A high-efficiency data fusion control method for heterogeneous wireless sensor networks based on fuzzy logic

Abstract

The present invention discloses a data fusion control method for heterogeneous wireless sensor networks based on fuzzy logic. The method comprises the following steps: first, at the source node level, the initial sensor data set is reduced in weight to remove redundancy and assigned weights; then, at the cluster head node level, the node type, node location, and the Pearson correlation coefficient of the sensor data set are considered, and data similarity is calculated through fuzzy decision making. The cluster head node makes a decision on whether to perform data fusion based on energy changes; and finally, the sensor data set is fused through weighted averaging. By applying fuzzy logic to the data fusion decision-making process, the present invention can efficiently perform data fusion. While reducing data transmission energy consumption, the method does not significantly increase data processing energy consumption by simplifying the rule table and calculation process in the fuzzy logic.

Description

Technical Field

[0001] The present invention belongs to the field of wireless sensor network data fusion, and in particular to a high-energy-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic. Background Art

[0002] Energy is a critical resource in WSNs, and achieving efficient energy utilization is a major challenge in this field. Fuzzy logic has been successfully applied to WSNs to address this issue, achieving promising results. Literature research indicates that much of the research focus has been on improving key WSN design factors, aiming to achieve more efficient energy use, which is crucial for extending the entire WSN lifecycle. Issues such as sensor node deployment, clustering, cluster head election, data fusion, and routing are all key factors that impact the WSN lifecycle and, consequently, the event detection and decision-making processes.

[0003] Mhemed et al. proposed a fuzzy logic cluster formation protocol, FLCFP, which uses a fuzzy logic inference system in the cluster formation process. Energy level, distance to the base station, and distance to the cluster head are defined as inputs to the fuzzy logic inference system, while opportunity is defined as the output. The authors demonstrated that using multiple parameters in cluster formation can reduce energy dissipation. BabuKaruppiah et al. proposed a data fusion-based data transmission protocol, FSPSDA, which uses fuzzy logic to find the shortest path. Parameters such as each node's trust level, power level, and distance to the cluster head are used as inputs to form fuzzy rules. The cluster head merges packets from the best and healthy nodes and rejects packets from the worst nodes. Bhushan et al. proposed a fuzzy attribute-based joint integrated scheduling and tree generation protocol, FAIJT, for fuzzy logic-based tree generation and parent node selection in heterogeneous networks. FAIJT primarily addresses the problem of selecting parent nodes for different node types in heterogeneous networks, merging different types of packets at the aggregation node to improve energy efficiency. The protocol prefers to select candidate nodes with the fewest dynamic neighbors, while fuzzy logic is applied when the number of dynamic neighbors is equal. However, the application of fuzzy logic in this work is limited to normalizing multiple inputs before making decisions. Consequently, the timing of data aggregation is inaccurate, and energy efficiency improvements are limited. The aforementioned protocol applies fuzzy logic to clustering, path generation, and data fusion decisions. While this application can improve network performance to a certain extent, its drawback is that excessive fuzzy rules complicate the computational process.

[0004] In summary, some rule simplification methods should be considered to reduce the complexity of rules, thereby better reducing energy consumption and extending the network life cycle, which has become an urgent problem to be solved in this field. Summary of the Invention

[0005] In order to solve the technical problems mentioned in the above background technology, the present invention proposes a high-energy-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic.

[0006] In order to achieve the above technical objectives, the technical solution of the present invention is:

[0007] A high-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic includes the following steps:

[0008] (1) The source node compares the similarities between the initial sensing data set, finds redundant sensing data, deletes them, and assigns weights to the deduplicated sensing data to obtain the deduplicated sensing data set;

[0009] (2) The cluster head node collects all the sensing data sets in the cluster in this round, and calculates the data similarity using fuzzy decision theory based on the sensing data sets and information such as the source nodes. When the data similarity is greater than the similarity threshold, it is determined that the sensing data collected in this round meets the data fusion conditions;

[0010] (3) When the collected sensor data meets the fusion conditions, the energy loss of transmitting the data without fusion and after fusion is calculated respectively. If the energy consumption of transmitting the data after fusion is significantly reduced, the cluster head node will fuse the given sensor data set by weighted averaging.

[0011] Furthermore, the specific process of step (1) is as follows:

[0012] (101) In the time slot j of a cycle, the source node node i The new sensory data perceived is recorded as a ij , after one cycle, node node i Form an initial sensing data set vector D i , first define a matching function to quantify the two sensing data a ij with a ik The degree of similarity or identity between them is defined as follows:

[0013]

[0014] Where δ is a threshold value, which is pre-set according to application requirements. Two data values ​​are redundant if and only if their matching function value is 1;

[0015] (102) Then, weights are assigned to the deduplicated sensing data, and the initial sensing data set D i The number of similar (according to the matching function) or identical data in the sensor data a is defined as ij The weight of wt(aij ) indicates that by comparing the similarities between the sensing data, if redundant sensing data is found, the new sensing data is deleted and the weight of the existing data is increased by 1. After deduplication of the initial sensing data set, the sensing data set D′ is obtained. i ;

[0016] In step (2), the data similarity calculation steps are as follows:

[0017] (201) Let the three input parameters in the fuzzy decision index set be the node position index I1 = s i , node type index I2 = c i and sensor data collection metrics s i is the source node i The relative position within the cluster is used to represent the normalized node position information and is defined by the following formula:

[0018]

[0019] Among them, d i For node i Euclidean distance to the cluster head, d min and d max are the distances between the closest and farthest nodes in the cluster, i.e., d min =min{d(i,CH)},i∈Nodes,d max =max{d(i,CH)},i∈Nodes,d MIN and d MAX are the normalized minimum and maximum relative positions within the cluster, respectively.

[0020] c i is the source node i The relative type is used to measure whether the sensing data of the current node can be effectively integrated with the sensing data of other nodes. It is described by the similarity between the mode of the current node and the rest of the nodes. The specific definition is as follows:

[0021]

[0022] Among them, type i For node i Type of sensing data, type m is the mode of the sensed data types in all source nodes.

[0023] ρ(D′ i ,D′ j ) are two sets of sensor data after deduplication D′ i and D′ jThe Pearson correlation coefficient is defined as follows:

[0024]

[0025] Among them, conv(D′ i ,D′ j )=E(D′ i D′ j )-E(D′ i )E(D′ j ) is the covariance of the two sensor data sets associated with the weights.

[0026] For the three linguistic variables, node location index I1, node type index I2, and sensor data set index I3, the corresponding comment sets {low, medium, high} are defined. The triangular membership function is used to obtain the membership output of the input set {I′1, I′2, I′3} parameters for the comment set, which is recorded as: J1 = (I 11 ,I 12 ,I 13 ), J2=(I 21 ,I 22 ,I 23 ), J3=(I 31 ,I 32 ,I 33 ), then construct the membership fuzzy matrix J = [J1, J2, J3];

[0027] (202) The weight matrix A is calculated by the AHP method, and then the weight vector X′ is calculated by the eigenvector method. A ;

[0028] (203) Comprehensively considering various factors, fuzzy comprehensive evaluation is applied to obtain a comprehensive data similarity evaluation vector. The M(·,+) weighted average model is applied:

[0029]

[0030] (204) Finally, the final evaluation value F of the data similarity is obtained by defuzzification using the centroid method. sim , the calculation formula is as follows:

[0031]

[0032] Among them, N is the number of evaluation factors, y i It refers to the fuzzy value of the i-th comment in the comment set, μ(y i ) is the corresponding membership degree, the comment set V C ={poor, average, excellent}, the fuzzy values ​​are as follows: {poor: 1; average: 3; excellent: 6}.

[0033] The specific process of step (3) is as follows:

[0034] (301) The set of sensing data received by the cluster head node is packet = D′1, D′2, …, D′ k , calculate the energy consumption E of non-fusion transmission data all And the energy consumption of data transmission after fusion E agg , the formula is as follows:

[0035]

[0036] Among them, E T (m, d) is the energy consumed by a node to send m bits of data to another node with a distance d from it, E A (m) is the energy consumed by the node to fuse m bits of data.

[0037] (302) Then through E all and E agg Calculate the fusion judgment factor γ, which is defined as follows:

[0038]

[0039] Among them, E CH is the residual energy of the cluster head node. When the energy consumption reduced after data fusion reaches a certain proportion of the residual energy of the cluster head node, data fusion can be performed as soon as possible to reduce the amount of data transmitted in the network. That is, when λ>ρ, the cluster head node automatically fuses the data and then transmits it, where ρ=0.1.

[0040] (303) Determine all sensor data sets that can be aggregated, denoted as P = {D′1, D′2, … D′ k In the previous steps, the weights of each sensor data have been defined and calculated. When performing data fusion, the weighted average of all sensor data is obtained through the weights, which is the final data fusion result. The specific definition is as follows:

[0041]

[0042] When the cluster head node completes the data fusion of all the sensing data sets, it obtains the data packet packe′={data1,data2,…,data i}, and then the CH transmits the packet to the base station.

[0043] The beneficial effects brought about by adopting the above technical solution are:

[0044] (1) The present invention applies fuzzy logic to the data fusion decision process, taking node location, node type, and sensor data as inputs, simplifies fuzzy rules, calculates data similarity through fuzzy logic, and then completes the decision of whether to perform data fusion within controllable data processing energy consumption;

[0045] (2) The heterogeneity of energy and type of sensor nodes is taken into account at the same time. The sensed data is reduced and weighted at the source node level. The heterogeneity of node types is considered at the CH level through fuzzy logic, and the sensed data is safely and effectively fused. Data fusion is completed as early as possible while ensuring the correctness of the data, thereby effectively reducing the amount of data transmitted in the network. The diversity of nodes also makes the present invention more adaptable to complex practical application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a schematic diagram of a data sensing cycle of the present invention;

[0047] Figure 2 It is a schematic diagram of the source node data deduplication algorithm of the present invention;

[0048] Figure 3 It is a data similarity calculation flow chart of the present invention;

[0049] Figure 4 Schematic diagram of the triangular membership function of the present invention;

[0050] Figure 5 It is a schematic diagram of the data fusion process of heterogeneous nodes of the present invention. DETAILED DESCRIPTION

[0051] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0052] A high-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic includes the following steps:

[0053] Step 1: The source node compares the similarities between the initial sensing data set, finds redundant sensing data, deletes them, and assigns weights to the deduplicated sensing data to obtain the deduplicated sensing data set;

[0054] Step 2: The cluster head node collects all the sensor data sets in the cluster in this round, and calculates the data similarity using fuzzy decision theory based on the sensor data sets and information such as the source nodes. When the data similarity is greater than the similarity threshold, it is determined that the sensor data collected in this round meets the data fusion conditions;

[0055] Step 3: When the collected sensor data meets the fusion conditions, the energy loss of transmitting the data without fusion and after fusion is calculated respectively. If the energy consumption of transmitting the data after fusion is significantly reduced, the cluster head node will fuse the given sensor data set by weighted averaging.

[0056] In this embodiment, the above step 1 can be implemented by adopting the following preferred solution:

[0057] 101. Figure 1 As shown, in the time slot j of a cycle, the source node node i The new sensory data perceived is recorded as a ij , after one cycle, node node i Form an initial sensing data set vector D i , first define a matching function to quantify the two sensing data a ij with a ik The degree of similarity or identity between them is defined as follows:

[0058]

[0059] Where δ is a threshold value, which is pre-set according to application requirements. Two data values ​​are redundant if and only if their matching function value is 1;

[0060] 102. Then, weights are assigned to the deduplicated sensing data, and the initial sensing data set D i The number of similar (according to the matching function) or identical data in the sensor data a is defined as ij The weight of wt(a ij ) indicates that by comparing the similarities between the sensing data, if redundant sensing data is found, the new sensing data is deleted and the weight of the existing data is increased by 1, as shown in Figure 2 As shown, after deduplication of the initial sensing data set, the sensing data set D′ is obtained. i ;

[0061] In this embodiment, the following preferred solution can be used to implement the above step 2: Figure 3 As shown:

[0062] 201. Let the three input parameters in the fuzzy decision indicator set be the node position index I1 = s i , node type index I2 = c i and sensor data collection metrics s i is the source node i The relative position within the cluster is used to represent the normalized node position information and is defined by the following formula:

[0063]

[0064] Among them, d i For node i Euclidean distance to the cluster head, d min and d max are the distances between the closest and farthest nodes in the cluster, i.e., d min =min{d(i,CH)},i∈Nodes,d max =max{d(i,CH)},i∈Nodes,d MIN and d MAX are the normalized minimum and maximum relative positions within the cluster, respectively.

[0065] c i is the source node i The relative type is used to measure whether the sensing data of the current node can be effectively integrated with the sensing data of other nodes. It is described by the similarity between the mode of the current node and the rest of the nodes. The specific definition is as follows:

[0066]

[0067] Among them, type i For node i Type of sensing data, type m is the mode of the sensed data types in all source nodes.

[0068] ρ(D′ i ,D′ j ) are two sets of sensor data after deduplication D′ i and D′ j The Pearson correlation coefficient is defined as follows:

[0069]

[0070] Among them, cov(D′ i ,D′ j )=E(D′ i D′ j )-E(D′ i )E(D′ j ) is the covariance of the two sensor data sets associated with the weights.

[0071] For the three linguistic variables, node location index I1, node type index I2, and sensor data set index I3, the corresponding comment sets {low, medium, high} are defined. The triangular membership function is used to obtain the membership output of the input set {I′1, I′2, I′3} parameters for the comment set, which is recorded as: J1 = (I 11,I 12 ,I 13 ), J2=(I 21 ,I 22 ,I 23 ), J3=(I 31 ,I 32 ,I 33 ), then construct the membership fuzzy matrix J = [J1, J2, J3], the triangular membership function is as follows Figure 4 As shown;

[0072] 202. Calculate the weight matrix A by AHP, and then calculate the weight vector X′ by eigenvector method A ;

[0073] 203. Comprehensively consider all factors and apply fuzzy comprehensive evaluation to obtain a comprehensive data similarity evaluation vector. Apply the M(·,+) weighted average model:

[0074]

[0075] 204. Finally, the final evaluation value F of the data similarity is obtained by defuzzification using the centroid method. sim , the calculation formula is as follows:

[0076]

[0077] Among them, N is the number of evaluation factors, y i It refers to the fuzzy value of the i-th comment in the comment set, μ(y i ) is the corresponding membership degree, the comment set V C ={poor, average, excellent}, the fuzzy values ​​are as follows: {poor: 1; average: 3; excellent: 6}.

[0078] In this embodiment, the following preferred solution can be used to implement the above step 3:

[0079] 301. The sensing data set received by the cluster head node is packet = D′1, D′2, …, D′ k , calculate the energy consumption E of non-fusion transmission data all And the energy consumption of data transmission after fusion E agg , the formula is as follows:

[0080]

[0081] Among them, E T (m, d) is the energy consumed by a node to send m bits of data to another node with a distance d from it, E A (m) is the energy consumed by the node to fuse m bits of data.

[0082] 302, then through E all and E agg Calculate the fusion judgment factor γ, which is defined as follows:

[0083]

[0084] Among them, E CH is the residual energy of the cluster head node. When the energy consumption reduced after data fusion reaches a certain proportion of the residual energy of the cluster head node, data fusion can be performed as soon as possible to reduce the amount of data transmitted in the network. That is, when λ>ρ, the cluster head node automatically fuses the data and then transmits it, where ρ=0.1.

[0085] 303. Determine all sensor data sets that can be aggregated, denoted as P = {D′1, D′2, … D′ k In the previous steps, the weights of each sensor data have been defined and calculated. When performing data fusion, the weighted average of all sensor data is obtained through the weights, which is the final data fusion result. The specific definition is as follows:

[0086]

[0087] When the cluster head node completes the data fusion of all the sensing data sets, it obtains the data packet packe′={data1,data2,…,data i}, then CH transmits the packet to the base station, such as Figure 5 shown.

Claims

1. A high-efficiency data fusion control method for heterogeneous wireless sensor networks based on fuzzy logic, characterized by: The following steps are involved: (1) The source node compares the similarities between the initial sensing data set, finds redundant sensing data, deletes them, and assigns weights to the deduplicated sensing data to obtain the deduplicated sensing data set; (2) The cluster head node collects all the sensing data sets in the cluster in this round, and calculates the data similarity using fuzzy decision theory based on the sensing data sets and information such as the source nodes. When the data similarity is greater than the similarity threshold, it is determined that the sensing data collected in this round meets the data fusion conditions; (3) When the collected sensor data meets the fusion conditions, the energy loss of transmitting the data without fusion and after fusion is calculated respectively. If the energy consumption of transmitting the data after fusion is significantly reduced, the cluster head node will fuse the given sensor data set by weighted averaging.

2. The high-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic according to claim 1 is characterized in that: The specific process of step (1) is as follows: (101) In the time slot j of a cycle, the source node node i The new sensory data perceived is recorded as a ij , after one cycle, node node i Form an initial sensing data set vector D i , first define a matching function to quantify the two sensing data a ij with a ik The degree of similarity or identity between them is defined as follows: Where δ is a threshold value, which is pre-set according to application requirements. If and only if the matching function value of two data values ​​is 1, they are redundant; (102) Then, weights are assigned to the deduplicated sensing data, and the initial sensing data set D i The number of similar (according to the matching function) or identical data in the sensor data a is defined as ij The weight of wt(a ij ) indicates that by comparing the similarities between the sensing data, if redundant sensing data is found, the new sensing data is deleted and the weight of the existing data is increased by 1. After deduplication of the initial sensing data set, the sensing data set D′ is obtained. i。 3. The high-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic according to claim 1 is characterized in that: The steps for calculating data similarity are as follows: (201) Let the three input parameters in the fuzzy decision index set be the node position index I1 = s i , node type index I2 = c i and sensor data collection metrics s i is the source node i The relative position within the cluster is used to represent the normalized node position information and is defined by the following formula: Among them, d i For node i Euclidean distance to the cluster head, d min and d max are the distances between the closest and farthest nodes in the cluster, i.e., d min =min{d(i,CH)},i∈Nodes,d max =max{d(i,CH)},i∈Nodes,d MIN and d MAX are the normalized minimum and maximum relative positions within the cluster, c i is the source node i The relative type is used to measure whether the sensing data of the current node can be effectively integrated with the sensing data of other nodes. It is described by the similarity between the mode of the current node and the rest of the nodes. The specific definition is as follows: Among them, type i For node i Type of sensing data, type m is the mode of the sensed data types in all source nodes, ρ(D′ i ,D′ j ) are two sets of sensor data after deduplication D′ i and D′ j The Pearson correlation coefficient is defined as follows: Among them, cov(D′ i ,D′ j )=E(D′ i D′ j )-E(D′ i )E(D′ j ) is the covariance of the two sensor data sets associated with the weights, For the three linguistic variables, node location index I1, node type index I2, and sensor data set index I3, the corresponding comment sets {low, medium, high} are defined. The triangular membership function is used to obtain the membership output of the input set {I′1, I′2, I′3} parameters for the comment set, which is recorded as: J1 = (I 11 ,I 12 ,I 13 ), J2=(I 21 ,I 22 ,I 23 ), J3=(I 31 ,I 32 ,I 33 ), then construct the membership fuzzy matrix J = [J1, J2, J3]; (202) The weight matrix A is calculated by the AHP method, and then the weight vector X′ is calculated by the eigenvector method. A ; (203) Comprehensively considering various factors, fuzzy comprehensive evaluation is applied to obtain a comprehensive data similarity evaluation vector, and the M(·,+) weighted average model is applied: (204) Finally, the final evaluation value F of the data similarity is obtained by defuzzification using the centroid method. sim , the calculation formula is as follows: Among them, N is the number of evaluation factors, y i It refers to the fuzzy value of the i-th comment in the comment set, μ(y i ) is the corresponding membership degree, the comment set V C ={poor, average, excellent}, the fuzzy values ​​are as follows: {poor: 1; average: 3; excellent: 6}.

4. The high-efficiency heterogeneous wireless sensor network data fusion control method based on fuzzy logic according to claim 1 is characterized in that: The specific process of step (3) is as follows: (301) The set of sensing data received by the cluster head node is packet = D′1, D′2, …, D′ k , calculate the energy consumption E of non-fusion transmission data all And the energy consumption of data transmission after fusion E agg , the formula is as follows: Among them, E T (m, d) is the energy consumed by a node to send m bits of data to another node with a distance d from it, E A (m) is the energy consumed by the node to fuse m bits of data, (302) Then through E all and E agg Calculate the fusion judgment factor γ, which is defined as follows: Among them, E CH is the remaining energy of the cluster head node. When the energy consumption reduced after data fusion reaches a certain proportion of the remaining energy of the cluster head node, data fusion can be performed as soon as possible to reduce the amount of data transmitted in the network. That is, when λ>ρ, the cluster head node automatically fuses the data and then transmits it. Let ρ = 0.

1. (303) Determine all sensor data sets that can be aggregated, denoted as P = {D′1, D′2, … D′ k In the previous steps, the weights of each sensor data have been defined and calculated. When performing data fusion, the weighted average of all sensor data is obtained through the weights, which is the final data fusion result. The specific definition is as follows: When the cluster head node completes the data fusion of all the sensing data sets, it obtains the data packet packe′={data1,data2,…,data i }, and then the CH transmits the packet to the base station.

Images