Path planning method for mobile sink nodes based on MR-WSNs

Abstract

The present invention discloses a path planning method based on MR-WSNs mobile Sink nodes, including: clustering nodes in MR-WSNs based on preset indicators and threshold functions, establishing a cluster head node set, wherein the cluster head node is a traversal point of a mobile Sink node; constructing an optimal traversal point set of a mobile Sink node according to the location information of the cluster head node and the communication coverage area, and the mobile position of the sink point is based on the location distance that is most conducive to collecting cluster head information; using an improved ant colony algorithm, solving the optimal traversal point set of the cluster head node as a path planning problem, and obtaining the optimal mobile path of the mobile Sink node. The present invention can not only enable the Sink node to move to a suitable communication position, improve the efficiency of data transmission, but also enhance the stability of data transmission, and effectively avoid the energy hole problem of the network.

Description

Technical Field

[0001] The invention belongs to the technical field of wireless sensor networks, and in particular relates to a path planning method based on MR-WSNs mobile Sink nodes. Background Art

[0002] As wireless sensor networks (WSNs) have entered the stage of large-scale networking applications, large-scale WSNs (Large Wireless Sensor Networks, L-WSN) have been widely used in many fields, such as marine ranching, intelligent transportation and smart agriculture projects. Modern marine ranching needs to establish a monitoring network for wide-area water bodies and ecological environments. It is necessary to deploy multiple types of sensor nodes to conduct in-situ real-time monitoring of various environmental profile information such as marine water bodies, ecology, and biology, and establish marine ranch wireless sensor networks (MR-WSNs). There are many representative achievements in this field. Haitao C et al. designed a marine environment monitoring system based on the Internet of Things based on ZigBee and 4G wireless communication technology, which realized the collection and transmission of marine environmental data. However, due to the harsh environment of the sea area, the network stability is poor and it is not suitable for large-area aquaculture environments; Jin Guang et al. used LoRa technology to achieve large-scale data collection in the sea area with low power consumption in order to solve the problem of large island environment and inconvenient inspection. Zhang Qin et al. also designed a low-power, long-distance sea monitoring WSN network through LoRa technology. The experiment compared the Zig Bee multi-hop network. Both of them have improved in low power consumption and stability. However, since the sink node is fixed at a specific location, the sensor nodes in the network need to transmit data to the sink node after collecting data, which increases the data forwarding task of the neighboring nodes, resulting in excessive energy consumption of the neighboring nodes. Eventually, the nodes form energy holes due to energy exhaustion. In response to the energy hole problem, many scholars have proposed adding mobile nodes to WSN for data collection. For example, Yang Ying et al. proposed a data collection strategy for farmland wireless sensor networks based on mobile nodes, which extended the network life cycle, but the path planning for mobile nodes was simple and the improvement effect was not ideal. In response to the problem of poor planning effect of mobile nodes in wide-area wireless sensor networks, He Dong et al. proposed a path planning method based on energy perception, which achieved the optimal path selection by calculating the weight coefficient of the Sink node in the moving direction. Xu Jia et al. designed a low-energy trajectory constraint algorithm to minimize the mobile loss of the Sink node, but the algorithm was complex and shortened the network survival time. In order to maximize the network survival time, some scholars have used intelligent algorithms to plan the path of mobile sinks. For example, Xie Yinghui et al. proposed an algorithm that uses genetic algorithms to solve the optimal residence point of mobile sink nodes, and constructs mobile paths based on the residence points. Qiao Xuegong et al. proposed an improved gray wolf optimization algorithm based on a nonlinear dynamic change convergence factor based on a Gaussian decreasing strategy, and used TSP to solve the shortest mobile path of mobile beacon nodes. Feng Hao et al. used the firefly algorithm improved by Ying for mobile node path planning, which improved the algorithm solution accuracy and accelerated the convergence speed of the algorithm. In response to the problem of premature convergence of traditional ant colony algorithms, Ji Shanshan introduced genetic operators and proposed a WSN mobile sink path planning algorithm based on an enhanced ant colony algorithm, which improved the balance of network energy consumption.Xu Yuqiong et al. improved the problem of the lack of robustness of the traditional ant colony algorithm in path planning by adopting a multi-step selection strategy. Zhu Zhengwei et al. proposed a fusion immune algorithm and simulated annealing particle swarm algorithm for planning the movement path of nodes; Jian-Li Duan et al. proposed a hierarchical marine environment monitoring (HMMN-EC) model based on edge computing, where ships serve as edge computing nodes, receive and process data collected by sensor nodes, and forward them to shore base stations through drones, ultimately achieving efficient monitoring services in harsh marine environments.

[0003] The contribution proposed by the existing literature is to reduce the number of terminal nodes by using the long-distance and low-power characteristics of LoRa technology, introduce mobile sink points to reduce the "relay transmission hops" of L-WSN; design mobile node path planning algorithms to improve data collection efficiency and balance network node load energy consumption; introduce group puzzle algorithms to increase algorithm intelligence, reduce network energy consumption to a certain extent, and extend network life cycle. However, there are the following problems: ① The path planning algorithm proposed in the literature only considers the path length and running speed, but not the network scale, especially the mobile planning of sink points for specific targets in large-scale network scenarios, which weakens the advanced nature; ② The algorithm has a single number of mobile sink nodes, and does not take into account the robustness of the mobile nodes themselves, data transmission reliability, and engineering mobile scenarios. Comprehensive consideration; ③ The group puzzle algorithm used lacks pertinence, the optimization method considers few factors, and the algorithm computing power is not strong enough. Therefore, the current WSN path planning method is still not suitable for large-scale application scenarios such as marine ranches. Summary of the invention

[0004] The purpose of the present invention is to propose a path planning method for a mobile Sink node of a stable communication for a marine ranch wireless sensor network (MR-WSNs), comprising: clustering the nodes in the MR-WSNs based on a preset index and a threshold function, establishing a cluster head node set, wherein the cluster head node is a traversal point of a mobile Sink node; constructing an optimal traversal point set of a mobile Sink node according to the location information of the cluster head node and the communication coverage area, and the mobile position of the sink point is based on the location distance that is most conducive to collecting cluster head information (the present invention ignores the distance between the sink point and the information collection cluster head for research convenience); using an improved ant colony algorithm, solving the optimal traversal point set of the cluster head node as a path planning problem, and obtaining the optimal moving path of the mobile Sink node. The present invention can not only enable the Sink node to move to a suitable communication position, improve the efficiency of data transmission, but also enhance the stability of data transmission, and effectively avoid the energy hole problem of the network.

[0005] To achieve the above object, the present invention provides a path planning method based on MR-WSNs mobile Sink node, comprising:

[0006] Based on the preset indicators and threshold functions, the nodes in the marine ranch wireless sensor network are clustered and a cluster head node set is constructed.

[0007] Select traversal points according to preset conditions, optimize the cluster head node set, and establish the optimal traversal point set;

[0008] The improved ant colony algorithm is used to solve the path planning of the optimal traversal point set to obtain the optimal moving path.

[0009] Optionally, obtaining the cluster head node set includes:

[0010] Based on the monitoring area and the node communication radius, the number of cluster heads is obtained;

[0011] According to the number of cluster heads, based on the preset index and threshold function, the nodes in the ocean ranch wireless sensor network are clustered to obtain the cluster head node set.

[0012] Optionally, the number of cluster heads is:

[0013]

[0014] Where k is the number of cluster heads, d max is the maximum distance between the node and the mobile sink node, d min is the minimum distance between the node and the MS, R is the communication radius of the static node, L is the side length of the monitoring area, and n is the number of static nodes in the monitoring area.

[0015] Optionally, the preset indicators include: node energy factor, node density factor and node spacing factor;

[0016] The node energy factor is:

[0017]

[0018] Among them, E node (i) represents node s i The remaining energy

[0019] The node density factor is:

[0020]

[0021] Among them, N i Represents node s i The two-hop neighbor node of

[0022] The node spacing factor is:

[0023]

[0024] Among them, d i Represents node s i Distance to MS, d j It represents the distance from the cluster head participating in the election to the MS in this round, and R represents the radius of the center circle.

[0025] Optionally, the threshold function is:

[0026]

[0027] Among them, P i represents the proportion of cluster heads, r is the number of current execution rounds, G represents the set of nodes that were not cluster heads in the previous round, and γ i Represents the fitness factor constructed by energy factor, density factor and spacing factor.

[0028] Optionally, the preset condition for selecting the traversal point includes:

[0029] The moving distance of the MS is the shortest, the number of traversal points is the least, and each traversal point must be within the communication coverage of the node.

[0030] Optionally, establishing the optimal traversal point set includes:

[0031] S1. Input cluster head node set C, traverse point set Q, traverse zone set F;

[0032] S2. Determine whether the traversal areas of adjacent cluster head nodes intersect. If so, the intersection becomes a new traversal point. If not, find a traversal point on the boundary of the communication range R and satisfy

[0033] S3. Output the optimal traversal point and set Q;

[0034] Among them, q i and q j is the traversal point on the boundary of the communication range, c i and c j is the cluster head node, They represent the Euclidean distance of the traversal points and the Euclidean distance of the cluster head node respectively.

[0035] Optionally, using an improved ant colony algorithm to perform path planning and solving the traversal node set includes:

[0036] S1. Data initialization establishes a random rectangular coordinate system and imports the position coordinate values ​​of n cluster head nodes;

[0037] S2. Calculate the distance between any two cluster head nodes and obtain the initial matrix of heuristic information;

[0038] S3. Initialize population parameters;

[0039] S4. Calculate the initial pheromone concentration value;

[0040] S5. Randomly assign ants to each cluster head node;

[0041] S6. The ant selects the next node according to the improved transition probability;

[0042] S7. Using the improved pheromone update rule, update the pheromone on each path of the ant individual;

[0043] S8. Determine whether the preset number of iterations has been reached. If so, output the optimal path; if not, execute step S5.

[0044] Optionally, the improved pheromone update rule includes:

[0045] After each iteration, the path pheromone of the ants reaching the target point is updated according to the first preset rule;

[0046] Strengthen the pheromone on the path where the best ants pass according to the second preset rule;

[0047] According to the third preset rule, the pheromone on the path passed by the worst ant is weakened, and the upper and lower limits of the pheromone concentration are limited;

[0048] The first preset rule is:

[0049] τ ij (t+1)=(1-ρ)τ ij (t)+ρΔτ ij (t+1)

[0050] Among them, τ ij (t) represents the pheromone concentration of the ant on the path (i, j) at time t, ρ is the pheromone volatility coefficient, Δτ ij (t) represents the pheromone added on the path (i, j) after the ant iterates once;

[0051] The second preset rule is:

[0052] τ ij (t+1)=(1-ρ)τ ij (t)+σ·ρΔτ ij (t+1)

[0053] Among them, τ ij (t) represents the pheromone concentration of the ant on the path (i, j) at time t, ρ is the pheromone volatility coefficient, Δτ ij (t) represents the pheromone added on the path (i, j) after one iteration of the ant, and σ is the pheromone amplification coefficient;

[0054] The third preset rule is:

[0055] τ ij (t+1)=(1-ρ)τ ij (t)-σ·ρΔτ ij (t+1)

[0056] Among them, τ ij (t) represents the pheromone concentration of the ant on the path (i, j) at time t, ρ is the pheromone volatility coefficient, Δτ ij (t) represents the pheromone added on the path (i, j) after one iteration of the ant, and σ is the pheromone amplification coefficient.

[0057] Optionally, the improved transition probability is:

[0058]

[0059] Among them, j is the node, τ ij is the pheromone concentration, t is the maximum number of iterations, α is the pheromone inspiration factor, η ij is the heuristic function, β is the expected heuristic factor, is the transition probability, q0 is a dynamic selection factor, and q is the set threshold.

[0060] The present invention has the following beneficial effects:

[0061] The present invention clusters the nodes in MR-WSNs based on preset indicators and threshold functions, obtains the optimal cluster head node sequence set, reduces the number of traversed nodes, and thus reduces the traversal paths.

[0062] The present invention improves the ant colony algorithm for the application of MR-WSNs path planning, optimizes the initial pheromone concentration, state transfer and pheromone update strategy of the ant colony algorithm, and improves the convergence speed and optimization speed of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0064] Figure 1 A schematic diagram of a clustering algorithm according to an embodiment of the present invention;

[0065] Figure 2 A schematic diagram of a clustering method flow chart of an embodiment of the present invention;

[0066] Figure 3 A schematic diagram of a moving path of an MS according to an embodiment of the present invention;

[0067] Figure 4 This is a schematic diagram of an optimal path according to an embodiment of the present invention;

[0068] Figure 5 A schematic diagram of an algorithm for constructing a traversal point set according to an embodiment of the present invention;

[0069] Figure 6 A schematic diagram showing the change of path length with the number of nodes in an embodiment of the present invention;

[0070] Figure 7 A schematic diagram showing the variation of data transmission delay with the number of nodes according to an embodiment of the present invention;

[0071] Figure 8 A schematic diagram showing how throughput varies with the number of nodes according to an embodiment of the present invention;

[0072] Fig. 9 The figure is a flow chart of a path planning method based on MR-WSNs mobile Sink nodes according to an embodiment of the present invention. DETAILED DESCRIPTION

[0073] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0074] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0075] First, the system model in this embodiment is introduced;

[0076] Network Model

[0077] The research area of ​​this embodiment is located in the cage aquaculture area of ​​the ocean ranch in Xiayang Town, Xuwen. This embodiment describes the MR-WSNs deployed on the cages in this area as deploying n nodes {s1, s2, s3…, s n} and a mobile sink node (MS). Let s i represents the i-th common node, and 1≤i≤n. The MS moves along the planned path at a constant speed and collects cluster head node data during the movement. Each moving path of the MS consists of k EPs, k<n. The k EPs constitute the EPs set Q={q1,q2,q3,…,q k}. Let Z p represents a moving path of the MS. Assuming that the MS starts from q1, the path Z p It can be expressed as: Z p ={q1,q2,q3,…,q k ,q1}. Calculate the path Z according to formula (1) plength:

[0078] L(Z p ) = d 12 +d 12 +...+d k2 (1)

[0079] where d ij Indicates q i and q j The distance between them is calculated according to formula (2).

[0080]

[0081] For the convenience of research, this embodiment imposes the following constraints:

[0082] (1) Node {s1,s2,s3…,s n} is static. After randomly deploying in the area, the nodes no longer move. The nodes have the same initial energy E and communication radius R, and their geographical location information is known.

[0083] (2) When the MS moves into the communication range, the cluster head node directly transmits data to the MS at a fixed transmission rate. The initial position of the MS is at the center of the monitoring area. In addition, the MS has controllable mobility, and its energy, storage capacity and transmission distance are not restricted.

[0084] (3) The cluster head node only considers the death caused by energy exhaustion.

[0085] (4) All nodes do not consider the influence of natural factors such as ocean currents.

[0086] Energy consumption model:

[0087] Assume that each node in the network uses a fixed transmit power and receive power. When the MS moves from the starting point to the end point, it is defined as completing a round of movement. The minimum time interval of each round is represented by Δt. The life cycle of the entire network is defined as from network initialization to the exhaustion of power of the first cluster head node, as shown in formula (3).

[0088]

[0089] Among them, T k Indicates the time the MS stays at the kth traversal point.

[0090] The sensors sense the physical environment and transmit data packets to the MS using e t Represents the energy consumption of sending unit data, e r represents the energy consumption of receiving unit data, then the energy consumption of each round in MR-WSNs can be expressed as formula (4).

[0091]

[0092] Among them, any node s i Amount of data received and the amount of data sent The relationship between q is the node s i The total amount of data generated in a single round. Assuming that the energy consumption of data transmission in the network is related to the number of node transmission hops, the expression is shown in (5).

[0093]

[0094] Among them, h i Represents node s i The minimum number of hops to send data to the cluster head node, and if the current node s i is the cluster head node, then h i = 0. Therefore, for any node s i , its energy consumption during the entire network life cycle is shown in formula (6).

[0095]

[0096] Among them, p i is the total energy consumption of a single round of any node i. From formula (6), it can be seen that the multi-hop communication between sensor nodes affects the network energy consumption.

[0097] Therefore, selecting a suitable cluster head node is crucial.

[0098] Traditional ant colony algorithm for solving TSP problem

[0099] TSP Problem Description

[0100] The Traveling Salesman Problem (TSP) is a classic problem in the Combinatorial Optimization Problem (COP), which can be described as: the route needs to visit the location of the city only once and then return to the starting city to make the path shortest. The mathematical model can be described as: Define the objective function as F, and the mathematical model of the TSP problem of n cities is shown in formula (7):

[0101]

[0102] Among them, D(m,m+1) is the distance from city m to city m+1. In MR-WSNs, the process of MS traversing nodes to collect data in a certain order can be equivalent to the TSP problem. Common algorithms for the TSP problem include genetic algorithm (GA), ant colony algorithm (ACA), and particle swarm optimization (PSO). Among them, the classic ant colony algorithm has stable convergence characteristics and good search efficiency when solving the TSP problem, but it has the disadvantages of slow convergence speed and easy to fall into local optimality when solving the optimal solution. Therefore, this implementation case improves the problems existing in the solution process of the classic ant colony algorithm.

[0103] Traditional ant colony algorithm

[0104] Description of traditional ant colony algorithm

[0105] The ant colony algorithm (ACA) is a random heuristic search algorithm that simulates the foraging behavior of ants. During the path-finding process, ants release pheromones on the paths they pass through. The concentration of pheromones is inversely proportional to the corresponding path length. When choosing a path, ants will choose a path with a high pheromone concentration based on the state transition probability. As a large number of ant individuals continue to search, the accumulated pheromone concentration on the optimal path increases. Under the action of the positive feedback mechanism, the entire ant colony will eventually find an optimal path.

[0106] State transition probability

[0107] When ant colony members are searching for the shortest path, the path search is affected by two factors: pheromone concentration and heuristic function. Assuming that there are M ants at the starting point, the transfer probability formula of ant k selecting the next node from node i is shown in formula (8).

[0108]

[0109] Where α represents the pheromone heuristic factor, which reflects the influence of pheromone concentration on path selection; β represents the expected heuristic factor, which indicates the importance of prior path information in path selection; allowed k Represents the set of all next reachable nodes in path search; τ ij (t) represents the pheromone concentration; η ij (t) is the heuristic function, and its expression is (9).

[0110]

[0111] Where d ijRepresents the Euclidean distance between the current node i and the candidate node j.

[0112] Pheromone Update

[0113] Ants will leave a certain amount of pheromone in the process of moving. When the tth generation of ants completes the iterative search, the pheromone update rule is used to update the path. The specific update rule is shown in formula (10):

[0114] τ ij (t+1)=(1-ρ)τ ij (t)+Δτ ij (10)

[0115] Where ρ is the pheromone volatility coefficient.

[0116] The traditional ant colony algorithm has the following problems: ① The initial concentration of pheromones at each node is evenly distributed. In the early search, the ant colony will choose to walk in the opposite direction of the end point or form a loop, which leads to blindness in the early search, and the length of the found path is long and time-consuming. ② A large number of repeated paths will inevitably be generated during the initial search of the ant colony. Due to the limitations of the algorithm's taboo table, a large number of ant individuals will fall into a deadlock state, causing the ants to stop searching before reaching the target point, resulting in low path search efficiency and slow algorithm convergence.

[0117] Improved Ant Colony Algorithm (IACA)

[0118] Initial pheromone distribution is uneven

[0119] Improving the uneven distribution of initial pheromones is beneficial to improving the early convergence speed of the ant colony algorithm and reducing the blind search phenomenon of ants in the early stage. In order to improve the problem of path lengthening caused by the blindness of path planning, this embodiment stipulates that all nodes between the starting node and the target node are "excellent" nodes, whose initial pheromones are slightly larger than those of other nodes. The initial pheromone expression is (11):

[0120]

[0121] The initial information concentration is constructed by the inverse of the Euclidean distance between excellent nodes, as shown in formula (12):

[0122]

[0123] Among them, τ0 is a constant; C is the initial pheromone concentration of the "excellent" node; d ij is the inverse of the Euclidean distance between the "excellent" node and the mobile sink node, and Z is the set of favorable nodes. According to the zero point theorem, the global optimal path is often a path from the starting point to the target point. Therefore, the uneven distribution of initial pheromones is conducive to reducing the blind search path of ants, reducing the search time and improving the convergence speed.

[0124] The initial pheromone concentration of the traditional ant colony algorithm is generally 0 or a constant, which makes ants fall into a blind search state in the early stage, resulting in slow search speed and long search time in the early stage of the ant colony algorithm. Therefore, it is proposed to improve the initial pheromone concentration and use the inverse of the Euclidean distance of the excellent nodes to construct the pheromone concentration. According to its uneven characteristics, the probability of "excellent" nodes being selected is increased. The closer the distance of the "excellent" node to the mobile Sink node, the greater the probability of being selected. Conversely, the farther the distance, the lower the probability of being selected, thereby improving the early convergence speed of the ant colony algorithm.

[0125] Improved transition probability

[0126] The ant colony algorithm finally finds the global optimal solution through the continuous iteration of ants. Therefore, the more solutions the algorithm finds in the early stage and the better the quality of the solutions, the greater the probability of finding the global optimal solution in the later stage and the faster the convergence speed. The fundamental reason why ants enter the deadlock state is that the surrounding nodes have been walked through. Therefore, the discrete strategy is adopted in the early stage of the search to increase the congestion factor ξ t , the improved transition probability formula is shown in formula (13):

[0127]

[0128]

[0129] Among them, ξ t Depending on the value of the pheromone concentration, τ max represents the maximum pheromone concentration. By increasing the crowding factor, the number of ants entering the deadlock state each time they iterate the search path is reduced. In order to improve the search efficiency of the algorithm, a pseudo-random state transition strategy is adopted, and the expression is shown in formula (15):

[0130]

[0131]

[0132] Among them, δ is the dynamic adjustment coefficient, δ∈(0.5,1), N max is the maximum number of iterations set; N is the current number of iterations; q0 is a dynamic selection factor, q0∈(0,1), calculated according to formula (16). In the early stage of the algorithm, q0 takes a larger value, which can improve the search efficiency of the algorithm in the initial stage. In the later stage, q0 takes a smaller value, which is conducive to the random search of ants and ensures the global search ability of the algorithm.

[0133] Improved pheromone update rules

[0134] After an iteration, the pheromones on the path of the ants will accumulate and evaporate over time. In order to obtain a better path, the pheromones on the path need to be adjusted and updated. In the basic ant colony algorithm, the pheromone volatility coefficient ρ usually takes a fixed value, which will lead to blind update of global pheromones, which is not conducive to ensuring the global search capability in the early stage of the algorithm and the rapid convergence capability in the later stage of the algorithm. The pheromone update rule adopted in this embodiment is:

[0135] τ ij (t+1)=(1-ρ)τ ij (t)+ρΔτ ij (t+1) (17)

[0136] τ ij (t+1)=(1-ρ)τ ij (t)+σ·ρΔτ ij (t+1) (18)

[0137] τ ij (t+1)=(1-ρ)τ ij (t)-σ·ρΔτ ij (t+1) (19)

[0138] After each iteration, the pheromone of the path of the ants reaching the target point is updated according to formula (17); the pheromone on the path passed by the best ants is strengthened according to formula (18); the pheromone on the path passed by the worst ants is weakened according to formula (19), and the upper and lower limits of the pheromone concentration are limited, τ ij (t)∈[τ min ,τ max ], σ is the pheromone amplification coefficient, σ∈(0.4,1).

[0139] The globality of the algorithm is improved by using an adaptive pheromone volatility factor ρ. If the optimal solution of the ant colony algorithm is the same in T consecutive iterations, formula (20) is used to make adaptive adjustments to avoid falling into the local optimum. If the optimal solution obtained for T consecutive times is less than the historical optimal solution, formula (18) is used to update the pheromone.

[0140]

[0141] Where ρ min is the minimum value of ρ, ε is the pheromone attenuation coefficient, ε∈(0,1).

[0142] The following is an overall introduction to the improved ant colony algorithm in this embodiment, which includes the following steps:

[0143] W1: Data initialization establishes a random rectangular coordinate system and imports the position coordinate values ​​of n cluster head nodes; calculates the distance between any two cluster head nodes to obtain the distance matrix D, thereby obtaining the initial matrix of heuristic information Eta=1. / D; initializes population parameters, sets pheromone heuristic factor α=0.4, heuristic function factor β=0.6, pheromone volatility factor v_factor=0.8, initial crowding factor cro_factor=0.4, dynamic adjustment coefficient δ=0.7, pheromone amplification coefficient σ∈(0.4,1). The number of ants A_NUM set in the algorithm loop is determined by the problem scale n, and the maximum number of iterations N max ;

[0144] W2: Non-uniform initialization pheromone concentration: by formula and Complete the assignment of the initial information concentration, where τ0 is the initial pheromone concentration of the non-"excellent" node, which is an arbitrary constant, generally 0; C is the initial pheromone concentration of the "excellent" node; d ij is the reciprocal of the Euclidean distance between node i and the mobile sink node, and Z is the set of favorable nodes;

[0145] W3: Randomly assign ants to each cluster head node;

[0146] W4: The ant selects the next node according to the improved transfer probability; the probability calculation formula is as follows:

[0147]

[0148]

[0149]

[0150]

[0151] Among them, ξ t is the crowding factor, which depends on the value of pheromone concentration, τ max represents the maximum pheromone concentration, δ is the dynamic adjustment coefficient, δ∈(0.5,1), N max is the maximum number of iterations set, N is the current number of iterations, q0 is a dynamic selection factor, q0∈(0,1), and q is the set threshold.

[0152] W5: Calculate the path length according to the Euclidean distance formula and record the optimal path of this iteration;

[0153] W6: Update the pheromone on each path of the ant individual; use the improved pheromone update rule, the formula is as follows:

[0154] τ ij(t+1)=(1-ρ)τ ij (t)+ρΔτ ij (t+1)

[0155] τ ij (t+1)=(1-ρ)τ ij (t)+σ·ρΔτ ij (t+1)

[0156] τ ij (t+1)=(1-ρ)τ ij (t)-σ·ρΔτ ij (t+1)

[0157]

[0158] where τ ij (t) is the pheromone concentration, τ ij (t)∈[τ min ,τ max ], Δτ ij (t) is the sum of the pheromone concentrations released by all ants on the path connecting node i and node j, σ is the pheromone amplification coefficient, σ∈(0.4,1). ρ is the adaptive pheromone volatility factor, ρ min is the minimum value of the adaptive pheromone volatility factor, ε is the pheromone attenuation coefficient, ε∈(0,1);

[0159] W7: Determine whether the algorithm has reached the preset maximum number of iterations N max , if yes, then output the optimal path; if no, then execute step W3.

[0160] like Fig. 9 As shown, the path planning method based on MR-WSNs mobile Sink node proposed in this embodiment includes the following steps:

[0161] Based on the preset index and threshold function, the nodes in the wireless sensor network are clustered to obtain a set of cluster head nodes, where the cluster head node is the node where data is collected;

[0162] The improved ant colony algorithm is used to solve the cluster head node set of this embodiment to obtain the initial moving path of the mobile Sink node;

[0163] The cluster head nodes in the cluster head node set of this embodiment are traversed, the initial moving path of this embodiment is optimized, and the optimal moving path of the mobile Sink node of this embodiment is obtained.

[0164] Cluster head selection

[0165] In WR-WSN, if the MS directly traverses all nodes in the monitoring area, there will be problems such as long moving paths and high data transmission delay. Therefore, before path planning, the nodes in the network are first clustered, and ordinary nodes send data to the cluster head, and the MS only needs to access the cluster head node.

[0166] Number of cluster heads

[0167] After the network nodes are randomly arranged in the monitoring area, the number of cluster heads k is estimated based on the area of ​​the monitoring area and the node communication radius. First, the maximum distance d from the n nodes to the MS is calculated. max and the minimum distance d min :

[0168]

[0169] Among them, d(i,0) represents the node s i The distance from s0. Then the number of cluster heads k is calculated according to formula (22).

[0170]

[0171] Select cluster head

[0172] In order to ensure the rationality of cluster head node selection, cluster heads are selected according to the objective function value, which comprehensively considers three indicators: node residual energy, node density, and node spacing. i The energy factor f e (i) Calculation: see formula (23).

[0173]

[0174] Among them, E node (i) represents node s i The residual energy of n is the node s i The number of neighbor nodes.

[0175] In the actual test environment of MR-WSNs, nodes are randomly arranged in the monitoring area, and the number of neighbor nodes of each node is different. If the node is located in a dense area, the probability of becoming a cluster head should be increased; conversely, if the node is located in a sparse area, the probability of becoming a cluster head should be reduced. Node density factor D i The calculation of is shown in formula (23):

[0176]

[0177] Where N i Represents node s i The two-hop neighbor node of is defined as shown in formula (25).

[0178]

[0179] By considering the distance between cluster heads, the spacing factor F i Calculate according to formula (26).

[0180]

[0181] Among them, d i Represents node s i Distance to MS, d j Indicates the distance from the cluster head participating in the election to the MS in this round, n indicates the number of nodes participating in the election, and R indicates the radius of the central circle. The cluster head nodes generated by the election are evenly distributed in the monitoring area to reduce the occurrence of cluster head nodes concentrated in a certain local area. The threshold function used by the LEACH protocol is referenced, and energy factor, density factor, and spacing factor are added on its basis to construct a new cluster head selection threshold model, see formula (27).

[0182]

[0183] Among them, P i Represents the proportion of cluster heads, that is, P i = k / (n-1); r is the current execution round number; G represents the set of nodes that were not cluster heads in the previous round; if a node has been selected as a cluster head in the previous round, it will no longer participate in the cluster head election in this round; γ i represents the fitness factor constructed by energy factor, density factor and spacing factor, and its definition is as shown in formula (28).

[0184] γ i =α1f e (s i )+α2D i +α3F i (28)

[0185] As the network operation phase changes, the impact of energy on network operation continues to increase. Therefore, during the cluster head rotation process, the weight of the energy factor in the threshold model should also be gradually increased. The threshold model is adjusted by dynamic weight coefficients to alleviate the problem of decreased energy utilization in network operation and excessive energy consumption of cluster heads in the network. The energy weight coefficient is updated according to formula (29).

[0186]

[0187] Among them, E i For node s i The remaining energy, E init For node s i The relative residual energy E i / E init Represents node si As the network runs, the average node energy of the traversal point and the traversal area decreases, and α1 gradually increases.

[0188] The weight coefficient calculation expression (30) of the spacing factor and the density factor is as follows:

[0189]

[0190] During the network operation phase, in order to avoid energy waste caused by frequent cluster head rotation, the cluster head uses formula (31) to determine whether the current network needs to select a new cluster head:

[0191] γ i (p)≤ωγ max (i) (31)

[0192] Among them, γ i (p) is the objective function value of the current cluster head, γ max (i) is the maximum objective function value of ordinary nodes in the cluster, and ω∈(0,1) is used to control the frequency of cluster head rotation.

[0193] Clustering algorithm and process

[0194] Clustering algorithms such as Figure 1 The algorithm flow chart is shown in Figure 2 shown.

[0195] Constructing traversal point set

[0196] The data transmission delay is the time consumed by the MS on the entire path. Therefore, when the MS traverses all cluster head nodes in the network in sequence, the longer the path it traverses, the greater the corresponding delay. In order to further shorten the mobile path of the MS and reduce the data transmission delay, a data collection algorithm for constructing a traversal point set is proposed.

[0197] Problem Model

[0198] The sensor nodes are then deployed in the network, and the cluster head nodes exist in the set C. In TSP, the moving path of the MS consists of a series of line segments. The shortest moving distance model of the MS is shown in Equation (32).

[0199]

[0200] Where d(c i ,c j ) represents the cluster head node c i and c j The distance between.

[0201] The data transmission delay is expressed as the sum of the time consumed by the MS to move and the time it stays at the traversal point, as shown in formula (33).

[0202]

[0203] Where V represents the moving speed of MS, t k represents the time that the MS stays at the traversal point k, and Q is the set of traversal points. Under the condition that the MS traversal path length L is limited, the data transmission delay T min To obtain the minimum value, on the one hand, the traversal points of the MS should be reduced as much as possible. On the other hand, the more cluster head nodes visited, the better.

[0204] Algorithm Implementation

[0205] Step ①: Find the TSP path. Based on the cluster head node set C obtained in the previous section, use the improved ant colony algorithm to find a mobile path Z of the MS. p ={c1,c2,…,c m ,c1}, such as Figure 3 The MS moving path shown is Z1 = {c1, c2, c3, c4, c5, c6, c7, c1}.

[0206] Step ②: Select the traversal point EP. If you want to make the MS's moving distance shorter and the number of EPs smaller, the MS's traversal point should be at the boundary of the CH node's communication range. Figure 4 As shown in the figure, the face areas f1, f2, f3, f4, f5, f6, and f7 are all covered by a cluster head node communication area, and they constitute the traversal area set F. Assume that the path Z1 = {c1, c2, c3, c4, c5, c6, c7, c1} is the current shortest path, where points q1 and q2 are on the boundary of face areas f1 and f2 and According to the Euclidean distance formula, d(q1,q2)<d(c1,c2). Therefore, in order to shorten the MS path, each traversal point must be within the communication coverage of the node. At this time, Q={q1,q2,c3,c4,c5,c6,c7}.

[0207] Secondly, in order to reduce the number of EPs and improve data collection efficiency, it is necessary to determine whether each traversal area intersects. If so, the intersection point t m is the new EP, and the original cluster head node moves out of Q. Figure 4 In the figure, f3 and f4 intersect, and the intersection t1 is on the boundary of the communication areas of both. Obviously, when the MS is located at t1, it can simultaneously receive monitoring data from the traversal areas f3 and f4. Compared with traversing q3 and q4, the number of EPs is reduced, the mobile path is shortened, and the data delay is reduced. t1 is more suitable as a new EP. Therefore, after the above optimization, the EP set Q = {q1, q2, t1, t2, t3}, that is, the optimal path of the MS is Z1 = {q1, q2, t1, t2, t3, q1}.

[0208] Construct a point set traversal algorithm such as Figure 5 shown.

[0209] Experimental analysis and evaluation

[0210] Simulation Environment

[0211] In this section, the PMRM method is experimentally evaluated. The algorithm is implemented on the MATLAB R2020 simulation platform. The experimental platform is a desktop computer with Intel(R) Corei9-13900K CPU@4.3GHz, 16GB memory, and 1TB SSD. The software operating system is Windows 11. 150 to 300 nodes are randomly deployed in a 500m×500m rectangular area. The node communication radius R is 100m, and the initial energy of the node is randomly selected between 600 and 1200. The specific simulation parameters are shown in Table 1.

[0212] Table 1L-WSN environmental parameters

[0213]

[0214] The classic ACA algorithm, EDAMS algorithm and GMSDC algorithm are selected as references to analyze the performance of the planning algorithm of this embodiment. Among them, algorithms such as ACA use all nodes in the network as traversal points and do not build a node set. The PMRM proposed in this embodiment first clusters the nodes in the network, selects appropriate traversal points according to the distribution of cluster head nodes, then builds a node set, and then uses TSP to plan the mobile path of the MS. The specific parameter settings of the IACA algorithm are shown in Table 2.

[0215] Table 2 IACA simulation parameter settings

[0216]

[0217] Data analysis

[0218] MS moving path length

[0219] Figure 6It reflects the change of MS's path length with the number of nodes. As can be seen from the figure, the path length of each algorithm increases with the increase of the number of nodes. The main reason is that when the number of sensor nodes in the network increases, the MS needs to collect more data, which causes the path of MS movement to become longer. In addition, ACA has the longest path because the ACA algorithm directly traverses all nodes in the network; the EDAMS algorithm selects the best residence point and reduces the number of mobile Sink traversal points, so its path is significantly shorter than ACA. The path lengths of GMSDC and PMRM are similar. Compared with ACA and EDAMS algorithms, the path length is further shortened because GMSDC uses a genetic algorithm to find the optimal residence point and shortens the moving distance. This embodiment constructs a node set by considering the communication coverage area of ​​the cluster head node, and finally uses the improved ant colony algorithm to solve the TSP problem for the set to obtain the shortest path. When the number of nodes in the network is n=275, the path length of the PMRP algorithm is reduced by 52.1%, 41.9%, and 20.3% respectively compared with other algorithms.

[0220] Data collection delay

[0221] Figure 7 It reflects the variation of data collection delay with the number of nodes. As can be seen from the figure, compared with ACA, EDAMS and GMSDC, PMRM can effectively control the data collection delay, and the delay of the ACA algorithm is the highest. The reason is that the ACA algorithm requires traversing each sensor node and does not take into account the node communication coverage area, which prolongs the mobile path of the MS. Although the EDAMS algorithm and the GMSDC algorithm select the best residence point, they do not cluster the network, and the data is transmitted in multiple hops, which increases the data transmission delay of the node. Compared with EDAMS and GMSDC, when the number of nodes n = 275, the data delay of the PMRM algorithm is reduced by 42.0% and 38.5% respectively, and when the number of nodes n = 300, it is reduced by 35% and 31.6% respectively. Therefore, in scenarios with a large number of network nodes, the PMRM algorithm can effectively reduce the data transmission delay.

[0222] Throughput of mobile sink nodes

[0223] MS throughput refers to the amount of data collected by the MS per unit time. Figure 8 It reflects the change of throughput with the number of nodes. Figure 7It can be seen that the more nodes there are in the network, the higher the throughput is. The main reason is that the amount of data generated by the nodes increases with the number of nodes, which is in line with expectations. When the number of nodes n = 275, compared with the EDAMS and GMSDC algorithms, the throughput of MS increases by 10% and 10.6% respectively. Experiments have shown that since the PMRM algorithm comprehensively considers the communication coverage area between cluster head nodes, it can receive data from multiple cluster head nodes at the same time, and the energy hole problem no longer exists. The larger the network scale, the better the effect of the algorithm in this embodiment.

[0224] The PMRM proposed in this embodiment can effectively alleviate the problem of D energy holes that are prone to occur in MR-WSNs. Since PMRM first optimizes the initial pheromone concentration, state transition and pheromone update strategy of the ant colony algorithm, the convergence speed and optimization speed of the algorithm are improved. Secondly, the network is clustered, and the optimal traversal points are selected according to the location of the cluster head node, the communication radius and the communication coverage area, and the EPs set is constructed, which reduces the number of traversal nodes and thus reduces the traversal path. Through simulation experiment comparison, the PMRM proposed in this embodiment has obvious improvements in path length, data transmission delay and data throughput, and enhances the robustness of the network. Finally, this embodiment makes data transmission more reliable by moving the MS to a suitable communication position, avoiding the problem of energy holes in the network.

[0225] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A path planning method based on MR-WSNs mobile Sink node, characterized in that: include: Based on the preset indicators and threshold functions, the nodes in the marine ranch wireless sensor network are clustered and a cluster head node set is constructed. Select traversal points according to preset conditions, optimize the cluster head node set, and establish the optimal traversal point set; Using an improved ant colony algorithm to solve the path planning of the optimal traversal point set to obtain an optimal moving path; Acquiring the cluster head node set includes: Based on the monitoring area and the node communication radius, the number of cluster heads is obtained; According to the number of cluster heads, based on the preset index and threshold function, clustering the nodes in the MR-WSNs to obtain the cluster head node set; The preset conditions for selecting traversal points include: The moving distance of MS is the shortest, the number of traversal points is the least, and each traversal point must be within the communication coverage of the node; among them, MS is a mobile Sink node; The establishing of the optimal traversal point set comprises: S1. Input cluster head node set C, traverse point set Q, traverse zone set F; S2. Determine whether the traversal areas of adjacent cluster head nodes intersect. If so, the intersection becomes a new traversal point. If not, find a traversal point on the boundary of the communication range R and satisfy S3. Output the optimal traversal point set Q; Among them, q i and q j is the traversal point on the boundary of the communication range, c i and c j is the cluster head node, They represent the Euclidean distance of the traversal points and the Euclidean distance of the cluster head node respectively.

2. The path planning method based on MR-WSNs mobile Sink node according to claim 1 is characterized in that: The number of cluster heads is: Where k is the number of cluster heads, d max is the maximum distance between the node and the mobile sink node, d min is the minimum distance between the node and the MS, R is the communication radius of the static node, L is the side length of the monitoring area, and n is the number of static nodes in the monitoring area.

3. The path planning method based on MR-WSNs mobile Sink node according to claim 1, characterized in that: The preset indicators include: node energy factor, node density factor and node spacing factor; The node energy factor is: Among them, E node (i) represents node s i The remaining energy The node density factor is: Among them, N i Represents node s i The two-hop neighbor node of The node spacing factor is: Among them, d i Represents node s i Distance to MS, d j It represents the distance from the cluster head participating in the election to the MS in this round, and R represents the radius of the center circle.

4. The path planning method based on MR-WSNs mobile Sink node according to claim 1, characterized in that: The threshold function is: Among them, P i represents the proportion of cluster heads, r is the number of current execution rounds, G represents the set of nodes that were not cluster heads in the previous round, and γ i Represents the fitness factor constructed by energy factor, density factor and spacing factor.

5. The path planning method based on MR-WSNs mobile Sink node according to claim 1 is characterized in that: Using the improved ant colony algorithm, the path planning solution for the optimal traversal point set includes: S1. Data initialization establishes a random rectangular coordinate system and imports the position coordinate values ​​of n cluster head nodes; S2. Calculate the distance between any two cluster head nodes and obtain the initial matrix of heuristic information; S3. Initialize population parameters; S4. Calculate the initial pheromone concentration value; S5. Randomly assign ants to each cluster head node; S6. The ant selects the next node according to the improved transition probability; S7. Using the improved pheromone update rule, update the pheromone on each path of the ant individual; S8. Determine whether the preset number of iterations has been reached. If so, output the optimal path; if not, execute step S5.

6. The path planning method based on MR-WSNs mobile Sink node according to claim 5 is characterized in that: The improved pheromone update rules include: After each iteration, the path pheromone of the ant reaching the target point is updated according to the first preset rule; Strengthen the pheromone on the path where the best ants pass according to the second preset rule; According to the third preset rule, the pheromone on the path passed by the worst ant is weakened, and the upper and lower limits of the pheromone concentration are limited; The first preset rule is: τij(t+1)=(1-ρ)τij(t)+ρΔτij(t+1) Among them, τ ij (t) represents the pheromone concentration of the ant on the path (i, j) at time t, ρ is the pheromone volatility coefficient, Δτ ij (t) represents the pheromone added on the path (i, j) after the ant iterates once; The second preset rule is: t ij (t+1)=(1-ρ)τ ij (t)+σ·ρΔτ ij (t+1) Among them, τ ij (t) represents the pheromone concentration of the ant on the path (i, j) at time t, ρ is the pheromone volatility coefficient, Δτ ij (t) represents the pheromone added on the path (i, j) after one iteration of the ant, and σ is the pheromone amplification coefficient; The third preset rule is: t ij (t+1)=(1-ρ)τ ij (t)-s·rΔτ ij (t+1) Among them, τ ij (t) represents the pheromone concentration of the ant on the path (i, j) at time t, ρ is the pheromone volatility coefficient, Δτ ij (t) represents the pheromone added on the path (i, j) after one iteration of the ant, and σ is the pheromone amplification coefficient.

7. The path planning method based on MR-WSNs mobile Sink node according to claim 5 is characterized in that: The improved transition probability is: Among them, j is the node, τ ij is the pheromone concentration, t is the maximum number of iterations, α is the pheromone inspiration factor, η ij is the heuristic function, β is the expected heuristic factor, is the transition probability, q0 is a dynamic selection factor, and q is the set threshold.

Images