An underwater sensor network clustering and parallel communication method

Abstract

The application discloses a kind of underwater sensor network clustering parallel communication methods based on repeated calculation and improved K-means clustering, the method is not considered underwater energy propagation model in the current underwater LEACH protocol based on K-means clustering, and the lack of cluster head node broadcast parallel transmission time slot allocation information will increase its energy consumption, influence network communication rate, proposes to use underwater energy propagation model in K-means clustering to optimize the number of clustering, and to determine cluster head node and network clustering;Meanwhile, each node obtains its parallel transmission time slot by repeated calculation, i.e. The application can be applied to underwater sensor networks in network structures such as underwater ecological monitoring, early warning detection, disaster monitoring, etc.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater acoustic communication, and particularly relates to a method for underwater sensor network clustering and parallel communication based on repeated calculation and improved K-means clustering. BACKGROUND

[0002] With the continuous exploration of human beings in the ocean, underwater communication technology is also developing, and sound waves are the only available information carrier for long-distance wireless communication in the ocean. The underwater acoustic channel has narrow frequency band, strong interference and long time delay; and underwater nodes are powered by limited batteries, so correct data transmission and low power consumption of the network become necessary considerations for underwater acoustic networks. Underwater sensor networks are underwater acoustic communication networks composed of multiple underwater nodes equipped with low-power sensors and base stations. Underwater nodes (hereinafter referred to as "nodes") collect underwater information through low-power sensors while continuously transmitting the collected information to the base station through the underwater acoustic communication module. Underwater sensor networks have good application prospects in activities such as ocean environment management, resource protection, disaster monitoring and marine military.

[0003] To reduce the energy consumption of underwater sensor network operation, in large-scale underwater sensor networks, a clustering communication method is often used: that is, to select part of the nodes as cluster head nodes to communicate directly with the base station, and the remaining nodes only communicate directly with the cluster head node of the cluster. In this way, the energy consumption of the nodes in the data transmission process is reduced, and the network lifetime is extended. In recent years, a number of research results have emerged for underwater sensor networks. Currently, there are routing protocols such as VBF, HHYBF, FBR, etc. for underwater sensor networks, as well as improvements such as using traditional wireless communication protocols such as LEACH and SEP for underwater sensor networks.

[0004] LEACH protocol is one of the classic clustering communication protocols for wireless sensor networks. In the LEACH protocol, the network operates in rounds, and each round is divided into two stages: cluster establishment and data transmission. When applying the LEACH protocol to underwater sensor networks, the main disadvantage is the randomness of the cluster establishment stage: the cluster head nodes are selected by random results, so it is possible that the cluster head nodes are far away from the base station, causing the cluster head nodes to consume more energy during the operation of the underwater sensor network; at the same time, the LEACH protocol requires the cluster head nodes to send control signaling in both the cluster establishment and data transmission stages, resulting in more energy consumption of the cluster head nodes and shorter network operation life. In order to overcome the above problems and extend the network operation life, researchers have introduced algorithms such as simulated annealing and K-means clustering to determine the cluster head nodes and network clustering in the clustering stage of the LEACH protocol, or used multi-hop routing algorithms to equalize the energy consumption of each cluster head node in the data transmission stage. However, in the underwater LEACH protocol based on K-means clustering, the following problems still exist:

[0005] 1. In the phase of establishing clustering under the LEACH protocol, the underwater energy propagation model is not considered or the terrestrial wireless propagation model is misused, resulting in that the node clustering result cannot make the network operation energy consumption minimum;

[0006] 2. The parallel communication of nodes is a necessary method to improve the network communication rate, but the transmission time slot allocated for parallel communication needs the cluster head node to broadcast control signaling in the cluster, and the broadcasting signaling process increases the energy consumption of the cluster head node and affects the network communication rate. SUMMARY

[0007] The purpose of the present application is to solve the defects in the prior art that the underwater LEACH protocol based on K-means clustering does not consider the underwater energy propagation model and the cluster head node broadcasting parallel transmission time slot allocation information increases its own energy consumption and affects the network communication rate, and provides an underwater sensor network clustering and parallel communication method based on repeated calculation and improved K-means clustering. The above method uses the underwater energy propagation model in K-means clustering to optimize the clustering number and determine the cluster head node and network clustering; at the same time, each node obtains its own parallel transmission time slot by repeated calculation, i.e. simultaneously and independently running the same determination algorithm, without the cluster head node broadcasting parallel transmission time slot allocation information, thereby reducing the energy consumption of the network and prolonging the network life.

[0008] The purpose of the present application can be achieved by adopting the following technical solutions:

[0009] A underwater sensor network clustering and parallel communication method based on repeated calculation and improved K-means clustering, applied to an underwater sensor network having a plurality of nodes capable of independently transmitting and receiving data, and the topology structure is as shown in Figure 1 . Each node communication mode is half duplex, and can only listen or send data on the channel at the same time. The present application specifically comprises the following steps:

[0010] S1, based on the underwater energy propagation model, calculating the optimal clustering number K according to the number N of nodes, the energy consumption of sending and receiving data packets, and the distribution of nodes in the underwater area opt ;

[0011] S2, initializing the iteration variable e=1, setting the maximum iteration number as e max , and setting the initial energy of the node as E0;

[0012] S3, randomly selecting K opt nodes as initial cluster head nodes CH i , i=1,2,…,K opt , i is the clustering number;

[0013] S4, respectively calculating the non-cluster head nodes Cluster head node CH of cluster j j Euclidean distance wherein, is a non-cluster head node in cluster i with intra-cluster serial number m, m = 1, 2, …, (N-K opt ); when e = 1, let i = 0;

[0014] i = 1, 2, …, K opt is defined in step S3, but i = 0 has not been defined before.

[0015] S5, search j that makes the Euclidean distance minimum, and record it as j min

[0016]

[0017] add the non-cluster head node to cluster j min ;

[0018] S6, perform steps S4-S5 on all non-cluster head nodes so that each non-cluster head node joins a cluster;

[0019] S7, according to the clustering result, calculate the sum of Euclidean distances of each node in each cluster to all other nodes in the cluster (all other nodes in the cluster include non-cluster head nodes or cluster head nodes CH i ); when calculating the Euclidean distance of cluster head node CH i to all other nodes in the cluster, define m = 0;

[0020] Before this, in step S4, the node transmission priority of non-cluster head nodes is defined as m = 1 ~ (N-K opt ), and m = 0 has not been defined before; therefore, the definition of m = 0 here makes that is, it can refer to non-cluster head nodes (m = 1, 2, …), or it can refer to cluster head nodes (m = 0).

[0021] S8, search m that makes minimum in cluster i, and record it as m min

[0022]

[0023] take the node corresponding to m min as the cluster head node of cluster i;

[0024] At this time, mmin It may point to a non-cluster head node or a cluster head node, so it is collectively referred to as a "node".

[0025] S9. Let e ​​= e + 1; if the clustering result or the cluster head node changes, and e <e max If the result is positive, proceed to step S4; otherwise, continue to step S10.

[0026] S10. Reset the non-cluster head nodes within each cluster. The sequence number m, m = 1 to N i And it is equivalent to making non-cluster head nodes To the cluster head node CH i The sequence number when the distances are arranged from nearest to farthest, N i This represents the number of non-cluster head nodes within cluster i;

[0027] S11, Non-cluster head node and cluster head node CH j Calculate the collision-free parallel transmission time of data packets for each node within and between clusters. and t i S12, Timer t starts counting, for each non-cluster head node. Based on the transmission time calculated in step S11 exist The current cluster head node CH i Send a b CH For bit-based data packets, calculate the non-cluster head nodes based on the underwater energy propagation model. Power of data packets

[0028]

[0029] in Indicates non-cluster head node To cluster head node CH i Euclidean distance, Indicates non-cluster head node The underwater depth, I0 represents the minimum sound intensity at which the data packet can be correctly received by the destination node, and the physical definition of sound intensity is power per unit area;

[0030] Calculate the energy consumption of sending data packets

[0031]

[0032] Among them, t TX (b CH ) for b CH The time taken to send a bit data packet;

[0033] The cluster head node CH i Receiving and fusing the energy consumption of 1-bit data of all non-cluster head nodes in the cluster i The energy consumption of data packets The calculation formula is as follows:

[0034]

[0035] e′ RX The cluster head node CH i Receiving and fusing the energy consumption of 1-bit data of all non-cluster head nodes in the cluster i

[0036] Updating the energy of the non-cluster head node according to formula (6), (7) The cluster head node CH i The cluster head node CH

[0037]

[0038] S13, set the timer t to zero, and restart the timing, each cluster head node CH i According to the sending time t calculated in step S11 i , send the data packet received in step S12 to the base station at t=t i ; according to the underwater energy propagation model, the same method of formula (3) (4) is used to calculate the energy consumption of CH i sending data packets

[0039]

[0040] wherein, The sending power of the cluster head node CH i , The Euclidean distance from the cluster head node CH i to the base station, The underwater depth of the cluster head node CH i , TX (b BS ) is the sending time of b BS bit data packet;

[0041] Updating the energy of the cluster head node CH i according to formula (9)

[0042]

[0043] S14, if all communication tasks of the underwater sensor network are completed, end; otherwise, jump to step S1.

[0044] Further, in the underwater sensor network during operation, the energy consumption of the node transmitting data packets is one of the main energy consumptions of the network. The energy consumption of the node transmitting data packets conforms to the underwater energy propagation model, which is mainly positively correlated with the distance between the sending and receiving nodes. Therefore, in the underwater sensor network, the total distance of all nodes transmitting data packets is often reduced by clustering nodes, thereby reducing the operating energy consumption of the network. Through the algorithm based on the underwater energy propagation model, the relationship between the network operating energy consumption and the clustering number can be constructed, and then the optimal clustering number K that minimizes the network energy consumption is calculated opt . The following method is used to calculate the optimal clustering number K opt in step S1

[0045] S1.1, according to the underwater energy propagation model, the total energy consumption E' of the underwater sensor network for sending a communication task is calculated as:

[0046] E' = E' TX + E' RX (10)

[0047] Wherein, E' TX is the energy consumption of all non-cluster head nodes and cluster head nodes CH i sending data packets:

[0048]

[0049] E' RX is the energy consumption of all cluster head nodes CH i receiving and fusing data packets

[0050]

[0051] S1.2: Calculate K opt :

[0052]

[0053] Further, in the above method of calculating the optimal clustering number K opt , when N underwater nodes in the static wireless sensor network are uniformly distributed in a two-dimensional square water area with a side length of l and a depth of H, and the base station is located at the center of the entire network, the formula for calculating E' is as follows:

[0054]

[0055] Wherein, the variable K represents the clustering number;

[0056] Calculate the optimal clustering number K of the network opt

[0057]

[0058] Further, the non-collision parallel transmission of the network should meet the condition that the data packets are received without collision at the destination node: taking data packets a1, a2 from two sending nodes, the same destination node as an example, if they are both received without collision at the destination node, the arrival time t'1, t'2 of a1, a2 at the destination node should meet:

[0059]

[0060] wherein t guard represents the protection time slot, t RX is the receiving time of the data packet. On the basis of the above formula, the non-cluster head node and the cluster head CH j calculate the sending time t and t i of the data packet of each node in the cluster and between the clusters without collision parallel transmission respectively, and the process is as follows:

[0061] S11.1, let the clustering number i = 1, set the set TS' i record the non-cluster head node in the clustering i sending the non-collision arrival time of the data packet to the cluster head node CH i of the cluster; arrange the elements in the set TS' i from small to large, let ts' k represent the element with the serial number k in TS' i , and let TS' i = {0, +∞};

[0062] S11.2, let the serial number of the non-cluster head node m = 1, calculate the non-collision arrival time t of the data packet sent by the non-cluster head node

[0063] calculate the transmission delay of the data packet sent by each non-cluster head node to the cluster head node in the clustering i; v represents the underwater sound speed, and d represents the Euclidean distance from the non-cluster head node i to the cluster head node CH

[0064] search for adjacent elements with serial numbers k min and k min +1 that meet the following formula:

[0065]

[0066] wherein tguard Indicates the protection time slot, t RX (b CH ) for b CH Time taken to receive bit data packets;

[0067] Equation (16) is further explained as follows: In step S11.1, the set TS′ of the arrival times of data packets is declared. i and initialize TS′ i = {0, +∞}. Therefore, the set TS′ is... i There are at least two elements in it;

[0068] In step S11.1, it was declared that Represents set TS′ i The element in, k is The indices are arranged in ascending order. Therefore, k and k+1 form the set TS′. i The index of adjacent elements in the array, given k (element). That is, we know k+1 (element) There is no non-causal statement;

[0069] In summary, k min +1 and k min For set TS′ i The sequence numbers of adjacent elements in equation (16) do not contain any non-causal expressions.

[0070] make Time of arrival Add to set TS′ i ; TS′ i All arrival times are sorted in ascending order; for the cluster with index i>1, its set TS′ i It records both the arrival time of data packets transmitted within the cluster and the arrival time of data packets from other sub-clusters, hence the collective term "arrival time";

[0071] Let m = m + 1, and execute step S11.2 until each non-cluster head node in cluster i is calculated. Collision-free arrival time of transmitted data packets

[0072] S11.3 Calculate each non-cluster head node within cluster i. Collision-free transmission time of data packets

[0073]

[0074] S11.4 Calculate each non-cluster head node within cluster i. To other cluster head nodes CH j Arrival time of transmitted data packets

[0075]

[0076] Data packet arrival time Add the corresponding cluster head node CH j The set TS′ j ;

[0077] S11.5. Let i = i + 1, and repeat steps S11.2-11.4 until each non-cluster head node in all clusters is calculated. Collision-free transmission time of data packets

[0078] S11.6 Calculate the cluster head node CH i The data packets arrive at the base station without collision at transmission time t. i :

[0079] Let i = 1; set set TS′ BS Record each cluster head node CH i Collision-free arrival time of data packets sent to the base station In the set TS′ BS Arranged in ascending order, using Represents set TS′ BS Let TS′ be the element with index k in the middle. BS ={0,+∞};

[0080] S11.7 Calculate CH for each cluster head node i Transmission delay to base station The index k that satisfies the following formula is searched. min k min +1 adjacent elements

[0081]

[0082] make Data packet arrival time Add to set TS′ BS ; TS′ BS middle Sort by size from smallest to largest;

[0083] Let i = i + 1, and execute S11.7 until the CH of each cluster head node is calculated. i Collision-free arrival time of data packets sent to the base station

[0084] S11.8 Calculate the cluster head node CH i Collision-free transmission time t of data packetsi

[0085]

[0086] The present application has the following advantages and effects relative to the prior art:

[0087] 1. In the process of calculating the clustering of the underwater sensor network node, the underwater energy propagation model is considered, and the optimal clustering number K is calculated to minimize the network operation energy consumption opt The node clustering result and cluster head node selection of the present application can significantly shorten the transmission distance when transmitting data, thereby significantly reducing the network operation energy consumption and prolonging the network operation life;

[0088] 2. In the process of determining the non-collision parallel transmission of data packets within and between network clusters, repeated calculation is adopted, each node calculates the transmission time of the data packet, and the cluster head does not need to broadcast signaling to issue the clustering result and time slot division information, which significantly reduces the energy consumption of the cluster head node in the broadcasting signaling process, significantly prolongs the network operation life, and significantly shortens the node waiting time for signaling, significantly improving the network communication rate;

[0089] 3. The parallel communication method adopted by the present application can realize non-collision parallel communication of nodes within and between underwater sensor network clusters, and improve the network communication rate. BRIEF DESCRIPTION OF DRAWINGS

[0090] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0091] Figure 1 is a schematic diagram of the underwater sensor network topology structure disclosed in the present application for the improved K-means clustering underwater sensor network clustering and parallel communication method;

[0092] Figure 2 is a flowchart of the improved K-means clustering underwater sensor network clustering and parallel communication method disclosed in the present application. DETAILED DESCRIPTION

[0093] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme of the embodiments of the present application will be described clearly and completely below in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0094] Embodiment 1

[0095] The embodiment discloses a method for underwater sensor network clustering and parallel communication based on repeated calculation and improved K-means clustering, and is applied to an underwater sensor network with multiple nodes capable of independently transmitting and receiving data. In the underwater sensor network, the number of nodes N=100, the underwater depth of the nodes H=50m, all the nodes are fixed and subject to uniform distribution in a square two-dimensional water area with a side length l=100m, a base station is located at the center of the entire network, the initial energy of the nodes is E0=40j, a non-cluster head node transmits a data packet to a cluster head node with a size b CH =50 bits, a cluster head node transmits a data packet to a base station with a size b BS =1000 bits, the minimum sound intensity at which a data packet can be correctly received by a destination node is I0=1w, the energy consumption of a cluster head node for receiving and fusing 1-bit data is e′ RX =0.005J, the guard time is t guard =0.2s, the underwater sound speed is v=1500m / s, the maximum iteration number of network iteration clustering is e max =2000, and each node is half-duplex in communication, and can only listen or transmit data on a channel at the same time. The method specifically comprises the following steps.

[0096] S1, based on the number of nodes N, the energy consumption of transmitting and receiving data packets and the distribution of nodes in the underwater area, the optimal clustering number K is calculated based on an underwater energy propagation model opt ;

[0097] S2, an iteration variable e=1 is initialized, the maximum iteration number is set to e max , and the initial energy E0 of the nodes is set;

[0098] S3, K opt nodes are randomly selected as initial cluster head nodes CH i , i=1,2,…,K opt , and i is a clustering serial number;

[0099] S4, the Euclidean distance of a non-cluster head node to a cluster head node CH j of a cluster j is calculated respectively wherein, is a non-cluster head node with an intra-cluster serial number m in the cluster i, m=1,2,…,(N-K opt ), when e=1, i=0 is set;

[0100] In step S3, i=1,2,…,K opt is defined, but i=0 has not been defined before.

[0101] S5, Searching for Euclidean Distance The smallest j is denoted as j min

[0102]

[0103] Non-cluster head nodes Add cluster j min ;

[0104] S6. For all non-cluster head nodes Perform steps S4-S5 to ensure that each non-cluster head node They were all added to a certain cluster;

[0105] S7. Based on the clustering results, calculate the distance from each node in each cluster to all other nodes in the cluster (all other nodes in the cluster include non-cluster head nodes). or cluster head node CH i The sum of Euclidean distances Calculate the cluster head node CH i to all other nodes within the cluster When m = 0, it is stipulated that m = 0;

[0106] Previously, in step S4, for non-cluster head nodes... The node sending priority m = 1 to (NK) is defined. opt The value of m = 0 was not defined before; therefore, it is stipulated here that m = 0 such that It can refer to non-cluster head nodes (m = 1, 2, ...) or cluster head nodes (m = 0).

[0107] S8, Search Clustering i-internal command The smallest m is denoted as m min

[0108]

[0109] m min The corresponding node serves as the cluster head node for cluster i;

[0110] At this time, m min It may point to a non-cluster head node or a cluster head node, so it is collectively referred to as a "node".

[0111] S9. Let e ​​= e + 1; if the clustering result or the cluster head node changes, and e <e max If the result is positive, proceed to step S4; otherwise, continue to step S10.

[0112] S10. Reset the non-cluster head nodes within each cluster. The sequence number m, m = 1 to N iand equal to the distance from the non-cluster head node to the cluster head node CH of the cluster i arranged from near to far, the serial number N i represents the number of non-cluster head nodes in the cluster i;

[0113] S11, the non-cluster head node and the cluster head node CH j calculate the sending time of each node in the cluster and the cluster for non-collision parallel transmission of data packets and t i ;

[0114] S12, the timer t starts timing, each non-cluster head node according to the sending time calculated in step S11 at to the cluster head node CH of the cluster i send a b CH bit data packet, according to the underwater energy propagation model, calculate the power of the non-cluster head node sending data packets

[0115]

[0116] where represents the Euclidean distance from the non-cluster head node to the cluster head node CH i , H is the underwater depth of the non-cluster head node , I0 is the minimum sound intensity at which the data packet can be correctly received by the destination node, and the sound intensity is physically defined as the power per unit area;

[0117] Calculate the energy consumption of sending data packets

[0118]

[0119] where t TX (b CH ) is the sending time of b CH bit data packet;

[0120] Calculate the energy consumption of the cluster head node CH i receiving and fusing all the data packets of the non-cluster head nodes in the cluster i The calculation formula is as follows:

[0121]

[0122] e′ RX is the energy consumption of the cluster head node CH i receiving and fusing 1 bit data.

[0123] update the energy of non-cluster head nodes according to formula (6), (7) and cluster head nodes CH i energy

[0124]

[0125] S13, set the timer t to zero, and restart the timer, each cluster head node CH i according to the sending time t calculated in step S11 i , sends the data packet received in step S12 to the base station at t = t i ; according to the underwater energy propagation model, the energy consumption of CH i sending the data packet is calculated by the same method as formula (3) (4)

[0126]

[0127] wherein, is the transmission power of cluster head node CH i , is the Euclidean distance from cluster head node CH i to the base station, t TX (b BS ) is the transmission time of b BS bit data packet;

[0128] update the energy of cluster head node CH i according to formula (9)

[0129]

[0130] S14, if all communication tasks of the underwater sensor network are completed, end; otherwise, jump to step S1.

[0131] Further, in the running process of the underwater sensor network, the energy consumption of node transmitting data packet is one of the main energy consumptions of the network. The energy consumption of node transmitting data packet conforms to the underwater energy propagation model, which is mainly positively correlated with the distance between the sending and receiving nodes. Therefore, in the underwater sensor network, the total distance of all nodes transmitting data packets is often reduced by clustering nodes, so as to reduce the running energy consumption of the network. Through the algorithm based on the underwater energy propagation model, the relationship between the network running energy consumption and the clustering number can be constructed, and then the optimal clustering number K opt that minimizes the network energy consumption can be calculated. The optimal clustering number K opt is calculated in step S1 by the following method:

[0132] ​S1.1, according to the underwater energy propagation model, the total energy consumption E' of the underwater sensor network for completing a communication task is calculated as:

[0133] E' = E' TX + E' RX (10)

[0134] wherein E' TX is the energy consumption of all non-cluster head nodes and cluster head nodes CH i for sending data packets:

[0135]

[0136] E' RX is the energy consumption of all cluster head nodes CH i for receiving and fusing data packets

[0137]

[0138] S1.2: K opt is calculated:

[0139]

[0140] In the embodiment, the total energy consumption E' is:

[0141]

[0142] wherein the variable K represents the number of clusters;

[0143] The optimal number of clusters K opt of the network

[0144]

[0145] Further, the conflict-free parallel transmission of the network should satisfy that the data packets are received conflict-free at the destination node: taking data packets a1, a2 from two sending nodes, the same destination node as an example, if they are both received conflict-free at the destination node, then the arrival time t'1, t'2 of a1, a2 at the destination node should satisfy

[0146]

[0147] wherein t guard represents the protection time slot, and t RX is the receiving time of the data packet. On the basis of the above formula, the non-cluster head nodes and the cluster head nodes CH j respectively calculate the sending time t and t iThe process is as follows:

[0148] S11.1 Let the cluster index i = 1, and set the set TS′ i Record each non-cluster head node in cluster i To the cluster head node CH i Collision-free arrival time of transmitted data packets In set TS′ i Arrange the data in ascending order, using ts′ k Indicates TS′ i Let TS′ be the element with index k in the middle. i ={0,+∞};

[0149] S11.2. Let the index of the non-cluster head node m = 1, and calculate the index of the non-cluster head node. Collision-free arrival time of transmitted data packets

[0150] Calculate each non-cluster head node within cluster i Transmission delay for sending data packets to the cluster head node v represents the speed of sound underwater. Indicates non-cluster head node To the cluster head node CH i The Euclidean distance;

[0151] The index k that satisfies the following formula is searched. min k min +1 adjacent elements

[0152]

[0153] In the formula, t guard Indicates the protection time slot, t RX (b CH ) for b CH Time taken to receive bit data packets;

[0154] The following is a further explanation of equation (16):

[0155] In step S11.1, the set TS′ of data packet arrival times is declared. i and initialize TS′ i ={0,+∞}. Therefore, the set TS′ is... i There are at least two elements in it;

[0156] In step S11.1, it was declared that Represents set TS′ i The element in, k is The indices are arranged in ascending order. Therefore, k and k+1 form the set TS′.i The index of adjacent elements in the array, given k (element). That is, we know k+1 (element) There is no non-causal statement;

[0157] In summary, k min With k min +1 represents the set TS′ i The sequence numbers of adjacent elements in equation (16) do not contain any non-causal expressions.

[0158] make Time of arrival Add to set TS′ i ; TS′ i All arrival times are sorted in ascending order; for the cluster with index i>1, its set TS′ i It records both the arrival time of data packets transmitted within the cluster and the arrival time of data packets from other sub-clusters, hence the collective term "arrival time";

[0159] Let m = m + 1, and execute step S11.2 until each non-cluster head node in cluster i is calculated. Collision-free arrival time of transmitted data packets

[0160] S11.3 Calculate each non-cluster head node within cluster i. Collision-free transmission time of data packets

[0161]

[0162] S11.4 Calculate each non-cluster head node within cluster i. To other cluster head nodes CH j Arrival time of transmitted data packets

[0163]

[0164] Data packet arrival time Add the corresponding cluster head node CH j The set TS′ j ;

[0165] S11.5. Let i = i + 1, and repeat steps S11.2-11.4 until each non-cluster head node in all clusters is calculated. Collision-free transmission time of data packets

[0166] S11.6 Calculate the cluster head node CH i The data packets arrive at the base station without collision at transmission time t. i:

[0167] Let i = 1; Set TS' BS Record each cluster head node CH i Conflict-free arrival time of data packet to base station In set TS' BS Arrange from small to large, use To represent the element with ordinal number k in set TS' BS , Let TS' BS = {0, +∞};

[0168] S11.7, Calculate the transmission delay of each cluster head node CH i To base station Search the adjacent element with ordinal number k min , k min +1 that satisfies the following formula

[0169]

[0170] Let Add the arrival time of data packet to set TS' ; Arrange BS In TS' BS From small to large;

[0171] Let i = i + 1, execute S11.7 until the conflict-free arrival time of data packet to base station of each cluster head node CH i Is calculated

[0172] S11.8, Calculate the conflict-free sending time t i Of data packet of cluster head node CH i

[0173]

[0174] ​According to the above method, the method (hereinafter referred to as "the method herein") and the original LEACH protocol (hereinafter referred to as "method 1") and GEC algorithm (hereinafter referred to as "method 2") are simulated for 10 rounds on the NS3 network simulator platform, and the experimental results of the average residual energy of nodes, the average number of surviving nodes, the round of the first dead node, and the average received data amount of the base station of the three methods are compared, as shown in Tables 1-4. Comparing the method herein with method 1 and method 2, it can be seen that with the increase of communication rounds, the average residual energy of nodes and the average number of surviving nodes of the method herein decrease the slowest, the average received data amount of the base station increases the fastest, and the round of the first dead node is the latest. The experimental results of the above performance indicators are better than those of method 1 and method 2. Therefore, the clustering method of the application can reduce the energy consumption of network operation, the node parallel communication algorithm of the application can reduce the energy consumption of nodes, prolong the life of cluster head nodes, and improve the communication rate of the network. In summary, the method proposed in the application is feasible.

[0175] Table 1. Average residual energy of nodes in 20 simulations

[0176] Run 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Method 1 * ]] 40 30.89 21.51 12.11 2.87 0.1 0 0 0 0 0 Method 2 40 33.58 26.80 20.28 13.77 6.67 0.41 0 0 0 0 Algorithm herein 40 34.89 28.67 23.52 17.56 12.1 6.24 0.71 0 0 0

[0177] Table 2. Average number of surviving nodes in 20 simulations

[0178] Run 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Method 1 * ]] 100 100 100 100 87 2 0 0 0 0 0 Method 2 100 100 100 100 100 76 4 0 0 0 0 Algorithm herein 100 100 100 100 100 99 92 6 0 0 0

[0179] Table 3. Round of the first dead node in 10 simulations

[0180] Run 1 2 3 4 5 6 7 8 9 10 Method 1 * ]] 664 734 691 728 671 648 601 728 673 633 Method 2 913 895 936 939 927 909 922 885 916 880 Algorithm herein 1030 1006 1007 1045 1025 1024 1006 1047 1049 959

[0181] Table 4. Average received data amount of the base station in 10 simulations (unit: 10 4 bit)

[0182] Run 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Method 1 * ]] 0 0.12 0.23 0.35 0.46 0.47 0.47 0.47 0.47 0.47 0.47 Method 2 0 0.80 1.56 2.30 3.06 3.81 4.68 4.76 4.76 4.76 4.76 Algorithm herein 0 0.95 1.82 2.75 3.61 4.55 5.41 5.61 5.61 5.61 5.61

[0183] Example 2

[0184] The embodiment continues to disclose a method for underwater sensor network clustering and parallel communication based on repeated calculation and improved K-means clustering, which is applied to an underwater sensor network with multiple nodes capable of independent data transmission and reception. In the underwater sensor network, the number of nodes N = 50, the underwater depth of the nodes H = 50 m, all the nodes are fixed and subject to uniform distribution in a square two-dimensional water area with a side length l = 400 m; the base station is located at the center of the entire network; the initial energy of the nodes E0 = 40 j, the size of the data packet sent by the non-cluster head node to the cluster head node b CH = 100 bits, the size of the data packet sent by the cluster head node to the base station b BS = 2000 bits, the minimum sound intensity at which the data packet can be correctly received by the destination node I0 = 1 w, the energy consumption e' of the cluster head node for receiving and fusing 1-bit data e' RX = 0.005 J, the guard time t guard = 0.2 s; the underwater sound speed v = 1500 m / s; the maximum number of iterations of the network iterative clustering e max = 2000; each node communicates in half-duplex mode and can only listen or send data on the channel at the same time. The present application specifically comprises the following steps:

[0185] S1, based on the number of nodes N, the energy consumption of sending and receiving data packets, and the distribution of nodes in the underwater area, the optimal clustering number K opt is calculated based on the underwater energy propagation model;

[0186] S2, initialize the iteration variable e = 1, set the maximum number of iterations e max , and set the initial energy of the nodes E0;

[0187] S3, randomly select K opt nodes as initial cluster head nodes CH i , i = 1, 2, …, K opt , i is the clustering sequence number;

[0188] S4, the Euclidean distance of the non-cluster head node to the cluster head node CH j of the clustering j is calculated respectively wherein, is the non-cluster head node with the intra-cluster sequence number m in the clustering i, m = 1, 2, …, (N-K opt ); when e = 1, let i = 0;

[0189] In step S3, i = 1, 2, …, K opt is defined, but i = 0 has not been defined before.

[0190] S5, search for the j that makes the Euclidean distance minimum, and record it as jmin

[0191]

[0192] Non-cluster head nodes Add cluster j min ;

[0193] S6. For all non-cluster head nodes Perform steps S4-S5 to ensure that each non-cluster head node They were all added to a certain cluster;

[0194] S7. Based on the clustering results, calculate the distance from each node in each cluster to all other nodes in the cluster (all other nodes in the cluster include non-cluster head nodes). or cluster head node CH i The sum of Euclidean distances Calculate the cluster head node CH i to all other nodes within the cluster When m = 0, it is stipulated that m = 0;

[0195] Previously, in step S4, for non-cluster head nodes... The node sending priority m = 1 to (NK) is defined. opt The value of m = 0 was not defined before; therefore, it is stipulated here that m = 0 such that It can refer to non-cluster head nodes (m = 1, 2, ...) or cluster head nodes (m = 0).

[0196] S8, Search Clustering i-internal command The smallest m is denoted as m min

[0197]

[0198] m min The corresponding node serves as the cluster head node for cluster i;

[0199] At this time, m min It may point to a non-cluster head node or a cluster head node, so it is collectively referred to as a "node".

[0200] S9. Let e ​​= e + 1; if the clustering result or the cluster head node changes, and e <e max If the result is positive, proceed to step S4; otherwise, continue to step S10.

[0201] S10. Reset the non-cluster head nodes within each cluster. The sequence number m, m = 1 to N i And it is equivalent to making non-cluster head nodes To the cluster head node CH iThe sequence number when the distances are arranged from nearest to farthest, N i This represents the number of non-cluster head nodes within cluster i;

[0202] S11, Non-cluster head node and cluster head node CH j Calculate the collision-free parallel transmission time of data packets for each node within and between clusters. and t i S12, Timer t starts counting, for each non-cluster head node. Based on the transmission time calculated in step S11 exist The current cluster head node CH i Send a b CH For bit-based data packets, calculate the non-cluster head nodes based on the underwater energy propagation model. Power of data packets

[0203]

[0204] in Indicates non-cluster head node To cluster head node CH i The Euclidean distance, where H is the non-cluster head node. The underwater depth is I0, which is the minimum sound intensity at which the data packet can be correctly received by the destination node. Sound intensity is physically defined as power per unit area.

[0205] Calculate the energy consumption of sending data packets

[0206]

[0207] Among them, t TX (b CH ) for b CH The time taken to send a bit data packet;

[0208] Calculate the cluster head node CH i Receive and merge all non-cluster head nodes in cluster i Data packet power consumption The calculation formula is as follows:

[0209]

[0210] e′ RX For cluster head node CH i Energy consumption for receiving and fusing 1 bit of data;

[0211] Update the non-cluster head nodes according to equations (6) and (7). energy and cluster head node CH i energy

[0212]

[0213] S13, set the timer t to zero, and restart the timer, each cluster head node CH i sending time t calculated according to step S11 i , t = t i to the base station; according to the underwater energy propagation model, the same method of formula (3) (4) is used to calculate CH i energy consumption of sending data packets

[0214]

[0215] wherein, is the transmission power of the cluster head node CH i , is the Euclidean distance from the cluster head node CH i to the base station, H is the underwater depth of the cluster head node CH i , t TX (b BS ) is the sending time of b BS bit data packets;

[0216] update the energy of the cluster head node CH i according to formula (9)

[0217]

[0218] S14, if all communication tasks of the underwater sensor network are completed, end; otherwise, jump to step S1.

[0219] In the running process of the underwater sensor network, the energy consumption of the node transmitting data packets is one of the main energy consumptions of the network. The energy consumption of the node transmitting data packets conforms to the underwater energy propagation model, and is mainly positively correlated with the distance between the sending and receiving nodes. Therefore, in the underwater sensor network, the total distance of all nodes transmitting data packets is often reduced by clustering the nodes, so as to reduce the running energy consumption of the network. Through the algorithm based on the underwater energy propagation model, the relationship between the network running energy consumption and the clustering number can be constructed, and then the optimal clustering number K opt that minimizes the network energy consumption can be calculated.

[0220] Collision-free parallel transmission in a network should satisfy the requirement that data packets are received without collision at the destination node. For example, taking data packets a1 and a2 from two sending nodes with the same destination node, if both are received without collision at the destination node, then the arrival times t′1 and t′2 of a1 and a2 at the destination node should satisfy the following condition:

[0221]

[0222] Among them, t guard Indicates the protection time slot, t RX This refers to the time taken to receive data packets. Based on the theory in the above formula, the non-cluster head node in step S11... and cluster head node CH j Calculate the collision-free parallel transmission time of data packets for each node within and between clusters. and t i The process is as follows:

[0223] S11.1 Let the cluster index i = 1, and set the set TS′ i Record each non-cluster head node in cluster i To the cluster head node CH i Collision-free arrival time of transmitted data packets In set TS′ i Arrange the data in ascending order, using ts′ k Indicates TS′ i Let TS′ be the element with index k in the middle. i ={0,+∞};

[0224] S11.2. Let the index of the non-cluster head node m = 1, and calculate the index of the non-cluster head node. Collision-free arrival time of transmitted data packets

[0225] Calculate each non-cluster head node within cluster i Transmission delay for sending data packets to the cluster head node v represents the speed of sound underwater. Indicates non-cluster head node To the cluster head node CH i The Euclidean distance;

[0226] The index k that satisfies the following formula is searched. min k min +1 adjacent elements

[0227]

[0228] In the formula, t guard Indicates the protection time slot, t RX(b CH ) for b CH Time taken to receive bit data packets;

[0229] The following is a further explanation of equation (16):

[0230] In step S11.1, the set TS′ of data packet arrival times is declared. i and initialize TS′ i ={0,+∞}. Therefore, the set TS′ is... i There are at least two elements in it;

[0231] In step S11.1, it was declared that Represents set TS′ i The element in, k is The indices are arranged in ascending order. Therefore, k and k+1 form the set TS′. i The index of adjacent elements in the array, given k (element). That is, we know k+1 (element) There is no non-causal statement;

[0232] In summary, k min With k min +1 represents the set TS′ i The sequence numbers of adjacent elements in equation (16) do not contain any non-causal expressions.

[0233] make Time of arrival Add to set TS′ i ; TS′ i All arrival times are sorted in ascending order; for the cluster with index i>1, its set TS′ i It records both the arrival time of data packets transmitted within the cluster and the arrival time of data packets from other sub-clusters, hence the collective term "arrival time";

[0234] Let m = m + 1, and execute step S11.2 until each non-cluster head node in cluster i is calculated. Collision-free arrival time of transmitted data packets

[0235] S11.3 Calculate each non-cluster head node within cluster i. Collision-free transmission time of data packets

[0236]

[0237] S11.4 Calculate each non-cluster head node within cluster i. To other cluster head nodes CH j Arrival time of transmitted data packets

[0238]

[0239] Data packet arrival time Add the corresponding cluster head node CH j The set TS′ j ;

[0240] S11.5. Let i = i + 1, and repeat steps S11.2-11.4 until each non-cluster head node in all clusters is calculated. Collision-free transmission time of data packets

[0241] S11.6 Calculate the cluster head node CH i The data packets arrive at the base station without collision at transmission time t. i :

[0242] Let i = 1; set set TS′ BS Record each cluster head node CH i Collision-free arrival time of data packets sent to the base station In the set TS′ BS Arranged in ascending order, using Represents set TS′ BS Let TS′ be the element with index k in the middle. BS ={0,+∞};

[0243] S11.7 Calculate CH for each cluster head node i Transmission delay to base station The index k that satisfies the following formula is searched. min k min +1 adjacent elements

[0244]

[0245] make Data packet arrival time Add to set TS′ BS ; TS′ BS middle Sort by size from smallest to largest;

[0246] Let i = i + 1, and execute S11.7 until the CH of each cluster head node is calculated. i Collision-free arrival time of data packets sent to the base station

[0247] S11.8 Calculate the cluster head node CH i Collision-free transmission time t of data packetsi

[0248]

[0249] According to the above method, the method (hereinafter referred to as "the method herein") and the original LEACH protocol (hereinafter referred to as "method 1") and GEC algorithm (hereinafter referred to as "method 2") are simulated for 10 rounds on the NS3 network simulator platform, and the experimental results of the average residual energy of nodes, the average number of surviving nodes, the round of the first dead node, and the average amount of data received by the base station of the three methods are compared, as shown in Tables 5-8. By comparing the method herein with method 1 and method 2, it can be seen that as the communication round increases, the average residual energy of nodes and the average number of surviving nodes of the method herein decrease the slowest, the average amount of data received by the base station increases the fastest, and the round of the first dead node is the latest. The experimental results of the above performance indicators are better than those of method 1 and method 2. Therefore, the clustering method of the present application can reduce the energy consumption of the network, the node parallel communication algorithm of the present application can reduce the energy consumption of the node, prolong the life of the cluster head node, and improve the communication speed of the network. Therefore, the method proposed in the present application is feasible.

[0250] Table 5. Average residual energy of nodes in 10 simulations

[0251]

[0252]

[0253] Table 6. Average number of surviving nodes in 10 simulations

[0254] Run 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Method 1 * ]] 50 50 50 44 2 0 0 0 0 0 0 Method 2 50 50 50 50 50 22 0 0 0 0 0 Algorithm herein 50 50 50 50 50 48 17 0 0 0 0

[0255] Table 7. Round of the first dead node in 10 simulations

[0256] Run 1 2 3 4 5 6 7 8 9 10 Method 1 * ]] 590 660 601 648 618 556 533 666 612 572 Method 2 879 835 904 905 877 860 886 851 866 838 Algorithm herein 997 969 969 1008 983 991 980 1020 1005 922

[0257] Table 8. Average amount of data received by the base station in 10 simulations (unit: 10 4 bit)

[0258] Run 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Method 1 0.00 0.07 0.14 0.21 0.27 0.28 0.28 0.28 0.28 0.28 0.28 Method 2 0.00 0.46 0.92 1.38 1.84 2.31 2.99 3.05 3.05 3.05 3.05 Algorithm herein 0.00 0.56 1.10 1.63 2.17 2.71 3.26 3.28 3.28 3.28 3.28

[0259] The above embodiments are the preferred embodiments of the present application, but the embodiments of the present application are not limited to the above embodiments, and any changes, modifications, substitutions, combinations, simplifications, etc. made without departing from the spirit and principles of the present application should be equivalent replacement manners and should be included in the protection scope of the present application.

Claims

1. A clustered parallel communication method for underwater sensor networks based on repeated computation and improved K-means clustering, applied to a static underwater sensor network with multiple nodes capable of independently transmitting and receiving data, characterized in that... The underwater sensor network clustered parallel communication method includes the following steps: S1. Based on the number of nodes N, the energy consumption for sending and receiving data packets, and the distribution of nodes in the underwater area, calculate the optimal cluster size K using the underwater energy propagation model. opt ; S2. Initialize the iteration variable e = 1, and set the maximum number of iterations to e. max Set the initial energy of the node to E0; S3, Randomly select K opt One node serves as the initial cluster head node CH i i = 1, 2, ..., K opt i is the cluster number; S4. Calculate the non-cluster head nodes separately. To the cluster head node CH of cluster j j Euclidean distance in, Let m be the non-cluster head node with index m in cluster i, where m = 1, 2, ..., (NK). opt When i = 1, let i = 0; S5, Searching for Euclidean Distance The smallest j is denoted as j min Non-cluster head nodes Add cluster j min ; S6. For all non-cluster head nodes Perform steps S4-S5 to ensure that each non-cluster head node They were all added to a certain cluster; S7. Based on the clustering results, calculate the sum of the Euclidean distances from each node in each cluster to all other nodes in the cluster. Calculate the cluster head node CH i to all other nodes within the cluster When m = 0, all other nodes in the cluster, including non-cluster head nodes, are included. or cluster head node CH i ; S8, Search Clustering i-internal command The smallest m is denoted as m min m min The corresponding node serves as the cluster head node for cluster i; S9. Let e ​​= e + 1. If the clustering result or the cluster head node changes, and e <e max If the result is positive, proceed to step S4; otherwise, continue to step S10. S10. Reset the non-cluster head nodes within each cluster. The sequence number m, m = 1 to N i And it is equivalent to making non-cluster head nodes To the cluster head node CH i The sequence number when the distances are arranged from nearest to farthest, N i This represents the number of non-cluster head nodes within cluster i; S11, Non-cluster head node and cluster head node CH j Calculate the collision-free parallel transmission time of data packets for each node within and between clusters. and t i ; S12, Timer t starts counting down for each non-cluster head node. Based on the transmission time calculated in step S11 exist The current cluster head node CH i Send a b CH For bit-based data packets, calculate the non-cluster head nodes based on the underwater energy propagation model. Power of data packets in Indicates non-cluster head node To cluster head node CH i Euclidean distance, Indicates non-cluster head node The underwater depth, I0 represents the minimum sound intensity at which the data packet can be correctly received by the destination node, and the sound intensity is physically defined as the power per unit area; Calculate the energy consumption of sending data packets Among them, t TX (b CH ) for b CH The time taken to send a bit data packet; Calculate the cluster head node CH i Receive and merge all non-cluster head nodes in cluster i Data packet power consumption The calculation formula is as follows: e′ RX For cluster head node CH i Energy consumption for receiving and fusing 1 bit of data; Update the non-cluster head nodes according to equations (6) and (7). energy and cluster head node CH i energy S13. Reset timer t to zero and restart timing for each cluster head node CH. i Based on the transmission time t calculated in step S11 i At t=t i The data packet received in step S12 is sent to the base station; based on the underwater energy propagation model, CH is calculated using the same method as equations (3) and (4). i Energy consumption for sending data packets in, For cluster head node CH i The transmission power, For cluster head node CH i Euclidean distance to the base station For cluster head node CH i underwater depth, t TX (b BS ) for b BS The time taken to send a bit data packet; Update the cluster head node CH according to equation (9). i energy S14. If all communication tasks of the static underwater sensor network are completed, then the process ends; otherwise, proceed to step S1.

2. The underwater sensor network clustered parallel communication method according to claim 1, characterized in that, In step S1, the optimal cluster size K is calculated using the following method. opt : S1.

1. Based on the underwater energy propagation model, the total energy consumption E′ for an underwater sensor network to complete one communication task is calculated as follows: E′=E′ TX +E′ RX (10) Among them, E′ TX For all non-cluster head nodes and cluster head node CH i Energy consumption for sending data packets: E′ RX For all cluster head nodes CH i Energy consumption for receiving and fusing data packets S1.2: Calculate K opt :

3. The underwater sensor network clustered parallel communication method according to claim 2, characterized in that, Step S1 calculates the optimal number of clusters K. opt During the process, when N underwater nodes in a static wireless sensor network are uniformly distributed within a two-dimensional square water area with side length l and depth H, and the base station is located at the center of the entire network, the formula for calculating E′ is as follows: Where the variable K represents the number of clusters; Calculate the optimal number of clusters K for the network opt 4. The underwater sensor network clustered parallel communication method according to claim 2, characterized in that, In step S11, the non-cluster head node and cluster head node CH j Calculate the collision-free parallel transmission time of data packets for each node within and between clusters. and t i The process is as follows: S11.1 Let the cluster index i = 1, and set the set TS′ i Record each non-cluster head node in cluster i To the cluster head node CH i Collision-free arrival time of transmitted data packets In the set TS′ i Arrange the data in ascending order, using ts′ k Indicates TS′ i Let TS′ be the element with index k in the middle. i ={0,+∞}; S11.

2. Let the index of the non-cluster head node m = 1, and calculate the index of the non-cluster head node. Collision-free arrival time of transmitted data packets Calculate each non-cluster head node within cluster i Transmission delay for sending data packets to the cluster head node v represents the speed of sound underwater. Indicates non-cluster head node To the cluster head node CH i The Euclidean distance; The index k that satisfies the following formula is searched. min k min +1 adjacent elements In the formula, t guard Indicates the protection time slot, t RX (b CH ) represents b CH The time taken to receive a 1-bit data packet; make Time of arrival Add to set TS′ i ; TS′ i Sort all arrival times in ascending order; Let m = m + 1, and execute step S11.2 until each non-cluster head node in cluster i is calculated. Collision-free arrival time of transmitted data packets S11.3 Calculate each non-cluster head node within cluster i. Collision-free transmission time of data packets S11.4 Calculate each non-cluster head node within cluster i. To other cluster head nodes CH j Arrival time of transmitted data packets Data packet arrival time Add the corresponding cluster head node CH j The set TS′ j ; S11.

5. Let i = i + 1, and repeat steps S11.2-11.4 until each non-cluster head node in all clusters is calculated. Collision-free transmission time of data packets S11.6 Calculate the cluster head node CH i The data packets arrive at the base station without collision at transmission time t. i : Let i = 1; set set TS′ BS Record each cluster head node CH i Collision-free arrival time of data packets sent to the base station In the set TS′ BS Arranged in ascending order, using Represents set TS′ BS Let TS′ be the element with index k in the middle. BS ={0,+∞}; S11.7 Calculate CH for each cluster head node i Transmission delay to base station The index k that satisfies the following formula is searched. min k min +1 adjacent elements make Data packet arrival time Add to set TS′ BS ; TS′ BS middle Sort by size from smallest to largest; Let i = i + 1, and execute S11.7 until the CH of each cluster head node is calculated. i Collision-free arrival time of data packets sent to the base station S11.8 Calculate the cluster head node CH i Collision-free transmission time t of data packets i

Images