Analysis Method for Nodes of Wireless Sensor Network Based on UAV Communication

Abstract

The present invention discloses an analysis method for wireless sensor network nodes based on drone communication, belonging to the technical field of data acquisition. The specific steps of the analysis method are as follows: (1) Collect and record the working state of the drone; (2) Construct and train a test model; (3) Plan the flight route of the drone. By using a genetic algorithm to update the flight path of the drone in real time, the present invention can greatly reduce its usage limitations, improve the data acquisition efficiency of the drone, and enhance the user experience of the staff.

Description

Technical Field

[0001] The present invention relates to the technical field of data acquisition, and particularly to an analysis method for wireless sensor network nodes based on UAV communication. Background Art

[0002] With the gradual maturity of UAV technology, the manufacturing cost and entry threshold have been reduced, and the consumer UAV market has exploded. The civilian UAV market is on the verge of explosion. In recent years, the application of UAV technology in wireless communication has become more and more extensive. UAVs have the characteristics of high flexibility and strong mobility, and can be used as mobile aerial base stations to assist ground base station communication.

[0003] After retrieval, Chinese Patent No. CN111752304A discloses a UAV data acquisition method and related equipment. Although this invention can ensure the shortest UAV data acquisition time and ensure that the UAV can be charged in time, it cannot plan the UAV flight path in real time, has great limitations in use, and reduces the working experience of staff. Therefore, we propose an analysis method for wireless sensor network nodes based on UAV communication. Summary of the Invention

[0004] The purpose of the present invention is to solve the defects existing in the prior art, and propose an analysis method for wireless sensor network nodes based on UAV communication.

[0005] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0006] An analysis method for wireless sensor network nodes based on UAV communication, and the specific steps of the analysis method are as follows:

[0007] (1) Collect and record the UAV working state: The UAV starts from the data center, collects data from each group of wireless sensors, and the computing platform collects the UAV working state in real time.

[0008] (2) Construct and train a test model: Construct a test model based on the positions of each group of wireless sensors, and train and optimize the test model.

[0009] (3) Plan the UAV flight route: Import the UAV working state into the test model to simulate and plan the UAV collection route, and notify the UAV to collect wireless sensor data according to the planned flight route.

[0010] As a further solution of the present invention, the specific steps of the training and optimization in step (2) are as follows:

[0011] Step 1: The computing platform communicates with the remote sensing satellite and receives the environmental information of the location where the wireless sensor is located collected by the remote sensing satellite.

[0012] Step 2: Import the environmental information into the test model, and at the same time mark the locations of each wireless sensor in the test model;

[0013] Step 3: Import the working state information of the drone into the test model. At the same time, the test model extracts the flight data of the drone and constructs an observation data set;

[0014] Step 4: Select an observation data from the observation data set as the verification data, and use this verification data to verify the accuracy of the test model through the root mean square error, and repeat this process multiple times;

[0015] Step 5: For each group of data, select any subset as the test set, and then take the remaining subset as the training set. Again, calculate the root mean square error until a prediction is made for each group of data, and output the data with the best prediction result as the optimal parameter;

[0016] Step 6: Standardize the training data set according to the optimal parameter. Finally, send the training samples into the learning network model, set the specific parameters of the model by yourself, and train the model using the long-term iteration method.

[0017] As a further solution of the present invention, the working state of the drone described in Step 3 includes the percentage of data collected by the wireless sensor, the position of the drone, and the battery power of the drone.

[0018] As a further solution of the present invention, the specific steps of the simulation planning in Step (3) are as follows:

[0019] The first step: The test model simulates the working state of the drone, and at the same time collects the time required for the drone to complete one collection of data within the wireless sensor node;

[0020] The second step: The test model plans a set of flight routes for the drone based on the positions of each group of wireless sensors and the environmental information, and at the same time enters them into the storage unit of the drone;

[0021] The third step: After the drone takes off from the data center, it sends a signal to activate the communication mode of the sensor nodes within its communication range. When the dormant sensor nodes receive the signal, they enter the active state and wait for the drone to fly over to transmit data, and re-enter the dormant state after completing the data transmission;

[0022] The fourth step: The drone performs simulated flight according to the flight route in the storage unit and conducts data transmission with each group of wireless sensors. The data transmission rate can be specifically expressed as:

[0023]

[0024] Among them, B is the available bandwidth for communication, Pi is the transmission power of the ground sensor node, ρ i is the channel power gain at a reference distance of 1 meter, σ 2 is the power of the channel noise, and h is the fixed flight altitude of the UAV during the process of collecting node data;

[0025] Step 5: Construct a data collection probability model and calculate the probability of successful UAV collection when the UAV collects data from wireless sensor nodes multiple times;

[0026] Step 6: The UAV detects the surrounding environmental information through the on-board sensors during flight and performs path optimization in real-time and in stages.

[0027] As a further solution of the present invention, the specific calculation formula for the probability of successful UAV collection in Step 5 is as follows:

[0028]

[0029]

[0030] Among them, ξ represents the parameter for evaluating the sensing performance, represents the distance between the UAV and the sensor node being collected at this time, represents the number of the i-th data collection when the UAV collects the data of node n, and g th represents the minimum probability threshold for successful data collection;

[0031] The specific calculation formula for its age of information in Step 5 is as follows:

[0032]

[0033] A i (t)| f =A i (t - 1)+1 (6)

[0034]

[0035] Among them, T i b is the moment when the UAV starts to collect the data of sensor node N i , and T i d is the moment when the UAV starts to transmit the data of sensor node N i to the data center. In addition, formula (5) represents the age of information corresponding to the sensor node when the collection is successful, formula (6) represents the age of information corresponding to the sensor node when the collection fails, and formula (7) represents the age of information of a single sensor node at time slot t.

[0036] As a further solution of the present invention, the specific steps of the path optimization in the sixth step are as follows:

[0037] S1: The drone adds the sensor nodes that have completed data collection to the sensor node queue Γ(x) of the drone flight trajectory, and represents the queue of sensor nodes that have not completed data collection as where x represents the x-th sensor node that has completed data collection, y represents the y-th sensor node that has not completed data collection, and both x and y are included in the set of natural numbers N;

[0038] S2: After the drone obtains the position information, the number of sensor nodes, and the surrounding environment information of the sensor nodes capable of collecting data, it collects the drone flight path information and adjusts it. The specific flight path is specifically represented as follows:

[0039]

[0040] where, represents the drone flight path, k represents the k-th takeoff when the drone completes data collection of all sensor nodes, and the drone flight path is composed of multiple sub-paths, and τ k，n represents the n-th sub-path in the drone flight path during the k-th takeoff;

[0041] S3: Represent the set of all flight paths of the drone as a population, and generate a population matrix in combination with the genetic algorithm. The specific matrix representation form is as follows:

[0042]

[0043] where P represents the population matrix;

[0044] S4: Collect the distance between two sensor nodes and the energy consumed by the drone flying over the two sensor nodes, and represent the drone flight trajectory through the two sets of collected data. The drone flight trajectory is represented as:

[0045]

[0046] s.t.∑E i,j <E i (10)

[0047] where L i,j represents the distance between two sensor nodes, E i,j represents the energy consumed by the drone flying over the two sensor nodes, and E i represents the initial energy of the drone;

[0048] S5: Collect the flight path quality of the UAV, quantify the attributes of each individual in the population through the fitness function, and calculate the flight energy consumption of the UAV between two nodes. The formula for calculating its flight energy consumption is as follows:

[0049] E i,j = μL i,j (11)

[0050] Among them, E i,j and L i,j For the explanation of the unknowns, refer to formula (10). μ represents the energy consumption coefficient;

[0051] S6: When the flight energy consumption E i,j is less than the initial energy E i , the UAV continues to fly to the next sensor node; otherwise, it will return to the data center. After the UAV returns to the data center, minimize the Aol of multiple tasks. The formula for minimizing Aol is as follows:

[0052]

[0053] P2: min A i

[0054] s.t. A i (t)| s , A i (t)| f (13)

[0055] Among them, represents the flight time of the UAV between two adjacent sensor nodes;

[0056] S7: Randomly select two individuals from the population P and Then, select a certain path τ and from the individuals a,t , τ b,t respectively, and then exchange them to obtain two new individuals. Among them,

[0057] S8: Randomly generate a coefficient η, When η < Υ, randomly select an individual Randomly select two paths τ k,c and τ k,d in the selected individual for exchange, and at the same time, k is incremented by one in sequence, and continue to execute the above steps until k > m;

[0058] S9: After the path optimization is completed, traverse each node starting from the path end point. If a certain node M can be connected to the starting point without obstacles, the nodes between the starting point and node M are redundant nodes. After the redundant nodes are confirmed, delete these redundant nodes and recalculate the fitness function of the path, and continuously optimize the path through continuous iteration. The specific calculation formula for the probability that an individual e is selected is as follows:

[0059]

[0060] where G is the fitness matrix;

[0061] S10: Establish a new matrix by selecting an individual from P C times with a probability of At the same time, select the individual with the maximum fitness and splice it with to form a new population to complete the iteration of the flight path and save the optimal flight path.

[0062] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0063] 1. Compared with the previous acquisition methods, in the analysis method of wireless sensor network nodes based on UAV communication of the present invention, during the flight of the UAV, the surrounding environmental information is detected by the on-board sensors. At the same time, the set of all flight paths of the UAV is represented as a population, and a population matrix is generated in combination with the genetic algorithm. At the same time, the distance between two sensor nodes and the energy consumed by the UAV flying over the two sensor nodes are collected. The flight trajectory of the UAV is represented by the two sets of collected data, and then the quality of the UAV flight path is collected. The attributes of each individual in the population are quantified through the fitness function, and the redundant nodes existing in the UAV flight path are deleted and the fitness function of the path is recalculated, and the path is continuously optimized through continuous iteration. By using the genetic algorithm to update the UAV flight path in real time, its use limitations can be greatly reduced, and at the same time, the UAV data acquisition efficiency can be improved, and the working experience of the staff can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention.

[0065] Figure 1 It is a flowchart of the analysis method of wireless sensor network nodes based on UAV communication proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] Refer to Figure 1, an analysis method for nodes of a wireless sensor network based on UAV communication, the specific steps of this analysis method are as follows:

[0067] Collect the working status of the UAV and record: The UAV departs from the data center, conducts data collection on each group of wireless sensors, and the computing platform collects the working status of the UAV in real time.

[0068] Construct and train a test model: Construct a test model based on the positions of each group of wireless sensors and train and optimize this test model.

[0069] Specifically, the computing platform communicates with a remote sensing satellite, receives the environmental information of the location where the wireless sensors are located collected by the remote sensing satellite, and at the same time imports the environmental information into the test model. At the same time, mark the positions of each wireless sensor in the test model. At the same time, the test model selects an observation data from the observation dataset as the validation data, and uses this validation data to verify the accuracy of the test model through the root mean square error. Repeat this many times. For each group of data, select any subset as the test set, and then take the remaining subset as the training set, and count the root mean square error again until a prediction is made for each group of data, and output the data with the best prediction result as the optimal parameter. At the same time, standardize the training dataset according to the optimal parameter, and finally send the training samples to the learning network model, set the specific parameters of the model by yourself, and train this model using the long-term iteration method.

[0070] It should be further noted that the working status of the UAV includes the percentage of data already collected by the wireless sensors, the UAV position, and the UAV battery power.

[0071] Plan the UAV flight route: Import the UAV working status into the test model to simulate and plan the UAV collection route, and notify the UAV to collect wireless sensor data according to the planned flight route.

[0072] Specifically, the test model simulates the working state of the drone, and at the same time collects the time required for the drone to complete one acquisition of the data within the wireless sensor node. According to the positions of each group of wireless sensors and the environmental information, a set of flight routes are planned for the drone and recorded in the drone's storage unit. After taking off from the data center, the drone sends a signal to activate the communication mode of the sensor nodes within its communication range. When the sensor nodes in the sleep state receive the signal, they enter the active state and wait for the drone to fly over to transmit data, and then re-enter the sleep state after completing the data transmission. The drone performs a simulated flight according to the flight routes in the storage unit and conducts data transmission with each group of wireless sensors, constructs a data acquisition probability model, and calculates the probability of successful acquisition by the drone when it performs multiple acquisitions of the data in the wireless sensor node. At the same time, during the flight, the drone detects the surrounding environmental information through the on-board sensors and optimizes the path in real time in stages.

[0073] It should be further noted that the specific data transmission rate can be expressed as:

[0074]

[0075] where B is the available bandwidth of the communication, Pi is the transmission power of the ground sensor node, ρ i is the channel power gain at a reference distance of 1 meter, σ 2 is the power of the channel noise, and h is the fixed flight altitude of the drone during the process of collecting data from the node;

[0076] The specific calculation formula for the successful acquisition probability of the drone is as follows:

[0077]

[0078] where ξ represents the parameter for evaluating the sensing performance, represents the distance between the drone and the sensor node being collected at this time, represents the number of the i-th data acquisition when the drone collects the data of node n, and g th represents the minimum probability threshold for successful data acquisition;

[0079] The specific calculation formula for its age of information is as follows:

[0080]

[0081] A i (t)| f =A i (t - 1)+1 (6)

[0082]

[0083] where, T i b is the moment when the UAV starts to collect data from the sensor node N i , and T i d is the moment when the UAV starts to transmit the data of the sensor node N i to the data center. In addition, formula (5) represents the age of information corresponding to the sensor node when the collection is successful, formula (6) represents the age of information corresponding to the sensor node when the collection fails, and formula (7) represents the age of information of a single sensor node in time slot t.

[0084] In addition, it should be further noted that the UAV adds the sensor nodes that have completed data collection to the sensor node queue Γ(x) of the UAV flight trajectory, and represents the queue of sensor nodes that have not completed collection as where x represents the x-th sensor node that has completed data collection, y represents the y-th sensor node that has not completed collection, both x and y are included in the set of natural numbers N. At the same time, after the UAV obtains the position information, the number of sensor nodes, and the surrounding environment information of the sensor nodes that can collect data, it collects the UAV flight path information and adjusts it. The set of all UAV flight paths is represented as a population, and a population matrix is generated in combination with the genetic algorithm. At the same time, the distance between two sensor nodes and the energy consumed by the UAV flying over the two sensor nodes are collected. The UAV flight trajectory is represented by the two sets of collected data, and then the quality of the UAV flight path is collected. The attributes of each individual in the population are quantified through the fitness function. At the same time, the flight energy consumption of the UAV between two nodes is calculated. When the flight energy consumption E i,j is less than the initial energy E i , the UAV continues to fly to the next sensor node, otherwise it will return to the data center. When the UAV returns to the data center, the Aol of multiple tasks is minimized. The computing platform randomly selects two individuals from the population P and then selects a certain path segment τ and from the individuals a,t 、τ b,t respectively, and then exchanges them to obtain two new individuals. Among them, then a coefficient η is randomly generated, when η < Υ, a random individual is selected and two path segments τ k,c and τ k,dExchange them while incrementing k by one in sequence, and continue to execute the above steps until k > m. After the path optimization is completed, traverse each node starting from the path end point. If a certain node M can be connected to the starting point without obstacles, the nodes between the starting point and node M are redundant nodes. After the redundant nodes are confirmed, delete these redundant nodes and recalculate the fitness function of the path, and continuously optimize the path through continuous iteration. By probability Select C individuals from P to establish a new matrix At the same time, select the individual with the maximum fitness and Concatenate to form a new population To complete the iteration of the flight path and save the optimal flight path.

[0085] It should be further noted that the specific flight path is specifically represented as follows:

[0086]

[0087] Among them, represents the UAV flight path, k represents the number of takeoffs of the UAV when all sensor node data is collected, and the UAV flight path consists of multiple sub-paths, and τ k，n represents the nth sub-path in the UAV flight path during the kth takeoff;

[0088] Its specific matrix representation form is as follows:

[0089]

[0090] Among them, P represents the population matrix;

[0091] Its UAV flight trajectory is expressed as:

[0092]

[0093] s.t.∑E i,j <E i (10)

[0094] Among them, L i,j represents the distance between two sensor nodes, E i,j represents the energy consumed by the UAV flying over two sensor nodes, and E i represents the initial energy of the UAV;

[0095] Its flight energy consumption calculation formula is as follows:

[0096] E i,j =μL i,j (11)

[0097] Among them, E i,j and L i,j For the explanation of the unknown quantity, refer to formula (10). μ represents the energy consumption coefficient. The formula for minimizing Aol is as follows:

[0098]

[0099] P2: min A i

[0100] s.t. A i (t)| s , A i (t)| f (13)

[0101] Among them, represents the probability that the individual e of the flight time between two adjacent sensor nodes of the UAV is selected. The specific calculation formula is as follows:

[0102]

[0103] Among them, G is the fitness matrix.

Claims

1. Analysis method for nodes of a wireless sensor network based on drone communication, characterized in that, The specific steps of this analysis method are as follows: (1) Collect and record the working status of the drone: The drone departs from the data center, collects data from each group of wireless sensors, and the computing platform collects the working status of the drone in real time; (2) Construct and train the test model: Construct a test model based on the positions of each group of wireless sensors and train and optimize the test model; (3) Plan the flight route of the drone: Import the working status of the drone into the test model to simulate and plan the data collection route of the drone, and notify the drone to collect wireless sensor data according to the completed flight route; The specific steps of the simulation planning in step (3) are as follows: The first step: The test model simulates the working status of the drone and simultaneously collects the time required for the drone to complete one collection of the data within the wireless sensor node; The second step: The test model plans a set of flight routes for the drone based on the positions of each group of wireless sensors and environmental information, and simultaneously enters them into the storage unit of the drone; The third step: After the drone takes off from the data center, it sends a signal to activate the communication mode of the sensor nodes within its communication range. When the sensor nodes in the sleep state receive the signal, they enter the active state and wait for the drone to fly over to transmit data, and re-enter the sleep state after completing the data transmission; The fourth step: The drone performs a simulated flight according to the flight route in the storage unit and transmits data with each group of wireless sensors. The data transmission rate can be specifically expressed as: where B is the available bandwidth for communication, Pi is the transmission power of the ground sensor node, ρ i is the channel power gain at a reference distance of 1 meter, σ 2 is the power of the channel noise, and h is the fixed flight altitude of the UAV during the process of collecting node data; The fifth step: Construct a data collection probability model, and calculate the probability of successful collection by the drone when collecting the data in the wireless sensor nodes multiple times. At the same time, calculate the age of the data information of the successfully collected sensor nodes; The sixth step: The drone detects the surrounding environmental information through the on-board sensors during flight and performs path optimization in real time and in stages; The specific steps of the path optimization in the sixth step are as follows: S1: The drone adds the sensor nodes that have completed data collection to the sensor node queue Γ(x) of the drone flight trajectory, and represents the sensor node queue that has not completed collection as where x represents the x-th sensor node that has completed data collection, y represents the y-th sensor node that has not completed collection, and both x and y are included in the set of natural numbers N; S2: After the drone obtains the position information of the sensor nodes capable of collecting data, the number of sensor nodes, and the surrounding environmental information, it collects the drone flight path information and adjusts it. The specific flight path is specifically expressed as follows: Among them, represents the flight path of the UAV, k represents the k-th takeoff when the UAV completes the data collection of all sensor nodes, and the flight path of the UAV consists of multiple sub-paths, and τ k，n represents the n-th sub-path in the flight path of the UAV during the k-th takeoff; in the S3: Represent the set of all flight paths of the drone as a population, and generate a population matrix in combination with the genetic algorithm. The specific matrix representation form is as follows: Among them, P represents the population matrix; S4: Collect the distance between two sensor nodes and the energy consumed by the drone flying over the two sensor nodes, and represent the drone flight trajectory through the two sets of collected data. The drone flight trajectory is expressed as: Among them, L i,j represents the distance between two sensor nodes, and E i,j represents the energy consumed by the UAV flying over two sensor nodes, and E i represents the initial energy of the UAV; S5: Collect the quality of the drone flight path, quantify the attributes of each individual in the population through the fitness function, and calculate the flight energy consumption of the drone between two nodes. The flight energy consumption calculation formula is as follows: E i,j = μL i,j (11) Among them, E i,j and L i,j For the explanation of the unknowns, refer to formula (10), where μ represents the energy consumption coefficient; S6: When the flight energy consumption E i,j is less than the initial energy E i , the UAV continues to fly to the next sensor node; otherwise, it returns to the data center. After the UAV returns to the data center, the Aol of multiple tasks is minimized, and its Aol minimization processing formula is as follows: Among them, represents the flight time of the drone between two adjacent sensor nodes; S7: Randomly select two individuals from the population P and Then, select a certain path τ and from the individuals respectively a,t 、τ b,t , and then exchange them to obtain two new individuals. Among them, S8: Randomly generate a coefficient η, When η < Υ, randomly select an individual Randomly select two segments of paths τ in the individual k,c and τ k,d Exchange them. At the same time, increment k by one in sequence, and continue to execute the above steps until k > m; S9: After the path optimization is completed, traverse each node starting from the end point of the path. If a certain node M can be connected to the starting point without obstacles, the nodes between the starting point and node M are redundant nodes. After the redundant nodes are confirmed, delete these redundant nodes and recalculate the fitness function of the path, and continuously optimize the path through continuous iteration. The specific calculation formula for the probability that an individual e is selected is as follows: where G is the fitness matrix; S10: Establish a new matrix by selecting individuals from P C times with probability Meanwhile, select the individual with the maximum fitness and concatenate them to form a new population to complete the iteration of the flight path and save the optimal flight path.

2. The analysis method of a wireless sensor network node based on drone communication according to claim 1, wherein, The specific steps of the training optimization described in step (2) are as follows: Step 1: Calculate the communication connection between the platform and the remote sensing satellite, and receive the environmental information of the location where the wireless sensor is collected by the remote sensing satellite; Step 2: Import the environmental information into the test model, and at the same time mark the location of each wireless sensor in the test model; Step 3: Import the UAV working state information into the test model, and at the same time the test model extracts the UAV flight data and constructs an observation data set; Step 4: Select an observation data from the observation data set as the verification data, and use this verification data to verify the accuracy of the test model through the root mean square error, and repeat this many times; Step 5: For each group of data, select any subset as the test set, and then take the remaining subset as the training set, and count the root mean square error again until each group of data is predicted once, and output the data with the best prediction result as the optimal parameter; Step 6: Standardize the training data set according to the optimal parameters, and finally send the training samples to the learning network model, set the specific parameters of the model by yourself, and train the model using the long-term iteration method.

3. The analysis method of a wireless sensor network node based on drone communication according to claim 2, wherein, The UAV working state described in step 3 includes the percentage of data collected by the wireless sensor, the UAV position, and the UAV battery power.

4. The analysis method of a wireless sensor network node based on drone communication according to claim 1, wherein, The specific calculation formula for the UAV acquisition success probability described in step 5 is as follows: Among them, ξ represents the parameter for evaluating the sensing performance, representing the distance between the UAV and the sensor node being collected at this time, indicating the number of the i-th data collection when the UAV collects data from node n, g th representing the minimum probability threshold for successful data collection; The specific calculation formula for its information age described in step 5 is as follows: A i (t)| f = A i (t - 1)+1 (6) Among them, T i b is the moment when the UAV starts to collect data from the sensor node N i and T i d is the moment when the UAV starts to transmit data from the sensor node N i to the data center. In addition, formula (5) represents the age of information corresponding to the sensor node when the collection is successful, formula (6) represents the age of information corresponding to the sensor node when the collection fails, and formula (7) represents the age of information of a single sensor node in time slot t.

Images