A node credibility evaluation method for underwater acoustic sensing networks based on node behavior

Abstract

The present invention provides a node credibility assessment method for an underwater acoustic sensor network based on node behavior. The method uses node behavior and interaction success rate during information interaction as trust factors, aggregates trust factors using information entropy, and obtains direct trust, indirect trust, and historical trust. It then calculates node trust and establishes a trust assessment model based on node behavior. This allows for rapid and accurate detection of malicious nodes, effective identification and isolation, and ensures the safety and reliability of nodes in routing paths. The present invention detects tampering attacks, deception attacks, and black hole malicious nodes in underwater acoustic sensor networks. Combined with opportunistic routing protocols, it can rapidly and accurately detect malicious nodes, effectively identify and isolate them, ensuring the safety and reliability of nodes in routing paths.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater acoustic sensor network security, and in particular to a method for evaluating the credibility of nodes in a sensor network through abnormal node behavior. Background Art

[0002] Underwater acoustic sensor networks are a typical application of wireless sensor networks. As an ideal medium in the marine environment, they can monitor the target sea area over a large area and in real time. They have broad application prospects in the fields of marine resource exploration, marine environment monitoring, and marine safety assurance.

[0003] As a wireless self-organizing network, underwater acoustic sensor networks feature open transmission media, inter-node collaborative algorithms, and fuzzy defense boundaries. These networks are vulnerable to various attacks, such as Wormhole, Hello-Flood, and SelectiveForward attacks. Furthermore, due to the low transmission rate, high bit error rate, extended latency, narrow bandwidth, and rapid node energy consumption of underwater acoustic channels, the network is more susceptible to selfish nodes or captive nodes. Consequently, the trustworthiness of all nodes in a link cannot be guaranteed during network communication, compromising network robustness.

[0004] To ensure the normal operation of underwater acoustic sensor networks in the presence of malicious nodes, many researchers at home and abroad have proposed various methods for node trust assessment to safeguard the network's basic functionality. Jiang Jinfang et al. proposed an attack-resistant trust model based on multidimensional trust metrics (ARTMM), which takes into account the characteristics of the underwater acoustic channel and node mobility. However, this algorithm suffers from high computational complexity, the trust evidence generation process fails to account for the influence of malicious nodes, and the fuzzy set definition is subjective, making it unsuitable for dynamic underwater acoustic sensor networks. Han Guangjie et al. proposed a trust model based on cloud theory (TMC) to address these issues. This model, based on traditional fuzzy sets and probability statistics, can better assess the uncertainty of trust relationships. However, the model design makes many assumptions and places high demands on the location information of each node in the underwater acoustic sensor network. Lei Shu et al. proposed a collaborative trust model (STMS) based on SVM. This model divides the network into clusters, with nodes within clusters classified as MCHs, SCHs, and CMs. The k-means and SVM algorithms are used to generate the trust assessment model. However, this model assumes that the probability of both the MCH and SCH in the same cluster being attacked simultaneously is zero, making it incapable of defending against multi-node coordinated attacks.

[0005] The above research demonstrates that existing trust models for underwater acoustic sensor networks largely fail to consider the unique characteristics of the underwater acoustic environment and network resource constraints during their design. Consequently, these trust models have limitations in their application to underwater acoustic sensor networks. This paper, incorporating the complex characteristics of underwater acoustic channels, updates node trust assessments in real time as network communications progress. This method uses node behavior and interaction success rates during information exchange as trust factors, and incorporates both spatial and temporal dimensions to propose a node trustworthiness assessment method for underwater acoustic sensor networks based on node behavior. Summary of the Invention

[0006] In order to overcome the shortcomings of the existing technology, the present invention provides a method for evaluating the credibility of underwater acoustic sensor network nodes based on node behavior. In order to defend against malicious nodes that become the best next-hop nodes in the candidate set through attacks such as camouflage, monitoring, and deception, and to obtain the credibility of underwater acoustic sensor network nodes, a method for evaluating the trust of underwater acoustic sensor network nodes based on node behavior is proposed. The present invention uses the node behavior and interaction success rate during the information interaction process as trust factors, and then uses information entropy to aggregate the trust factors to obtain direct trust, indirect trust, and historical trust, and then obtains the node trust, and establishes a trust evaluation model based on node behavior. In this way, malicious nodes can be detected quickly and accurately, and effectively identified and isolated, ensuring the safety and reliability of nodes in the routing path.

[0007] Aiming at the location deception vulnerability existing in underwater acoustic sensor networks, selfish nodes and malicious nodes in the network exploit it, causing the network to crash and shorten the network lifespan. In this paper, a node credibility assessment method for underwater acoustic sensor networks based on node behavior is proposed.

[0008] The steps of the technical solution adopted by the present invention to solve its technical problem are as follows:

[0009] The first step is to calculate the direct trust of the node;

[0010] Direct trust mainly includes two aspects: information interaction trust and node behavior trust in the process of node network communication; direct trust The calculation formula is as follows:

[0011]

[0012] Among them, r ij and b ij Refers to information interaction trust and node behavior trust respectively, and They refer to the adaptive weights of information interaction trust and node behavior trust obtained through the concept of information entropy;

[0013] (1) Trust in information interaction

[0014] The trust in information interaction depends on the number of successful and failed direct information interactions between the sending node i and its neighbor node j. Therefore, the expectation of successful information interaction can be considered as the trust in information interaction between nodes. The trust in information interaction follows a beta distribution.

[0015] Therefore, the information interaction trust r ij Expressed as:

[0016]

[0017] Among them, r ij represents the trust degree of information interaction of evaluated node j that evaluated node i believes; α ij and β ij Respectively represent the number of successful interactions and failed interactions between the evaluation node i and the evaluated node j in the past; the probability density function

[0018] (2) Node behavior trust

[0019] The node behavior trust refers to whether the node has abnormal behavior during the network communication process; it is used to perceive the suspicious behavior of the node and adjust the node trust in time; the trust of the evaluated node behavior is calculated based on the frequency of abnormal behavior of the evaluated node counted by the evaluation node during the monitoring process; b ij It represents the node behavior trust of the evaluated node j as considered by the evaluating node i. The node behavior trust is expressed as:

[0020]

[0021] in, The node behavior trust of the evaluated node j as considered by the evaluating node i in the previous time period;

[0022] The second step is to statistically calculate the trust of the nodes in the spatial dimension;

[0023] Indirect trust refers to the direct trust of the evaluated node fed back by the evaluating node through other neighboring nodes of the evaluated node. It is a trust evaluation of the node in the spatial dimension. Specifically, when the evaluating node i needs to further evaluate the trust of the evaluated node j, the evaluating node i broadcasts a query message to its neighboring nodes to obtain the recommendation information of the evaluated node j. Once the query message is received, the recommending node k (the common neighboring node of the evaluating node i and the evaluated node j) returns the direct trust of the evaluated node j to the evaluating node i as the recommendation information. Then the trust of node j fed back by the recommending node k is Expressed as:

[0024]

[0025] in, is the direct trust of evaluating node k on the evaluated node j.

[0026] The n neighboring nodes of the evaluation node i act as recommendation nodes to recommend the evaluated node j. The trust levels fed back to the evaluation node i are: Evaluation node i uses information entropy to weight the trust feedback from the recommendation node, and then sums the values ​​obtained by multiplying their respective weights by the trust to obtain the indirect trust of evaluation node i on the evaluated node j;

[0027] The third step is to calculate the trust of the nodes in the time dimension.

[0028] Historical trustworthiness: Due to the dynamic changes of underwater acoustic sensor networks and the selective attacks of malicious nodes, the node trustworthiness evaluation will also change with the network communication and time cycle changes. Therefore, the node's historical record trustworthiness must be used as the node's trustworthiness evaluation in the time dimension.

[0029] Historical trust depends on the comprehensive trust of the previous update cycle and previous historical trust Therefore, in the present invention, information entropy is first used to distribute the weights of the comprehensive trust degree of the previous update cycle and the previous historical trust degree in the historical trust degree, and then the comprehensive trust degree of the previous update cycle and the previous historical trust degree are aggregated using information entropy;

[0030] The fourth step is to comprehensively evaluate the trustworthiness of the nodes;

[0031] In the trust model based on node behavior, node trust includes direct trust and node trust in spatial and temporal dimensions. Therefore, the evaluation of the comprehensive trust of underwater nodes requires the aggregation of direct trust, indirect trust, and historical trust of nodes. The information entropy is used to aggregate and calculate the comprehensive trust of nodes.

[0032] Step 5: Identify and isolate malicious nodes;

[0033] After the node comprehensive trust evaluation stage, the evaluation node makes a decision based on the comprehensive trust of the evaluated node; the setting of the threshold in the trust model based on node behavior refers to the trust interval division process in the trust model based on information entropy, and the TH value is set to 0.6; the given preset threshold is compared with the node comprehensive trust, and a malicious node refers to an evaluated node whose comprehensive trust is assessed by the evaluation node to be lower than the threshold; then, the evaluated node with a comprehensive trust lower than the threshold is identified as a malicious node by the evaluation node, and in the subsequent network communication process, the evaluation node will automatically avoid the node assessed as a malicious node.

[0034] Based on the identification results of network nodes, trusted nodes are allowed to participate in normal routing operations and forward and receive data; untrusted nodes are not allowed to participate in normal routing operations, and the data forwarded by them will be discarded by normal network nodes.

[0035] In the first step, the adaptive weight calculation formula is as follows:

[0036]

[0037]

[0038] H(r ij )=-r ij log2 r ij -(1-r ij )log2(1-r ij )

[0039] H(b ij )=-b ij log2 b ij -(1-b ij )log2(1-b ij )

[0040] Among them, H(r ij ) and H(b ij ) are the information entropy of information interaction trust and node behavior trust, respectively.

[0041] In the second step, first calculate Entropy:

[0042]

[0043] in, It represents the direct trust degree of evaluated node j at the recommending node k when the recommending node k recommends the evaluated node j to the evaluating node i;

[0044] Use information entropy to calculate the weight w of recommendation trust k :

[0045]

[0046] Finally, indirect trust The calculation results are expressed as:

[0047]

[0048] in, w represents the direct trust of evaluated node j at the recommending node k when the recommending node k recommends the evaluated node j to the evaluating node i; krepresents the trust weight of the recommended node k at the evaluation node i.

[0049] In the third step, the historical trust that is updated over time is The calculation is as follows:

[0050]

[0051] in, and They represent the comprehensive trust of the previous update cycle and the historical trust in the latest cycle, w # represents the adaptive weight of the comprehensive trust in the previous update cycle, w * Represents the adaptive weight of the previous historical trust; the calculation formula is as follows:

[0052]

[0053]

[0054]

[0055]

[0056] in, and They represent the comprehensive trust in the previous update cycle and the information entropy of the previous historical trust, respectively.

[0057] In the fourth step, the comprehensive trust includes the node's direct trust, indirect trust and historical trust. ij Information entropy is still used for aggregation calculation, and the calculation formula is as follows:

[0058]

[0059] Among them, w d , w i and w h are the adaptive weights of the node’s direct trust, indirect trust, and historical trust, respectively. and are the direct trust, indirect trust, and historical trust of the node respectively; the calculation formula is as follows:

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066] in, and They represent the information entropy of the node's direct trust, indirect trust, and historical trust respectively.

[0067] The beneficial effect of the present invention lies in the node behavior-based credibility assessment method for underwater acoustic sensor networks proposed by the present invention. In a complex, changeable and vulnerable underwater environment, the model proposes corresponding abnormal behavior detection for potential security threats to underwater acoustic sensor networks, and calculates node trust by integrating direct node interactions, recommendations from other nodes and historical node behaviors, thereby isolating identified malicious nodes. Afterwards, the trust model can be introduced into the opportunistic routing protocol, and the node completes the trust assessment process by broadcasting trust probe packets and updates the node indirect trust; at the same time, during normal routing operations and data transmission processes, the node trust is updated in real time by monitoring the node behavior, and malicious nodes are promptly identified and isolated. In order to detect tampering attacks, deception attacks and black hole malicious nodes in underwater acoustic sensor networks, the present invention, combined with the opportunistic routing protocol, can quickly and accurately detect malicious nodes, and effectively identify and isolate them, ensuring the safety and reliability of nodes in the routing path. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a schematic diagram of the average trust level of relay nodes in a normal network.

[0069] Figure 2 It is a diagram of the trust degree of malicious nodes in tampering attacks.

[0070] Figure 3 This is a diagram of the trust level of malicious nodes in a deception attack.

[0071] Figure 4 This is a diagram of the trust level of malicious nodes in a black hole attack. DETAILED DESCRIPTION

[0072] The present invention will be further described below with reference to the accompanying drawings and examples.

[0073] In the trust model based on node behavior, node trust includes direct trust, indirect trust and historical trust. Therefore, the calculation of the comprehensive trust of underwater nodes requires the aggregation of direct trust, indirect trust and historical trust of nodes.

[0074] The specific implementation steps are as follows:

[0075] Step 1: Calculate the direct trust of the node;

[0076] The calculation of direct trust includes information interaction trust r ij and node behavior trust b ij Aggregate calculation of the two parts. Then, direct trust is to convert the information interaction trust r ij and node behavior trust b ij Obtained by further summing through weight distribution.

[0077] In order to make the given trust model applicable to networks of different uses in different scenarios and reduce the impact of human subjective factors on the accuracy of the trust model, the present invention uses information entropy to dynamically allocate the weights of information interaction trust and node behavior trust.

[0078] Direct trust The calculation formula is as follows:

[0079]

[0080] Among them, r ij and b ij Refers to information interaction trust and node behavior trust respectively, and They refer to the adaptive weights of information interaction trust and node behavior trust obtained through the concept of information entropy. ij ) and H(b ij ) are the information entropy of information interaction trust and node behavior trust, respectively. The calculation formula is as follows:

[0081] H(r ij )=-r ij log2 r ij -(1-r ij )log2(1-r ij )

[0082] H(b ij )=-b ij log2 b ij -(1-b ij )log2(1-b ij )

[0083]

[0084]

[0085] Step 2: Calculate the indirect trust of the node;

[0086] The calculation of indirect trust is mainly the aggregate calculation of the trust of the evaluated node given by the recommending node to the evaluating node. Assume that there are n recommending nodes that recommend evaluated node j to the evaluating node i, and there are n indirect trusts. In this process, it is not possible to subjectively judge whether a certain recommendation node is trustworthy or not and the weight of each recommendation node is excellent. Therefore, information entropy is introduced to assign weights to different recommendation nodes. This not only takes the differences between different recommendation nodes into account, but also enhances the adaptability of the model. First, calculate Entropy:

[0087]

[0088] in, It represents the direct trust degree of evaluated node j at the recommending node k when the recommending node k recommends the evaluated node j to the evaluating node i;

[0089] Information entropy reflects the degree of disorder of information, while the information entropy of each recommended trust reflects the degree of difference between them, that is, the degree to which each recommended trust deviates from the overall recommended trust set. The behavior of malicious nodes deliberately devaluing legitimate nodes in the underwater acoustic sensor network or maliciously praising other malicious nodes will cause the recommended trust of the node evaluated by this malicious node to deviate from the actual trust of the node to a certain extent. Information entropy can be used to identify this recommended node, thereby reducing its impact on the objectivity and accuracy of the node trust. Generally speaking, the smaller the difference between the recommended trust of the evaluated node given by the recommending node, the more objective the recommendation of each recommending node to the evaluated node. Therefore, the weight w of the recommended trust can be calculated using information entropy. k :

[0090]

[0091] Finally, indirect trust The calculation results are expressed as:

[0092]

[0093] in, w represents the direct trust of evaluated node j at the recommending node k when the recommending node k recommends the evaluated node j to the evaluating node i; k represents the trust weight of the recommendation node k at the evaluation node i;

[0094] Step 3: Calculation of historical trustworthiness of nodes

[0095] The calculation of historical trust mainly depends on the comprehensive trust of the previous update cycle and previous historical trust In this process, if the weight of the comprehensive trust in the most recent update cycle is high, then the historical trust of the malicious node will change the most during this time period, which will cause the comprehensive trust of the node in the most recent update cycle to have a great influence on the historical trust, and it will not be able to well reflect the performance of the node in the entire stage. If the weight of the previous historical trust is high, it will not be able to well evaluate the situation of the node in the most recent update cycle, so the information entropy is used to distribute the weight. For the historical trust updated over time, The calculation is as follows:

[0096]

[0097] in, and They represent the comprehensive trust of the previous update cycle and the historical trust in the latest cycle, w # represents the adaptive weight of the comprehensive trust in the previous update cycle, w * An adaptive weight representing the previous historical trust; and They represent the comprehensive trust of the previous update cycle and the information entropy of the previous historical trust, respectively. The calculation formula is as follows:

[0098]

[0099]

[0100]

[0101]

[0102] Step 4: Calculation of the comprehensive trust of the node

[0103] The comprehensive trust includes the node's direct trust, indirect trust and historical trust. ij Information entropy is still used for aggregation calculation, and the calculation formula is as follows:

[0104]

[0105] Among them, w d , w i and w h are the adaptive weights of the node’s direct trust, indirect trust, and historical trust, respectively. and are the direct trust, indirect trust and historical trust of the node respectively; and The information entropy representing the direct trust, indirect trust and historical trust of a node is calculated as follows:

[0106]

[0107]

[0108]

[0109]

[0110]

[0111]

[0112] Step 5: Identify and isolate malicious nodes;

[0113] After calculating the node's comprehensive trustworthiness, the evaluation node can make decisions based on the comprehensive trustworthiness of the evaluated node. Nodes with a comprehensive trustworthiness below a threshold (0.6) are identified as untrustworthy, while nodes with a comprehensive trustworthiness greater than or equal to the threshold are identified as trusted. Based on the network node's identification results, trusted nodes are allowed to participate in normal routing operations, forwarding and receiving data. Untrustworthy nodes are not allowed to participate in normal routing operations, and data forwarded by them is discarded by normal network nodes.

[0114] The average trust value of normal relay nodes in the underwater acoustic sensor network of the present invention is as follows: Figure 1 As shown in the figure, it shows that under normal network conditions, the trust of network nodes is stable above the threshold, but the change in trust is not obvious, which also prevents some malicious attack nodes from gaining recognition from network nodes through short-term good performance during the network communication process. The average trust of malicious nodes in the underwater acoustic sensor network of the present invention is shown in Figure 2. Figure 2 As shown, the average trust value of malicious nodes in the underwater acoustic sensor network of the present invention performing deception attacks is as follows: Figure 3 As shown, the average trust value of malicious nodes in the underwater acoustic sensor network of the present invention performing black hole attack is as follows: Figure 4 As shown in the figure, in an attacked network, the trustworthiness of malicious nodes immediately drops below the threshold. At the same time, as information is exchanged, the trustworthiness of nodes continues to decline, and the decline is greater than normal. This also verifies that the present invention can quickly and accurately detect malicious nodes, effectively identify and isolate them, and ensure the security and reliability of network nodes.

Claims

1. A node credibility assessment method for underwater acoustic sensing networks based on node behavior, characterized in that The steps include: The first step is to calculate the direct trust of the node; Direct trust includes two aspects: information interaction trust and node behavior trust during node network communication; direct trust The calculation formula is as follows: Among them, r ij and b ij Refers to information interaction trust and node behavior trust respectively, and They refer to the adaptive weights of information interaction trust and node behavior trust obtained through the concept of information entropy; (1) Trust in information interaction The trust in information interaction depends on the number of successful and failed direct information interactions between the sending node i and its neighbor node j. Therefore, the expectation of successful information interaction can be considered as the trust in information interaction between nodes. The trust in information interaction follows a beta distribution. Therefore, the information interaction trust r ij Expressed as: Among them, r ij represents the trust degree of information interaction of evaluated node j that evaluated node i believes; α ij and β ij Respectively represent the number of successful interactions and failed interactions between the evaluation node i and the evaluated node j in the past; the probability density function (2) Node behavior trust The node behavior trust refers to whether the node has abnormal behavior during the network communication process; it is used to perceive the suspicious behavior of the node and adjust the node trust in time; the trust of the evaluated node behavior is calculated based on the frequency of abnormal behavior of the evaluated node counted by the evaluation node during the monitoring process; b ij It represents the node behavior trust of the evaluated node j as considered by the evaluating node i. The node behavior trust is expressed as: in, The node behavior trust of the evaluated node j as considered by the evaluating node i in the previous time period; The second step is to statistically calculate the trust of the nodes in the spatial dimension; Indirect trust refers to the direct trust of the evaluating node on the evaluated node fed back by other neighboring nodes of the evaluated node. It is a trust evaluation of the node in the spatial dimension. Specifically, when the evaluating node i needs to further evaluate the trust of the evaluated node j, the evaluating node i broadcasts a query message to its neighboring nodes to obtain the recommendation information of the evaluated node j. Once the query message is received, the recommending node k is recommended. The recommending node k is a common neighboring node of the evaluating node i and the evaluated node j. It returns its direct trust in the evaluated node j as the recommendation information to the evaluating node i. Then the trust of node j fed back by the recommending node k is Expressed as: in, is the direct trust of evaluating node k in the evaluated node j; The n neighboring nodes of the evaluation node i act as recommendation nodes to recommend the evaluated node j. The trust levels fed back to the evaluation node i are: Evaluation node i uses information entropy to weight the trust feedback from the recommendation node, and then sums the values ​​obtained by multiplying their respective weights by the trust to obtain the indirect trust of evaluation node i on the evaluated node j; The third step is to calculate the trust of the nodes in the time dimension. Historical trustworthiness: Due to the dynamic changes of underwater acoustic sensor networks and the selective attacks of malicious nodes, the node trustworthiness evaluation will also change with the network communication and time cycle changes. Therefore, the node's historical record trustworthiness must be used as the node's trustworthiness evaluation in the time dimension. Historical trust depends on the comprehensive trust of the previous update cycle and previous historical trust Therefore, we first use information entropy to distribute the weights of the comprehensive trust in the previous update cycle and the previous historical trust in the historical trust, and then aggregate the comprehensive trust in the previous update cycle and the previous historical trust using information entropy; The fourth step is to comprehensively evaluate the trustworthiness of the nodes; In the trust model based on node behavior, node trust includes direct trust and node trust in spatial and temporal dimensions. Therefore, the evaluation of the comprehensive trust of underwater nodes requires the aggregation of direct trust, indirect trust, and historical trust of nodes. The information entropy is used to aggregate and calculate the comprehensive trust of nodes. Step 5: Identify and isolate malicious nodes; After the node comprehensive trust evaluation phase, the evaluation node makes a decision based on the comprehensive trust of the evaluated node. The threshold setting in the trust model based on node behavior refers to the trust interval division process in the trust model based on information entropy, and the TH value is set to 0.

6. The given preset threshold is compared with the node comprehensive trust. A malicious node refers to an evaluated node whose comprehensive trust is assessed by the evaluation node to be lower than the threshold. Based on the identification results of network nodes, trusted nodes are allowed to participate in normal routing operations and forward and receive data; untrusted nodes are not allowed to participate in normal routing operations, and the data forwarded by them will be discarded by normal network nodes.

2. The node credibility assessment method for underwater acoustic sensor networks based on node behavior according to claim 1 is characterized by: In the first step, the adaptive weight calculation formula is as follows: H(r ij )=-r ij log2r ij -(1-r ij )log2(1-r ij ) H(b ij )=-b ij log2b ij -(1-b ij )log2(1-b ij ) Among them, H(r ij ) and H(b ij ) are the information entropy of information interaction trust and node behavior trust, respectively.

3. The node credibility assessment method for underwater acoustic sensor networks based on node behavior according to claim 1 is characterized by: In the second step, first calculate Entropy: in, It represents the direct trust degree of evaluated node j at the recommending node k when the recommending node k recommends the evaluated node j to the evaluating node i; Use information entropy to calculate the weight w of recommendation trust k : Finally, indirect trust The calculation results are expressed as: in, w represents the direct trust of evaluated node j at the recommending node k when the recommending node k recommends the evaluated node j to the evaluating node i; k represents the trust weight of the recommended node k at the evaluation node i.

4. The node credibility assessment method for underwater acoustic sensor networks based on node behavior according to claim 1 is characterized by: In the third step, the historical trust that is updated over time is The calculation is as follows: in, and They represent the comprehensive trust of the previous update cycle and the historical trust in the latest cycle, w # represents the adaptive weight of the comprehensive trust in the previous update cycle, w * Represents the adaptive weight of the previous historical trust; the calculation formula is as follows: in, and They represent the comprehensive trust in the previous update cycle and the information entropy of the previous historical trust, respectively.

5. The node credibility assessment method for underwater acoustic sensor network based on node behavior according to claim 1 is characterized by: In the fourth step, the comprehensive trust includes the node's direct trust, indirect trust and historical trust. ij Information entropy is still used for aggregation calculation, and the calculation formula is as follows: Among them, w d , w i and w h are the adaptive weights of the node’s direct trust, indirect trust, and historical trust, respectively. and are the direct trust, indirect trust, and historical trust of the node respectively; the calculation formula is as follows: in, and They represent the information entropy of the node's direct trust, indirect trust, and historical trust respectively.

Images