SP-DPC clustering-based high-energy-efficiency WSN topology control method

Abstract

The invention discloses a high energy efficiency WSN topology control method based on SP-DPC clustering. The method is based on an SP-DPC clustering algorithm, SP-DPC is an algorithm based on density peak clustering, the problem that DPC-MND is poor in performance when facing manifold cluster data sets is solved, meanwhile, the thought of transmission probability allocation is introduced, and associated allocation errors when non-cluster-head nodes are classified into clusters are prevented. Meanwhile, during cluster head election, the distance from the elected node to the base station is used as a dynamic factor, and the problem of energy holes around the base station is solved. In the routing path design stage, unified calculation is carried out by the base station. A path weight matrix is constructed by integrating factors such as node energy and distance, and an optimal path is selected through a Dijkstra algorithm. Compared with other protocols, the protocol provided by the invention has remarkable performance improvement in the aspect of energy efficiency.

Description

Technical Field

[0001] The invention belongs to the field of wireless sensor network routing security, in particular to a high-energy-efficiency WSN topology control method based on SP-DPC clustering. Background Art

[0002] When planning nodes and selecting routes, traditional network architectures often prioritize end-to-end latency, overall network throughput, and data security, while rarely considering energy consumption as a key consideration. This means that each sensor node must work together to not only accurately capture and process the monitored environmental information, but also ensure that this data can be transmitted to the base station in an efficient and highly reliable manner.

[0003] According to the research review by Dai S et al., cluster routing protocols perform better than flat routing protocols in terms of energy consumption, reliability and other aspects in large-scale wireless sensor networks. The design goal of cluster routing protocols is to select cluster head nodes by designing excellent clustering algorithms. Cluster head nodes take on the responsibility of data forwarding, thereby reducing redundant data transmission and reducing the number of communications. Finally, a reasonable routing algorithm is designed, that is, the communication path of cluster head nodes between clusters.

[0004] The most classic clustering routing protocol is the LEACH protocol, which has epoch-making research significance. It is a self-organizing clustering algorithm. However, this protocol is randomly selected, which will lead to excessive energy consumption in large-scale scenarios. Zhao Jia proposed to use the DPC-MND algorithm based on mutual proximity to cluster and reduce node energy consumption. This is a density-based algorithm. It has made certain improvements in energy consumption, but there are still problems such as energy holes. Many scholars have proposed a variety of improved density peak clustering algorithms. Chen Lei, Zhang Xinyuan and Gong C tried to combine the K nearest neighbor algorithm with the density peak algorithm. Ma Zhenming and Hou J incorporated the mutual proximity idea into the density peak algorithm. The SP-DPC (Density Peaks Clustering Algorithm Based On Shared Neighbor Degree And Probability Assignment) clustering algorithm combines their advantages. Compared with other algorithms, it is more accurate and can accurately classify clusters of various shapes. At the same time, it solves the problem that the density peak algorithm is prone to joint allocation errors.

[0005] In summary, the design of secure topology control methods has become a research hotspot in wireless sensor networks in recent decades. How to balance the energy consumption of nodes and take into account the security of data transmission has become an urgent problem to be solved in this field. Summary of the invention

[0006] To solve the technical problems mentioned in the above background art, the present invention proposes an energy-efficient WSN topology control method based on SP-DPC clustering.

[0007] To achieve the above technical objectives, the technical solution of the present invention is as follows:

[0008] An energy-efficient WSN topology control method based on SP-DPC clustering includes the following steps:

[0009] (1) The clustering algorithm uses the SP-DPC algorithm. First, determine the number of clusters in advance, and allocate cluster center nodes through the algorithm.

[0010] (2) For non-cluster center nodes, a two-step allocation method is used to solve the allocation associated error problem of the DPC-MND algorithm. In the first step, cluster core sample points are allocated by means of transfer probability allocation. In the second step, the unallocated sample points are added to the nearest cluster.

[0011] (3) In the cluster head selection stage, consider the communication energy consumption between the cluster head node and the members within the cluster and the energy loss of wireless communication to construct a cluster head election factor to select the cluster head.

[0012] (4) In the designed routing stage, according to the different positions of the nodes, construct a path cost decision matrix, and calculate the path with the highest weighted value as the routing path through the Dijkstra single-source shortest path algorithm.

[0013] Further, in step (1), the method for allocating cluster center nodes is as follows:

[0014] (101) Input the node set Nodes, the neighborhood number k, and the number of clusters M.

[0015] (102) Normalize the entire data set and calculate the distance matrix.

[0016] (103) Calculate the shared neighborhood matrix Snd set, and the calculation formula is as follows:

[0017]

[0018] Where |SNN(x i ,x j )| is the number of shared neighbors between nodes x i , x j , d ij represents the distance from node x i to x j , and represents the sum of the distances from the sensor node x j and its K nearest neighbors to x i . The introduction of shared proximity makes clustering more accurate.

[0019] (104) Calculate the local density of each sensor node and the relative distance The calculation formula is as follows:

[0020]

[0021] where X is the set of all sensor nodes. is the sum of the shared proximities of sensor node x i with all sensor nodes (including itself).

[0022]

[0023] δ i refers to the relative distance between node x i and each of the other nodes, which is defined as the relative energy distance in WSN.

[0024] (105) Calculate the evaluation value of each sensor node Sort the comprehensive evaluation values in descending order. Select the top M large values of the cluster numbers as the density peak set S. The evaluation value The calculation formula is as follows:

[0025]

[0026] where E resi is the remaining energy of sensor node x i , d itosink is the distance from x i to the base station. The higher d itosink , the lower the value of the decision value.

[0027] Furthermore, in step (2), the allocation method of non-cluster center nodes is as follows:

[0028] (201) Initialize an empty queue Q and add the nodes in the peak density node set S to the queue

[0029] (202) Take out and delete the first element first from the queue

[0030] (203) Traverse each neighbor node near in the K-nearest neighbor set KNN(first) of the first element. If near has no cluster label and first meets the conditions for transfer probability allocation to the near node, then assign the near node to the cluster where the first node is located and add near to the queue Q. If not, traverse the next neighbor node. The calculation formulas for the transfer probability and the transfer probability allocation method are as follows:

[0031]

[0032] Among them, Lb(x j ) is the cluster where node x j is located. If it is -1, no transfer probability assignment is performed. Otherwise, it is judged whether to assign according to the formula. K is the number of neighbors, and n is the total number of sensor nodes. X is the set of all sensor nodes.

[0033] (204) When the queue Q is empty, end and output the result. Otherwise, jump to step (202).

[0034] (205) Calculate for each node outside the set C and add it to the cluster where the peak node with the largest relative proximity to it in the density peak set S is located.

[0035] (206) Output the clustering result

[0036] Furthermore, in step (3), the construction method of the cluster head election factor is as follows:

[0037] (301) In order to reduce the communication energy consumption between the cluster head node and the members within the cluster and considering that the energy loss of wireless communication is proportional to the fourth power of the transmission distance, the cluster head election factor is calculated according to the following formula:

[0038]

[0039] Among them, a, b, c are weight coefficients, E res is the remaining energy of the sensor, is the average value of the remaining energy of the sensors within the cluster. is the average value of the distance from the node to the base station, d toBS is the distance from the node to the base station. is the average value of the distance from the node to the selected cluster center node in the above text, d cluster is the distance from the node to the selected cluster center node in the above text. Among them, a + b + c = 1, and the weights are set as a = 0.5, b = 0.25, c = 0.25.

[0040] Furthermore, in step (4), the method of designing the routing is as follows:

[0041] (401) This protocol uses the multi-hop routing method to be transmitted to the base station. The base station is set at the center of the square area, and concentric circles are drawn around the base station. The distance between the arcs is set as d, and the area is divided into {L1, L2,..., Ln} layers from the inside to the outside in turn. The cluster head node is k intermediate areas away from the aggregation node, and the distance between each layer is fixed, which is d.

[0042] (402) According to the first-order radio model, the node directly transmits a data packet of length n to the base station in a single-hop manner, and the energy consumption can be expressed by the following formula:

[0043] E sg = E elec ·n + E mp ·n·(k·d) 4 #(4 - 1)

[0044] If the cluster head node reaches the sink node through m-hop forwarding, then this data packet is forwarded m - 1 times. Therefore, the energy consumption for multi-hop reaching the base station is as follows:

[0045] E mp = m·E elec ·n + m·E fs ·n·d 2 + (m - 1)·E elec ·n #(4 - 2)

[0046] (403) To determine the value range of d, it is necessary to satisfy that the energy consumption of single-hop is higher than that of multi-hop:

[0047] E mg < E sg #(4 - 3)

[0048] E elec ·n + E mp ·n·k 4 ·d 4 > m·E elec ·n + m·E fs ·n·d 2 + (m - 1)·E elec ·n #(4 - 4)

[0049] E mp 、E fs and E elec are fixed values in a radio model. Since m-hop is multi-hop, m should be greater than or equal to 2. Therefore, substituting m = 500 / (2*d) into formula (3 - 21), the value range of d is obtained:

[0050]

[0051] (404) Construct a routing cost decision matrix D[i][j] to represent the weight between two nodes. i, j = 0, 1, 2,..., n. Each node records its own level. It is stipulated that the cluster head node transmits data only in the direction facing the base station. The path weight between each node is calculated by the following formula:

[0052]

[0053] Among them, is the energy threshold in the energy model, and d is the distance from node i to j. E resj is the remaining energy of node j. a, b, c, d, and m, n, p, q are all weight coefficients. θ is the angle between sensor node i and j, and α is the angle between sensor node j and the base station. level is the hierarchical label of each sensor node, recording the layer where the node is located. The cluster head node transfers data from the outermost layer to the innermost layer, and selects the node with the highest decision value as the next-hop routing node.

[0054] (405) Use the Dijkstra algorithm to generate the shortest path. Starting from the starting node M, find the node N that is closest to M and has not been processed yet, and set it as the current relay node;

[0055] (406) Check all other nodes connected through node N. If it is found that the total distance to these nodes via node N is less than the distance directly from M, update the shortest path estimate values of these nodes;

[0056] (407) Add node N to the set of visited nodes, indicating that the shortest path of node N has been considered;

[0057] (408) Execute step (405), continue to select the next closest unprocessed node as the new relay node, and repeat the above process until all nodes are included in the visited set.

[0058] Beneficial effects brought by adopting the above technical solutions:

[0059] (1) The SP-DPC proposed by the present invention is an algorithm based on density peak clustering, which solves the problem that DPC-MND performs poorly in the face of manifold-like cluster datasets. At the same time, the idea of transfer probability assignment is introduced to prevent the associated assignment error when non-cluster head nodes are classified into clusters. In addition, when electing cluster heads, the distance from the elected node to the base station is also used as a dynamic factor to solve the problem of energy holes around the base station.

[0060] (2) In the stage of designing the routing path, unified calculation is performed by the base station. A path weight matrix is constructed by comprehensively considering factors such as node energy and distance, and the optimal path is selected through the Dijkstra algorithm, achieving a significant improvement in energy efficiency. Description of the Drawings

[0061] Figure 1 is the network clustering flowchart of the present invention;

[0062] Figure 2 is the hierarchical routing structure design diagram of the present invention;

[0063] Figure 3It is a conceptual diagram of the shared proximity of the present invention; Detailed implementation manners

[0064] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0065] An energy-efficient WSN topology control method based on SP-DPC clustering includes the following steps:

[0066] Step 1: The clustering algorithm adopts the SP-DPC algorithm. First, determine the number of clusters in advance, and allocate the cluster center nodes through the algorithm.

[0067] Step 2: For non-cluster center nodes, a two-step allocation method is adopted to solve the allocation associated error problem of the DPC-MND algorithm. In the first step, the cluster core sample points are allocated by the way of transfer probability, and in the second step, the unallocated sample points are added to the nearest cluster.

[0068] Step 3: In the cluster head selection stage, consider the communication energy consumption between the cluster head node and the cluster members and the energy loss of wireless communication to construct a cluster head election factor to select the cluster head.

[0069] Step 4: In the designed routing stage, according to the different positions of the nodes, construct a path cost decision matrix, and calculate the path with the highest weighted value as the routing path through the Dijkstra single-source shortest path algorithm.

[0070] In this embodiment, the above step 1 can be implemented by adopting the following preferred scheme:

[0071] 101. Input the node set Nodes, the proximity number k, and the number of clusters M

[0072] 102. Normalize the entire data set and calculate the distance matrix.

[0073] 103. Calculate the shared proximity matrix Snd set, and the calculation formula is as follows:

[0074]

[0075] Where |SNN(x i , x j )| is the number of shared neighbors between nodes x i , x j , d ij represents the distance from node x i to x j , represents the sum of the distances from the sensor node x j and its K nearest neighbors to x i . The introduction of shared proximity makes the clustering more accurate. The conceptual diagram of the shared proximity is asFigure 3 as shown

[0076] 104. Calculate the local density of each sensor node and the relative distance The calculation formula is as follows:

[0077]

[0078] where X is the set of all sensor nodes. is the sum of the shared proximities of sensor node x i with all sensor nodes (including itself).

[0079]

[0080] δ i refers to the relative distance between node x i and each of the remaining nodes, which is defined as the relative energy distance in WSN.

[0081] 105. Calculate the evaluation value of each sensor node Sort the comprehensive evaluation values in descending order. Select the top M large values of the number of clusters as the density peak set S. The evaluation value The calculation formula is as follows:

[0082]

[0083] where E resi is the remaining energy of sensor node x i and d itosink is the distance from x i to the base station. The higher d itosink , the lower the value of the decision value.

[0084] In this embodiment, the above step 2 can be implemented by the following preferred scheme:

[0085] 201. Initialize an empty queue Q, and add the nodes in the peak density node set S to the queue. 202. Take out and delete the head element first from the queue

[0086] 203. Traverse each neighbor node near in the K-nearest neighbor set KNN(first) of the first element. If near has no cluster label and first satisfies the condition of transfer probability assignment for the near node, then assign the near node to the cluster where the first node is located and add near to the queue Q. If not, traverse the next neighbor node. The calculation formulas for the transfer probability and the transfer probability assignment method are as follows:

[0087]

[0088] Among them, Lb(x j ) is the cluster where node x j is located. If it is -1, no transfer probability assignment is performed. Otherwise, it is judged whether to assign according to the formula. K is the number of neighbors, and n is the total number of sensor nodes. X is the set of all sensor nodes.

[0089] 204. When the queue Q is empty, end and output the result. Otherwise, jump to step (202).

[0090] 205. Calculate for each node outside the set C and add it to the cluster where the peak node with the largest relative proximity to it in the density peak set S is located.

[0091] 206. Output the clustering result

[0092] In this embodiment, the above step 3 can be implemented by the following preferred scheme:

[0093] 301. In order to reduce the communication energy consumption between the cluster head node and the cluster members and considering that the energy loss of wireless communication is proportional to the fourth power of the transmission distance, the cluster head election factor is calculated according to the following formula:

[0095]

[0096] Among them, a, b, and c are weight coefficients, E res is the remaining energy of the sensor, is the average value of the remaining energy of the sensors in the cluster. is the average value of the distance from the node to the base station, d toBS is the distance from the node to the base station. is the average value of the distance from the node to the selected cluster center node in the above text, d cluster is the distance from the node to the selected cluster center node in the above text. Among them, a + b + c = 1, and the weights are set as a = 0.5, b = 0.25, and c = 0.25.

[0097] In this embodiment, the above step 4 can be implemented by the following preferred scheme:

[0098] 401. There are usually two ways for the cluster head node to reach the base station, one is single-hop and the other is multi-hop. This protocol uses the multi-hop routing method to be transmitted to the base station. The base station is set in the center of the square area, and concentric circles are drawn around the base station. The distance between the arcs is set to d, and the area is divided into {L1, L2,..., Ln} layers from the inside to the outside in turn, as Figure 2As shown in the figure, the sensor nodes in the figure are transmitted from the outermost layer to the innermost layer. The nodes in each cluster first perform data aggregation and forward all the data to the cluster head. The cluster head communicates with the base station. The distance between the cluster head node and the aggregation node is k intermediate regions, and the distance between each layer is fixed at d.

[0099] 402. According to the first-order radio model, the node directly transmits a data packet of length n to the base station in a single-hop form, and the energy consumption can be expressed by the following formula:

[0100] E sg = E elec ·n + E mp ·n·(k·d) 4 #(4 - 1)

[0101] If the cluster head node reaches the aggregation node through m hops of forwarding, then this data packet is forwarded m - 1 times. Therefore, the energy consumption for reaching the base station through multiple hops is as follows:

[0102] E mp = m·E elec ·n + m·E fs ·n·d 2 + (m - 1)·E elec ·n#(4 - 2)

[0103] 403. To determine the value range of d, it is necessary to satisfy that the energy consumption of a single hop is higher than that of multiple hops:

[0104] E mg < E sg #(4 - 3)

[0105] E elec ·n + E mp ·n·k 4 ·d 4 > m·E elec ·n + m·E fs ·n·d 2 + (m - 1)·E elec ·n#(4 - 4)

[0106] E mp 、E fs and E elec are fixed values in a radio model. Since m hops are multiple hops, m should be greater than or equal to 2. Therefore, substituting m = 500 / (2*d) into formula (3 - 21), the value range of d is obtained:

[0107]

[0108] 404. Construct a routing cost decision matrix D[i][j] to represent the weight between two nodes. i, j = 0, 1, 2,..., n. Each node records its own level. It is stipulated that the cluster head node transmits data only in the direction facing the base station. The path weight between each node is calculated by the following formula:

[0109]

[0110] Among them, is the energy threshold in the energy model, d is the distance from node i to j. E resj is the remaining energy of node j. a, b, c, d and m, n, p, q are all weight coefficients. θ is the angle between sensor node i and j, and α is the angle between sensor node j and the base station. level is the level label of each sensor node, recording the layer where the node is located. The cluster head node transfers data from the outermost layer to the innermost layer, and selects the node with the highest decision value as the next-hop routing node.

[0111] 405. Use the Dijkstra algorithm to generate the shortest path. Starting from the starting node M, find the node N that is closest to M and has not been processed, and set it as the current relay node;

[0112] 406. Check all other nodes connected through node N. If it is found that the total distance to these nodes via node N is less than the distance directly from M, then update the shortest path estimate value of these nodes;

[0113] 407. Add node N to the set of visited nodes, indicating that the shortest path of node N has been considered;

[0114] 408. Execute step 405, continue to select the next closest unprocessed node as the new relay node, and repeat the above process until all nodes are included in the visited set.

Claims

1. A high-energy-efficiency WSN topology control method based on SP-DPC clustering, characterized in that: Bag The following steps are included: (1) The clustering algorithm uses the SP-DPC algorithm. First, the number of clusters is determined in advance, and the cluster center nodes are allocated through the algorithm. The specific steps are as follows: (101) Input node set Nodes, number of neighbors k, number of clusters M (102) The entire data set is normalized and the distance matrix is ​​calculated. (103) Calculate the shared proximity matrix Snd set, the calculation formula is as follows: Where |SNN(x i ,x j )| is node x i , x j The number of shared neighbors between ij Represents node x i to x j The distance Represents sensor node x j and its K nearest neighbors to x i The introduction of shared proximity makes clustering more accurate. (104) Calculate the local density of each sensor node and relative distance The calculation formula is as follows: Where X is the set of all sensor nodes. is the sensor node x i The sum of shared proximity with all sensor nodes (including itself). δ i refers to node x i The relative distance to other nodes is defined as the relative energy distance in WSN. (105) Calculate the evaluation value of each sensor node Sort the comprehensive evaluation values ​​in descending order. Select the top M values ​​of the cluster number as the density peak set S. Evaluation value The calculation formula is as follows: Where E resi For sensor node x i The remaining energy, d itosink For x i The distance to the base station. itosink The higher it is, the lower the numerical value of the decision value. (2) For non-cluster center nodes, a two-step allocation method is used to solve the allocation error problem of the DPC-MND algorithm. The first step is to allocate cluster core sample points by transferring probability allocation, and the second step is to add the unallocated sample points to the nearest cluster. The specific steps are as follows: (201) Initialize an empty queue Q and add the nodes in the peak density node set S to the queue. (202) Queue out and delete the first element first (203) Traverse each neighbor node near in the K nearest neighbor set KNN(first) of the first element. If near has no cluster label and first satisfies the condition of transfer probability allocation to the near node, then assign the near node to the cluster where the first node is located and add near to the queue Q. If it does not meet the condition, traverse the next neighbor node. The calculation formulas of transfer probability and transfer probability allocation method are as follows: Among them, Lb(x j ) is the node x j If the cluster is -1, no transmission probability allocation is performed. Otherwise, it is determined whether to allocate according to the formula. K is the number of neighbors, n is the total number of sensor nodes, and X is the set of all sensor nodes. (204) When the queue Q is empty, end and output the result, otherwise jump to step (202). (205) Calculate each node outside the set C and add it to the cluster where the peak node with the greatest relative proximity to it in the density peak set S is located. (206) Output clustering results (3) In the cluster head selection stage, the cluster head election does not use the cluster center of the density peak algorithm. In order to reduce the communication energy consumption between the cluster head node and the members in the cluster and considering that the energy loss of wireless communication is proportional to the fourth power of the transmission distance, the cluster head election factor is calculated according to the following formula: Where a, b, c are weight coefficients, E res is the remaining energy of the sensor, is the mean residual energy of the sensors in the cluster. is the mean distance from the node to the base station, d toBS is the distance from the node to the base station. is the mean distance from the node to the cluster center node selected above, d cluster is the distance from the node to the center node of the cluster selected above, where a+b+c=1, and the weights are set to a=0.5, b=0.25, and c=0.25 respectively. (4) In the routing design phase, a path cost decision matrix is ​​constructed based on the different locations of the nodes, and the path with the highest weight is calculated using the Dijkstra single-source shortest path algorithm as the routing path. The specific steps are as follows: (401) This protocol uses multi-hop routing to transmit to the base station. The base station is set at the center of the square area and concentric circles are drawn around the base station. The distance between arcs is set to d. The area is divided into {L1, L2, ..., Ln} layers from the inside to the outside. The cluster head node is k intermediate areas away from the sink node. The distance between each layer is fixed, which is d. (402) According to the first-order radio model, a node directly transmits a data packet of length n to a base station in a single hop. The energy consumption can be expressed by the following formula: E sg =E elec ·n+E mp ·n·(k·d) 4 # (4-1) If the cluster head node reaches the sink node through m hops, then the data packet is forwarded m-1 times, so the energy consumption of multiple hops to reach the base station is as follows: E mp =m·E elec ·n+m·E fs ·n·d 2 +(m-1)·E elec ·n# (4-2) (403) In order to determine the value range of d, it is necessary to satisfy that the energy consumption of a single hop is higher than that of multiple hops: AND mg <E sg # (4-3) E elec ·n+E mp ·n·k 4 ·d 4 >m·E elec ·n+m·E fs ·n·d 2 +(m-1)·E elec ·n# (4-4) E mp 、E fs and E elec In the radio model of Section 1, it is a fixed value. Since m hops are multi-hop, m should be greater than or equal to 2. Therefore, m = 500 / (2*d) is substituted into formula (4-5) to obtain the value range of d: (404) Construct a routing cost decision matrix D[i][j] to represent the weight between two nodes. i,j = 0, 1, 2, ..., n. Each node records its own level, and stipulates that the cluster head node only transmits data in the direction facing the base station. The path weight between each node is calculated by the following formula: in, is the energy threshold in the energy model, and d is the distance from node i to node j. resj is the residual energy of node j, a, b, c, d and m, n, p, q are all weight coefficients. θ is the angle between sensor node i and j, and α is the angle between sensor node j and the base station. level is the level label of each sensor node, recording the layer where the node is located. The cluster head node transfers data from the outermost layer to the innermost layer, and selects the node with the highest decision value as the next hop routing node. (405) Use Dijkstra algorithm to generate the shortest path, starting from the starting node M, find the node N that is closest to M and has not been processed, and set it as the current relay node; (406) Check all other nodes connected through node N. If it is found that the total distance to these nodes via node N is less than the distance directly starting from M, update the shortest path estimates of these nodes; (407) adding node N to the set of visited nodes, indicating that the shortest path to node N has been considered; (408) Execute step (405), continue to select the next nearest unprocessed node as a new relay node, and repeat the above process until all nodes are included in the visited set.

Images