proposes an effective reasoning and calculation method for expanding trust relationships. After identifying highly credible participants, the data submitted by the credible participants as they move is used as a secondary baseline value to verify the trustworthiness of other participants. This method allows the trustworthiness of many participants to be acquired. Only a small amount of baseline data needs to be collected by drones to infer the trustworthiness of many participants, effectively reducing the cost of trust identification.

[0060] (2) Retrospective Trust Evaluation

[0061] Although active trust evaluation methods can effectively identify the trustworthiness of participants, in order to make the trust evaluation system proposed in this patent more efficient and targeted, some participants are asked to submit historical perception data that has been obtained but not submitted in the past, and historical data of specific time and place are retrieved from the system as a benchmark value. This allows the historical data submitted by participants in the past to be compared to determine the trustworthiness of the participants. This allows the trustworthiness of some participants whose trustworthiness needs to be determined to be obtained in a targeted manner, which has not been possible in previous studies.

[0062] The following describes in detail the proactive and retrospective crowd-sensing trust management method according to an embodiment of the present application with reference to FIG1 to FIG5 .

[0063] As shown in FIG1 , the active and retrospective crowd-sensing trust management method according to an embodiment of the present application includes the following steps:

[0064] S1, obtain the task submission data D of all participants all , the edge nodes normalize each dimension of data so that its value range is between [0,1].

[0065] Specifically, first obtain the task submission data D of all participants all, and consider that the data range for each dimension may vary significantly. For example, the temperature data for a certain area ranges from 15°C to 25°C, and the daily rainfall ranges from 0mm to 120mm. If the raw data is used directly, the data with larger ranges will be given a larger weight. Malicious actors can maintain their reputation by only providing authentic data for large ranges and false data for other data, thus failing to reflect the reliability of the data.

[0066] Therefore, this step uses edge nodes to standardize each dimension of data so that its value range is between [0, 1].

[0067] Specifically, the following formula was used to standardize each dimension of data for each participant:

[0068] in, is the normalized value of the j-dimension data of the i-th participant; D i,j The j-th dimension data submitted by the i-th participant; u j is the maximum value of the j-th dimension data; l j is the minimum value of the j-th dimension data; i is the i-th participant; j is the j-th dimension data.

[0069] S2: For areas with uncertain participant reliability, send drones to the areas with uncertain participant reliability to collect data, and record the collected data as real data D real , and the real data D real Each dimension of data is standardized.

[0070] Figure 4 is a schematic diagram of the active data collection process of the drone in the embodiment of the present application. The data collected by the drone in the area with uncertain participant reliability is recorded as the real data D real , use formula (1) to calculate the real data D real Each dimension of is also normalized.

[0071] Based on the data submitted by participants all Real data collected by drones real The sum of the difference values ​​of each dimension of the standardized data is used to judge the credibility of the participants.

[0072] Specifically, the data D submitted by the participants is calculated using the following formula: all Real data collected by drones real The sum of the difference values ​​of each dimension of the standardized data,

[0073] Among them, Δ i is the sum of the difference values; The normalized value of the j-th dimension data submitted by the i-th participant; is the normalized value of the j-th dimension data of the true value; k is the current dimension, and m is the total dimension.

[0074] Δ i The larger the value, the lower the credibility of the data submitted by the participant, and the greater the possibility that the participant is a malicious participant; Δ i The smaller the value, the closer the data submitted by the participant is to the actual value collected by the drone, and the higher the reliability.

[0075] Furthermore, for areas where drones cannot be deployed due to weather conditions, geographical environment, or service costs, if certain participants are determined to be completely trustworthy during long-term collaboration, the data collected by these participants can also be used as a secondary benchmark to measure the credibility of other participants in the area. Figure 5 is a schematic diagram of trusted participants moving and collecting data in an embodiment of the present application. In this case, the average of the data collected by all trusted participants in the same area can be used as the actual data for the mission, and the trustworthiness can be assessed using the same formula as above.

[0076] That is, when the data submitted by a participant in the past is judged to be credible many times, the participant is judged to be a credible participant, and the average of the data collected by all credible participants in the same area is taken as the real data D of the task. real .

[0077] S3, calculate the ratio of each participant’s difference value to the true value δ i In order to determine the difference δ i The influence of the size of the trust value is used to define the cutoff point θ. Then δ i Compare it with θ and judge the trust value trend of the data submitted by the participant based on the comparison result.

[0078] When δ i When ≤θ, the difference is judged to be within the acceptable range, the participant is trustworthy, and its trust value should be increased;

[0079] When δ i When θ > θ, it is judged that the data provided by the participant is false and the trust value should be reduced.

[0080] S4, sets the trust value update formula according to the trust value trends of the reliable and dishonest behaviors of the participants in the actual scenario.

[0081] The trust value update formula and update rules should follow the following principles: In real-world scenarios, a participant's reputation tends to increase slowly after some reliable behavior, but should decrease sharply after some dishonest behavior. Furthermore, by consistently providing reliable data, a participant can gain a high reputation, while continuously providing false data will result in a very low reputation.

[0082] The trust value update formula is:

[0083] in, is the updated trust value; r i is the trust value of the i-th participant, θ is the median selected after sorting the difference values ​​of all participants, δ i is the ratio of the difference value to the true value.

[0084] Function S(x) is a sigmoid function, and its graph conforms to the changing pattern of the trust value. The characteristics of function S(x) are as follows: when x > 0, S(x) > 0.5, and the function value increases as x increases. When x approaches ∞, S(x) converges to 1. When x < 0, S(x) < 0.5, and the function value decreases as x decreases. When x approaches -∞, S(x) converges to 0.

[0085] α1 and β1 are used to control the rate of increase and convergence of the reputation value; α2 and β2 are used to control the rate of decrease and convergence of the reputation value.

[0086] S5, submits the data that has not been submitted by the participants in history to the upper level, traces the data of each historical period, and obtains the real data that has been determined at a specific time and place, so as to obtain a comparison between historical data and real data, trace back and compare the previous behavior of the participants, and analyze and judge the credibility of the participants.

[0087] Specifically, some participants, for some reason, may have already collected data but not submitted it. Later, by submitting a request to a trusted sensor device, the previously unsubmitted data can be submitted. By tracing back data from various historical periods and obtaining real data from a specific time and location, a comparison between historical data and real data can be made. By retrospectively comparing a participant's past behavior, the trustworthiness of that participant can be analyzed and determined.

[0088] According to the behavioral characteristics of people in real life and considering the impact of time on trust, new events, that is, the records submitted most recently, should be given higher weights, while records submitted more recently should be given lower weights. Based on this, a time decay function μ(h) is set to reasonably weight the data submitted at different times.

[0089] Among them, h is the current h-th timestamp, and v is the maximum timestamp.

[0090] S6: Find the cutoff value θ from the database for calculating the reputation value of other participants in the same period when collecting data in the area at each timestamp. tj , and then calculate the increase or decrease in the reputation value at each timestamp. tj is the cutoff value between credible and uncredible for the ratio of difference values ​​of all other participants who participated in the same task at timestamp tj.

[0091] The following formula is used to calculate the increase or decrease in the reputation value at each timestamp:

[0092] in, is the increase or decrease in reputation value; i is the i-th participant, tj is the tj-th timestamp, θ tj is the cutoff value between credible and uncredible for the difference ratio of all other participants who participated in the same task at time stamp tj, ρ i,tj For participant P at timestamp tj i The percentage of difference between the submitted data and the determined true data.

[0093] S7, calculate the total increase or decrease in the participant's reputation value based on the time decay factor and the increase or decrease in the reputation value within each timestamp, and judge the trust trend of the participant based on the increase or decrease in the total reputation value.

[0094] After taking the time decay factor into consideration in the above formula (5), the participant P i The total credit value increase or decrease is,

[0095] in, is the total credit value increase or decrease; μ(t) is the time decay function value in the t-th timestamp; tv is the maximum value of the timestamp, is the increase or decrease in the reputation value of the i-th participant at the tj-th timestamp.

[0096] This means that the increase or decrease in a participant's reputation value within the most recent timestamp will be given greater weight, as recently submitted data better reflects the participant's dynamic trust trends. If a participant has demonstrated high integrity in recent data submissions, it can be assumed that they are becoming more trustworthy, so recent data should be given a higher weight.

[0097] like This indicates that the participant may be trustworthy and has a high probability of being trustworthy;

[0098] like This indicates that the probability that the participant is a malicious participant is greater.

[0099] The active and retrospective crowd perception trust management method provided by the embodiment of the present application has high applicability to application systems in various business scenarios and is not limited by specific products and fields. The method of the present application is used to identify malicious participants in the system and improve the quality of crowd perception data. It can solve the problem that malicious participants in crowd perception submit forged and false data, which affects the quality of services provided by the crowd perception platform. In crowd perception, if you want to gain the trust of participants, you need to determine whether the data reported by the participants is true. The best way is to obtain the real data of the participant's collection area through another credible channel, and compare the real data with the data submitted by the participants. If they are consistent, it means that the data submitted by the participant is true, and the trust of the participant should be improved. Otherwise, it means that the data submitted by the participant is false, and its trust needs to be reduced.

[0100] The proactive and retrospective crowd-sensing trust management method provided in the embodiments of the present application has the following beneficial effects:

[0101] (1) Traditional trust detection methods in crowd intelligence perception mostly use the aggregated results of the data submitted by all participants as the real results for comparison, such as the average, median, etc. This application adopts an active trust evaluation method, using drones to fly to the area to be collected to collect real data, and using the collected data as a comparison benchmark for the real data, thereby verifying the credibility of the data submitted by the participants, solving the problem of difficulty in obtaining real data in traditional methods. The comparison benchmark obtained by this application is more objective, true and accurate. In traditional methods, if there are a majority of malicious participants, the aggregated results will deviate from the true value, and it will lose its meaning as reference data. The active trust evaluation method adopted in this application can resist the joint attack of malicious participants, is not affected by the existence of malicious participants, and the judgment results are more accurate.

[0102] (2) Compared with the traditional method of sending many participants to collect task data for each task, the method of this application can only send one or two trusted participants to the area to collect data after obtaining enough trusted participants, which greatly saves costs and resources.

[0103] (3) Compared with the traditional method of conducting trust tests on a group of participants, the retrospective trust evaluation method proposed in this application solves the problem that the traditional method is difficult to quickly obtain the credibility of a specific participant. This application recalculates the results of the participant's previous tasks and assigns a higher weight to the newly participated tasks, making the trust judgment more efficient, fast and accurate. This application can judge the credibility of a participant very efficiently and quickly. This is very useful when the system wants to quickly know whether a specific participant is trustworthy.

[0104] (4) Compared with the traditional method that does not make full use of the trusted participants identified in the system, resulting in low verification efficiency, this application adopts the method of moving the trusted participants with shortcomings in long-term cooperation to the area to be detected, and using the data collected by the trusted participants as the real data for comparison. On the one hand, it saves system resources, and on the other hand, it can flexibly expand the detection range.

[0105] (5) The trust detection inference model and algorithm proposed in this application are consistent with real-world scenarios: a participant's reputation tends to grow slowly after some reliable behavior, but should drop sharply after some dishonest behavior. By continuously providing reliable data, a participant can gain a high reputation, while if they continue to provide false data, they will receive a very low reputation.

[0106] (6) This application improves the recognition rate of malicious participants in the crowd-sensing system. Through two trust detection methods, active and retrospective, the true results of the task can be obtained. A series of trust derivation and evolution formulas are proposed, which can quickly reduce the trust value of malicious participants to 0. In addition, the system will not be affected by the joint attack of malicious participants, thereby improving the recognition rate of malicious participants.

[0107] (7) This application reduces resource consumption in the crowd-sensing system. When the system obtains enough trusted participants in the later stage, the trusted participants move into the area to collect data, reducing the use of drones. At the same time, the system can distribute tasks to a small number of participants, greatly reducing the system's resource consumption.

[0108] (8) This application improves the accuracy of task results in the crowd-sensing system: by distributing tasks to participants with high trust values, the task submission results can be made closer to the true value, thereby improving the accuracy of the task results.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of this application, and are not intended to limit them. Although this application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some of the technical features therein with equivalents; and these modifications or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of this application. Any changes or replacements that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be covered by the scope of protection of this application.

Claims

1. A crowd-sensing trust management method based on active and retrospective intelligence, comprising: S1, obtain the task submission data D of all participants all , the edge nodes standardize each dimension of data so that its value range is between [0,1]; S2: For areas with uncertain participant reliability, send drones to the areas with uncertain participant reliability to collect data, and record the collected data as real data D real ; and for real data D real Each dimension of data is standardized; according to the data submitted by the participants all Real data collected by drones real The sum of the difference values ​​of each dimension of the standardized data is used to judge the credibility of the participants; S3, calculate the ratio of each participant’s difference value to the true value δ i , define the dividing point θ; set δ i Compare it with θ, and determine the trust value trend of the data submitted by the participant based on the comparison result; S4, according to the trust value trends of the reliable and dishonest behaviors of the participants in the actual scenario, set the trust value update formula; S5, submit the data that the participant has not submitted in the past to the upper level, trace the data of each historical period, and obtain the real data that has been determined at a specific time and place, so as to obtain the comparison between the historical data and the real data, trace back and compare the previous behavior of the participant, and analyze and judge the credibility of the participant; S6: Find the cutoff value θ from the database for calculating the reputation value of other participants in the same period when collecting data in the area at each timestamp. tj , and then calculate the increase or decrease of the reputation value at each timestamp; where θ tj is the cutoff value between credible and uncredible for the ratio of difference values ​​of all other participants who participated in the same task at timestamp tj; S7, calculating the total increase or decrease in the participant's reputation value according to the time decay factor and the increase or decrease in the reputation value within each timestamp, and judging the trust trend of the participant according to the total increase or decrease in the reputation value.

2. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In S1, the following formula is used to standardize each dimension of data of each participant: in, is the normalized value of the data of the jth dimension after the i-th participant is normalized; D i,j The j-th dimension data submitted by the i-th participant; u j is the maximum value of the j-th dimension data; l j is the minimum value of the j-th dimension data; i is the i-th participant; j is the j-th dimension data.

3. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In S2, the data D submitted by the participant is calculated using the following formula: all Real data collected by drones real The sum of the difference values ​​of each dimension of the standardized data, Among them, Δ i is the sum of the difference values; The normalized value of the j-th dimension data submitted by the i-th participant; is the normalized value of the j-th dimension data of the true value; k is the current dimension, and m is the total dimension; Δ i The larger the value, the lower the credibility of the data submitted by the participant, and the greater the possibility that the participant is a malicious participant; Δ i The smaller it is, the closer the data submitted by the participant is to the actual value collected by the drone, and the higher the reliability.

4. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In said S3, When δ i When ≤θ, the difference is judged to be within the acceptable range, the participant is trustworthy, and its trust value should be increased; When δ i >θ, it is judged that the data provided by the participant is false, and the trust value should be reduced.

5. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In S4, the trust value update formula is: in, is the updated trust value; r i is the trust value of the ith participant, θ is the median selected after sorting the differences of all participants, δ i is the ratio of the difference value to the true value; the function S(x) is the Sigmoid function, and the graph of this function conforms to the law of change of the trust value; α1 and β1 are used to control the rate of increase of the reputation value and the convergence boundary; α2 and β2 are used to control the rate of decrease of the reputation value and the convergence boundary.

6. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: When the data submitted by a participant in the past is judged to be credible multiple times, the participant is judged to be a credible participant, and the average of the data collected by all credible participants in the same area is taken as the real data D of the task. real .

7. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In S5, a time decay function μ(h) is added to reasonably weight the data submitted at different times. Among them, h is the current h-th timestamp, and v is the maximum timestamp.

8. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In S6, the following formula is used to calculate the increase or decrease of the credit value in each timestamp: in, is the amount of credit increase or decrease; i is the i-th participant, tj is the tj-th timestamp, θ tj is the credible and uncredible cutoff value of the difference ratio of all other participants who participated in the same task at timestamp tj, ρ i,tj For participant P at timestamp tj i The difference between the submitted data and the determined true data.

9. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In S7, the following formula is used to calculate the participant P i The total credit value increase or decrease, in, is the total credit value increase or decrease; μ(t) is the time decay function value in the tth timestamp; tv is the maximum value of the timestamp, It is the increase or decrease of the reputation value of the i-th participant at the tj-th timestamp.

10. The method for trust management based on active and retrospective crowd intelligence perception according to claim 1, wherein: In said S7, if It indicates that the participant may be credible and the probability of being credible is high; if This indicates that the probability that the participant is a malicious participant is greater.

Images