CN103812696B - A kind of Internet of things node credit assessment method based on shuffled frog leaping algorithm - Google Patents

Abstract

An IoT node reputation evaluation method based on the hybrid leapfrog algorithm, which analyzes the local characteristics of nodes in the IoT, calculates the importance of nodes in the autonomous domain of the IoT, and uses the calculated node importance as the basis for node screening. The hybrid leapfrog algorithm is used to cluster the nodes, and a class of nodes with high importance is selected as the neighbor nodes for reputation evaluation. According to the reputation evaluation algorithm, the neighbor nodes are used to evaluate the reputation of the nodes that need to be evaluated. According to the current reputation of the node and The historical reputation calculates a more accurate node reputation value, sets a threshold, and compares the node reputation value with the threshold to determine whether the node is credible; when the reputation value is lower than the set threshold, it is determined that the node is an untrustworthy node, otherwise Those are credible nodes; the invention can effectively avoid the problem that untrustworthy nodes interfere with the evaluation results in the traditional reputation evaluation system.

Description

A Reputation Evaluation Method for Internet of Things Nodes Based on Hybrid Leapfrog Algorithm

technical field

The invention relates to the field of communication technology, in particular to a reputation evaluation method for Internet of Things nodes based on a hybrid leapfrog algorithm.

Background technique

At present, research on the credibility of distributed systems mainly focuses on two aspects: trust management and trust evaluation. The essence of trust management is an access control model and method based on authentication and authorization. The trust evaluation method is based on the recommended trust relationship between entities, and evaluates the credibility of entities based on their own experience, and then makes decisions based on the credibility. Classical reputation management techniques include reputation averaging schemes, Bayesian networks, and cluster filtering. It is very important to establish trust between nodes and domains. Generally speaking, the trust level can be evaluated by the behavior of neighbor nodes and the information they generate for corresponding events. However, nodes are very vulnerable to attacks, and then provide unreliable or malicious feedback, affecting the real node reputation value.

The disadvantages of existing methods are:

1. Credit rating evaluation lacks indirect reputation recommendation

The usual evaluation of reputation level needs to consider both subjective and objective factors. For example, the evaluation of the credibility of resources to users is not only determined by the direct reputation obtained locally, but also the indirect reputation recommendation (reputation) obtained by the computing agency; and the node The evaluation of the reliability is not only determined by the users who use the node, but also needs to consider the reputation recommendations of the node by users of other self-made domains. Comprehensively consider the reputation of users, nodes, and subjective and objective factors in the reputation evaluation process.

2. Unreliable nodes interfere with reputation evaluation

Due to the highly heterogeneous and highly mixed characteristics of the Internet of Things itself, nodes in the autonomous domain are likely to suffer from illegal intrusion attacks, thus making wrong reputation rating evaluations, interfering with the final evaluation results, and affecting the accuracy of the evaluation results.

Contents of the invention

Aiming at the problem that untrustworthy nodes may interfere with evaluation results in traditional reputation evaluation systems, the invention proposes a reputation evaluation method for Internet of Things nodes based on a hybrid leapfrog algorithm.

The technical scheme adopted in the present invention is: a method for evaluating the reputation of Internet of Things nodes based on the hybrid leapfrog algorithm, and the method includes the following steps:

Step 1. Analyze the local characteristics of nodes in the Internet of Things, and calculate the importance of nodes in the autonomous domain of the Internet of Things;

Step 2. Use the calculated node importance as the basis for node screening, and use the hybrid leapfrog algorithm to cluster the nodes;

Step 3, selecting a class of nodes with higher node importance obtained in step 2 as neighbor nodes for reputation evaluation;

Step 4, according to the reputation evaluation algorithm, use the neighbor nodes obtained in step 3 to perform reputation evaluation on the node to be evaluated;

Step 5. Calculate a more accurate node reputation value based on the node's current reputation and historical reputation;

Step 6. Set a threshold, compare the node reputation value obtained in step 5 with the threshold to determine whether the node is credible; when the reputation value is lower than the set threshold, it is determined that the node is an untrustworthy node, otherwise it is credible node.

The calculation method of the node importance in the step 1 comprises the following steps:

Step 201, represent the network as a binary group YK, EY, and represent the node set as , the set of edges connecting nodes is expressed as , where n and m represent the number of nodes and edges of the network, the more important the edges connecting nodes are, the more important the nodes are;

Step 202, use the formula to calculate the edge weights, where Indicates the number of edges connected to node i;

Step 203, summing the weights of the edges connected to node i That is, the weight of node i;

Step 204, use the formula Calculate the importance of node i, where .

The leapfrog hybrid algorithm in the step 2 comprises the following steps:

Step 301, based on the importance of nodes, use the hybrid leapfrog algorithm to screen neighbor nodes, each individual frog can indicate the importance of a node and using the formula Calculate the fitness of the frog;

Step 302. Randomly initialize a frog group consisting of P frogs , i=1, 2,...P;

Step 303, arrange in descending order according to the calculated fitness of each frog, and the individual frog with the best function value is set to ;

Step 304, divide the entire frog population into F groups, each group contains G frogs, therefore , the first frog enters the first group, the second frog enters the second group, the Fth frog enters the Fth group, and then the F+1th frog enters the first group, and the F+2th frog Frogs enter the second group, and so on, until all frogs are divided;

Step 305, after the group division is completed, a local deep search is performed on each group, and the individuals with the best and worst fitness in each group are with , each iteration for the worst fitness Go on, the update strategy is: frog moving distance , update the worst frog position ( )in, yes a random number between, is the maximum distance that frogs are allowed to move, update the frog individuals with the worst fitness in the group through the above formula, and perform the set number of local searches for each group;

Step 306 : Merge the clusters that have undergone local deep search to form a new cluster, and judge whether the termination condition of the algorithm is satisfied, and complete the screening of reliable neighbor nodes.

The reputation evaluation in step 4 specifically includes the following steps:

According to the nodes obtained after screening, they are used as neighbor nodes. During the _TP cycle, there will be multiple neighbor nodes observing the evaluated node. Each node has and maintains its own neighbor node list, which contains the neighbor node ID number. and reputation value and other information, when node i sends a data packet to node n, the intermediate node j needs to forward it, and the node calculates the node reliability through the existing monitoring conditions , The calculation formula is as follows: ,in Indicates the number of data packets that node i requests node j to forward; Indicates that j is the number of data packets forwarded by i, in the cycle Within, the more data packets forwarded by j, the higher the reliability.

In order to prevent nodes with high reputation values from providing too high a right to speak, resulting in subjective bias, the global reputation R is introduced as a parameter to reduce risks and reduce subjective bias. The specific formula is as follows:

Among them, i is the neighbor node of N, S _N is the set of neighbor nodes of N, is to provide the reputation value of the listening node, T is the reliability threshold, the behavior of the node below the threshold will be considered as a bad node, and the introduced parameter ,in is the adjustment function, , M _iki is the total transaction amount of node i, is according to the local reputation expectation of node i relative to its neighbor nodes, and .

Beneficial effects of the present invention: the present invention filters the nodes in the network before evaluating the reputation of the nodes, and obtains nodes with higher importance as neighbors, which better avoids the interference of unreliable neighbors on the reputation evaluation results, and improves The accuracy of reputation evaluation is improved; the invention uses the hybrid leapfrog algorithm to search locally and globally, and then screens nodes, which has faster convergence and stronger robustness.

Description of drawings

Fig. 1 is a structural block diagram of the present invention.

detailed description

As shown in the figure, a method for evaluating the reputation of IoT nodes based on the hybrid leapfrog algorithm, the method includes the following steps:

Step 1. Analyze the local characteristics of nodes in the Internet of Things, and calculate the importance of nodes in the autonomous domain of the Internet of Things;

Step 2. Use the calculated node importance as the basis for node screening, and use the hybrid leapfrog algorithm to cluster the nodes;

Step 3, selecting a class of nodes with higher node importance obtained in step 2 as neighbor nodes for reputation evaluation;

Step 4, according to the reputation evaluation algorithm, use the neighbor nodes obtained in step 3 to perform reputation evaluation on the node to be evaluated;

Step 5. Calculate a more accurate node reputation value based on the node's current reputation and historical reputation;

Step 6. Set a threshold, compare the node reputation value obtained in step 5 with the threshold to determine whether the node is credible; when the reputation value is lower than the set threshold, it is determined that the node is an untrustworthy node, otherwise it is credible node.

The calculation method of the node importance in the step 1 comprises the following steps:

Step 201, represent the network as a binary group YK, EY, and represent the node set as , the set of edges connecting nodes is expressed as , where n and m represent the number of nodes and edges of the network, the more important the edges connecting nodes are, the more important the nodes are;

Step 202, use the formula to calculate the edge weights, where Indicates the number of edges connected to node i;

Step 203, summing the weights of the edges connected to node i That is, the weight of node i;

Step 204, use the formula Calculate the importance of node i, where .

The leapfrog hybrid algorithm in the step 2 comprises the following steps:

Step 301, based on the importance of nodes, use the hybrid leapfrog algorithm to screen neighbor nodes, each individual frog can indicate the importance of a node and using the formula Calculate the fitness of the frog;

Step 302. Randomly initialize a frog group consisting of P frogs , i=1, 2,...P;

Step 303, arrange in descending order according to the calculated fitness of each frog, and the individual frog with the best function value is set to ;

Step 304, divide the entire frog population into F groups, each group contains G frogs, therefore , the first frog enters the first group, the second frog enters the second group, the Fth frog enters the Fth group, and then the F+1th frog enters the first group, and the F+2th frog Frogs enter the second group, and so on, until all frogs are divided;

Step 305, after the group division is completed, a local deep search is performed on each group, and the individuals with the best and worst fitness in each group are with , each iteration for the worst fitness Go on, the update strategy is: frog moving distance , update the worst frog position ( )in, yes a random number between, is the maximum distance that frogs are allowed to move, update the frog individuals with the worst fitness in the group through the above formula, and perform the set number of local searches for each group;

Step 306 : Merge the clusters that have undergone local deep search to form a new cluster, and judge whether the termination condition of the algorithm is satisfied, and complete the screening of reliable neighbor nodes.

The reputation evaluation in step 4 specifically includes the following steps:

According to the nodes obtained after screening, they are used as neighbor nodes. During the _TP cycle, there will be multiple neighbor nodes observing the evaluated node. Each node has and maintains its own neighbor node list, which contains the neighbor node ID number. and reputation value and other information, when node i sends a data packet to node n, the intermediate node j needs to forward it, and the node calculates the node reliability through the existing monitoring conditions , The calculation formula is as follows: ,in Indicates the number of data packets that node i requests node j to forward; Indicates that j is the number of data packets forwarded by i, in the cycle Within, the more data packets forwarded by j, the higher the reliability.

In order to prevent nodes with high reputation values from providing too high a right to speak, resulting in subjective bias, the global reputation R is introduced as a parameter to reduce risks and reduce subjective bias. The specific formula is as follows:

Among them, i is the neighbor node of N, S _N is the set of neighbor nodes of N, is to provide the reputation value of the listening node, T is the reliability threshold, the behavior of the node below the threshold will be considered as a bad node, and the introduced parameter ,in is the adjustment function, , M _iki is the total transaction amount of node i, is according to the local reputation expectation of node i relative to its neighbor nodes, and .

The larger the number of neighbor nodes participating in reputation evaluation and the larger the amount of data transmission, the more accurate the global reputation expectation will be.

Nodes in the Internet of Things are relatively cheap because of their mobility. It would be unfair to be identified as an unreliable node passively forwarding packets due to accidental reasons. Therefore, when we calculate the current reputation value, we also need to take the historical reputation value into account to form a new reputation value. The specific formula is as follows:

, therefore, according to different autonomous domain environments, we can weigh the proportion of historical reputation value to current node reputation by adjusting the α factor.

The present invention filters the nodes in the network before evaluating the reputation of the nodes, and obtains nodes with higher importance as neighbors after filtering, which better avoids the interference of unreliable neighbors on the reputation evaluation results, and improves the accuracy of reputation evaluation ; The present invention uses the hybrid leapfrog algorithm to search locally and globally, and then screen nodes, which has faster convergence and stronger robustness.

Claims (4)

1. A node reputation evaluation method of the Internet of things based on a mixed frog-leaping algorithm is characterized by comprising the following steps: the method comprises the following steps:

step 1, analyzing local characteristics of nodes in the Internet of things, and calculating the importance of the nodes in the autonomous domain of the Internet of things;

step 2, clustering the nodes by using a mixed frog-leaping algorithm by taking the calculated importance of the nodes as a basis for node screening;

step 3, selecting the nodes with higher node importance obtained in the step 2 as neighbor nodes for reputation evaluation;

step 4, using the neighbor node obtained in the step 3 to evaluate the credit of the node to be evaluated according to a credit evaluation algorithm;

step 5, calculating a more accurate node reputation value according to the current reputation and the historical reputation of the node;

step 6, setting a threshold, and comparing the node reputation value obtained in the step 5 with the threshold to judge whether the node is credible; when the credit value is lower than a set threshold value, judging that the node is an untrusted node, and judging that the node is a trusted node if the node is not an untrusted node;

the frog jump mixing algorithm comprises the following steps: 301, taking the importance of the nodes as a screening basis, screening the neighbor nodes by using a mixed frog leaping algorithm, wherein each frog individual is selectedCan represent the importance of a nodeAnd using formulasCalculating the fitness of the frog;

step 302, randomly initializing a frog group consisting of P frogs，i=1，2，……P；

Step 303, arranging the frogs in descending order according to the calculated fitness of each frog, and setting the frog individual with the optimal function value as the frog individual；

Step 304, divide the entire frog group into F groups, each group containing G frogs, thusThe first frog enters the second frog1 clan, the second frog enters the 2 nd clan, the F-th frog enters the F-th clan, the F + 1-th frog enters the 1 st clan, the F + 2-th frog enters the 2 nd clan, and so on until all the frogs are divided;

step 305, after the population division is completed, local deep search is performed on each population, and the individuals with the optimal and worst fitness in each population areAndeach iteration for worst fitnessThe update strategy is carried out as follows: frog travel distanceUpdating the worst frog position（) Wherein,is thatA random number in between, and a random number,the frog individual with the worst fitness in the clan is updated through the formula, and each clan executes the set local search times;

and step 306, merging the families subjected to the local depth search to form a new family, judging whether the termination condition of the algorithm is met, and finishing screening reliable neighbor nodes.

2. The internet of things node reputation evaluation method based on the mixed frog-leaping algorithm according to claim 1, characterized in that: the method for calculating the importance of the nodes in the step 1 comprises the following steps:

step 201, representing the network as a binary YK, EY, and the node set as a binary YKThe set of edges connecting the nodes is represented asWherein n and m represent the number of nodes and edges of the network, and the more important the edges connecting the nodes are, the more important the nodes are;

step 202, utilizing a formulaTo calculate the weight of the edge, whereinRepresenting the number of edges connected to node i;

step 203, summing the weights of the edges connected to the node iI.e. representing the weight of node i;

step 204, using formulaCalculating the importance of the node i, wherein。

3. The internet of things node reputation evaluation method based on the mixed frog-leaping algorithm according to claim 1, characterized in that: the reputation evaluation in step 4 specifically includes the following steps:

according to the nodes obtained after screening, taking the nodes as neighbor nodes at T_PIn a period, a plurality of neighbor nodes observe an evaluated node, each node has and maintains a neighbor node list of the node, the list comprises information such as neighbor node ID numbers and credit values, when a node i sends a data packet to a node n, an intermediate node j is required to forward the data packet, and the node calculates the node reliability through the existing monitoring conditions，The calculation formula of (a) is as follows:whereinIndicating the number of the data packets which are requested to be forwarded by the node j by the node i;denotes that j is the number of i forwarding packets, in cyclesAnd j forwards more data packets with higher reliability.

4. The internet of things node reputation evaluation method based on the mixed frog-leaping algorithm according to claim 3, characterized in that: in order to avoid providing too high floor by nodes with high reputation value and causing subjective deviation, a global reputation R is introduced as a parameter for reducing risk, and a specific formula for reducing the subjective deviation is as follows:

where i is a neighbor node of N, S_NIs a set of neighbor nodes for N,the parameters are introduced that the reputation value of the monitoring node is provided, T is a reliability threshold value, the node behavior below the threshold value can be considered as a bad nodeWhereinIn order to adjust the function of the adjustment,，M_ikiis the total amount of the transaction for node i,is based on the local reputation expectation of node i relative to its neighbor nodes, an。