A Clustering and Cluster Head Election Method for Wireless Sensor Nodes Based on Network Partitioning

Abstract

The present invention addresses the problem of premature node death caused by uneven energy consumption among nodes in a wireless sensor network. From the perspective of grid partitioning, the present invention considers the clustering and cluster head election of wireless sensor nodes, and proposes a method for clustering and cluster head election of wireless sensor nodes based on grid partitioning with energy load balancing. The sensor nodes are divided into different grid regions for clustering, and a dynamic cluster head election mechanism is adopted to reduce the collection of duplicate data. Moreover, the cluster head nodes are evenly distributed through the uniform distribution of the grids, thereby solving the problems of unreasonable clustering and uneven energy consumption of network nodes.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless sensing, and particularly relates to a method for clustering wireless sensor nodes and electing cluster heads based on network partitioning. Background Art

[0002] A wireless sensor network is an application-oriented network, which is widely used in military monitoring, industry, agriculture, medical health monitoring and guardianship and other fields. Due to the limitations of wireless sensors themselves and the environment, it is difficult to provide energy to them from the outside world. Therefore, extending the lifespan of the wireless sensor network through effective energy-saving measures is an important part of the entire wireless sensor network. Dividing the network into multiple clusters and balancing the energy consumption of each node is one of the important methods to improve the energy efficiency of the network.

[0003] In current research, the LEACH algorithm is a classic clustering routing algorithm. Compared with traditional static routing, it is more suitable for large-scale networks with high real-time requirements at the present stage and can effectively extend the network survival period. However, the defect of LEACH is that the remaining energy of nodes and the energy consumption balance of nodes are not considered during cluster head election.

[0004] Based on the LEACH algorithm, many scholars have conducted research, such as the DEEC algorithm, the LEACH-C algorithm, the HEED algorithm, etc. On the basis of the LEACH's round-robin selection of cluster heads, the DEEC algorithm sets a selection probability based on the remaining energy of nodes. Through this method, nodes with more remaining energy are more likely to be selected as cluster head nodes. Therefore, the selected nodes have enough energy to undertake the additional tasks of being cluster head nodes. These algorithms are simple and effective, but they do not consider the problem of the location distribution of cluster head nodes. The inappropriate distribution of cluster head nodes will also lead to waste of energy consumption of nodes within the cluster.

[0005] The LEACH-C algorithm is an improved method of the LEACH algorithm. When selecting cluster head nodes, the LEACH-C algorithm takes into account the communication distance between nodes and the remaining energy of nodes, and solves the randomness of the communication method and cluster head node selection in the LEACH algorithm. However, the LEACH-C algorithm is too dependent on the base station, increasing the energy consumption of the base station and being unfavorable to the stability of the network.

[0006] Aiming at the problem of uneven selection of cluster heads in the LEACH algorithm, the HEED algorithm selects nodes with higher energy as cluster head nodes each time it iterates. The HEED method can ensure that the cluster head nodes are evenly distributed in the network, and all cluster sizes are similar. However, in each iteration of the HEED algorithm, at the cost of communication overhead, the energy consumption of nodes increases. As the number of iterations in the initial stage increases, the communication overhead and network delay increase, resulting in a significant reduction in the network energy consumption balance.

[0007] In summary, the classical clustering routing algorithm has certain limitations in applicability. Implementing node clustering and cluster head election from the perspective of node data incurs high computational costs and serious data redundancy, leading to uneven network energy consumption, premature death of some nodes, and a short network lifespan. Summary of the Invention

[0008] The objective of the present invention is to address the deficiencies of the above prior art and propose a method for clustering wireless sensor nodes and electing cluster heads based on network partitioning, with the following steps:

[0009] a. Establish a planar coordinate system based on the edge of the wireless sensor network area;

[0010] Assume that the vertex coordinates of the area are the starting point O(0, 0), the lower right point A(x max , 0), the upper left point B(0, y max ), and the upper right point C(x max , y max ); Set the total number of wireless sensor nodes in the area to m, set the coordinates of the wireless sensor nodes in the area to (x p , y p ), 0 < p ≤ m, set the maximum communication distance between sensor nodes within a cluster to the low-power wireless communication critical distance d0, and d0 can be obtained from the data transmission energy consumption model of the wireless sensor network.

[0011]

[0012] Among them, E tx (k, d) is the energy consumption when a node sends k bit of data, with the unit nJ / bit; E elec is the basic power consumption coefficient of the transceiver circuit, with the unit: nJ / bit; e fs and e mp are the power amplification parameters of the signal amplifier, e fs is the energy consumption for transmitting 1 bit of data in free space, with the unit: pJ / bit / m 2 , e mp is the energy consumption for transmitting 1 bit of data with multipath attenuation, with the unit pJ / bit / m 2 ; d is the distance between two nodes, with the unit m; d0 is the boundary condition threshold for distinguishing between the two models, with the unit: m, and its specific calculation method is

[0013]

[0014] Let the distance between the sensor node p within the cluster (with the position coordinates (x p , y p )) and the starting point of a certain grid be d , where p Among them

[0015]

[0016] Among them, i is the grid number, and i ∈ (1, +∞).

[0017] Secondly, let the number of sensor nodes in a certain cluster be n i , and the optimal number of nodes in the cluster is n, which can be obtained according to the optimal number of cluster heads k opt .

[0018] The optimal number of cluster heads is:

[0019]

[0020] Among them, N is the number of all wireless sensor nodes in the target area of the wireless sensor network, e fs and e mp are the power amplification parameters of the signal amplifier, M is the area of the target area of the wireless sensor network, and d toBS is the distance between the base station and the center point of the target area of the wireless sensor network.

[0021] The optimal number of nodes n in the cluster is:

[0022]

[0023] Therefore, the grid division constraint condition d p ≤ d0 ∩ n i ≤ n is used to perform grid division on the coverage range of the wireless sensor network.

[0024] b. Divide the first grid;

[0025] Take the origin O(0, 0) of the established coordinate system as the starting point of grid division, and use as the grid side length in the x-axis direction, and search for sensor nodes (x p , y p ) in the y-axis direction. When the communication distance between sensor nodes and the number of sensor nodes n in the cluster i satisfy the constraint conditions of the communication critical distance d0 and the optimal number of nodes n in the cluster d p ≤ d0 ∩ n i ≤ n, that is, if the distance between the sensor nodes already searched and the starting point of the grid (the origin O(0, 0) of the coordinate system) satisfies d p < d0, y p < y max , and the number of nodes n already searched i < n, then mark the sensor node (x p , y p ) as (x 1j , y1j )(where j is the number of sensor nodes in the grid, j ∈ (0, n i )) and continue to search for sensor nodes in the y-axis direction (x p+1 , y p+1 ); if the distance d of the sensor node from the origin p = d0, y p < y max , The number of nodes searched n i < n or the distance d of the sensor node from the origin p < d0, y p < y max , The number of nodes searched n i = n, then use the difference between the maximum ordinate y 1j of the searched sensor nodes and the ordinate 0 of the starting point of the grid (y 1j - 0) as the grid side length in the y-axis direction.

[0026] c. Divide the second grid;

[0027] Subsequently, use the maximum ordinate (0, y 1j ) of the sensor nodes in the previous grid as the starting point (0, y 20 ) of the second grid, that is, y 20 = y 1j , and use as the grid side length in the x-axis direction, and search for sensor nodes (x p , y p ) in the y-axis direction. When the communication distance between sensor nodes and the number of sensor nodes n i in the cluster satisfy the constraint conditions of the communication critical distance d0 and the optimal number n of nodes in the cluster, d p ≤ d0 ∩ n i ≤ n, that is, if the distance between the searched sensor nodes and the starting point ((0, y 20 )) of the second grid satisfies d p < d0, y p < y max , and the number of nodes searched n i < n, then mark the sensor node (x p , y p ) as (x 2j , y 2j ) and continue to search for sensor nodes (x p+1 , y p+1 ) in the y-axis direction; if the distance d of the sensor point from the origin p = d0, y p < y max , The number of nodes n that have been searched i <n or the distance d between the sensor node and the origin p <d0, y p <y max , The number of nodes n that have been searched i = n, then use the maximum ordinate y of the searched sensor nodes 2j and the ordinate y of the starting point of the grid 20 The difference (y 2j - y 1j ) is the length of the grid side in the y-axis direction.

[0028] d, divide the remaining grids;

[0029] Repeat the above operations to divide the third grid, and so on.

[0030] e. Divide the last grid

[0031] When the ordinate y of the sensor node aj is the maximum ordinate in the interval, where a is the number of grids in the interval, then stop the search and divide the interval above.

[0032] f. Divide the interval grid;

[0033] Take the abscissa as the starting point of the grid division, and use as the grid side length in the x-axis direction to divide the grids in the interval, and repeat steps c, d, and e until y ij = y max ends.

[0034] g. Divide the remaining several regions;

[0035] Repeat the above interval's four steps to divide

[0036] the grids of each interval such as.

[0037] h. Elect cluster heads within the cluster according to the grid division and clustering;

[0038] According to the data transmission energy consumption model of the wireless sensor network, it can be obtained that the energy consumption of the sensor node is related to the transmission distance. According to the formula When a certain node is used as the cluster head node, the total energy consumption of communication between the nodes in the cluster is the smallest when the sum of the squares of the distances to the other nodes in the cluster is the smallest.

[0039] Beneficial effects:

[0040] By forming clusters through grid division, compared with the classical Leach algorithm that requires re-clustering every round in a polling manner, it reduces the collection of duplicate data, lowers the computational cost of the clustering algorithm, and at the same time adopts a dynamic cluster head election mechanism, extending the network lifetime. Description of the Drawings

[0041] Figure 1 It is a flowchart of a method for clustering wireless sensor nodes and electing cluster heads based on network division.

[0042] Figure 2 It is a schematic diagram of dividing the first grid.

[0043] Figure 3 It is a schematic diagram of dividing the second grid.

[0044] Figure 4 Schematic diagram of dividing grids in intervals.

[0045] Figure 5 Schematic diagram of the clustering effect of grid division.

[0046] Figure 6 Flowchart of the method for electing cluster heads within a cluster. Detailed implementation manners

[0047] To deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with embodiments and drawings. The embodiments are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0048] Embodiment example:

[0049] Considering the clustering and cluster head election of wireless sensor nodes from the perspective of grid division, a method for clustering wireless sensor nodes and electing cluster heads based on grid division with energy load balance is proposed. The sensor nodes are divided into different grid regions for clustering, and a dynamic cluster head election mechanism is adopted, reducing the collection of duplicate data, and the cluster head nodes are evenly distributed through the uniform distribution of grids, thus solving the problems of unreasonable clustering and uneven energy consumption of network nodes.

[0050] As Figure 1 shown,

[0051] For the process of grid clustering, it is necessary to first determine the coordinate system of the wireless network coverage area, divide the grid intervals, and then perform grid division on each interval to form clusters of wireless sensor nodes. The following describes the specific division steps.

[0052] As Figure 2 ,3 , as shown in FIGS. 4 and 5,

[0053] Take the origin O(0, 0) of the established coordinate system as the starting point for grid division, and use as the grid side length in the x-axis direction, and search for sensor nodes (x p , y p ) in the y-axis direction. When the communication distance between sensor nodes and the number n of sensor nodes within the cluster i satisfy the constraint conditions of the communication critical distance d0 and the optimal number n of nodes within the cluster, i.e., d p ≤d0 ∩ n i ≤n, that is, if the distance between the searched sensor node and the grid starting point (the origin O(0, 0) of the coordinate system) satisfies d p <d0, y p [[ID=XX]]<y max , and the number n of the searched nodes i <n, then mark the sensor node (x p , y p ) as (x 1j , y 1j ) and continue to search for sensor nodes (x p+1 , y p+1 ) in the y-axis direction; if the distance d of the sensor node from the origin p =d0, y p <y max , the number n of the searched nodes i <n or the distance d of the sensor node from the origin p <d0, y p <y max , the number n of the searched nodes i =n, then use the difference between the maximum ordinate y 1j of the searched sensor nodes and the ordinate 0 of the grid starting point (y 1j -0) as the grid side length in the y-axis direction, thereby dividing the first grid.

[0054] Where: d0 is the boundary condition threshold of the wireless communication energy consumption model in formula (1); n is the optimal number of nodes within the cluster, which can be obtained from the total number N of all nodes in the action area of the wireless sensor nodes and the optimal number k of cluster heads opt .

[0055] The optimal number of cluster heads is: [[ID=7XX]]

[0056]

[0057] The optimal number n of nodes within the cluster is:

[0058] It should be noted that there are some placeholders in the original text that seem to be incomplete or have incorrect variable names (e.g., "XX" in the translation). Please check and correct the original text for a more accurate translation.

[0059] Subsequently, using the maximum ordinate (0, y 1j ) of the sensor nodes in the previous grid as the starting point (0, y 20 ) of the second grid, that is, y 20 = y 1j , and using as the side length of the grid in the x-axis direction, search for sensor nodes (x p , y p ) in the y-axis direction. When the communication distance between sensor nodes and the number n i of the sensor nodes within the cluster satisfy the constraint conditions of the communication critical distance d0 and the optimal number n of nodes within the cluster, d p ≤ d0 ∩ n i ≤ n, that is, if the distance between the sensor nodes already searched and the starting point of the second grid ((0, y 20 )) satisfies d p < d0, y p < y max , and the number of nodes already searched n i < n, then mark the sensor node (x p , y p ) as (x 2j , y 2j ) and continue to search for sensor nodes (x p+1 , y p+1 ) in the y-axis direction; if the distance between the sensor point and the origin d p = d0, y p < y max , the number of nodes already searched n i < n or the distance between the sensor node and the origin d p < d0, y p < y max , the number of nodes already searched n i = n, then use the difference (y 2j - y 20 ) between the maximum ordinate y 2j of the sensor nodes already searched and the ordinate y 1j ) of the starting point of this grid as the side length of the grid in the y-axis direction, so as to divide the second grid.

[0060] Then repeat the above operations to divide the third grid, and so on.

[0061] When the ordinate y aj of the sensor node is the maximum ordinate within the interval, then stop the search and divide The last grid on the interval.

[0062] Take the abscissa as the starting point of grid division, and use as the side length of the grid in the x-axis direction to divide the grid within the interval. Repeat the above three steps, as Figure 3 shown.

[0063] Repeat the above steps to divide the grids of each interval, and the grid division is completed.

[0064] As Figure 6 shown,

[0065] After the grid division is clustered, the cluster head needs to be elected within the cluster. First, add the nodes with remaining energy greater than the average energy in each cluster to the candidate cluster heads of their clusters. Then, according to the position information of the nodes within the cluster, calculate the distances from each node within the cluster to other nodes. Finally, select the node with the smallest sum of distances to other nodes from the candidate cluster heads as the cluster head node of the cluster.

[0066] The average energy within the cluster (p≥j, p≤n), where p is the number of nodes within the cluster; set the coordinates of each node for grid clustering to be (x i0 , y i0 ), (x i1 , y i1 ), …, (x ij , y ij ), …, (x ip , y ip ), and let the function of the sum of distances to each node within the grid be f(x, y).

[0067] First, compare the remaining energy E ij of each node in each cluster with the average energy of the nodes within the cluster. If E ij > E avg , then add this node to the candidate cluster head nodes of its cluster, calculate the array of the sum of distances from the candidate cluster head nodes to other nodes in the grid through Equation 4, and sort it from small to large.

[0068]

[0069] Thus, the cluster head node is elected according to the sum of distances from the candidate cluster head nodes to other nodes in the grid.

Claims

1. A method for clustering wireless sensor nodes and electing cluster heads based on network partitioning, characterized in that, The steps are as follows; a. Establish a plane coordinate system according to the edge of the wireless sensor network area; Suppose the vertex coordinates of the assumed area are the starting point O(0, 0), the lower right point A(x max , 0), the upper left point B(0, y max ), and the upper right point C(x max , y max ); assume that the total number of wireless sensor nodes in this area is m, and assume that the coordinates of wireless sensor node p in the area are (x p , y p ), 0 < p ≤ m. Assume that the maximum communication distance between sensor nodes within a cluster is the low-power wireless communication critical distance d0, and d0 can be obtained from the data transmission energy consumption model of the wireless sensor network: Among them, E tx (k, d) is the energy consumption when the node sends k bits of data, with the unit of nJ / bit; E elec is the basic power consumption coefficient of the transceiver circuit, with the unit: nJ / bit; e fs and e mp are the power amplification parameters of the signal amplifier, and e fs is the energy consumption for transmitting 1 bit of data in free space, unit: pJ / bit / m 2 , and e mp is the energy consumption for transmitting 1 bit of data with multipath attenuation, unit pJ / bit / m 2 ; d is the distance between two nodes, unit: m; d0 is the boundary condition threshold for distinguishing between the two models, unit: m, and its specific calculation method is: Let the distance between the in-cluster sensor node p(x p , y p ) and the starting point of a certain grid be d , where p ​ where i is the grid number, and i ∈ (1, +∞) Secondly, let the number of sensor nodes in a certain cluster be n i , and the optimal number of nodes in the cluster is n, which can be obtained according to the optimal number of cluster heads k opt obtained; The optimal number of cluster heads is: Where N is the number of all wireless sensor nodes in the target area of the wireless sensor network, M is the area of the target area of the wireless sensor network, and d toBS is the distance between the base station and the center point of the target area of the wireless sensor network; The optimal number n of nodes within a cluster is: Therefore, the grid division constraint condition d can be obtained p ≤ d0 ∩ n i ≤ n; b. Divide the first grid; Take the origin O(0, 0) of the established coordinate system as the starting point for grid division, and use as the grid side length in the x-axis direction, and search for sensor nodes (x p , y p ) in the y-axis direction. When the communication distance between sensor nodes and the number n of sensor nodes within the cluster i satisfy the constraint conditions of the communication critical distance d0 and the optimal number n of nodes within the cluster d p ≤ d0 ∩ n i ≤ n, that is, if the distance between the searched sensor node and the origin O(0, 0) of the coordinate system satisfies d p < d0, y p < y max , and the number n of searched nodes i < n, then mark the sensor node (x p , y p ) as (x 1j , y 1j ), where j is the sensor node number within the grid, j ∈ (0, n i ), and continue to search for sensor nodes (x p+1 , y p+1 ) in the y-axis direction; if the distance d of the sensor node from the origin p = d0, y p < y max , the number n of searched nodes i < n or the distance d of the sensor node from the origin p < d0, y p < y max , the number n of searched nodes i = n, then use the difference (y 1j - 0) between the maximum ordinate y 1j of the searched sensor nodes and the ordinate 0 of the starting point of this grid as the length of the grid side length in the y-axis direction; c. Divide the second grid; Subsequently, use the maximum ordinate of the sensor nodes in the previous grid (0, y 1j ) as the starting point of the second grid (0, y 20 ), that is, y 20 = y 1j . Take as the side length of the grid in the x-axis direction, and search for sensor nodes (x p , y p ) in the y-axis direction. When the communication distance between sensor nodes and the number n i of the sensor nodes within the cluster satisfy the constraint conditions of the communication critical distance d0 and the optimal number n of nodes within the cluster d p ≤ d0 ∩ n i ≤ n, that is, if the distance between the already searched sensor nodes and the starting point of the second grid (0, y 20 ) satisfies d p < d0, y p < y max , and the number of already searched nodes n i < n, then mark the sensor node (x p , y p ) as (x 2j , y 2j ) and continue to search for sensor nodes (x p+1 , y p+1 ) in the y-axis direction; if the distance between the sensor point and the origin d p = d0, y p < y max , the number of already searched nodes n i < n or the distance between the sensor node and the origin d p < d0, y p < y max , the number of already searched nodes n i = n, then use the difference (y 2j - y 20 ) between the maximum ordinate y 2j of the already searched sensor nodes and the ordinate y 1j of the starting point of this grid as the side length of the grid in the y-axis direction; d. Divide the remaining grids; Repeat the above operations to divide the third grid, and so on; e. Divide the last grid When the vertical coordinate y of the sensor node aj is the maximum vertical coordinate within the interval, where a is the number of grids within the interval, then stop the search and divide the interval; f. Division Interval grid; Take the abscissa as the starting point of grid division, and use as the grid side length in the x-axis direction to divide the grid within the interval. Repeat steps c, d, and e until y ij = y max ends; g. Divide several other areas; Repeat the above four steps for each interval to divide the grids of each interval; h. Conduct intra-cluster cluster head election based on the grid-based clustering; According to the data transmission energy consumption model of the wireless sensor network, it can be obtained that the energy consumption of the sensor node is related to the transmission distance. According to the formula When a certain node is used as the cluster head node and the sum of the squares of the distances from the remaining nodes in the cluster is the smallest, the total energy consumption of the communication between the nodes in the cluster is smaller.

2. The method for clustering wireless sensor nodes and electing cluster heads based on network partitioning according to claim 1, wherein The steps of the intra-cluster cluster head election are as follows; a. Nodes with remaining energy greater than the average energy in each cluster are used as candidate cluster heads for their clusters; b. According to the location information of the nodes within the cluster, calculate the distances from each candidate cluster head node to other nodes; c. Select the node with the minimum sum of the squares of the distances to other nodes from the candidate cluster head nodes as the cluster head node of the cluster.

Images