A hierarchical scheduling method for wireless sensor networks based on node importance
By introducing node importance hierarchical scheduling method and African vulture optimization algorithm in wireless sensor network, the problem of node allocation scheduling and tracking accuracy in high-density wireless sensor network is solved, and the balance of energy consumption and accuracy and the extension of network life is achieved. .
Patent Information
- Application Number
- CN202211498683.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-28
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-11-28
AI Technical Summary
In the high-density, non-uniform wireless sensor network scenarios arranged at a large scale, the network faces node allocation and scheduling problems when performing tracking tasks, resulting in excessive energy consumption and unstable tracking accuracy, especially the differences in redundant nodes and region density. The cyber hollow problem caused.
Adopting the wireless sensor network hierarchical scheduling method based on node importance, the task cluster size and node importance level are adaptively adjusted through automatic clustering, perceived signal strength decision, cluster head prediction position, node status information broadcast and African vulture optimization algorithm. , optimize node allocation and data transmission.
The balance between network energy consumption and tracking accuracy is achieved, the network life is extended, the network coverage quality is improved, the complexity of task cluster node selection is reduced, and the relationship between tracking accuracy and energy consumption is optimized by dynamically adjusting the task cluster size. .
Smart Images

Figure CN115866706B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of wireless sensor networks, and in particular to a wireless sensor network hierarchical scheduling method based on node importance. Background Art
[0002] WSN is a self-organizing network composed of many resource-constrained sensor nodes, which can meet the needs of users to obtain detailed and reliable information about the physical world at any time, place and environmental conditions. It has broad research prospects and potential application value in many fields such as national defense, military, environmental monitoring, smart home, etc. Among them, the tracking technology based on WSN as a key means of regional monitoring has been widely used in tracking targets such as animals, humans, and vehicles, and the objects have also been involved in all aspects of life.
[0003] In the scenario of large-scale randomly deployed high-density, non-uniform wireless sensor networks, the network execution tracking task faces the problem of node allocation and scheduling. On the one hand, due to the dense deployment, there are a large number of redundant nodes in the network, which will cause unnecessary energy consumption and shorten the network life. How to deal with redundant nodes has always been a key issue in the design of node scheduling mechanism; on the other hand, due to the randomness of node deployment, the network presents a relatively dense or sparse phenomenon in some areas, which leads to differences in the importance of nodes to the area, and the impact of this difference on the node scheduling mechanism will be manifested as the speed at which network holes appear. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide a wireless sensor network hierarchical scheduling method based on node importance, which can balance the energy consumption and tracking accuracy of the network.
[0005] The technical solution adopted by the present invention to solve the technical problem is: to provide a wireless sensor network hierarchical scheduling method based on node importance, comprising the following steps:
[0006] When a target appears in the monitoring area, the sensor nodes automatically cluster to form an initial task cluster;
[0007] Each cluster member in the task cluster makes a transmission decision based on its own node importance level and perceived signal strength, and determines whether its perceived data needs to be transmitted to the cluster head;
[0008] The cluster head fuses the received target perception data, obtains the target state estimation information, and predicts the target position at the next moment based on the estimated value;
[0009] The cluster head broadcasts a request to obtain node status information within the communication radius, and the sensor node that receives the request sends its status information to the cluster head;
[0010] The cluster head adaptively adjusts the task cluster size at the next moment according to the recent tracking quality level and the distribution of nodes near the predicted target location;
[0011] Based on the received node status information, the cluster head determines the candidate node set of the task cluster at the next moment according to the target predicted position, and further determines the candidate cluster head set and the candidate cluster member set based on the candidate task cluster node set according to the task cluster size at the next moment and the node importance level of each sensor node in the candidate node set;
[0012] The preset algorithm is used to select the cluster head and cluster members at the next moment from the candidate cluster head set and the candidate cluster member set.
[0013] The wireless sensor network hierarchical scheduling method based on node importance also includes determining the node importance level of each sensor node based on a fuzzy logic system with regional node density and neighbor node energy level as input.
[0014] When each cluster member in the task cluster makes a transmission decision based on its own node importance level and perceived signal strength, the cluster member with a lower node importance level has a greater probability of data transmission, and the cluster member with a higher perceived signal strength has a greater probability of data transmission.
[0015] The sensing data includes the distance between the sensor node itself and the target and the node remaining energy information with a timestamp.
[0016] After the cluster head broadcasts a request to obtain node status information within the communication radius, if no data packet is received from any sensor node in reply within a preset time, the cluster head broadcasts the request to obtain node status information again. If no data packet is received from any sensor node in reply within the next preset time, a message to stop working is broadcast to terminate the current target tracking task.
[0017] The cluster head adaptively adjusts the size of the task cluster at the next moment according to the recent tracking quality level and the node distribution near the predicted target position. Specifically, when the recent tracking quality level decreases, the size of the task cluster is increased at the next moment; when the recent tracking quality level increases, the size of the task cluster is reduced at the next moment; the sparser the distribution of sensor nodes and the farther the distance from the estimated target, the larger the size of the task cluster at the next moment; the denser the distribution of sensor nodes and the closer the distance from the estimated target, the smaller the size of the task cluster at the next moment.
[0018] When determining the candidate cluster head set and the candidate cluster member set, the node corresponding to the lowest node importance level in the candidate node set is used to construct the candidate cluster head set, and the candidate nodes are sorted from low to high according to the node importance level, and are selected from the first one in sequence until the total number reaches the task cluster size required at the next moment. The node importance level corresponding to the last sensor node is recorded as rankI. high , then the node importance level rankI high The corresponding candidate nodes construct the candidate cluster member set; the node importance level is lower than rankI high All candidate nodes will be selected into the task cluster node set at the next moment, and the remaining task cluster nodes will be selected from the candidate cluster member set.
[0019] When the preset algorithm is used to select the cluster head and cluster member at the next moment from the candidate cluster head set and the candidate cluster member set, the discrete African vulture optimization algorithm is used, and its fitness function is F=λ1tanh(E ncp )+λ2tanh(Be)+λ3E level , where E ncp represents the energy consumption of executing the tracking task, Be represents the remaining energy balance, and E level represents the energy level of candidate cluster heads, λ1,λ2,λ3 are weight coefficients, and satisfy
[0020] When the discrete African vulture optimization algorithm is used to select the cluster head and cluster member at the next moment from the candidate cluster head set and the candidate cluster member set, specifically:
[0021] Sort the sensor nodes in the candidate cluster head set and candidate cluster member set from low to high according to the node importance level, and place the first in the candidate cluster head set and the first in the candidate cluster member set. The bit is 1, and the rest are 0, where is the number of remaining task cluster nodes required at time t+1; in the individual, the 1 in the candidate cluster head set part indicates that the sensor node corresponding to this position is selected as the cluster head, and the 1 in the candidate cluster member set part indicates that the sensor node corresponding to this position is selected into the task cluster;
[0022] The remaining individuals are determined by cyclic shift, specifically: the candidate cluster head set part and the candidate cluster member set part are cyclically shifted respectively, the candidate cluster head set part is moved one bit at a time, and the candidate cluster member set part is randomly moved n_shift bits to obtain a new individual;
[0023] Based on the discrete African vulture optimization algorithm, the individual positions are iteratively updated. The position vector of each vulture individual corresponds to a task cluster and a cluster head selection scheme.
[0024] Beneficial Effects
[0025] Due to the adoption of the above technical scheme, the present invention has the following advantages and positive effects compared with the prior art: the present invention evaluates the node importance level according to the neighbor node energy level and the regional node density. The lower the node importance level, the lower the importance of the node to the region. When selecting the task cluster node, it will be preferred, thereby ensuring the network coverage quality and extending the overall life of the network. Secondly, the present invention adaptively adjusts the task cluster size according to the recent tracking quality level and the distribution of nodes around the predicted target. When the tracking accuracy level is high, the task cluster size is appropriately reduced in exchange for a decrease in energy consumption. When the tracking accuracy level is low, the task cluster size is increased to ensure tracking accuracy. Based on the node importance level, the present invention hierarchically calls nodes to participate in the tracking task. Low-importance nodes will be preferred. This hierarchical scheduling mechanism also realizes the lightweight of the task cluster node selection algorithm. In the process of determining the specific task cluster node set, the present invention adopts a discretized African vulture optimization algorithm, and the setting of its fitness function considers three factors: the residual energy balance of the candidate task cluster node set, the network predicted energy consumption and the cluster head energy level, the individual position vector corresponds to the selection scheme of the task cluster node, and the optimal strategy is determined by continuously seeking the best in a dynamic environment. The present invention is suitable for tracking targets in a large-scale randomly deployed high-density, non-uniform wireless sensor network scenario, and has high use value. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 is a schematic diagram of a target tracking scenario in an embodiment of the present invention;
[0027] Figure 2 is a flow chart of a wireless sensor network hierarchical scheduling method based on node importance according to an embodiment of the present invention;
[0028] Figure 3 This is a schematic diagram of the first individual initialization when the BAVOA algorithm is used in an embodiment of the present invention;
[0029] Figure 4 It is a schematic diagram of cyclic shift when the BAVOA algorithm is adopted in an implementation manner of the present invention. DETAILED DESCRIPTION
[0030] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.
[0031] The embodiment of the present invention relates to a wireless sensor network hierarchical scheduling method based on node importance, which can be applied to Figure 1In the target tracking scenario shown in Figure 2 As shown, the specific steps include:
[0032] Network deployment:
[0033] A large number of sensor nodes are randomly and unevenly deployed in the monitoring area of WSN. Once the nodes are successfully deployed, they enter a dormant state. The sensing radius of the node is R s , the communication radius is R c , in order to ensure network coverage, that is, network connectivity, let R c ≥2R s The target appears randomly in the monitoring area and moves. The corresponding sensor nodes are activated according to the target position to perform the tracking task. The target movement is a dynamic and continuous change process, but the target state can be considered to remain unchanged in a sufficiently short time interval. Therefore, the target movement process is divided into several continuous time intervals, each time interval is recorded as time t (t = 1, 2, 3, ...), and the target state remains unchanged in each time.
[0034] Initial clustering:
[0035] When a target appears in the monitoring area, the perceived signal strength is greater than the threshold I threshold The sensor nodes will be activated. Threshold I threshold It is related to the density of deployed sensor nodes ρ and the sensitivity of the sensor itself γ:
[0036] I threshold ∝(ρ,γ) (1)
[0037] Wait for a broadcast delay time T delay After that, the activation node is in R c The first node to broadcast the "Competition Message" will serve as the cluster head of the initial task cluster and the fusion center of the initial task cluster. At the same time, the activated node that receives the corresponding broadcast message replies with a confirmation message and becomes a member of the initial task cluster. Broadcast delay time T delay and the node residual energy E res And the node perceived signal strength I is related to:
[0038] T delay ∝(E res ,I) (2)
[0039] Node importance level classification:
[0040] Since sensor nodes are randomly and non-uniformly arranged, the importance of different nodes to the region varies. If the importance of nodes to the region is not considered and they are indiscriminately selected to join the task cluster, it is very easy to cause network holes to appear early, thereby increasing the probability of target loss. Therefore, this implementation method proposes a concept of "node importance" to characterize the importance of nodes to the region. The lower the node importance, the smaller the impact of node death on the network life. The node importance level rankI is divided into two factors:
[0041] Regional node density. Regional node density reflects the degree of node substitution. The higher the regional node density, the smaller the node spacing, the greater the corresponding node redundancy, the stronger the node substitution, and the death of a few nodes has little impact on the network life, that is, the importance of the nodes is low; while for low-density areas, the importance of nodes to the area is relatively high, and the death of some nodes will cause network holes. The regional node density of node i can be measured by the ratio of the neighbor node density within the perception range of node i to the average node density of the network.
[0042] Neighbor node energy level. The neighbor node energy level characterizes the energy level of the area to which the node belongs and its tolerance for tracking tasks. Due to the dual randomness of node layout and target movement, there are differences in the energy levels of different areas of the network, and such differences largely affect the time when network holes appear. For node i with a higher neighbor node energy level, it dies early due to frequent tracking tasks, but due to the higher energy level of the neighbor nodes, the area's tolerance for tracking tasks is still at a high level, and network holes will not form quickly with the death of node i. On the contrary, for node i with a lower neighbor node energy level, after it dies, the neighbor nodes have insufficient tolerance for tracking tasks due to their lower energy, and network holes are very likely to appear. Therefore, the lower the neighbor node energy level, the more important the node is to the area. For node i, the energy level of its neighbor nodes can be measured by the ratio of the average remaining energy of the neighbor nodes within the sensing range of node i to its own remaining energy.
[0043] Based on the fuzzy logic system, the regional node density and neighbor node energy level are used as inputs, and the input variables are divided into two levels: high (H) and low (L). The output variable is the node importance level, which is divided into three levels: high (H), medium (M), and low (L). The corresponding fuzzy rules are set as follows:
[0044] Table 1 Fuzzy rules
[0045] Regional node density Neighbor node energy level Node importance level L L H L H M H L M H H L
[0046] Cluster members make transmission decisions:
[0047] In the task cluster, the cluster head acts as the data fusion center, and cluster members need to send their sensed data to the cluster head node. At each moment, cluster member nodes make a transmission decision before executing the transmission operation to determine whether their sensed data needs to be transmitted. There are two factors that affect the transmission decision DTrans: the node importance level rankI and the sensed signal strength I:
[0048] DTrans∝(rankI,I) (3)
[0049] A low level of perception signal strength I means that the target perception signal received by the sensor node has been attenuated to a large extent, the signal carries a large error, and even the effective information is submerged by noise. Therefore, transmitting such data to the cluster head for fusion processing will only consume energy and cause a decrease in tracking accuracy, and should be eliminated. Relatively speaking, the higher the I, the smaller the error carried by the signal and the more reliable the information. On the other hand, the node importance level represents the importance of the node to the region. Calling a node with a higher node importance level means that the network will pay a higher price. Therefore, the "cost-benefit" of data transmission behavior can be regarded as a trade-off between the node importance level and the perception signal strength. If the benefit is greater than the cost, the transmission operation is performed.
[0050] Target state estimation and prediction:
[0051] The target moves randomly in the monitoring area, and the network uses dynamic clustering to track the target according to the target position. For energy saving considerations, nodes that do not perform tracking tasks are often in a dormant state, so the network needs to predict the target position at the next moment to activate the corresponding nodes in advance.
[0052] At time t, the current cluster head CH t From the last moment cluster head CH t-1 After receiving the target estimation information, the target detection task begins. When the set detection timer ends, the cluster member sends the sensing data to the cluster head, reporting the distance between it and the target, the node remaining energy information with timestamp, etc. t The data is fused to obtain the target state information at time t, and based on this information, the target state at time t+1 is predicted.
[0053] If CH t In T max No data packet is received from any cluster member. t The "CollectAgain" message will be broadcasted, and the nodes that receive the broadcast message and can sense the target will report the sensed data. max If no data packet is received within 10 seconds, CH tBroadcast the "Stop Work" message to terminate the current target tracking task.
[0054] Information acquisition:
[0055] Selecting appropriate nodes to achieve optimal clustering requires knowing the status information of related sensor nodes. Therefore, obtaining the required information through the node interaction process becomes a key link in organizing clusters. t In its communication radius R c The "Message Gather" message is broadcast within the range, and the node that receives the message sends a message to CH t Send data and report its own status information. t The scheduling decision at time t+1 is made based on this information.
[0056] Adaptive adjustment of task cluster size:
[0057] The task cluster size is positively correlated with the tracking accuracy. However, due to the trade-off between tracking accuracy and energy consumption, it is often sought to reduce node calls within a certain tracking accuracy range to reduce energy consumption. Therefore, it is necessary to adaptively adjust the task cluster size. task It is related to the recent tracking quality level quality_tracking and the distribution of nodes near the target ND:
[0058] N task =f(quality_tracking,ND) (4)
[0059] Recent tracking quality level quality_tracking. This factor reflects the change of recent tracking quality level. If the recent tracking quality level decreases, the task cluster size should be increased at the next moment to improve tracking accuracy. If the recent tracking quality level increases, the task cluster size can be appropriately reduced at the next moment to reduce energy consumption. The recent tracking quality level can be measured by the ratio of the mean tracking error in two consecutive time windows.
[0060] Node distribution ND. This factor reflects the distribution of sensor nodes near the estimated target. Due to the uneven layout of sensor nodes, the nodes in the network are distributed sparsely and densely, and the number of sensor nodes that need to be called to achieve the same tracking accuracy also changes accordingly. The sparser the distribution of sensor nodes, the farther away from the estimated target, and the more sensor nodes need to be called at the next moment. Node distribution and estimated target R s The density of sensor nodes within the range is ρ t and the average distance from the sensor node to the estimated target related:
[0061]
[0062] Determine CCH and CCM:
[0063] According to the predicted target position information at the next moment, the distance to the predicted target R s The sensor nodes within the range can sense the target. These nodes constitute the candidate node set of the task cluster at the next moment, and the candidate nodes are divided into node importance levels. From the perspective of a single node, the lower the node importance level, the weaker the node's impact on the region. Therefore, low-importance nodes should be used to perform tracking tasks first, and hierarchical scheduling based on node importance should be achieved. For this reason, two points are achieved through the decision module: the cluster head selects from the nodes with the lowest importance level and gives priority to nodes with lower importance levels, that is, determining CCH and CCM.
[0064] Candidate cluster head set CCH. After the node importance level division, each candidate node corresponds to a node importance level label. Among them, the candidate nodes corresponding to the lowest node importance level construct the candidate cluster head set CCH, from which the cluster head at the next moment is selected.
[0065] Candidate cluster member set CCM. Since the hierarchical scheduling mechanism based on node importance is adopted, low-importance nodes will be given priority to perform tracking tasks. Therefore, the candidate nodes are arranged in order from low to high node importance, starting from the first node and selecting them in sequence until the total number reaches the task cluster size required at the next moment. The node importance level corresponding to the last node is recorded as rankI high , then rankI high The corresponding candidate nodes construct the candidate cluster member set CCM. The node importance level is lower than rankI high All candidate nodes will be selected into the task cluster at the next moment, and the remaining task cluster nodes will be selected from CCM.
[0066] Select task cluster and cluster head:
[0067] The African vulture optimization algorithm (BAVOA) is a new type of intelligent optimization algorithm that optimizes problems by simulating the foraging and navigation behaviors of African vultures. The African vulture optimization algorithm is discretized and applied to the selection of task cluster nodes and cluster heads.
[0068] (1) Fitness function
[0069] Energy consumption ncp This factor reflects the energy consumption of the candidate task cluster. Since the energy of sensor nodes is limited, this is also the most critical factor restricting the network life. Therefore, if you want to extend the network life, you need to reduce the energy consumption of each tracking task. At the current moment, the cluster head CH t Based on the candidate solution, calculate the predicted energy consumption of the solution.
[0070] The remaining energy balance degree Be. This factor reflects the uniformity of the energy distribution of the candidate task cluster. Due to the randomness of target movement, the pursuit of energy consumption minimization alone will still cause some nodes to be called multiple times and die prematurely, which is not conducive to maximizing the network life. Therefore, in addition to considering energy efficiency, the idea of energy balance should also be taken into consideration. The remaining energy balance degree of the candidate solution can be measured by the variance of the predicted remaining energy of the candidate task cluster node set after performing the tracking task.
[0071] Energy level of candidate cluster head E level This factor reflects the energy level of the candidate cluster head node. Since the energy consumption of the cluster head node in the tracking task cluster is much higher than that of the ordinary cluster members, it is necessary to avoid selecting nodes with too low residual energy as cluster heads. The energy level of the candidate cluster head can be measured by the ratio of the predicted energy consumption of the candidate cluster head to the residual energy.
[0072] Since the dimensions and magnitudes of various factors are different, normalization is required. The specific quantification method is as follows:
[0073] F1=tanh(E ncp ) (6)
[0074] F2=tanh(Be) (7)
[0075] F3=E level (8)
[0076] Based on the linear weighting idea, the above three variables are assigned corresponding weights, and the fitness function F for task cluster selection is established as follows:
[0077] F=λ1F1+λ2F2+λ3F3 (9)
[0078] in, For each candidate solution, the corresponding fitness function value will be calculated, and the solution with the smallest value is the current optimal solution.
[0079] (2)BAVOA initialization
[0080] The individual initialization of BAVOA algorithm is realized by using cyclic shift.
[0081] Step 1: Initialize the first individual. Figure 1 As shown, the nodes of the CCH and CCM sets are sorted from low to high according to the importance of the nodes, and the first node in the CCH and the first node in the CCM are placed. bit is 1 and the rest are 0. is the number of remaining task cluster nodes required at time t+1. In the individual, the 1 in the CCH part indicates that the node corresponding to this bit is selected as the cluster head, and the 1 in the CCM part indicates that the node corresponding to this bit is selected into the task cluster;
[0082] Step 2: Circular shift determines the remaining individuals. The CCH and CCM parts are cyclically shifted. Figure 2 As shown, based on the current individual, the CCH part moves one bit at a time, and the CCM part randomly moves n_shift bits to obtain a new individual. Indicates the number of nodes contained in the CCH set at time t+1.
[0083] If CCH and CCM are the same set, only CCM is used for cyclic movement to generate the required individuals, and the node corresponding to the first 1 in each individual is the candidate cluster head of the current solution.
[0084] (3) Exchange move operator
[0085] Based on the African vulture optimization algorithm, the individual positions are iteratively updated. The position vector of each vulture individual corresponds to a task cluster and a cluster head selection scheme. Since the node selection scheme is a discrete result, the AVOA algorithm needs to be discretized. The position vector is discretized through the V-type transfer function. The specific formula is as follows:
[0086]
[0087] Let the position of the i-th vulture in the t-th iteration obtained by the AVOA algorithm at the t+1-th iteration be and the tth iteration position The difference is Right now:
[0088]
[0089] The corresponding position vector transformation rules are as follows:
[0090]
[0091] Among them, the random number r∈[0,1].
[0092] Let Θ be the exchange operator, then the position of the i-th vulture in the t+1-th iteration is If you remember The CCH and CCM parts are The corresponding two parts are Then the commutative operator represented by Θ is:
[0093] 1) exist Randomly select two different bits from the corresponding positions of 1 in the , and perform the swap operation;
[0094] 2) exist Randomly select two different bits from the corresponding positions of 1 in the , and perform the swap operation;
[0095] Clustering:
[0096] After determining the optimal scheduling strategy at time t+1, the cluster head CH at time t t Send the activation control command to the task cluster node at time t+1, and transmit the predicted target state information and other data to the cluster head CH at time t+1 t+1 To make scheduling decisions. t Report the current target status information to the Sink.
[0097] In order to balance the tracking accuracy and network energy consumption, the present invention proposes the concept of node importance level according to the neighbor node status information. The node importance level is also an important basis for sensor node scheduling, which can comprehensively reflect the importance of the node relative to the region. The wireless sensor network hierarchical scheduling mechanism based on node importance proposed by the present invention achieves two points: first, low-importance nodes are given priority to perform tracking tasks, which ensures the network coverage quality and delays the emergence of network holes; second, selection is made according to the node importance level, which reduces the scale of nodes to be selected, reduces the complexity of clustering, and realizes the lightweight node scheduling algorithm.
[0098] The present invention realizes adaptive adjustment of task cluster size according to the recent tracking quality level and node distribution. Since the task cluster size is positively correlated with the tracking accuracy, the network can better achieve the balance between tracking accuracy and energy consumption by adjusting the task cluster size in real time. When the tracking accuracy level is low, the task cluster size is increased to improve the tracking accuracy. When the tracking accuracy level is high, the task cluster size is appropriately reduced to reduce energy consumption.
[0099] The present invention designs a node scheduling method based on a discretized African vulture optimization algorithm. Individual position vector initialization is achieved through cyclic shifting, and each position vector represents a node selection scheme and corresponds to a fitness function value. The African vulture optimization algorithm realizes a process of continuous optimization in a dynamic environment, through which an optimal strategy, i.e., the individual position with the best fitness function value, can be determined. At the same time, the African vulture optimization algorithm has a fast convergence speed and high solution accuracy, which can ensure that the algorithm has high real-time performance.
Claims
1. A wireless sensor network hierarchical scheduling method based on node importance, characterized in that: The following steps are involved: When a target appears in the monitoring area, the sensor nodes automatically cluster to form an initial task cluster; Each cluster member in the task cluster makes a transmission decision based on its own node importance level and perceived signal strength, and determines whether its perceived data needs to be transmitted to the cluster head; The cluster head fuses the received target perception data, obtains the target state estimation information, and predicts the target position at the next moment based on the estimated value; The cluster head broadcasts a request to obtain node status information within the communication radius, and the sensor node that receives the request sends its status information to the cluster head; The cluster head adaptively adjusts the task cluster size at the next moment according to the recent tracking quality level and the distribution of nodes near the predicted target location; Based on the received node status information, the cluster head determines the candidate node set of the task cluster at the next moment according to the target predicted position, and further determines the candidate cluster head set and the candidate cluster member set based on the candidate task cluster node set according to the task cluster size at the next moment and the node importance level of each sensor node in the candidate node set; The discrete African vulture optimization algorithm is used to select the cluster head and cluster members at the next moment from the candidate cluster head set and candidate cluster member set. Specifically: Sort the sensor nodes in the candidate cluster head set and candidate cluster member set from low to high according to the node importance level, and place the first in the candidate cluster head set and the first in the candidate cluster member set. The bit is 1, and the rest are 0, where is the number of remaining task cluster nodes required at time t+1; in the individual, the 1 in the candidate cluster head set part indicates that the sensor node corresponding to this position is selected as the cluster head, and the 1 in the candidate cluster member set part indicates that the sensor node corresponding to this position is selected into the task cluster; wherein, time t+1 is the next moment after time t, and time t is any moment in the target movement process; the remaining individuals are determined by cyclic shift, specifically: the candidate cluster head set part and the candidate cluster member set part are cyclically shifted respectively, the candidate cluster head set part moves one position at a time, and the candidate cluster member set part randomly moves n_shift positions to obtain a new individual; Based on the discrete African vulture optimization algorithm, the individual positions are iteratively updated. The position vector of each vulture individual corresponds to a task cluster and a cluster head selection scheme. The fitness function of the discrete African vulture optimization algorithm is F = λ1tanh(E ncp )+λ2tanh(Be)+λ3E level , where E ncp represents the energy consumption of executing the tracking task, Be represents the remaining energy balance, and E level represents the energy level of candidate cluster heads, λ1,λ2,λ3 are weight coefficients, and satisfy 2. The wireless sensor network hierarchical scheduling method based on node importance according to claim 1 is characterized in that: The invention also includes determining the node importance level of each sensor node based on a fuzzy logic system with the regional node density and the neighbor node energy level as input.
3. The wireless sensor network hierarchical scheduling method based on node importance according to claim 1 is characterized in that: When each cluster member in the task cluster makes a transmission decision based on its own node importance level and perceived signal strength, the cluster member with a lower node importance level has a greater probability of data transmission, and the cluster member with a higher perceived signal strength has a greater probability of data transmission.
4. The wireless sensor network hierarchical scheduling method based on node importance according to claim 1 is characterized in that: The sensing data includes the distance between the sensor node itself and the target and the node remaining energy information with a timestamp.
5. The wireless sensor network hierarchical scheduling method based on node importance according to claim 1 is characterized in that: After the cluster head broadcasts a request to obtain node status information within the communication radius, if no data packet is received from any sensor node in reply within a preset time, the cluster head broadcasts the request to obtain node status information again. If no data packet is received from any sensor node in reply within the next preset time, a message to stop working is broadcast to terminate the current target tracking task.
6. The wireless sensor network hierarchical scheduling method based on node importance according to claim 1 is characterized in that: The cluster head adaptively adjusts the size of the task cluster at the next moment according to the recent tracking quality level and the node distribution near the predicted target position. Specifically, when the recent tracking quality level decreases, the size of the task cluster is increased at the next moment; when the recent tracking quality level increases, the size of the task cluster is reduced at the next moment; the sparser the distribution of sensor nodes and the farther the distance from the estimated target, the larger the size of the task cluster at the next moment; the denser the distribution of sensor nodes and the closer the distance from the estimated target, the smaller the size of the task cluster at the next moment.
7. The wireless sensor network hierarchical scheduling method based on node importance according to claim 1 is characterized in that: When determining the candidate cluster head set and the candidate cluster member set, the node corresponding to the lowest node importance level in the candidate node set is used to construct the candidate cluster head set, and the candidate nodes are sorted from low to high according to the node importance level, and are selected from the first one in sequence until the total number reaches the task cluster size required at the next moment. The node importance level corresponding to the last sensor node is recorded as rankI. high , then the node importance level rankI high The corresponding candidate nodes construct the candidate cluster member set; the node importance level is lower than rankI high All candidate nodes will be selected into the task cluster node set at the next moment, and the remaining task cluster nodes will be selected from the candidate cluster member set.
