A Low-Energy-Priority Election and Hierarchical Multipath Uneven Clustering Method

Abstract

A low-energy priority election-based hierarchical multi-path non-uniform clustering method, including a cluster head election phase. In the cluster head election phase, candidate nodes are graded according to the minimum number of hops from the candidate nodes to the base station, and each candidate node participates in the election by integrating the remaining energy and the path quality parameter. A potential competition node set is established for each candidate node. During the cluster head election process, each candidate node sends a confirmation message to confirm whether it enters the election. If no response is received within the specified threshold time, the corresponding candidate node withdraws from the election. When a certain candidate node wins the election, it sends a completion message to the remaining candidate nodes, and the remaining candidate nodes withdraw from the election. After the election ends, the cluster head with the minimum communication cost is selected and the corresponding cluster head is notified, and then it enters the data transmission phase, and single-hop communication is used for intra-cluster communication. The present invention reduces the energy consumption of ordinary nodes in a wireless sensor network and improves the network lifetime.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless sensors, and particularly relates to a low-energy priority election hierarchical multi-path non-uniform clustering method. Background Art

[0002] Currently, the application of new-generation information integration technologies such as the Internet of Things, cloud computing, big data, and CPS has been gradually widespread in various fields, including important fields such as urban construction, national defense, environmental monitoring and management, agriculture, and healthcare. As the foundation of the above applications, wireless sensor networks play an important role in the above information integration process. A wireless sensor network consists of autonomous wireless nodes randomly distributed in space in a network domain, which can collect various types of sensed data, such as noise, temperature and humidity, light intensity, etc., and finally transmit it to the sink node. The full-scenario application of wireless sensor networks determines that it needs to face various extreme environments. In fact, it is often difficult and expensive to replace sensor nodes in these environments. Usually in these cases, sensor nodes need to work continuously for a long time, and the battery life cannot be guaranteed. At the same time, its huge number of nodes also makes it unrealistic to manually maintain and optimize the wireless sensor network one by one. For this reason, these strict limitations, that is, sensors with low memory, low processing power, and low battery power, have attracted people's attention. Just because of this, developing a WSN routing protocol with high energy efficiency and long lifespan is a very urgent problem.

[0003] According to the logical structure of WSN, routing protocols can be divided into flat routing protocols and hierarchical routing protocols. Among flat routing protocols, the flooding routing protocol is the most classic. The node that receives the information will broadcast the information to all its neighbor nodes. Although it has no requirement for the topology structure of WSN and has high robustness, there are problems such as information implosion and large resource consumption. Although there are improvements such as SPIN and Gossiping routing protocols based on it, the problem of low resource utilization efficiency has not been well solved. The core of this problem is that the relationship between all nodes is equal, which makes the routing of nodes too complicated and there is no clear information propagation route.

[0004] And clustering the nodes in hierarchical routing protocols well solves this problem. Among them, the classic LEACH uses data compression technology and clustering dynamic routing, making the compressed data propagate along a specific route, greatly improving the resource utilization efficiency. Although hierarchical routing protocols have made great improvements compared with flat routing protocols, there are still many problems that need to be solved urgently, such as the "hot spot" problem near the base station and how to reasonably select cluster heads. In addition, there are also energy routing protocols and query-based routing protocols, etc.

[0005] At present, many mature routing protocols have been widely used, but these routing protocols have some problems that have not been considered to a certain extent. For example, the "barrel effect" is not considered in the hierarchical routing algorithm, that is, no protection strategy is made for low-energy nodes, which causes these nodes to die prematurely. Secondly, in the clustering stage, a fixed competition radius and insufficient election parameters are used for nodes. Therefore, in order to solve the above two problems, it is urgent to optimize the WSN micro level. Finally, the existing technology is still not perfect in the study of multi-hop in WSN. Although it is aware of the waste of network resources by multi-hop, the optimization of multi-hop is too rough and does not take into account the two-sided nature of multi-hop. After analyzing the multi-hop transmission mode of traditional routing protocols, it is found that: in traditional multi-hop routing, relay nodes consume a lot of energy to forward data from other nodes. And the more relay nodes there are in multi-hop routing, the more serious the situation is for the overall network that a large amount of energy is used to forward duplicate data, but on the other hand, multi-hop communication can slow down the death time of nodes far away from the base station, thereby extending the life cycle of the network. Therefore, it is of great significance to reasonably balance the overall energy of the multi-hop network and the energy of the nodes in the network to extend the life cycle of WSN. Summary of the invention

[0006] To this end, the present invention provides a low-energy priority election hierarchical multi-path non-uniform clustering method to solve the problem that the existing protocol cannot completely balance the energy consumption of all nodes, and the overloaded nodes consume energy too quickly, resulting in low energy, thereby affecting the network life cycle.

[0007] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: a low-energy priority election hierarchical multi-path non-uniform clustering method, including a cluster head election stage, in which candidate nodes are classified according to the minimum number of hops from the candidate nodes to the base station, and each candidate node participates in the election based on the residual energy and path quality parameters;

[0008] A set of potential competing nodes is established for each candidate node. During the cluster head election process, each candidate node sends a confirmation message to confirm whether to enter the election. If no response is received within the specified threshold time, the corresponding candidate node withdraws from the election.

[0009] When a candidate node wins the election, it sends a completion message to the remaining candidate nodes, and the remaining candidate nodes withdraw from the election;

[0010] After the election, the cluster head with the lowest communication cost is selected and notified to the corresponding cluster head, and then the data transmission phase begins, and the intra-cluster communication adopts single-hop communication.

[0011] As a preferred scheme of the low-energy-first election hierarchical multi-path non-uniform clustering method, before the election of cluster heads, the WSN network is initialized, and the candidate nodes determine their neighbor nodes and their own levels;

[0012] After all candidate nodes are initialized, the candidate nodes start to update the competition radius, then update their own competition parameters, and then enter the cluster head election stage.

[0013] As an optimized solution of the low-energy priority election hierarchical multi-path non-uniform clustering method, the candidate nodes adopt a competition radius that changes with the topology, and the competition radius of the candidate nodes changes positively with the load borne;

[0014] Take the ratio of the remaining energy of the candidate node to the number of elements in itself and the current set of neighbor nodes as the average available energy of the neighbors.

[0015] As an optimized solution of the low-energy priority election hierarchical multi-path non-uniform clustering method, each time an election is held, the candidate nodes will update the average available energy of the neighbors and the historical maximum average available energy of the neighbors;

[0016] Determine the current competition radius from the past historical period of the candidate node, and use the obtained result as the normalized average available energy of the neighbors;

[0017] Transform the obtained normalized average available energy of the neighbors using the cosine offset model to obtain the competition radius formula of the cluster.

[0018] As an optimized solution of the low-energy priority election hierarchical multi-path non-uniform clustering method, the competition radius of the cluster is jointly determined by the level where the candidate node is located, the normalized average available energy of the neighbors, and the remaining energy;

[0019] Preset a threshold level, and change the competition radius when it is greater than the threshold level.

[0020] As an optimized solution of the low-energy priority election hierarchical multi-path non-uniform clustering method, preset the two-dimensional election parameters of the candidate nodes, and take the remaining energy of the candidate node as the first parameter for cluster head election; at the same time, take the routing quality parameter that measures the routing quality of the candidate node as the second parameter for cluster head election;

[0021] Preset the actual forward distance. The actual forward distance of a route is the difference between the distance from the sender to the base station and the distance from the receiver to the base station.

[0022] As an optimized solution of the low-energy priority election hierarchical multi-path non-uniform clustering method, if there is a communication internal node C of node A on the line connecting node A to the base station S, node C is called the optimal next-hop node at a distance of d AC When reaching the base station through the next-hop node, the total transmission distance reaches the minimum and the actual forward distance of this route is the longest.

[0023] As an optimal solution of the low - energy - first election - graded multi - path non - uniform clustering method, each node obtains corresponding votes according to its remaining energy. At the beginning of the voting, all candidate nodes with levels greater than 1 broadcast Vote_msg, and Vote_msg includes the fitness value, competition radius, and level of the node;

[0024] When node S j receives the Vote_msg from node S i , it first determines whether they are at the same level. If not, it discards it; otherwise, it starts to compare the distance d j between node S i and node S ij with the competition radius . If the distance d ij is less than the competition radius , it identifies node S i as a neighbor node of node S j ; if the distance d ij is less than , it adds node S i to the potential set of competing nodes of node S j .

[0025] As an optimal solution of the low - energy - first election - graded multi - path non - uniform clustering method, the comparison logic of two - dimensional campaign parameters is as follows:

[0026] i) When comparing node s i with node s j , if any dimension of the two - dimensional campaign parameters of node s i is in an advantageous position and the other dimension is at least tied, then node s i directly wins the campaign;

[0027] ii) When comparing node s i with node s j , if one dimension of the two - dimensional campaign parameters of node s i is at a disadvantage, and the difference in the disadvantaged dimension dim1 between the two is not greater than the corresponding threshold and the difference in the advantageous dimension dim2 of node s i is greater than the corresponding threshold , then node s i s i wins the campaign;

[0028] iii) If the two - dimensional campaign parameters of the compared nodes are the same, the node with the smaller node ID wins the campaign;

[0029] iv) If i, ii, and iii are not satisfied, node s i withdraws from the campaign and sends a withdrawal election message.

[0030] As an optimal solution of the low - energy - first election - graded multi - path non - uniform clustering method, find all the suitable next hops of cluster head I that are less than the channel threshold d0, and allow cluster head I to communicate with these next hops alternately according to a preset rule;

[0031] Single - hop communication is adopted within the cluster, and alternate multi - hop communication is selected between clusters.

[0032] The present invention has the following advantages: In the cluster - head election stage, in order to improve the lifespan of the WSN, and at the same time for the topological changes that occur during the lifespan of the WSN, two - dimensional competition parameters and competition radii that adaptively change are given to each node; The present invention adaptively adjusts clustering according to the survival status of WSN nodes, enabling the cluster head to have a higher fitness for the WSN environment and at the same time having a more flexible load - bearing strategy to prevent nodes from dying prematurely due to excessive load; The present invention proposes to balance the waste of network resources and node energy consumption in multi - hop communication and gives a communication scheme for alternate multi - hop; Simulation results show that compared with traditional routing algorithms such as LEACH, HEED, and EEUC, the present invention has improved performance. In the network stability period, the lifespan of the algorithm is extended by 80%, and during the entire lifespan of the WSN, the node performance during the survival period of the algorithm is maintained at a relatively high level and increased by about 43.7%. Brief Description of the Drawings

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, without creative efforts, other implementation drawings can also be obtained based on the provided drawings.

[0034] Figure 1 Schematic diagram of the competition radius changing with topology in the low - energy - first election - graded multi - path non - uniform clustering method provided by the embodiment of the present invention;

[0035] Figure 2 Schematic diagram of the actual forward distance in the low - energy - first election - graded multi - path non - uniform clustering method provided by the embodiment of the present invention;

[0036] Figure 3 Schematic diagram of alternate multi - hop in the low - energy - first election - graded multi - path non - uniform clustering method provided by the embodiment of the present invention. Detailed Embodiments

[0037] The following specific embodiments illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0038] In an embodiment of the present invention, consider a network formed by N randomly deployed sensor nodes, and its application scenario is periodic data collection. Let s i represent the i-th node, and the corresponding node set is S = {s1, s2, …, s N}, |S| = N, (x i , y i ) corresponds to the two-dimensional space coordinates of s i in the set. Then the Euclidean distance between two sensor nodes is:

[0039]

[0040] On this basis, assume that: the sensor nodes in the WSN are homogeneous, the communication radius is R c and they are randomly distributed in a circular area with a radius of M. The sensor nodes perform the following operations: sensing, data collection, data processing, and transmission. They can adjust the transmission power through power control methods, and at the same time can find the distance between nodes through signal strength. The sink node has no restrictions on any resources, such as: memory, energy, computing resources. When any nodes are within each other's communication range, their communication is not restricted.

[0041] The energy consumption is considered in three aspects. One is the energy consumption for sending data during wireless communication; the second is the energy consumption for receiving data during wireless communication; the third is the energy consumption for fusing data. At the same time, a simplified energy model in the prior art is adopted, and the energy consumption required for transmitting l bits of data over a distance d is expressed as follows:

[0042]

[0043] The energy consumption for receiving each l bits of data is:

[0044] E R = lE elec (3)

[0045] In the above two formulas, E T (l, d) represents the energy required to send l bits of data between two nodes at a distance of d, and E R is the energy required to receive l bits of data. E elccDenote the power consumption of the transmitter and receiver for processing unit-bit data as ε fs and ε mp are respectively the power consumption of the signal amplifier for processing unit-bit data per unit distance in the free-space channel model and the multipath attenuation channel model. Denote the threshold for differentiating between the free-space channel model and the multipath attenuation channel model. When the transmission distance is less than d0, the free-space fading model is selected, and the energy required for power amplification is ε fs . When the transmission distance is greater than d0, the multipath attenuation model is adopted, and the energy required for power amplification is ε mp .

[0046] In this embodiment, the energy consumption for fusing every l bits of data is:

[0047] E DF = lE DA (4)

[0048] In formula (4), E DF denotes the energy consumption for fusing every l bits of data, and E DA is the power consumption for fusing unit-bit data.

[0049] In this embodiment, the entire circular monitoring area R with a radius of M is divided into K (R = Kr, where r is the ring spacing and also the communication radius) uniformly spaced concentric ring regions, and the center of the ring is the base station (Sink). The i-th ring region can be expressed as:

[0050]

[0051] where the node s i determines its own level according to its distance d iToSink from the base station and the forwarding and transmission of information is to transmit messages layer by layer from the outer level to the inner level until they are sent to the base station. The node level The specific calculation formula is as follows:

[0052]

[0053] In this embodiment, a routing algorithm for protecting low-energy nodes is given. This algorithm seeks to consider nodes differently to extend the network lifetime. The routing algorithm adaptively adjusts the node clustering conditions according to the WSN network conditions and takes certain protective measures for low-energy nodes during clustering, aiming to balance the energy of nodes in the same ring to prevent premature death of low-energy nodes.

[0054] At the beginning of the algorithm, it is necessary to initialize the WSN network. At this stage, the node s i needs to determine its own neighbor nodes and its own level h siAfter all nodes are initialized, the nodes start to update their competition radii, then update their competition parameters, and finally enter the stage of cluster head election. After the cluster head election is completed, if it is found that there is no optimized routing scheme in the network, it will switch to the PSO optimization part; otherwise, it will start to enter the data dissemination stage of the WSN.

[0055] In traditional technologies, the competition radius of each cluster head is determined at the beginning and does not change during the life cycle of the nodes. However, this mechanism does not take into account that as nodes in the WSN die, the number of neighbor nodes around each node also changes. If the node competition radius remains unchanged, it is easy to cause an excessive load on the cluster heads of the node group in the low-energy area, resulting in premature death. The improved mechanism is to pay attention to the situation around each node to determine its competition radius. See Figure 1 , in the present invention, a competition radius that changes with the topology is adopted. The node competition radius changes positively with the load it can bear.

[0056] Combined with the existing technology, a clustering method of equidistant circular division can be obtained. The cluster radii in the same ring are equal and are positively correlated with the distance from the base station, as shown in Equation (7).

[0057]

[0058] The size of a node competition radius should be related to the upper limit of the load that the node can bear, that is, related to the energy level of the node and the topological environment of the surrounding network. Therefore, a parameter of the average available energy of neighbors is named to represent the situation of the load that the node can bear.

[0059] Specifically, for node s i , the remaining energy E r and the ratio of the number of elements NeiNum i in its own and the current neighbor node set are called the average available energy of neighbors. The mathematical description is as follows:

[0060] AvgNeiE i = E r / (NeiNum i + 1) (8)

[0061] And the initial value of the average available energy of neighbors AvgNeiE i of node s i and the historical maximum average available energy of neighbors HisAvgNeiE i is set as the ratio of the initial energy of the node to the maximum number of neighbors MaxNeiNum i :

[0062] HisAvgNeiE i= AvgNeiE i = E i / (MaxNeiNum i + 1) (9)

[0063] Then, during each subsequent election, node s i will update its average neighbor available energy AvgNeiE i , as shown in Equation (8), and update its historical maximum average neighbor available energy HisAvgNeiE i , as shown in Equation (10).

[0064] HisAvgNeiE i = max(E r / (NeiNum i + 1), HisAvgNeiE i ) (10)

[0065] If only this one parameter is used to measure the competition radius of a node in each period, it is not accurate. Therefore, this parameter needs to be compared with HisAvgNeiE i to determine the current competition radius from the previous historical periods of node s i . The result obtained is called the Normalized Average Neighbor available Energy (NAvgNeiE), as shown in Equation (11).

[0066]

[0067] Then, the normalized average neighbor available energy NAvgNeiE i of node s i obtained is transformed using the cosine offset model, and the following competition radius formula can be obtained, as shown in Equation (12):

[0068]

[0069] The competition radius r c of a cluster is jointly determined by the level h where node s i is located, the normalized average neighbor available energy NAvgNeiE i , and the remaining energy E r . Among them, h o is the threshold level, is a natural number greater than 1. Since r hThe distance to the base station has been considered, so it is not considered repeatedly in the above formula. Since the inner-layer nodes need to forward data as relay nodes, in order to reduce their energy consumption, a threshold layer is specified. Only when the layer number is greater than this layer will its contention radius change.

[0070] In this embodiment, the cluster head plays an important role in fusing and forwarding the data of all nodes in an area. In addition to having relatively high battery energy, the routing it has also needs to be excellent enough. Traditional routing algorithms usually use the remaining energy of nodes as the competition parameter for the cluster head. Although the selected cluster head has the most remaining energy, its routing metrics may not be the best.

[0071] The present invention defines a two-dimensional competition parameter CP i . To become a cluster head, the remaining energy of the node is the most critical measurement criterion. Therefore, the remaining energy E of the node r is used as the first parameter for cluster head election. At the same time, the parameter that measures the quality of the node's routing, that is, the routing quality parameter R i is used as the second parameter for cluster head election. The following will give an explanation: An excellent routing can consume less energy when transmitting the same amount of data, and the parameter EECR is used to measure the quality of the routing. First, the definition of EECR will be given, and then the rationality proof of EECR will be given.

[0072] In this embodiment, the actual forward distance is defined, that is, the actual forward distance (AFD) of a routing is the distance from the sender s i to the base station minus the distance from the receiver s j to the base station The mathematical description is as follows:

[0073] AFD = d iToSink - d jToSink = d iToSink - d j′ToSink (13)

[0074] Specifically, AFD is the distance that the data actually shortens from the base station through this routing. As Figure 2 shown, the node s i sends data to the node s j . Although it transmits a distance of d ij , in fact, the data packet only shortens the distance of d ij’ from the base station, that is, the distance of AFD. Obviously, the conversion efficiency of the energy consumed by its transmission and the distance it actually shortens from the base station is extremely low. Therefore, the present invention combines the actual forward distance of this routing with the energy E consumed for transmitting data sThe conversion efficiency is called the effective distance energy consumption ratio, simply referred to as the energy efficiency ratio (Effective distance energy consumption ratio, EECR).

[0075] EECR = AFD / E s (14)

[0076] It can be seen that when the distance to the next hop is the same, the larger the EECR, the better the route. However, because the energy E consumed at different transmission distances d ij is different, it causes the "distortion" phenomenon of EECR. In order to more objectively and truly reflect the quality of the route, the present invention needs to standardize the EECR at different transmission distances and then compare them. And the standard for each different route is the optimal route of the optimal next hop at the transmission distance of this route. At s the medium distance d Figure 2 the optimal next hop is point P, and at the same time its AFD is also the longest. ij In this embodiment, low-energy nodes are emphasized in the initial design. Therefore, a voting strategy is adopted, and it is hoped that low-energy nodes have a larger proportion to select a cluster head that is more friendly to most low-energy nodes, so as to extend the lifespan of more low-energy nodes as much as possible.

[0077] And the cluster head election parameters of the participating nodes are used as the reference basis for the voting nodes.

[0078] Let s

[0079] be an arbitrary participating node, and s i calculates its competition radius according to its own level and Equation (12). The radius of the area is denoted as i where MaxVote is the upper limit of the number of votes that each node can have, to prevent a node with too low energy from having too many votes, resulting in the influence of other low-energy nodes being reduced, and thus an unreasonable cluster head being selected. Each node obtains the corresponding number of votes (Votes) according to its remaining energy E

[0080] The specific formula for calculating the number of votes is as follows: r Votes

[0081] Votes i = min(E i / E r , MaxVote) (15)

[0082] The pseudocode of the voting process is shown in Algorithm 1. At the beginning of the voting, all nodes s with a level greater than 1 iBroadcast Vote_msg, where Vote_msg includes the fitness value, competition radius, and level of the node, as shown in lines 1 - 3. When node S j receives the Vote_msg from node S i , it will first determine whether they are at the same level. If not, it will discard them. Otherwise, it will start to compare the distance d ij between them with their competition radius. If d ij is less than , then it is determined that S i is a neighbor node of S j ; at the same time, if d ij is less than , then S i will be added to the set of potential competing nodes of S j . The above process is shown in lines 4 - 11. Lines 12 - 19 show the voting logic of the node. When node S i has the highest main competition parameter itself, it will vote for itself. If there is a node S n in its neighbor node set whose main competition parameter is higher than that of other nodes (including S i ), then S i will vote its own ballot for S n . Lines 20 - 26 indicate that if the number of votes obtained by node S i exceeds the threshold T.votenum, then this node will enter the next stage of the cluster head election process; otherwise, it will enter the sleep state to save energy.

[0083]

[0084] The following describes the algorithm for competitively selecting the cluster head. Algorithm 2 gives the pseudocode of the algorithm executed by any node s i during the cluster head election process. Because in the voting stage, each node s will establish its own set of potential competing nodes s.EC, so during the election, the participating nodes do not need to broadcast their own information again. They only need to send a confirmation message named Status_msg to confirm whether node S j enters the election. If no response from S j is received within the specified threshold time T.time, then by default S j quits the election and removes it from EC. In II(8 - 37), the process of selecting the final cluster head is shown. If the set of competing nodes s i .EC is empty, then s i directly declares itself as the cluster head. Otherwise, the final cluster head is selected through the CP comparison in II(12 - 37) from s i .EC: The comparison logic of the two-dimensional competition parameter is as follows:

[0085] i) When s i compared with s j and in the two-dimensional election parameters of node s i if either dimension has an advantage and the other dimension is at least equal, then s i directly wins the election, as shown in II(12 - 15).

[0086] ii) When s i compared with s j and in the two-dimensional election parameters of node s i if one dimension is at a disadvantage and the difference in the disadvantaged dimension dim1 between the two is not greater than the corresponding threshold and the difference in its advantageous dimension dim2 is greater than the corresponding threshold then s i wins the election, as shown in II(16 - 19).

[0087] iii) If the two-dimensional election parameters of the compared nodes are the same, then the node with the smaller node ID wins the election, as shown in II(20 - 23).

[0088] iv) If i, ii, and iii are not satisfied, then node s i withdraws from the election and sends the Withdrew_Election message, as shown in II(24 - 27).

[0089] When a node wins the election and sends the FINAL_CH message to other nodes. A node s j receives a FINAL_CH message from its election set EC, s j will withdraw from the election, send the Withdrew_Election message, and the nodes in the election set that contain s j will remove s j after receiving this message, as shown in II(28 - 37). The above steps are carried out until all nodes complete the election.

[0090] As for after the election ends, the process of incorporating cluster nodes is similar to the prior art. Ordinary nodes select the cluster head with the lowest communication cost and notify that cluster head. After this process ends, it enters the data transmission phase, where in-cluster communication uses single-hop communication.

[0091]

[0092]

[0093] Previously, traditional multi-hop routing protocols could also greatly save the communication energy consumption of long-distance nodes to a large extent through multi-hop, but they did not consider the fact that multi-hop led to relay nodes repeating the forwarding of the same data, resulting in energy waste in the overall network. In this section, based on the conclusions proven in the previous section, the traditional multi-hop was improved from the perspectives of balancing network resources and node energy consumption, thus proposing a rotating multi-hop scheme.

[0094] See Figure 3 , the rotating multi-hop transmission finds all the appropriate next hops of cluster head I that are less than the channel threshold d0, and allows cluster head I to communicate with these next hops alternately according to a certain rule, and in this way, takes into account the energy consumption of the cluster head node and the problem of energy waste in the network. The requirements for the next hop of node I can be obtained:

[0095] ComIns I ={s j |d Ij ≤d0} (16)

[0096]

[0097] where d0 is the channel threshold. If the distance d j between node s Ij and node I is less than d0, then node s j is the appropriate next hop of I, and ComIns I is the set of appropriate next hops of node I. And H I is the set of the next hop levels of node I.

[0098] As Figure 3 shown, assume that the next hops of node I that are all less than the channel threshold d0 are A, B, and the base station respectively. When I communicates with A, B, and the base station in three-hop communication, at this time, the data of I is forwarded three times in the network, as shown by the red line. Although the energy of node I is saved, it costs 2 times of forwarding energy for the whole network. When I communicates with B and the base station in two-hop communication, at this time, the data of I is forwarded two times in the network, as shown by the blue line, and the whole network costs 1 time of forwarding energy. When I communicates directly with the base station, as shown by the black line, there is no extra energy waste in the network at this time, but this will cause node I to sacrifice its own energy too much, which will lead to the energy consumption of node I being too fast and will also affect the life cycle of the whole network. The purpose of rotating multi-hop is that cluster head I will select the next hop level in its set of appropriate next hop nodes according to a certain rule to communicate in order to balance the resource waste and node energy consumption in the network.

[0099] If these three communication methods are combined according to a certain rule, and this rule is called ι i, then the set of all possible rules of the cluster head node is ι. The combination of the communication rules ι of all cluster head nodes in the network is called a rotation scheme called μ i , the set of all solutions is u.

[0100] Optimization formula for alternating multiple hops(FALM)

[0101] The target formula 1 can be obtained:

[0102]

[0103] Where E T (μ i ) is in μ i Node energy consumption under the scheme, Net w (μ i ) is in μ i The round-robin multi-hop scheme aims to minimize the waste of network energy while ensuring the life of all nodes. Therefore, in the target formula 1, the node energy consumption E T (μ i ) is taken into account, and network waste Net w (μ i ) to reduce the energy consumption of relay nodes. At the same time, it is assumed that there is a rotation scheme that can achieve a goal: to minimize the sum of node energy consumption and network energy waste. The goal is to make a scheme that saves node energy and reduces network energy waste to the minimum.

[0104] In common routing protocols, cluster heads only communicate with each other, which results in the problem of excessive energy consumption of the relayed cluster heads. The advantage of multi-path transmission is that it can balance the energy consumption in the WSN network. The present invention proposes a cluster head s that is deformed after combining alternating multi-hop with multi-path transmission. i Alternating communications between different layers and adopting multi-path transmission within the same layer can balance the consumption of relay nodes and the network at the same time.

[0105] The present invention converts node s i The communication node at the next hop level of the current communication is called Cluster head nodes i The set of nodes in the communication with its current level When communicating, s i Send to any node s in the communication j The data accounts for i The ratio of the total amount of data to be sent is called s j Accounts i The flow distribution is calculated as follows:

[0106] λ ij = DATA iToj / DATA i (19)

[0107] Where DATA iToj is the data volume sent to any in - communication node s i and DATA j is the total data volume required to be sent by s i We can get: i

[0108]

[0109] If node s i is a cluster - member (CM), then its traffic distribution λ i = λ ij = 1, and s j is its cluster - head node. As long as the above two equations are met, there are countless combinations of traffic distributions λ i from s to its in - communication nodes, and there must be a certain combination of traffic distributions that can effectively improve the lifetime of the WSN network. Therefore, some researchers have proposed a multi - path routing formula based on sequential quadratic programming (SQP) in multi - paths, namely EB - DEFL, to find such an effective solution. i

[0110] In this embodiment, the first objective function of energy - balanced multi - path routing is as follows:

[0111] Objective formula 2:

[0112]

[0113] Where θ i (λ) is the lifetime of node s i under this traffic distribution combination λ, and ∈ i (λ) is the energy consumption of node s i under this traffic distribution combination λ. In objective formula 2, the first term of the equation is to maximize the lifetime of the node with the minimum lifetime, the second term in the equation is to reduce the total consumption of all nodes, and η is a trade - off parameter. EB - DEFL is applicable to multi - path routing of nodes, but does not consider how to balance node energy consumption and network resources. Therefore, it is necessary to transform the formula in combination with objective 1.

[0114] ​​Among them, the target formulas 1 and 2 are fused to obtain the formula for the energy balance life maximization routing problem for alternate multi-hop (EB-FALM), that is, the target formula 3:

[0115]

[0116] Among them,

[0117] According to the EB-FALM formula, it can be known that maximizing the node with the minimum life cycle and minimizing the transmission energy consumption in the entire WSN are the key parts of optimization. Analyzed microscopically, the transmission energy consumption of a node is affected by two factors: 1. The distance to the next-hop node; 2. The size of the data exchanged. As the convergence point of all data within a cluster and needing to transmit a large amount of data to its next hop, the cluster head node (CH) consumes much more energy than the cluster member nodes, so its life will be shorter than other nodes. At the same time, in the second half of the target formula 3, the network transmission of the entire WSN and the energy consumption wasted by the network are also considered macroscopically.

[0118] The more relay nodes there are on a route, except for being beneficial to the first node sending data, for any relay node in the route, its energy consumption will far exceed normal, which belongs to a waste of network energy. Therefore, from the perspective of the entire network, this is not conducive to extending the life cycle of the WSN. For the proposed alternate multi-hop communication scheme, the node selects a reasonable next-hop node after weighing its own energy consumption and the energy consumption of the network, and extends the life cycle of the network on the premise of ensuring the life of the node itself.

[0119] In this embodiment, single-hop communication is adopted within the cluster, and only alternate multi-hop communication is selected between clusters. And because when the cluster head alternates multi-hop, its current alternate next hops are all at the same level, and there will be no situation of different levels, that is, the delay is the same. And according to the experiment, the multi-hop times of any node will not exceed 4 times, and the delay is less than the standard multi-hop algorithm.

[0120] The simulation results of the technical solution of the present invention show that compared with the traditional routing algorithms such as LEACH, HEED, and EEUC, the present invention has improved performance. In the network stable period stage, the life cycle of the algorithm is extended by 80%, and during the entire life cycle of the WSN, the node performance during the survival period of the algorithm is maintained at a relatively high level and increased by about 43.7%.

[0121] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made thereto based on the present invention, which will be obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection required by the present invention.

Claims

1. A low-energy priority election-based hierarchical multi-path non-uniform clustering method, characterized in that It includes a cluster head election phase, in which the candidate nodes are graded according to the minimum number of hops from the candidate nodes to the base station, and each candidate node participates in the election by integrating the remaining energy and the routing quality parameter. A set of potential competing nodes is established for each candidate node. In the cluster head election process, each candidate node sends a confirmation message to confirm whether to enter the election. If no response is received within the specified threshold time, the corresponding candidate node withdraws from the election. When a certain candidate node wins the election, it sends a completion message to the remaining candidate nodes, and the remaining candidate nodes withdraw from the election. After the election ends, the cluster head with the minimum communication cost is selected and the corresponding cluster head is notified, and then it enters the data transmission phase, and single-hop communication is adopted for intra-cluster communication. Define the actual forward distance, that is, the actual forward distance of a route is the distance from the sender s i to the base station minus the distance from the receiver s j to the base station Each node obtains corresponding votes according to its remaining energy. At the beginning of the voting, all candidate nodes with a level greater than 1 broadcast Vote_msg, and Vote_msg includes the fitness value, competition radius and level of the node. When node S j receives a Vote_msg from S i node, first determine whether they are at the same level. If not, discard it; otherwise, start calculating the distance d j between node S i and node S ij and compare it with the competition radius If the distance d ij is less than the competition radius then identify node S i as a neighbor node of node S j If the distance d ij is less than then add node S i to the set of potential competing nodes of node S j ; Two-dimensional election parameters of the candidate nodes are preset, and the remaining energy of the candidate nodes is used as the first parameter for cluster head election. The routing quality parameter that measures the routing quality of the candidate nodes is used as the second parameter for cluster head election; the routing quality parameter is measured by the energy efficiency conversion ratio (EECR) parameter, and the EECR is the conversion efficiency of the actual forward distance of the routing and the energy consumed for transmitting data; the comparison logic of the two-dimensional election parameters is: i) When node s i compares with node s j and any dimension of the two-dimensional campaign parameter of node s i is in an advantageous position in one dimension and at least on a par in the other dimension, then node s i directly wins the campaign; ii) When node s i is compared with node s j , node s i has a disadvantage in one dimension of its two-dimensional election parameters, and the difference in the disadvantageous dimension dim1 between the two is not greater than the corresponding threshold T.CP dim1 , and the difference in the advantageous dimension dim2 of node s i is greater than the corresponding threshold T.CP dim2 , then node s i s i wins the election; iii) If the two-dimensional election parameters of the compared nodes are the same, the node with the smaller node ID wins the election. iv) If i, ii, and iii are not satisfied, node s i Withdraw from the campaign and send a withdrawal election message; Find all the suitable next hops of cluster head I that are less than the channel threshold d0, and allow cluster head I to communicate with these next hops alternately according to a preset rule. Single-hop communication is adopted within the cluster, and alternate multi-hop communication is selected between clusters.

2. The low-energy-preferred election hierarchical multi-path non-uniform clustering method according to claim 1, wherein Before the cluster head election, it also includes initializing the WSN network, and the candidate nodes determine their neighbor nodes and their own levels. When all candidate nodes are initialized, the candidate nodes start to update the competition radius, then update their competition parameters, and then enter the cluster head election phase.

3. A low-energy priority election-based hierarchical multi-path non-uniform clustering method according to claim 2, characterized in that The candidate nodes adopt a competition radius that changes with the topology, and the competition radius of the candidate nodes changes positively with the load borne. The ratio of the remaining energy of the candidate node to the number of elements in its own and the current neighbor node set is used as the average available energy of the neighbors.

4. A low-energy priority election-based hierarchical multi-path non-uniform clustering method according to claim 3, characterized in that Each time an election is held, the candidate node updates the average available energy of the neighbors and the historical maximum average available energy of the neighbors. Determine the current competition radius from the historical period of the candidate node, and use the obtained result as the normalized average available energy of the neighbors. The obtained normalized average available energy of the neighbors is transformed by using a cosine offset model to obtain the competition radius formula of the cluster.

5. A low-energy priority election-based hierarchical multi-path non-uniform clustering method according to claim 4, characterized in that The competition radius of the cluster is jointly determined by the level of the candidate node, the normalized average available energy of the neighbors and the remaining energy. A threshold level is preset, and the competition radius is changed when it is greater than the threshold level.

6. The low-energy priority election-based hierarchical multi-path non-uniform clustering method according to claim 1, characterized in that If there is a communication internal node C of node A on the line connecting node A to base station S, node C is called the optimal next-hop node at a distance of d AC When reaching the base station through the next-hop node, the total transmission distance is minimized and the actual forward distance of this route is the longest.

Images