A partitioned collaborative sensing method for large-scale Internet of Things

Abstract

The present invention proposes a partitioned collaborative perception method for large-scale Internet of Things, including the following steps: S1. Divide the perception area of ​​the node; S2. The node state is divided into working and sleeping, and the state is converted by the node wake-up probability; S3. Each node is provided with a request storage table to store the perceived request information and transmit the service request information; S4. Each node is provided with a preferred direction, and the optimal direction is modified according to the perceived request information and the request storage table; S5. After the optimal direction perception is completed, the node will calculate the selection operator of each direction, and select the direction represented by the largest selection operator for perception; S6. After the perception operation is completed, the node will wake up other nodes in the perception direction and transmit the request information; S7. After the transmission operation is completed, each node will calculate the node's capacity value and modify the wake-up probability, and filter out a part of the inefficient nodes through the convolution method. The method can improve the network's perception accuracy of service requests and reduce network energy consumption.

Description

[Technical field]

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a partition collaborative sensing method for large-scale Internet of Things. [Background technology]

[0002] At present, researchers have adopted hardware and software node optimization methods to deal with the problems of low node perception accuracy and high energy consumption in the IoT perception method. For hardware, researchers have innovated the battery structure and perception method of the node, which can effectively reduce energy consumption and accuracy, but the cost is high and it is difficult to use on a large scale. For software, researchers mainly optimize the scheduling mechanism of the node and improve the node optimization method, or innovate the perception method in combination with self-organizing networks. This method can have better performance in perception accuracy or network energy consumption, but it is difficult to achieve a balance between the two. In recent years, machine learning and deep learning methods have been continuously applied to the Internet of Things, and good progress has been made in some scheduling algorithms and feature extraction, but there is no good application in the underlying perception technology of the Internet of Things.

[0003] In view of the heterogeneous distribution of sensing nodes in the deployment scenarios of manufacturing Internet of Things networks, the dynamic change characteristics of uncertain events and the dynamic change characteristics of the availability of sensing nodes, it is necessary to study the node scheduling optimization mechanism for reliable perception of uncertain events and the efficient optimization method with fast convergence. Aiming at the problem of node perception in large-scale Internet of Things, a partitioned collaborative perception method for large-scale Internet of Things is proposed. [Summary of the invention]

[0004] The purpose of the present invention is to solve the problems in the prior art and to propose a partition collaborative perception method for large-scale Internet of Things, which can improve the network's perception accuracy of service requests and reduce the network's energy consumption.

[0005] To achieve the above object, the present invention proposes a partition collaborative sensing method for large-scale Internet of Things, comprising the following steps:

[0006] S1. Divide the node's perception area into different areas, each area represents a perception direction, and the perception of each direction is independent of each other;

[0007] S2. The node state is divided into two states: working and sleeping. The mutual conversion between the two states is controlled by the node wake-up probability;

[0008] S3. Each node has a request storage table to store the perceived request information and transmit the service request information;

[0009] S4. Each node has a preference position, which is modified according to the perceived request information and the request save table;

[0010] S5. After the preferred orientation perception is completed, the node will calculate the selection operator of each orientation and select the orientation represented by the largest selection operator for perception;

[0011] S6. After a round of sensing operation is completed, the node will wake up other nodes in the sensing direction and transmit request information;

[0012] S7. After the transmission operation is completed, each node will calculate the node's capability value, modify the wake-up probability according to the node's capability value, and filter out some inefficient nodes.

[0013] Preferably, the properties of the node are as follows:

[0014] (Nid,N_X,N_Y,Rc,Rs,PreOri,Pwake,Per,Sleep,RSF,Wtime,Em)

[0015] Among them, Nid represents the node number; N_X and N_Y represent the geographical coordinates of the node; Rc and Rs represent the communication radius and perception radius respectively; PreOri represents the number of the preferred direction; Pwake represents the wake-up probability of the node; Per and Sleep represent the perception and sleep flags respectively, and satisfy the following relationship:

[0016]

[0017]

[0018] RSF represents the perception flags of eight directions; Wtime represents the waiting time of eight search areas, and Em is the energy of the node.

[0019] Preferably, in step S1, each divided area has the same size.

[0020] As a preference, in step S3, in each round of sensing operation, node N i All nodes N will sense or receive service requests. i The request information will be recorded in the request storage table RST i The recorded information is represented as a set in the node, and its elements are as follows:

[0021] (Rid,R_X,R_Y,Ori,Rw,t)

[0022] Where Rid is the number of the service request; R_X and R_Y represent the geographical coordinates of the service request; Ori represents the location number of the perception Rid; Rw represents the method of obtaining the service request; t represents the time the request is stored in the node;

[0023]

[0024] As a preferred method, in step S4, during the sensing process, node Ni senses the service request with the preferred orientation first. The specific process is as follows: First, the RST i The requested coordinates in S1 are updated; then the corresponding positions in the dot matrix are marked as 1 according to the coordinates; the HNN network is used to determine the preferred direction of the node; by converting the dot matrix into vector noi by row, the HNN network is used to simulate noi with the dot matrix of the eight search directions divided in S1, and the result of the HNN network is modified with a small probability using probability p, and the result obtained is the next round of node N i The preferred direction; the probability p is a probability set separately to prevent the model from overfitting;

[0025]

[0026] Where R_X and R_Y represent RST i The requested coordinates in N_X and N_Y represent the node N i Coordinates, Rs i Represents node N i Perception radius.

[0027] Preferably, in step S5, the value of the selection operator of the node region is determined according to the search frequency and waiting time of each node, and the formula is as follows:

[0028]

[0029] Among them, S(i,k) represents the node N i The value of the selection operator of the k-th search area, where Rnum represents the value of the selection operator of the k-th search area at node N i The number of requests stored in ; Freq(i,k) represents the total number of requests in region k; PFReq(i,k) represents the total number of requests with region k and Rw 1; RSF(i,k) represents the number of requests stored in node N i Perceptual hallmarks of the mid-k region; Represents node N i The waiting time for the search area k.

[0030] Preferably, the node capability value Ab is calculated by selecting the value of the operator S according to the following formula:

[0031]

[0032] Where M is the control factor that controls the degree of difference between Ab, T represents the time from the beginning of model perception to the present, and S(i,m) represents the node N i The value of the selection operator of the mth search area; Freq(i,m) represents the node Ni The total number of requests in region m; Represents node N i The shortest waiting time among the eight search directions; after calculating the selection operator, the direction represented by the largest selection operator is enabled for perception. In this way, a perception direction is added on the basis of the preferred direction, which not only allows the search area to be continuously updated, but also prevents the model from falling into local search.

[0033] Preferably, in step S7, the awakening probability of the node is calculated using the following formula to control the awakening probability between [a, b]:

[0034]

[0035] Among them Ab i Represents node N i Ability value, Ab min and Ab max Indicates the minimum and maximum capacity values ​​of all nodes;

[0036] Then, the inefficient nodes are filtered out by convolution, the Ab of all nodes are obtained, and they are combined into a 12×12 matrix, called feature map C1; then the following steps are performed:

[0037] Step a: Use a 3×3 convolution kernel Perform convolution to obtain a 10×10 feature map named C2, which obtains the average capacity value of each 3×3 region;

[0038] Step b: A 2×2 normed majority pooling operation is performed on C2, and the result is used to represent the horizontal efficiency of the 2×2 region. The resulting feature map is called C3.

[0039] Step c: Through convolution and mode pooling, feature map C3 is obtained. By comparing the data in C3, it can be determined that the eigenvalues ​​with smaller eigenvalues ​​are areas with weaker perception ability;

[0040] Step d: According to the corresponding relationship in step c, select the nodes with lower capability values ​​in C1, and organize their capability values ​​into a dot matrix C4;

[0041] Step e: Divide each C4 into 4 feature maps with a size of 2×2, each feature map contains the capability values ​​of 4 nodes; perform convolution operation on each feature map; the convolution operation is the sum of the capability values ​​of the remaining three nodes after removing its own capability value;

[0042] Step f: By comparing the capacity values ​​in C5, select the node with the smallest capacity value and number it in C6; C6 contains the 4 results selected from each C4;

[0043] Step g: The node number recorded in C6 forms the final output result C7. According to the node number in C7, C1 sets the Pwake of the recorded node number to 0.

[0044] Beneficial effects of the present invention: The present invention divides the self-organizing area of ​​the sensing nodes in the Internet of Things and introduces the idea of ​​orientation, thereby improving the network's perception accuracy of service requests; based on the preferred orientation, a sensing orientation is added according to the selection operator, which improves the flexibility of the sensing area, not only updates the search area, but also avoids local search. By controlling the node state through adaptive wake-up probability, the number of working nodes in the model can be effectively controlled, and the convolution screening method can further reduce the number of nodes while ensuring the perception accuracy to reduce the network's energy consumption.

[0045] 1. The present invention divides the perception area of ​​the perception node into eight parts, each of which is represented by a perception direction. This division method refers to the cell sensitivity mechanism of human vision, making the node's perception more directional. The HNN neural network is used to determine the preferred direction of the perception node, which can make the determination of the direction more accurate. The preferred direction allows the node to quickly lock the area with frequent service requests, capture a large number of service requests in a short time, and improve the perception rate of the network.

[0046] 2. The present invention designs eight search operators for the eight search directions divided by the perception nodes. By calculating the search operator, the present invention adds a perception direction on the basis of the preferred direction, which not only increases the perception area, allowing the search area of ​​the perception node to be continuously updated, but also avoids the perception node from falling into local search. At the same time, the design of dual perception directions makes the perception of the node more flexible, and each perception direction has its own function, which improves the fault tolerance of the perception network and also improves the perception rate of the network.

[0047] 3. The present invention designs a dynamic node switching mode for large-scale perception networks. First, an adaptive wake-up probability is designed for each node, which can stabilize the number of nodes in the network and allow the number to change with the frequency of service requests. At the same time, a capability value is designed for each node. The convolution method is used to extract the perception capability of the area through the capability value of the node, and the nodes that are not helpful for the next round of perception can be cleverly screened out in areas with low perception capabilities, thereby reducing the number of working nodes while ensuring the network perception capability, and further reducing network energy consumption.

[0048] The features and advantages of the present invention will be described in detail by way of examples with reference to the accompanying drawings.

Brief Description of the Drawings

[0049] Figure 1It is a flow chart of a partition collaborative perception method for large-scale Internet of Things of the present invention;

[0050] Figure 2 It is the design diagram of the node;

[0051] Figure 3 It is a schematic diagram of using HNN network to determine the preferred orientation of nodes;

[0052] Figure 4 is a schematic diagram of the direction-sensitive perception pattern;

[0053] Figure 5 It is a schematic diagram of energy consumption comparison of each algorithm;

[0054] Figure 6 This is a schematic diagram comparing the perception rates of various algorithms;

[0055] Figure 7 This is a comparison diagram of the number of node wake-ups of each algorithm;

[0056] Figure 8 It is a schematic diagram of the 2×2 norm majority pooling operation performed on C2 to obtain C3;

[0057] Fig. 9 This is a schematic diagram of feature map C3 obtained through convolution and majority pooling;

[0058] Fig.10 It is a schematic diagram of performing convolution operations on each feature map of C4. [Specific implementation method]

[0059] The present invention proposes a partitioned collaborative sensing method for large-scale Internet of Things, and can be applied to self-organizing Internet of Things. Figure 1 The algorithm flow includes the following steps: 1) initializing the node state; 2) modifying the sensing position of the node; 3) sensing the request information; 4) waking up the surrounding nodes and sharing the request information; 5) modifying the wake-up probability of the node; 6) modifying the state of the node; 7) screening the awakened nodes.

[0060] The algorithm design of the present invention comprises the following steps:

[0061] S1. Divide the node's perception area into different areas, each area represents a perception direction, and the perception of each direction is independent of each other;

[0062] S2. The node state is divided into two states: working and sleeping. The mutual conversion between the two states is controlled by the node wake-up probability;

[0063] S3. Each node has a request storage table to store the perceived request information and transmit the service request information;

[0064] S4. Each node has a preference position, which is modified according to the perceived request information and the request save table;

[0065] S5. After the preferred orientation perception is completed, the node will calculate the selection operator of each orientation and select the orientation represented by the largest selection operator for perception;

[0066] S6. After a round of sensing operation is completed, the node will wake up other nodes in the sensing direction and transmit request information;

[0067] S7. After the transmission operation is completed, each node will calculate the node's capability value, modify the wake-up probability according to the node's capability value, and filter out some inefficient nodes.

[0068] The implementation method of the technical solution of the present invention is as follows:

[0069] 1. Node design

[0070] The sensing area of ​​each node is divided into eight areas, each of which is of the same size. The sensing of each area is independent of each other. In each round of sensing, two areas will be opened for sensing. The design of the node is as follows: Figure 2 shown.

[0071] Each node sets a preference direction, adjusts the preference direction according to the request status perceived by the node and the transmission status of other nodes, wakes up the nodes in the preference direction within the communication range, and transmits the stored request information. The properties of the node are as follows:

[0072] (Nid,N_X,N_Y,Rc,Rs,PreOri,Pwake,Per,Sleep,RSF,Wtime,Em)

[0073] Among them, Nid represents the node number; N_X and N_Y represent the geographical coordinates of the node; Rc and Rs represent the communication radius and perception radius respectively; PreOri represents the number of the preferred direction; Pwake represents the wake-up probability of the node; Per and Sleep represent the perception and sleep flags respectively, and satisfy the following relationship:

[0074]

[0075]

[0076] RSF represents the perception flags of eight directions; Wtime represents the waiting time of eight search areas; Em is the energy of the node.

[0077] For each node, the present invention designs a request save table (RST) to store and transmit service request information. i Each node will sense or receive service requests; the node will record the request information in the RST i In order to reflect the current perception situation, the stored information will only be remembered for 3 rounds of perception time. The recorded information is represented as a set in the node. Its elements are as follows:

[0078] (Rid,R_X,R_Y,Ori,Rw,t)

[0079] Among them, Rid is the number of the service request; R_X and R_Y represent the geographical coordinates of the service request; Ori represents the orientation number of the perception Rid; Rw represents the method of obtaining the service request; t represents the time the request is stored in the node.

[0080]

[0081] 2. Node's preferred location

[0082] Design a preference direction for each node. During the perception process, Ni prioritizes the service request based on the preference direction.

[0083] like Figure 3 As shown, the Hopfield neural network (HNN) is used to determine the preferred orientation of the node.

[0084] First, RST is calculated according to the following formula: i Update the requested coordinates in the dot matrix; then mark the corresponding position as 1 according to the coordinates;

[0085]

[0086] Where R_X and R_Y represent RST i The requested coordinates in N_X and N_Y represent the node N i Coordinates, Rs i Represents node N i Perception radius.

[0087] By converting the dot matrix into vector noi by row, HNN is used to simulate noi with the dot matrix v1-v8 of the 8 search areas. In order to avoid overfitting, the result of HNN is corrected using probability p, where probability p is a probability set separately to prevent overfitting of the model. The result is the next round of N i PreOri.

[0088] 3. Regional weight of the node

[0089] In order to prevent the preference direction from being too concentrated and causing local search, the value of the node area selection operator is designed according to the search frequency and waiting time of each node. The formula is shown in the figure below:

[0090]

[0091] S(i,k) represents node N i The value of the selection operator of the k-th search area, where Rnum represents the value of the selection operator of the k-th search area at node N i The number of requests stored in ; Freq(i,k) represents the total number of requests in region k; PFReq(i,k) represents the total number of requests with region k and Rw 1; RSF(i,k) represents the number of requests stored in node N i Perceptual hallmarks of the mid-k region; Represents node N i The waiting time for the search area k.

[0092] 4. Node orientation-sensitive perception mode

[0093] The present invention proposes a location-sensitive perception mode, the core idea of ​​which is to continuously use the preferred location and selection operator to select the location-aware service request.

[0094] See also Figure 4 , the area of ​​the preferred direction and N A The area with the highest value of the selection operator is represented by the arrow area and the shaded area. A In the working state, the arrow area and the shaded area are opened in turn to sense service requests R1 and R2. Then node N A Receiving Node N B and node N C RST B and RSTc. Node N A According to the information received and perceived, RST A Add information. Then, node N A Determine your preferred orientation and calculate the selection operator of the search area. After perception, node N A Will RST A Send to nodes N that can be contacted within the two search areas D and node N E .

[0095] 5. Node state switching based on convolution method

[0096] The node capability value Ab is calculated using the value of the selection operator S according to the following formula.

[0097]

[0098] Where M is the control factor that controls the degree of difference between Ab. T represents the time from the beginning of model perception to the present, and S(i,m) represents the node N i The value of the selection operator of the mth search area; Freq(i,m) represents the node N i The total number of requests in region m; Represents node N i The shortest waiting time among the eight search directions.

[0099] The node wake-up probability is calculated using the following formula to control the probability between [a, b]:

[0100]

[0101] Among them Ab i Represents node N i Ability value, Ab min and Ab max Indicates the minimum and maximum capacity values ​​among all nodes.

[0102] Then, we use convolution to filter out inefficient nodes, obtain the Ab of all nodes, and combine them into a 12×12 matrix, called feature map C1. Then we proceed as follows:

[0103] Step a: Use a 3×3 convolution kernel Convolution is performed to obtain a 10×10 feature map named C2, which obtains the average capacity value of each 3×3 region.

[0104] Step b: The 2×2 normed majority pooling operation is performed on C2. The result is used to represent the average efficiency of the 2×2 region. The obtained feature map is mapped to C3, as shown in Figure 8 This operation can better distinguish different regions by feature values.

[0105] Step c: The feature map C3 is obtained by convolution and majority pooling. The corresponding relationship is as follows Fig. 9 As shown, the eigenvalues ​​with smaller eigenvalues ​​are areas with weaker perception ability.

[0106] Step d: According to the corresponding relationship in step c, the nodes with lower capability values ​​in C1 are screened out, and their capability values ​​are combined into a dot matrix C4.

[0107] Step e: Divide each C4 into 4 feature maps with a size of 2×2. Each feature map contains the capability values ​​of 4 nodes. Perform a convolution operation on each feature map. The convolution operation is the sum of the capability values ​​of the remaining three nodes after removing their own capability values, such as Fig.10 shown.

[0108] Step f: By comparing the capability values ​​in C5, select the node with the smallest capability value to C6. C6 contains 4 results selected from each C4.

[0109] Step g: The node Nid recorded in C6 forms the final output result C7. According to the Nid in C7, the model sets the Pwake of the recorded Nid to 0.

[0110] 6. Experimental Results

[0111] In the experiment, a 400*400 (M) IoT area was set up, and 144 sensor nodes were evenly distributed in the area. The performance of the nodes in the experiment was consistent. The communication range was 32M, the perception range was 16M, and the initial preferred direction was numbered as a random integer from 1 to 8. The perception range of the node was evenly divided into 8 parts.

[0112] The properties requested in the experiment are as follows:

[0113] (Rid,R_X,R_Y,Rq,Rt)

[0114] Where Rid is the service request number. R_X and R_Y represent the geographic coordinates of the service request. Rq and Rt represent the coverage radius and perception flag of the request respectively.

[0115]

[0116] The number of independent experiments for each example is 30. The algorithm adopted by the present invention is recorded as OSPA algorithm, and the invented OSPA algorithm is compared with DSPA, NIA_BA, CAOP, and ISOS algorithms in experiments.

[0117] 6.1 Energy Consumption

[0118] For each algorithm, only 30% of the nodes are initially set to be awakened, and the number of requests is set to 1000, 1500, 2000, and 2500 respectively. The results are shown in Figure 2. Figure 5 As shown, corresponding to Figure 5 Figures a, b, c, and d in the figure. As can be seen from the figure, the energy consumption of the OSPA algorithm is significantly better than that of the other four algorithms. This is mainly due to the reasonable division of the perception range of the nodes by the present invention, which reduces the energy consumption of the sending and receiving modules. The adaptive wake-up probability and convolution method effectively reduce the number of working nodes. When the number of service requests increases, OSPA continues to widen the gap with other algorithms. This shows that the OSPA algorithm is more competitive in discovering large-scale service requests.

[0119] 6.2 Perception Rate Comparison Experiment

[0120] For each algorithm, only 30% of the nodes are initially set to be awakened, and the number of requests is set to 1000, 1500, 2000, and 2500. The results are shown in Figure 2. Figure 6 As shown, corresponding to Figure 6 Figures a, b, c, and d in the figure. The experimental results show that OSPA has the fastest perception rate increase, and can quickly rise to nearly 1. This is because the preference direction quickly searches for areas with dense service requests to improve the perception rate. Secondly, the selection operator increases the perception area, making the perception more global and improving the fault tolerance rate. This shows that OSPA will show its advantages in the environment of large-scale service requests.

[0121] 6.3 Number of nodes awakened

[0122] For each algorithm, only 30% of the nodes are initially set to be awakened, and the number of requests is set to 1000, 1500, 2000, and 2500. The results are shown in Figure 2. Figure 7 As shown, corresponding to Figure 7 Figures a, b, c, and d in the figure. As can be seen from the figure, the OSPA algorithm can effectively control the number of working nodes, thanks to the implementation of adaptive wake-up probability. Secondly, the convolution method modifies the useless nodes in the working nodes to sleep state, further reducing the number of working nodes. This shows that in large-scale IoT scenarios, the OSPA algorithm can effectively control the number of nodes.

[0123] The above embodiments are intended to illustrate the present invention, not to limit the present invention. Any solution that is a simple transformation of the present invention belongs to the protection scope of the present invention.

Claims

1. A partition collaborative perception method for large-scale Internet of Things, characterized by: The following steps are involved: S1. Divide the node's perception area into different areas, each area represents a perception direction, and the perception of each direction is independent of each other; S2. The node state is divided into two states: working and sleeping. The mutual conversion between the two states is controlled by the node wake-up probability; S3. Each node has a request storage table to store the perceived request information and transmit the service request information; S4. Each node has a preference position, which is modified according to the perceived request information and the request save table; S5. After the preferred orientation perception is completed, the node will calculate the selection operator of each orientation and select the orientation represented by the largest selection operator for perception; S6. After a round of sensing operation is completed, the node will wake up other nodes in the sensing direction and transmit request information; S7. After the transmission operation is completed, each node calculates the node's capability value, modifies the wake-up probability according to the node's capability value, and filters out some inefficient nodes; In step S3, in each round of sensing operation, node N i All nodes will sense or receive service requests. N i The request information will be recorded in the request storage table RST i The recorded information is represented as a set in the node, and its elements are as follows: （ Rid , R_X , R_Y , Ori , R , t ）； in, Rid The number of the service request; R_X and R_Y The geographical coordinates representing the service request; Ori Expressing Perception Rid The position number of R Indicates the method of obtaining service requests; t Indicates the time the request is stored in the node; ； In step S4, during the perception process, node Ni perceives the service request with the preferred direction first. The specific process is as follows: First, the RST i Update the requested coordinates in the dot matrix; then mark the corresponding position as 1 according to the coordinates; use the HNN network to determine the preferred orientation of the node; convert the dot matrix into a vector by row noi , using the HNN network noi Simulate the eight search positions divided in S1 using the probability p The result of the HNN network is corrected, and the result obtained is the next round of nodes N i Preferred location; ； in R_X and R_Y Indicates RST i The requested coordinates in N_X and N_Y Representation Node N i The coordinates of Rs i Representation Node N i Perception radius; In step S7, the inefficient nodes are screened using the convolution method to obtain the capacity values ​​of all nodes. Ab , and combine them into a 12×12 matrix, called feature map C1; then perform the following steps: Step a: Use a 3×3 convolution kernel Perform convolution to obtain a 10×10 feature map named C2, which obtains the average capacity value of each 3×3 region; Step b: A 2×2 normed majority pooling operation is performed on C2, and the result is used to represent the horizontal efficiency of the 2×2 region. The resulting feature map is called C3. Step c: Through convolution and mode pooling, feature map C3 is obtained; by comparing the data in C3, it can be determined that the eigenvalues ​​with smaller eigenvalues ​​are areas with weaker perception ability; Step d: According to the corresponding relationship in step c, select the nodes with lower capability values ​​in C1, and organize their capability values ​​into a dot matrix C4; Step e: Divide each C4 into 4 feature maps with a size of 2×2, and each feature map contains the capability values ​​of 4 nodes; perform convolution operation on each feature map to obtain C5; the convolution operation is the sum of the capability values ​​of the remaining three nodes after removing their own capability values; Step f: By comparing the capacity values ​​in C5, select the node with the smallest capacity value and number it in C6; C6 contains the 4 results selected from each C4; Step g: The node numbers recorded in C6 form the final output result C7. According to the node numbers in C7, the model will record the node numbers. Pwake Set to 0; Pwake represents the wake-up probability of the node.

2. The partition collaborative sensing method for large-scale Internet of Things according to claim 1, characterized in that: The properties of the node are as follows: ( Nid , N_X , N_Y , Rc , Rs , PreOri , Pwake , Per , Sleep , RSF , Wtime , E )； in, Nid Indicates the node number; N_X and N_Y Represents the geographical coordinates of the node; Rc and Rs They represent the communication radius and the perception radius respectively; PreOri A number indicating the preferred direction; Pwake represents the wake-up probability of the node; Per and Sleep Represent the perception and sleep signs respectively, and satisfy the following relationship: ； ； RSF Perceptual signs representing the eight directions; Wtime Indicates the waiting time for eight search areas, E is the energy of the node.

3. The partition collaborative sensing method for large-scale Internet of Things according to claim 1, characterized in that: In step S1, each divided area has the same size.

4. The partition collaborative sensing method for large-scale Internet of Things according to claim 2, characterized in that: In step S5, the value of the selection operator of the node area is determined according to the search frequency and waiting time of each node, and the formula is as follows: ； in, S ( i,k ) represents a node N i No. k The value of the selection operator of the search area, where Rnum Indicates that at the node N i The number of requests stored in ; Freq(i,k) Indicates area k The total number of requests in; PFreq(i,k) Indicates the area k and R The total number of requests that is 1; RSF(i,k) Representation Node N i Perceptual landmarks of the mid-k region; Representation Node N i Search area k waiting time.

5. The partition collaborative sensing method for large-scale Internet of Things according to claim 4, characterized in that: According to the following formula, use the selection operator S The value of the node capacity is calculated by Ab : ； in, M Yes Control Ab The controlling factors of the difference between T Indicates the time from the beginning of model perception to the present. Representation Node N i No. m The value of the selection operator for the search area; Representation Node N i Central Area m The total number of requests in; Representation Node N i The shortest waiting time among the eight search directions; after calculating the selection operator, the direction represented by the largest selection operator is turned on for perception.

6. A partition collaborative sensing method for large-scale Internet of Things according to claim 5, characterized in that: In step S7, the awakening probability of the node is calculated using the following formula to control the awakening probability between [a, b]: ； in Ab i Representation Node N i The ability value, Ab min and Ab max Indicates the minimum and maximum capacity values ​​among all nodes.

Images