Trust evaluation method based on resource awareness
By constructing a multi-dimensional trust index system and an improved CRITIC-entropy weight fusion model, the trust weight is dynamically adjusted, which solves the problems of accuracy and robustness of trust assessment between IoT devices and improves the trusted interaction and resource scheduling capabilities between devices.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-05
- Publication Date
- 2026-03-27
AI Technical Summary
Existing trust management methods lack accuracy and robustness in trust assessment among IoT devices, cannot adapt to dynamic network changes, do not fully consider device resource status, and are susceptible to false recommendations and attacks, resulting in inaccurate trust assessment results.
A multi-dimensional trust index system is constructed, including direct trust, computational idle rate, recommendation trust, and social relationship degree. An improved CRITIC-entropy weight fusion weighting model is used to generate adaptive trust weights. Trust assessment is carried out by dynamically updating the comprehensive trust value between devices and combining real-time resources and interactive feedback.
It significantly improves trust differentiation capabilities, increases task delegation success rate, and enhances trusted interaction and resource scheduling performance in the social IoT under interference from various types of malicious devices.
Smart Images
Figure CN121750282A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of communication, and relates to a trust evaluation method based on resource perception. BACKGROUND
[0002] With the development of the Internet of Things and the continuous progress of computing technology, intelligent devices therein are equipped with sensing, processing and communication functions, enabling them to provide various applications and services, which are expanding in various fields, including personal, industrial and commercial fields. Due to the rapid increase in the number of Internet of Things devices, the network discovery and navigation capabilities of devices when consuming and using services with each other are beginning to be limited. In order to improve the efficiency of information and service discovery, recent research has introduced social properties into Internet of Things devices, which enables different forms of cooperation between Internet of Things devices in an autonomous manner, and thus the Social Internet of Things (SIoT) paradigm of the next generation of Internet of Things has emerged in this background. In the SIoT, Internet of Things devices can establish social relationships with each other without human intervention and cooperate on this basis.
[0003] Compared with the traditional Internet of Things architecture, the SIoT has advantages in discovering, allocating and sharing network information and resources, for the following reasons: (1) Internet of Things devices have similar social characteristics to humans, which will help them to establish social relationships autonomously; (2) the structure of the SIoT enhances the navigability and scalability of the network, which will improve service discovery and resource acquisition; (3) the theoretical methods of social networks can be integrated into the management of Internet of Things devices, including identity recognition, information sharing, etc.
[0004] The development of SIoT allows trillions of devices to act as autonomous agents for requesting and providing services, ensuring effective device and information discovery in a trust-oriented manner and achieving scalability similar to human networks. This is achieved by building trust through interactions between friends and by leveraging and extending existing SIoT network social network models. However, maintaining trustworthy social relationships, along with security and privacy concerns, may limit the further evolution of the SIoT paradigm. For example, in SIoT scenarios, service requesters (or principals) have a responsibility to assess the trustworthiness of the service providers (or trustees) offering the requested services. However, malicious devices offering fraudulent services or making false recommendations about trustees to obtain a valid set of services can undermine the availability and integrity of SIoT services. While some research has proposed encrypted and unencrypted solutions to address these issues, trust and reputation problems are difficult to handle with these solutions. Therefore, an effective SIoT trust management framework is needed to address misbehaving SIoT devices by restricting their services and selecting only trustworthy and reliable devices before relying on their services.
[0005] Existing trust management methods typically rely on fixed weight allocation schemes, failing to fully consider the information differences between different trust indicators and unable to adaptively adjust weights according to dynamic network changes, resulting in a lack of accuracy and robustness in trust assessment results. Secondly, most studies ignore the actual available resources of devices when assessing trust values, especially variations in computing resources. Social IoT devices generally have limited energy and fluctuating loads; changes in resource status often directly affect task execution success rates. In traditional trust models, even if a device obtains a high trust score, it may be unable to complete the task due to insufficient computing power, leading to delegation failure. Thirdly, existing recommendation trust mechanisms are susceptible to false recommendations, collusion attacks, and unfair scoring behavior, lacking effective identification of the reliability of recommendation sources and struggling to maintain stable trust differentiation capabilities when the proportion of malicious devices is high. Furthermore, existing models have limited utilization of social relationships, failing to effectively integrate the multi-layered social attributes between devices, resulting in trust assessments that cannot comprehensively reflect the true behavioral characteristics of devices in the network. Summary of the Invention
[0006] In view of this, the purpose of this invention is to provide a resource-aware trust assessment method. To achieve the above objectives, the present invention provides the following technical solution: A resource-aware trust assessment method, comprising the following steps: S1. Based on the interaction behavior and resource constraints between devices, construct a system including direct trust. Calculate idle rate Recommendation and trust and social relationships A multidimensional trust indicator system; S2. An improved CRITIC-entropy weight fusion weighting model is adopted to generate adaptive trust weights based on the discreteness, heterogeneity and information content of multidimensional trust indicators, and to calculate the comprehensive trust value between devices based on the dynamic trust weights. S3. In each time slot, update direct trust, remaining device resources, indirect recommendation information and social relationship degree based on newly generated interaction records, and dynamically adjust the comprehensive trust value between devices through a fusion weighting method.
[0007] Furthermore, in step S1, the direct trust index in the multi-dimensional trust index system is based on the direct interaction history between the two devices. At any moment For equipment Direct trust is represented as:
[0008] in, Indicates task Contribution level, collection Indicates equipment To the equipment The set of tasks to be delegated The initial value is , indicating equipment For equipment Complete uncertainty, with continuous interaction, the device For equipment Direct trust is increasing.
[0009] Furthermore, in step S1, the computing idle rate index in the multi-dimensional trust index system is based on the device's maximum computing capacity and current occupied load, reflecting the device's resource capacity available for executing delegated tasks, and is defined as follows:
[0010] In the formula, Indicates device exist Calculate the idle rate at any given time. Indicates equipment The maximum computing resources, Indicates equipment exist Used computing resources at any given time.
[0011] Furthermore, in step S1, the recommendation trust index in the multi-dimensional trust index system is calculated based on the deviation between the recommendation (the direct trust of other devices in this device) and the actual credibility, and is expressed as follows:
[0012] In the formula, It is a weighting factor, if the device For equipment If a rating is provided, then Otherwise, it is 0; Indicates equipment exist Constantly monitor the equipment Direct trust; Represents the set of all devices; It is an exponential penalty factor used to enhance the penalty effect on deviations from public credibility; Indicates equipment exist Social credibility at any given moment refers to the trust capital a node accumulates within the group, calculated based on the accumulated trust feedback a device receives in the network.
[0013] In the formula, Indicates the device at the time of the last iteration. Reliability, This is a penalty decay factor used to adjust the convergence speed of the results based on the number of scoring samples.
[0014] Furthermore, in step S1, the social relationship index in the multidimensional trust index system represents a relatively fluctuating and dynamic social relationship, calculated based on interactions within the system:
[0015] In the formula, Indicates an indicator function, Indicates the indication conditions, i.e., the equipment To the equipment Successful interactive tasks, Indicates a specific time slot, It is a local constant.
[0016] Furthermore, in step S2, the process of generating adaptive weights using the improved CRITIC-entropy weight fusion weighting model is as follows: First, Z-score standardization is performed on each trust indicator, and the standard deviation of the indicator and its correlation with other indicators are calculated. Based on the information content of the indicators, a CRITIC weight vector is constructed:
[0017] in This represents the total number of indicators considered in the multidimensional trust indicator system. This indicates the amount of information in each indicator, reflecting the evaluation capability of each indicator. This represents the standardized Z-Score value of the j-th indicator. This represents the correlation coefficient between the indicators. Represents the standard deviation function; Then, the information entropy is calculated. Information entropy reflects the uncertainty of the index in the sample, and is applied to the index matrix. medium elements The index is obtained by performing Min-Max normalization. Then calculate its entropy value:
[0018] In the formula, Indicates the number of trust samples. Represents the normalization coefficient; Entropy weight normalization yields the weights .
[0019] Furthermore, in step S2, when calculating the comprehensive trust value between devices based on dynamic trust weights, an adaptive fusion model based on minimizing KL divergence is used to find a new and optimal set of weights. The process is as follows: First, an optimization objective function based on KL divergence is established, with optimal weights. and and The objective is to minimize the sum of the weighted KL divergences, which is expressed as:
[0020] in, express relatively Information bias, express relatively Information bias; It is a preference coefficient representing the degree of confidence in the results of the CRITIC method, and the constraints are satisfied. If the weights calculated by CRITIC Larger, then Increase; if the weights calculated by CRITIC Larger, then Decrease; The Lagrange multiplier method is used to solve the objective function of a convex optimization problem, i.e., to establish the Lagrange function. Taking the partial derivative with respect to it and setting it to 0, we get... The unique optimal solution:
[0021] Among them, the solution obtained For a weighted geometric mean, the denominator is a normalization term, and all... The sum is 1; at the same time, if or If any one of them approaches 0, then the final... All approach 0; Ultimately, each device node For target equipment The overall trust value is expressed as a weighted linear combination of various indicators:
[0022] in Indicates equipment For equipment The Trust index value.
[0023] Furthermore, in step S3, the adaptive update process of the trust value is as follows: based on the task execution feedback of the current time slot. Update the direct trust value by adjusting the success decay and failure penalty factors:
[0024] in, , , The decay factor representing the success rate of a task. The decay factor representing a failed task; Indicates the current time. Indicates the time threshold. The decay rate parameter indicates the success rate of the interaction. The parameter representing the decay rate of failed interactions, if and If the interval between them exceeds the time threshold, then the task... It will not be considered in the CoT calculation; Simultaneously, the idle rate is dynamically updated based on changes in real-time load demand and remaining resources, and is dynamically adjusted based on changes in task resource usage.
[0025] In the formula, , These represent the idle computing resources of device j in the current time slot and the idle computing resources in the previous time slot, respectively. This indicates the task load generated in this time slot. This indicates the computing resources released in this time slot; At the network level, the system collects information from neighboring nodes to recalculate node reliability and credibility, thereby dynamically adjusting indirect trust. It also updates social relationship degree hourly based on interaction frequency and stability. The updated indicators are used to construct a new matrix, which is then input into the CRITIC-entropy weight fusion model to redistribute weights and calculate a comprehensive trust value, achieving accurate, dynamic, and adaptive assessment of device trust status.
[0026] The beneficial effects of this invention are as follows: This invention addresses the problems of relying on a single trust indicator, which is insufficient to comprehensively depict the trustworthiness of devices; fixed weights that cannot adapt to dynamically changing network environments; insufficient consideration of the dynamic nature of device resources, leading to erroneous decisions due to resource shortages; and the vulnerability of recommendation information to attacks, affecting the accuracy of assessments. It proposes a resource-aware adaptive trust assessment method that integrates multiple dimensions of indicators. By simultaneously introducing multiple dimensions such as computing resource status, recommendation reliability, credibility, and social relationship, it comprehensively considers device characteristics from different perspectives, providing a more comprehensive and accurate reflection of device trustworthiness. An improved weight determination method based on indicator correlation—the entropy-weight fusion algorithm—dynamically adjusts the weights of each trust indicator according to the real-time network status, endowing the trust assessment process with dynamic adaptive capabilities, ensuring that trust values are both stable and effectively distinguish between different devices.
[0027] This method can significantly improve trust differentiation capabilities under interference from various types of malicious devices, increase the success rate of task delegation, and enhance the trusted interaction and resource scheduling performance of the social Internet of Things.
[0028] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description
[0029] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Fig. 1 This is a schematic diagram of the overall process of a resource-aware trust assessment method according to an embodiment of the present invention; Fig. 2 This is a schematic diagram of the social network architecture according to an embodiment of the present invention. Detailed Implementation
[0030] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.
[0031] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.
[0032] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.
[0033] Please see Figs. 1-2 This is a resource-aware trust assessment method.
[0034] Example This embodiment describes in detail a resource-aware trust assessment method, such as... Fig. 1 As shown, it specifically includes the following steps: Step 1: Based on the interaction behavior between devices and resource constraints, construct a multi-dimensional trust index system that includes direct trust, computational idle rate, recommendation trust, and social relationship degree; like Fig. 2 The diagram illustrating the social network architecture includes two planes: the physical device layer and the social network layer. The physical device layer consists of the actual existing devices and physical connections, while the social network layer represents the logical social relationships formed between these devices. There are three types of nodes in the network. At any given time, some nodes will generate task requests. White nodes are normal, trusted nodes, while red nodes are untrusted devices. Task nodes must avoid interacting with untrusted devices.
[0035] The construction of a multi-dimensional trust indicator system specifically includes: firstly using... This represents a set of devices in a network. Inter-device interaction is defined as a task delegated by the requesting party to the receiving party. To indicate, among them, the task It contains detailed information such as the operations the receiving device should perform and the required workload. Indicator variables are used to specify this information. To indicate the task Whether it is completed, when This indicates that the task is complete, signifying a successful interaction. Direct trust is used here. ), calculate idle rate ( ),reliability( ), public credibility ( Social relations Five trust metrics quantify trust values, using a quintuple. express.
[0036] Direct trust is built on the direct interaction history between the two devices, using Indicates equipment At any moment For equipment Direct trust. Computational space rate refers to the remaining computing load of a device. Since the amount of remaining computing resources directly affects whether a task can be completed, devices in the network tend to assign tasks to devices with abundant computing resources. Reliability reflects a node's ability to judge fairness when recommending others, demonstrating the objective fairness of the recommendation behavior; while credibility refers to the trust capital a node has accumulated within the group, reflecting the consensus of the network group on whether the node "possesses trustworthy recommendation qualifications." Social relations refer to several inherent social connections in the network, and devices tend to cooperate with devices similar to themselves. Generally, the trust value between devices in the network is expressed as:
[0037] in It is an aggregation operator. By aggregating multi-dimensional trust indicators to obtain a comprehensive trust value, it enables accurate trust assessment of devices in the network.
[0038] Step 1.1: Direct trust is built on the direct interaction history between the two devices. At any moment For equipment Direct trust is represented as:
[0039] in Indicates task Contribution level, collection Indicates equipment To the equipment The set of tasks to be delegated The initial value is 0.5, indicating that the device... For equipment Complete uncertainty, with continuous interaction, the device For equipment There will be more accurate direct trust estimates.
[0040] Step 1.2: Based on the device's maximum computing capacity and current load, construct a computing idle rate index to reflect the device's resource capacity available for executing delegated tasks. Normalize this index to the interval [0,1], and define it as follows:
[0041] Indicates equipment exist Calculate the idle rate at any given time. Indicates equipment The maximum computing resources, Indicates equipment exist Used computing resources at any given time.
[0042] Step 1.3: Reliability reflects a node's ability to judge fairness when recommending others, demonstrating the objective fairness of the recommendation behavior. This is calculated based on the deviation between the recommendation and actual credibility.
[0043] in, Indicates equipment exist Social credibility at all times It is a weighting factor, if the device For equipment If a rating is provided, then Otherwise, it is 0. It is an exponential penalty factor used to enhance the penalty effect on deviations from public credibility.
[0044] Public credibility refers to the trust capital a node accumulates within the group, reflecting the consensus of the network group on whether the node "possesses the qualifications to make trustworthy recommendations." It is calculated based on the cumulative trust feedback a device receives within the network.
[0045] Indicates the device at the time of the last iteration. Reliability, The penalty decay factor is used to adjust the convergence speed of the results based on the number of scoring samples. The credibility score is also in the range of [0,1], with higher values indicating better performance. The more trustworthy they are.
[0046] Step 1.4: Based on the various social relationships in SIoT, including Co-owner Relationship (OOR), Location Relationship (C-LOR), Collaborative Relationship (C-WOR), Master-Slave Relationship (POR), and Dynamic Relationship (SOR) based on interaction frequency, the first four social relationships can be regarded as stable relationships between devices, while SOR is a relatively fluctuating and dynamic social relationship, which can be calculated based on the interactions in the system:
[0047] Step 2: By inputting multidimensional trust indicators into the improved CRITIC-entropy weight fusion model, weights are calculated based on the discreteness, heterogeneity, and information content of the indicators to generate adaptive trust weights, and the comprehensive trust value between devices is calculated based on the dynamic trust weights. Step 2.1: First, standardize each trust indicator using Z-score, calculate the standard deviation of the indicator and its correlation with other indicators, and construct a CRITIC weight vector based on the information content of the indicators:
[0048] in This represents the total number of indicators considered in the multidimensional trust indicator system. This indicates the amount of information in each indicator, reflecting the evaluation capability of each indicator. This represents the standardized Z-Score value of the j-th indicator. This represents the correlation coefficient between the indicators. This represents the standard deviation function.
[0049] Information entropy reflects the uncertainty of an indicator in a sample, and is applied to the elements of the indicator matrix. The index is obtained by performing Min-Max normalization. Then calculate its entropy value:
[0050] In the formula, Indicates the number of trust samples. This represents the normalization coefficient.
[0051] Entropy weight normalization yields the weights:
[0052] Step 2.2: To balance information intensity and distributional differences, a new, optimal set of weights needs to be found. This set of weights W should be consistent with the CRITIC weights. And entropy weight To maintain minimal information deviation, this embodiment further proposes an adaptive fusion model based on minimizing KL divergence. First, an optimization objective function based on KL divergence is established, aiming to make the optimal weight W consistent with... and The sum of weighted KL divergences is minimized:
[0053] in , indicating that W is relative Information bias, Indicates W relative to Information bias. It is a preference coefficient representing the degree of confidence in the results of the CRITIC method, and the constraints are satisfied. If the weights calculated by CRITIC are more dispersed (i.e. (large), then Automatic scaling increases the model's trust in CRITIC's results, and vice versa. This makes the model completely data-driven, eliminating all subjective parameters.
[0054] Solving for this weight is a convex optimization problem. This embodiment uses the Lagrange multiplier method to solve it, that is, to establish the Lagrange function. Taking the partial derivative with respect to it and setting it to 0, we get... The unique optimal solution:
[0055] The solution is a weighted geometric mean, with a normalization term in the denominator, ensuring that all... The sum is 1. This formula also has consistency; if... or If any one of them approaches 0, then the final... Both will approach 0, which avoids the conflict that arises when one method considers it extremely unimportant while another considers it extremely important.
[0056] Ultimately, each device node For target equipment The overall trust value is expressed as a weighted linear combination of various indicators:
[0057] in Indicates equipment For equipment The Trust index value.
[0058] Step 3: Adaptively update the trust value based on interactive feedback. In each time slot of the network, update the direct trust, remaining device resources, indirect recommendation information and social relationship degree according to the newly generated interaction records, and dynamically obtain new weights through the fusion weighting method to complete the real-time update of the comprehensive trust value between devices.
[0059] Step 3.1: The adaptive update of the trust value is a multi-dimensional, dynamic, closed-loop process. This process is first based on the task execution feedback of the current time slot. Update the direct trust value by adjusting the success decay and failure penalty factors:
[0060] in , . The decay factor representing the success rate of a task. This represents the decay factor for failed tasks. Indicates the current time. Indicates the time threshold, if and If the interval between them exceeds the time threshold, then the task... Tasks from the distant past will not be considered in the CoT calculation; that is, tasks from a long time ago cannot help reflect the current trust level of the service provider. The attenuation rate is determined. The closer to 0, the faster the decay. Simultaneously, the idle rate is dynamically updated based on real-time load demand and changes in remaining resources, and is dynamically adjusted based on changes in task resource usage.
[0061] In the formula, , These represent the idle computing resources of device j in the current time slot and the idle computing resources in the previous time slot, respectively. This indicates the task load generated in this time slot. This indicates the computing resources released in this time slot.
[0062] At the network level, the system recalculates recommendation credibility and node reliability by collecting information from neighboring nodes, enabling dynamic adjustment of recommendation trust. It also updates the social relationship degree, centered on dynamic SOR, hourly based on interaction frequency and stability. These updated indicators are constructed into a new matrix, which is input into the CRITIC-entropy weight fusion model to redistribute weights and calculate a comprehensive trust value, thereby achieving accurate, dynamic, and adaptive evaluation of device trust status.
[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A resource-aware trust assessment method, characterized in that: The method includes the following steps: S1. Based on the interaction behavior and resource constraints between devices, construct a system including direct trust. Calculate idle rate ,reliability Public credibility and social relationships A multidimensional trust indicator system; S2. An improved weight determination method based on index correlation—the entropy weight fusion weighting model—is adopted to generate adaptive trust weights based on the discreteness, heterogeneity, and information content of multidimensional trust indicators, and to calculate the comprehensive trust value between devices based on the dynamic trust weights. S3. In each time slot, update direct trust, remaining device resources, indirect recommendation information and social relationship degree based on newly generated interaction records, and dynamically adjust the comprehensive trust value between devices through a fusion weighting method.
2. The resource-aware trust assessment method according to claim 1, characterized in that: In step S1, first use This represents a set of devices in a network. Inter-device interaction is defined as a task delegated by the requesting party to the receiving party, represented as... ; Use indicator variables To indicate the task Whether it is completed, when This indicates that the task is complete; The multidimensional trust indicator system adopts a five-tuple. In this representation, the trust value between devices in the network is expressed as: ,in It is an aggregation operator.
3. The resource-aware trust assessment method according to claim 2, characterized in that: The direct trust index in the multidimensional trust index system is based on the direct interaction history between two devices. At any moment For equipment Direct trust is represented as: in, Indicates task Contribution level, collection Indicates equipment To the equipment The set of tasks to be delegated The initial value is , indicating equipment For equipment Complete uncertainty, with continuous interaction, the device For equipment Direct trust is increasing.
4. The resource-aware trust assessment method according to claim 2, characterized in that: The computing idle rate metric in the multidimensional trust index system is based on the device's maximum computing capacity and current load, reflecting the device's resource capacity available for executing delegated tasks. Its definition is: In the formula, Indicates equipment exist Calculate the idle rate at any given time. Indicates equipment The maximum computing resources, Indicates equipment exist Used computing resources at any given time.
5. The resource-aware trust assessment method according to claim 2, characterized in that: The reliability index in the multidimensional trust index system is calculated based on the deviation between other devices' direct trust in the device and its actual credibility, and is expressed as follows: In the formula, It is a weighting factor, if the device For equipment If a rating is provided, then Otherwise, it is 0; Indicates equipment exist Constantly monitor the equipment Direct trust; Represents the set of all devices; It is an exponential penalty factor used to enhance the penalty effect on deviations from public credibility; Indicates equipment exist Social credibility at all times.
6. The resource-aware trust assessment method according to claim 2, characterized in that: In the multidimensional trust index system, social credibility refers to the trust capital accumulated by a node within a group, which is calculated based on the accumulated trust feedback obtained by the device in the network. In the formula, Indicates the device at the time of the last iteration. Reliability, This is a penalty decay factor used to adjust the convergence speed of the results based on the number of scoring samples.
7. The resource-aware trust assessment method according to claim 2, characterized in that: The social relationship index in the multidimensional trust index system represents a relatively fluctuating and dynamic social relationship, calculated based on interactions within the system. In the formula, Indicates an indicator function, It is a local constant.
8. The resource-aware trust assessment method according to claim 1, characterized in that: In step S2, the process of generating adaptive weights using the improved CRITIC-entropy weight fusion weighting model is as follows: First, Z-score standardization is performed on each trust indicator, and the standard deviation of the indicator and its correlation with other indicators are calculated. Based on the information content of the indicators, a CRITIC weight vector is constructed: in This represents the total number of indicators considered in the multidimensional trust indicator system. This indicates the amount of information in each indicator, reflecting the evaluation capability of each indicator. This represents the standardized Z-Score value of the j-th indicator. This represents the correlation coefficient between the indicators. Represents the standard deviation function; Then, the information entropy is calculated. Information entropy reflects the uncertainty of the index in the sample, and is applied to the index matrix. medium elements The index is obtained by performing Min-Max normalization. Then calculate its entropy value: In the formula, Indicates the number of trust samples. Represents the normalization coefficient; Entropy weight normalization yields the weights .
9. The resource-aware trust assessment method according to claim 8, characterized in that: In step S2, when calculating the comprehensive trust value between devices based on dynamic trust weights, an adaptive fusion model based on minimizing KL divergence is used to find a new and optimal set of weights. The process is as follows: First, an optimization objective function based on KL divergence is established, with optimal weights. and and The objective is to minimize the sum of the weighted KL divergences, which is expressed as: in, express relatively Information bias, express relatively Information bias; It is a preference coefficient representing the degree of confidence in the results of the CRITIC method, and the constraints are satisfied. If the weights calculated by CRITIC Larger, then Increase; if the weights calculated by CRITIC Larger, then Decrease; The Lagrange multiplier method is used to solve the objective function of a convex optimization problem, i.e., to establish the Lagrange function. Taking the partial derivative with respect to it and setting it to 0, we get... The unique optimal solution: Among them, the solution obtained For a weighted geometric mean, the denominator is a normalization term, and all... The sum is 1; at the same time, if or If any one of them approaches 0, then the final... All approach 0; Ultimately, each device node For target equipment The overall trust value is expressed as a weighted linear combination of various indicators: in Indicates equipment For equipment The Trust index value.
10. The resource-aware trust assessment method according to claim 9, characterized in that: In step S3, the adaptive update process of the trust value is as follows: based on the task execution feedback of the current time slot. Update the direct trust value by adjusting the success decay and failure penalty factors: in, , , The decay factor representing the success rate of a task. The decay factor representing a failed task; Indicates the current time. Indicates the time threshold. The decay rate parameter indicates the success rate of the interaction. The parameter representing the decay rate of failed interactions, if and If the interval between them exceeds the time threshold, then the task... It will not be considered in the CoT calculation; Simultaneously, the idle rate is dynamically updated based on changes in real-time load demand and remaining resources, and is dynamically adjusted based on changes in task resource usage. In the formula, , These represent the idle computing resources of device j in the current time slot and the idle computing resources in the previous time slot, respectively. This indicates the newly generated computational load in this time slot. This indicates the computing resources released in this time slot; At the network level, the system collects information from neighboring nodes to recalculate node reliability and credibility, thereby dynamically adjusting indirect trust. It also updates social relationship degree hourly based on interaction frequency and stability. The updated indicators are used to construct a new matrix, which is then input into the CRITIC-entropy weight fusion model to redistribute weights and calculate a comprehensive trust value, achieving accurate, dynamic, and adaptive assessment of device trust status.