CN115297497B - High-efficiency energy-saving clustering method based on biological heuristic algorithm - Google Patents
High-efficiency energy-saving clustering method based on biological heuristic algorithm Download PDFInfo
- Publication number
- CN115297497B CN115297497B CN202211223207.6A CN202211223207A CN115297497B CN 115297497 B CN115297497 B CN 115297497B CN 202211223207 A CN202211223207 A CN 202211223207A CN 115297497 B CN115297497 B CN 115297497B
- Authority
- CN
- China
- Prior art keywords
- suburb
- wolf
- cluster
- group
- clustering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
- H04W40/10—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources based on available power or energy
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/24—Connectivity information management, e.g. connectivity discovery or connectivity update
- H04W40/32—Connectivity information management, e.g. connectivity discovery or connectivity update for defining a routing cluster membership
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a high-efficiency energy-saving clustering method based on a biological heuristic algorithm, which comprises the following steps: 1. setting the number of clustersN a opt Inputting the node positions of ammunition to be clustered, 2, clustering according to the number of the ammunition to be clusteredN a opt Clustering ammunition nodes to be clustered, and distributing the nodes with the closest Euclidean distance to the same cluster to obtain an initial cluster; 3. clustering adjustment is carried out on the initial clusters and the mass center position is determined according to a clustering algorithm based on HCOWG, and a plurality of final clusters are obtained; 4. and determining a cluster head for each final cluster according to a cluster head selection algorithm based on HCOWG, and finishing clustering. The method has the advantages of strong network reliability, shortest clustering time, less total selection of the number of cluster head nodes and better performance than the current popular algorithm in the aspect of total energy consumption.
Description
Technical Field
The invention relates to the technical field of clustering methods, in particular to a high-efficiency energy-saving clustering method based on a biological heuristic algorithm.
Background
Under the background of internet of things in a battlefield, the development of ammunition is developing towards informatization, intellectualization and networking, wherein the most representative is patrol missile ad hoc network, and patrol missiles are taken as new concept ammunition capable of performing 'patrol flight' and 'standby' on a target area to execute various combat tasks. The intelligent ammunition is a product combining an advanced unmanned aerial vehicle technology and a guided missile technology, can quickly reach a target area, performs tasks such as reconnaissance and monitoring, target positioning, air blocking, accurate striking, damage effect evaluation and the like, and is an intelligent ammunition with distinct characteristics and capable of meeting the requirements of future informatization combat. The patrol bombs can form ammunition clusters through networking, the ammunition clusters are called networked ammunitions, the networked ammunitions can emerge multiplication effect in the actual combat process, and meanwhile, the system has strong damage resistance and control autonomy, so that the system has extremely superior combat capability.
The clustering is a network structure division method driven by tasks, communication, computation and resources in multiple ways, and aims to balance the computation pressure of each node, reasonably distribute network resources, and enable the nodes with better resources to act as more computation tasks, so as to improve the stability of the network. In an ammunition ad hoc network, clustering is crucial to efficiently managing the network topology.
Disclosure of Invention
The invention provides a high-efficiency energy-saving clustering method based on a biological heuristic algorithm, which aims to effectively solve the clustering problem of the patrol missile self-networking in a satellite refusing environment.
In order to solve the technical problems, the invention adopts the following technical scheme:
a high-efficiency energy-saving clustering method based on a biological heuristic algorithm comprises the following steps;
step 1, setting the clustering numberN a opt Inputting the positions of ammunition nodes to be clustered;
step 2, according to the number of clusters to be clusteredN a opt Clustering ammunition nodes to be clustered, and allocating nodes with the closest Euclidean distances to the phasesObtaining an initial cluster from the same cluster;
step 3, adjusting the initial clusters and determining the centroid position according to a clustering algorithm based on HCOWG to obtain a plurality of final clusters;
and 4, determining a cluster head for each final cluster according to a cluster head selection algorithm based on HCOWG, and finishing clustering.
Further, the clustering algorithm based on HCOWG in step 3 comprises the following steps:
step 3.1, taking each node position in the initial cluster as a suburb wolf, taking the suburb wolf in the same initial cluster as an initial wolf group, and obtaining a plurality of initial wolf groups;
3.2, calculating a new position of each suburb under the guidance of 3 optimal wolfs in the wolf group by using a wolf optimization algorithm to obtain the first updated suburb;
step 3.3, according toe=2-2t/MaxDT,CR=0.5×(sin(2π×0.25×t+π)×(t/MaxDT) + 1), whereintUpdating the parameters for the current first iteration number and the preset first maximum iteration number for MaxDTeAndCR;
step 3.4, randomly grouping the suburbs in each wolf group, and calculating the global optimal suburb of the wolf group by using a suburb optimization algorithmGPThe best wolf in each groupalphaAnd medium value suburb wolfcult;
And 3.5, updating all the suburbs for the second time, wherein the method for updating each suburb for the second time comprises the steps of calculating the value of each dimension of the current suburb, obtaining the updated suburb after all the dimensions are calculated, and calculating the value of each dimension by utilizing a random probability function to obtain the random probabilityrandIf, ifrand<CRIf so, acquiring the value of the current dimension by adopting an SGWO searching mode; otherwise, adopting a Gaussian global optimization growth operator to obtain the value of the current dimension;
step 3.6, calculating the social adaptation capacity of each suburb before and after the second updating of each suburb, selecting and reserving the better suburb by greedy, screening all suburbs, and finally obtaining all screened suburbs;
step 3.7, updating the global optimal suburb wolf according to all the screened suburb wolfsGP;
Step 3.8, the global optimum suburb wolfGPObtaining the centroid positions of all wolf groups as the centroids of the corresponding wolf groups, calculating the Euclidean distance between the screened suburb wolfs and each centroid in the step 3.6, and dividing the screened suburb wolfs into the wolf groups corresponding to the centroids closest to the selected suburb wolfs to obtain the iterative updated wolf groups;
step 3.9, adding 1 to the first iteration times, judging the first iteration times, if the first iteration times is smaller than a preset first maximum iteration times, turning to step 3.1, updating the initial wolf group into the wolf group updated in the current iteration in step 3.8, and otherwise, turning to the next step;
and 4.0, taking each wolf pack updated by the last iteration as a final cluster to obtain a plurality of final clusters.
Further, the number of clustersN a opt The calculation formula of (2) is as follows:
wherein the content of the first and second substances,Nas to the number of ammunition nodes to be clustered,is a set of edges that are to be considered,is the carrier frequency of the carrier wave,
in order to be the energy factor of the light,a quality parameter indicative of a quality of the communication channel,Mthe length of the represented air cube network region,indicating the average distance of the cluster head nodes from the base station,is a score of a member node.
Further, in step 3.5, the current dimension of the current wolf is obtained by adopting a Gaussian global optimization growth operatorThe formula of (1) is:in the formula (I), wherein,=GP-soc cr1 ,GPfor the current value of the global optimum suburb,soc cr1 one suburb randomly selected from the group of the current suburbs is showncr1The value of (a) is set to (b),rn 1 andrn 2 is a random number generated by a gaussian/normal distribution with a mean of 0 and a variance of 1.
Further, in step 3.5, the formula of the SGWO search mode is:
wherein, the first and the second end of the pipe are connected with each other,new cis the value of the current dimension of the current suburb wolf,temp c the state of the current suburb is shown,A 1 、A 2 andA 3 is a coefficient vectorAIn globally optimal suburbGPOptimum suburb wolf in groupalphaAnd medium value suburb wolfcultThe corresponding values under different conditions are obtained,NX 1 、NX 2 andNX 3 indicating that the current suburbs are respectively in the global optimal suburbsGPAnd the best suburb in groupalphaAnd medium value suburb wolfcultThe growth condition is obtained under the guidance of (1).
Further, the cluster head selection algorithm based on HCOWG comprises the following steps:
step 4.1, taking each cluster in the final clustering result as an initial wolf group, and taking each node in the cluster as an initial suburb in the wolf group;
step 4.2, calculating the fitness value of each suburb;
step 4.3, randomly grouping the suburbs in each wolf group, and calculating the global optimal suburb of each wolf group by using a suburb optimization algorithmGPThe best wolf in each groupalphaHarmony value suburbcult;
4.4, updating all the suburbs according to the SGWO searching mode to obtain the suburbs to be processed;
step 4.5, calculating the globally optimal suburbs of the suburbs to be processedGPOptimum suburb wolf in groupalphaAnd medium value suburb wolfcultNew state under guidance of fitness value、Andand find out、Andthe average value of the suburb is used for obtaining a new position of the current suburb, and all suburbs are updated again to obtain the iteration suburb;
step 4.6, adding 1 to the second iteration number, judging the second iteration number, if the second iteration number is smaller than a preset second maximum iteration number, turning to step 4.4, updating the suburb wolf to be processed into the current iteration suburb wolf in step 4.5, and meanwhile, updating the global optimal suburb wolfGPThe best wolf in each groupalphaHarmony value suburbcultUpdating, otherwise, turning to the next step;
and 4.7, taking the node corresponding to the global optimal suburb wolf of each wolf cluster obtained by the last iteration as a cluster head of the corresponding cluster of the wolf cluster.
Further, in step 4.5:
wherein, the first and the second end of the pipe are connected with each other,、andrespectively as global optimum suburb wolfGPAnd the best suburb in groupalphaAnd medium value suburb wolfcultThe value of the fitness value of (a) is,the current fitness value of the current suburb wolf,A 1 、A 2 andA 3 is a coefficient vectorAIn globally optimal suburbGPAnd the best suburb in groupalphaAnd medium value suburb wolfcultAnd (4) corresponding values under different conditions.
Further, in step 3.2, if any one of the obtained positions corresponding to the suburbs after the first update exceeds the search boundary, the position of the suburb after the first update is adjusted, and the adjusting method is as follows:
wherein the content of the first and second substances,in order to adjust the position of the suburb wolf,is the suburb wolf after the first update,is a wolf before the first update,andrespectively the upper and lower boundaries of the search area,is [0,1]]The random number of (1).
After the technical scheme is adopted, compared with the prior art, the invention has the following advantages:
(1) The invention provides a COA and GWO hybrid algorithm (HCOGW). Firstly, COA (ICOA) is improved, a Gaussian global optimization growth operator is provided in the growth process, the searching capability is improved, meanwhile, a scheme for dynamically adjusting the number of the wolfs in the group is provided, a small number of groups are adopted in the early stage of searching, the positive feedback effect of the global optimal solution is weakened, the searching capability is strengthened, a large number of groups are adopted in the later stage, the positive feedback effect of the global optimal solution is strengthened, the mining capability is strengthened, and the operability is improved. Then, GWO is improved, and a Simplified GWO (SGWO) is provided, so that the operability is improved while the advantage of strong local search capability is exerted. Finally, the ICOA and the SGWO are organically fused by adopting a sine crossing strategy, and the exploration and exploitation capacity of the growth of the suburb in the group is well balanced, so that the optimal optimization performance is obtained.
(2) An energy-saving clustering method SIC (sweep-integration-based clustering) is provided based on the HCOGW algorithm, and the service life of a network is effectively prolonged. Under a high dynamic topological structure, the scale and the number of clusters have great influence on the communication quality, so an analysis model is constructed in the text, the optimal cluster number is determined, and compared with popular clustering algorithms such as CBLADSR, ACO, EALC, GA, SOCS and the like, the SIC can have stronger network reliability and the clustering time is shortest according to simulation results.
(3) In the SIC algorithm, a new method for selecting the cluster head CH based on the adaptability value of the missile is provided. The core is that a node with high residual energy, more neighbor nodes and short intra-cluster distance is selected as a CH node and is responsible for data gathering and transmission, and the results show that the SIC is less on the selection of the number of cluster head nodes in general and the performance is superior to that of the current popular algorithm on the aspect of total energy consumption through simulation.
The present invention will be described in detail below with reference to the accompanying drawings and examples.
Drawings
FIG. 1 is a diagram illustrating the relationship between the number of clusters and the total number of nodes;
FIG. 2 is a diagram illustrating the relationship between clustering time and total node count;
FIG. 3 is a graph illustrating the number of surviving nodes as a function of cycle number;
FIG. 4 is a graph showing the variation of total energy consumption with cycle number.
Detailed Description
The principles and features of this invention are described below in conjunction with the following drawings, which are set forth to illustrate, but are not to be construed to limit the scope of the invention.
1. Grey wolf optimization algorithm
GWOO simulates the strict social level system and population predation behavior of the Huidong wolf population in nature. In the social classification system, the grey wolfs are sequentially from high to low、、Anda wolf. The predation behavior is tracking and approaching the prey, chasing and surrounding the prey, attacking and killing the prey. GWOO will be in the wolf group、、Andthe positions of the wolfs represent the first optimal solution, the second optimal solution, the third optimal solution and the candidate solution, respectively. The mathematical model of the gray wolf surround behavior is as follows:
wherein Dis is the distance between the gray wolf and the prey,tis the current number of iterations and,X p is the position vector of the prey and,Xis the location vector of the gray wolf. A and C are coefficient vectors, as in equations (3) and (4). The coefficient e decreases linearly from 2 to 0 during the iteration, as in equation (5), maxDT being the maximum number of iterations,r 1 andr 2 is [0,1]]The uniformly distributed random vector of (a) is,c 1 in order to be able to adjust the parameters,
representing the multiplication of the respective corresponding components of the two vectors. The mathematical model of hunting behavior is as follows:
wherein 、 、 Respectively representing the distance between the current wolf and the 3 optimal wolfs. Formula (12) representsThe wolf is on、Andthe new location of the move under the direction of the wolf, i.e. the new solution generated by GWO.
Compared with classical intelligent group algorithms such as PSO, GWOO has the following main characteristics: 1. GWO is guided with 3-headed optimal wolf (optimal solution)The wolf hunting has stronger local search capability, but is easy to fall into local optimum when solving the complex optimization problem; 2. GWO has only two parametersaAndc 1 the former adopts dynamic regulation mode, the latter adopts dynamic regulation modec 1 The constant 2 is often taken, and the adjusted parameters are few, so the operability is strong; 3. the principle is simple and easy to realize, but the updating is based on the calculation of dimension, and the calculation formulas (6) - (12) are needed, so the calculation complexity is higher; 4. the target function adopts a parallel computing mode, so the running speed is high.
2. Suburb wolf optimization algorithm
The COA includes 4 main steps of initialization parameters and wolf clusters, suburb growth and death, and being driven and received by the group.
First, set parameters such as the scale of the suburb groupNThe number of the wolfsN p Number of suburbsN c And MaxDT or the like, whereinN=N c ×N p . Then randomly initializing the suburb group, and randomizing operation formula (13) of the pth dimension of the pth group of the c suburbs. Finally calculating the suburb wolf of each headsocThe social fitness value fit is as follows:
whereinlb h Andub h respectively represents the lower bound and the upper bound of the k-dimension social status factor of the suburb wolf,h=1,2,……,D;Din order to search for the dimensions of the space,ris [0,1]]The random numbers are evenly distributed in the inner part, fa fitness function is represented.
Second, the optimum wolf in the groupalphaGroup culture trendcultAnd randomly selected two-headed suburb wolfcr1Andcr2affecting suburb wolfThe growth process is that the suburb wolf grows and receivesAndthe influence of (c). WhereincultThe calculation of (2) is shown in the formula (15), that is, each factor is the median of the social factors corresponding to all the wolfs in the group, socultAlso called as a medium value suburb wolf,andsee equations (16) and (17) for calculation. The suburb wolf grows in the following way:
wherein the content of the first and second substances,is a new solution obtained by the growth of the c-th suburb wolf in the group,r 3 andr 4 is [0,1]]Random numbers are uniformly distributed in the inner part. The social adaptation ability of the suburb wolf after the growth of the group is evaluated as shown in the formula (19),is a new fitness value.
And finally, adopting an iterative greedy algorithm to carry out the advantages and disadvantages as the formula (20), so that the newly generated better suburb participates in the growth of the rest suburbs in the group to accelerate the convergence speed.
Thirdly, the birth of the young suburb (pup) is influenced by the genetic genes and environmental factors of randomly selected parent suburbs, and is shown in an expression (21).
Wherein the content of the first and second substances,r h are uniformly distributed in [0,1]]A random number within;for randomly selected father suburb wolfIs/are as followshThe dimensions of the material are measured in the same way,for randomly selected suburb wolfIs/are as followshThe dimensions of the material are measured in the same way,h 1 andh 2 two randomly selected dimension labels are adopted to ensure that the gene of the suburb wolf of the parents can be inherited to the young wolf;R h is at the firsthRandomly generating variation values in a dimension decision variable range;P s andP a respectively, the dispersion probability and the association probability, which determine the genetic sum of the young wolfThe case of the mutation is represented by the formulas (22) and (23):
P s =1/D (22)
P a =(1- P s )/2 (23)
after birth, young wolves are evaluated for their social adaptation ability and are determined to be alive or dead according to their abilities, which are described in detail as follows: in terms of social adaptability, (1) when only one suburb is worse than the young wolf in the group, the suburb dies, the young wolf survives, and the age of the suburb is set to be 0, namely age =0; (2) When a plurality of suburbs are worse than young wolves in the group, the suburb with poor capability and the highest age dies, and the young wolves survive, and age =0 is set; (3) All the suburbs in the group are stronger than the young wolfs, and the young wolfs die.
The fourth step, in COA, the root of the West wolfP e Are driven away and admitted by the group. The suburbs are randomly allocated to the group at first, but sometimes some suburbs leave the group where the suburbs are located and join another group, the random expelling and admission is used for ensuring the diversity of the suburbs in the group and the information sharing among the groups,P e is calculated as shown in equation (24).
P e =0.005×N c 2 (24)
The pseudo code of COA is shown in algorithm 1.
Algorithm 1: determining global optimum suburb wolf
1: initialize: settingN p AndN c and parameters are equal, the suburb population is initialized randomly and grouped randomly,
2: for t=1 TO MaxDT do
4: determining optimal within a groupalphaWolf
5 calculating group culture trends according to equation (15)cult
6: for p=1 TON c do
7 according toalphaAndcultcarrying out the growth of the wolf in the suburb as shown in the formulas (16) - (18)
8, reserving more optimal suburbs and globally optimal suburbs by using greedy algorithm
9: end for
10, the young wolf is born and died, the birth is as formula (21), and the birth and death are determined according to the social ability
11: end for
12. The suburb adopts the probability of the formula (24) to be driven away and accommodated
13, updating the age of the suburb wolf,age=age+1
14: end for
15 outputting the globally optimal suburb wolf
From the above description, compared with the classical intelligent clustering algorithm such as PSO, the main advantages of COA are: 1. the system has a good search frame, namely a plurality of groups of structures, the suburbs are randomly driven and accommodated, so that suburb groups have diversity, stronger exploration capability is realized, and the complex optimization problem can be better solved; 2. the optimal suburb wolf in the group and the group culture trend guide the growth of each suburb wolf in the group, and the local search capability is enhanced; 3. the lives and deaths in the suburb group are influenced by genetic factors and environmental variation factors randomly selected from father suburbs, and the suburb lives and deaths operators enable the COA to have certain global search capacity.
The main defects are as follows: 1. the structure is complex, and the calculation complexity is high; 2. an iterative greedy algorithm is adopted, so that the stability is poor and the efficiency is low; 3. the number of parameters to be adjusted is large, and the operability is poor; 4. the search capability is enhanced by a plurality of groups of structures and the like, but the information sharing among the groups is insufficient, and the convergence speed in the later period of the search is slow.
3. Grey wolf and suburb hybrid optimization algorithm based on sine crossing strategy
The method comprises three contents, namely improving a suburb optimization algorithm to form an ICOA algorithm, simplifying a gray wolf optimization algorithm to form an SGWO algorithm, and finally mixing the ICOA algorithm and the SGWO algorithm by adopting a sine cross mixing strategy to form an HCOGW optimization algorithm.
First, the ICOA mainly includes a gaussian global optimization growth operator and a dynamic tuning group suburb number scheme. In order to improve the social adaptability, the information sharing degree between groups and the algorithm convergence speed after the growth of the wolf, inspired by the optimization approach strategy and Gaussian distribution in the global optimal harmony search algorithm proposed by Omran and the like, the group-internal wolf growth mode is improved, and a Gaussian global optimization approach growth operator is proposed. The global optimization operator makes full use of the state information of the current global optimal suburb of the suburb population, so that the suburb growth approaches to the current global optimal suburb, the mining capacity is improved, and the inter-group sharing of the group information is achieved through the global optimal solution. The Gaussian distribution is also called normal distribution, and compared with the uniform random distribution [0,1] of COA, the method can increase the search range and enhance the global search capability to a certain extent. In the suburb growth process, an optimization operator and a Gaussian distribution random factor are introduced into the formula (18), specifically see the formulas (44) and (45).
In the formula (44), the reaction mixture is,GPfor the current value of the global optimum suburb,soc cr1 one suburb randomly selected from the group of the current suburbs is showncr1The value of (a) is set to (b),rn 1 andrn 2 is a random number generated by a gaussian/normal distribution with a mean value of 0 and a variance of 1,indicates that each suburb is inAndunder the combined action of the two solutions, a new state (new solution) is generated by growth.
As can be seen from equations (44) and (45), the inter-individual difference is large and the range of the scaling factor (gaussian random number) is increased in the early stage of the search as compared with COA, so that the exploratory power is enhanced. In the later stage of the search, although Gaussian random numbers are adopted, the difference between individuals is small,andthe value of (2) is reduced, so that the search range is reduced, and the mining capability is enhanced. And because of adopting the global optimal solution to guide, the mining capability is enhanced, and the global optimal solution obtained by the previous group can act on the suburb growth of the next group, so that a positive feedback function of information sharing is formed, and the convergence speed is greatly accelerated. In addition, after all the suburbs grow up, the fitness values and the advantages and the disadvantages of the suburbs are calculated in parallel, and the running speed and the stability are improved.
COA has a plurality of parameters to be adjusted, and the operability is poor. For the above improved COA, there are mainly two parametersN c AndN p the impact on the optimization performance is large. In thatNUnder the fixation, ifN c Determine ifN p =N/N c I.e. byN p The larger the size of the hole is,N c the less the number of the growth operations, the less the overall solution effects are enhanced group by group, and the exploitation is strong; on the contrary, the growth operation is more,N p less, weakened global solution and weak exploitation. In order to improve the operability of COA and the like, the inventionN p AndN c and dynamically adjusting the parameters. Is provided withN=100, thenN p AndN c must be a factor of 100, and the number of wolfs per group cannot exceed 14, soN c But only 4, 5 and 10. Due to the fact thatN c It cannot be less than 3 because the group suburbs grow at least 3 suburbs, including two randomly chosen suburbs and the group-best suburbs. When in useN c The range of the selectable suburb wolfs is limited when =4, soN c The most likely values are 5 and 10.
In the later stage of the search,N c =5, thenN p =20, the number of groups is large, the positive feedback effect of the global solution is enhanced, and the local searching capability is enhanced; in the early stage of the search,N c =10, thenN p And (= 10), the number of groups is small, the positive feedback effect of the global solution is weakened, and the global searching capability is enhanced. Therefore, the operability is improved by dynamically adjusting the suburb number parameters of the group, and the exploration and exploitation capacities can be better balanced. In addition, after the parameters are dynamically adjusted, the parameters are randomly grouped, so that the process of expelling and accepting the wolf group can be omitted, and the parameters do not need to be adjustedP e The operability is also improved.
Secondly, the gray wolf optimization algorithm is simplified. In order to further solve the problems that COA is low in search efficiency and prone to falling into local optimum, the GWOO search mode is introduced. A simplified GWO search mode is provided, namely formulas (6) - (12) and a suburb wolf of a COA group are fused and simplified, and specifically formulas (46), (47) and (48) are provided.
Wherein the content of the first and second substances,temp c indicating the current suburb wolf in the groupcThe state of the suburb wolf is,NX 1 、NX 2 andNX 3 indicating that the current suburbs are respectively in the groupGP、alphaAndcultthe growth condition is obtained under the guidance of the suburb. FromAs can be seen from the expressions (46) - (48), the adjustable parameter vectors in the expressions (6), (7) and (8) are removedCThe advantages of GWO are retained and the disadvantages are overcome, i.e., in SGWO, the removalCWithout adjustmentc 1 And also omitCAnd (5) performing correlation calculation of the vectors. This simplified GWO, while ensuring its greater mining capacity, also increases operability and reduces computational complexity. In order to further simplify the calculation, the SGWO directly adopts guidance of the current global optimal suburb, the in-group optimal suburb and the medium suburb (group-internal culture tendency) in the COA to find the optimal solution without finding the second and third optimal suburbs in the group, so that the SGWO and the COA achieve an efficient fusion.
Finally, in order to balance the exploration and exploitation capacity of the growth of the suburb in the COA group, the ICOA and the SGWO are organically mixed by adopting a sine crossing strategy. The crossing strategy means that under the condition of certain probability, two solutions are crossed to obtain a new solution. When the probability is zero, the dimension values of the two solutions are not exchanged; when the probability is 1, the dimension values of the two solutions are all exchanged, and neither case can generate a new solution. The cross-over effect of the two solutions will only be best if the probability is an appropriate value. Wherein the probability model of sine function is used to make exploration and exploitation have good performance, i.e. the sine function is used to self-adaptively control the cross probabilityCR (small amplitude fluctuation in the early stage and large amplitude jump in the later stage around 0.5) can take into account the diversity and convergence of the suburb wolf in the group, and the calculation mode is shown in an expression (49).
The sine crossing strategy adopted by the invention is shown in algorithm 3 inUnder the probability of (1), the second in the groupcSuburb wolfDValues of some dimensions in the dimensions are obtained by adopting an SGWO searching mode; on the contrary, the first in the suburb wolfcSuburb wolfDThe other dimension values in the dimension are calculated by adopting a Gaussian global approach optimization growth algorithmAnd (5) sub-obtaining. This crossover strategy has the following characteristics: 1. the information exchange of various suburbs in the group is enhanced; 2. the two new solutions are crossed, and the information of the two new solutions is fused to obtain other two new solutions, so that the diversity of the new solutions is enhanced, and the probability of falling into local optimum is reduced; 3. in the early stageCRThe amplitude fluctuation is small near 0.5, a new solution is generated by Gaussian global optimization approach operation and GWO operation in an approximate equal probability mode, the diversity is strong, and the exploration capability is strengthened; at a later stage inCRThe large-amplitude jump is around 0.5, the generated new solution is mainly one of the operations, the diversity is reduced, and the mining capability is enhanced. The pseudo code of the growth operator of the sine crossing strategy is shown in algorithm 2.
Algorithm 2: growth operator for sinusoidal crossing strategy
1, calculating the sine cross probability (49)
2, selecting the firstpGroup IIIcSuburb wolf
3: for j=1 TODdo
5 calculation formula (46), (47), (48)
7: else
8 adopting a Gaussian global optimization growth operator formula (45)
9: end if
10: end for
A grey wolf and suburb hybrid optimization algorithm based on the sine crossing strategy is shown as algorithm 3.
Algorithm 3: hybrid optimization algorithm HCOGW for suburb wolfs and gray wolfs
Input, parameter CR, population quantity
Output globally optimal solutionGP
1: let t =0, randomly initializing the wolf population and grouping
2, calculating the social adaptability of the suburb wolf and obtaining the current global optimal suburb wolfGP
3: while (t = 1 to MaxDT) do
4: for t=1 to MaxDTdo
5: if t> MaxDT/2
6. N c =5
7. else
8. N c =10
9. end if
10. t=t+1
11. Dynamically adjusting parameters e and e according to equations (5) and (49)CR
12. Random grouping
Calculating the in-group optimum according to equations (16) - (18)alphaWolf and group cultural trendscult
14 calculating the suburb component growth process according to equations (44) and (45)
15, selecting the firstpGroup III c All-weather wolf
16. for j=1 TO D do
17. if rand<CR
18. Calculation formulas (46), (47) and (48)
20. else
21. Formula (45) using a Gaussian global optimization growth operator
22. end if
23. end for
24, boundary parallel control processing, and parallel computing of social adaptability of each suburb
25. Greedy selection of retention optimal wolf and update global optimal wolfGP
26. Randomly selecting father suburb wolf and variation value of environmental influence to generate new suburb wolf according to formula (21), calculating social adaptability, and reserving better suburb wolf compared with the oldest and the worst social adaptability
27. Survival of If New-born suburb wolf
28. age =0, and updates the global optimal suburb wolfGP
29. else
30. Death of new-born wolf
31. end if
32. Updating age of wolf, age = age +1
33. end for
34. end while
35. return GP
The sine crossing strategy organically integrates two search modes, and exploration and exploitation capacity in the growth process is well balanced.
From algorithm 3, HCOWG differs from COA mainly as follows: 1. a growth mode adopts a sine cross strategy to fuse a Gaussian global optimization approach mode and an SGWO mode; 2. the fitness value of the suburb wolf in the group is calculated in parallel; 3. dynamically adjusting a parameter scheme; 4. abandons the process of expelling and accepting the suburb wolf.
4. Group intelligence-based efficient energy-saving clustering algorithm
SIC clustering is a network structure division method driven by tasks, communication, calculation and resources in multiple ways, and aims to balance the calculation pressure of each node, reasonably distribute network resources, and enable the nodes with better resources to play more calculation tasks, thereby improving the stability of the network. In ammunition ad hoc networks, clustering is crucial to efficiently managing the network topology. In this section, the selection process of the cluster and the cluster head CH is a key step to reduce energy consumption, so that energy utilization rate can be balanced and network lifetime can be increased. In addition, the chapter mainly provides a clustering scheme based on suburb and grey wolf hybrid optimization, a cluster head selection algorithm and an efficient routing transmission algorithm.
4.1 System model
In a clustering scheme, a group of flying projectilesA plurality of groups of ammunition are formed and distributed in a certain area in the air. Assume that there are N number of cruise bombs in the area. Clustered collectionsIs defined asIn which N is p Indicates the number of clusters, N c Is the number of ammunition nodes in a cluster. The definition of the network map isWhereinShown is a set of vertices of a sub-cartridge,the representation is an edge set. If the signal-to-noise ratio (SNR) between the ammunition i and j is less than a specified minimum threshold, i and j are represented as adjacent nodes.
4.2 channel model
This section focuses on introducing a channel model between ammunition, since a flying round is a "combination" of an unmanned plane and a missile, reference is mainly made to a U2U channel model of the unmanned plane on the channel model, and when an ammunition node transmits a signal to j, the power received at j can be expressed as:
wherein the content of the first and second substances,P ij to representiTojThe energy of the transmission of (a) is,for the power gain of a small-scale fading channel,d ij is the distance between the two nodes, and the distance between the two nodes,representing the loss exponent of the average path. FromiTojThe signal-to-noise ratio of (c) is:
wherein the content of the first and second substances,shown is additional white gaussian noise, the U2U channel is typically controlled by a line-of-sight (LoS) link. Therefore, it is fromiTojThe path loss of (a) can be considered as free space propagation, and the loss can be expressed as:
wherein the content of the first and second substances,the carrier frequency of the U2U channel is indicated,cindicating the speed of light.
Second, we propose a U2BS (drone to base station) channel model in which the communication link has a certain probability of LoS or non line of sight (NLoS) conditions. The probability parameters are the height of the cruise missile and the elevation angle between the cruise missile and the ground base station (or the naval vessel base station). Assuming that the ammunition has a height ofz i And the distance between the estimated position of the ammunition and the BS isr i,BS Then the probability of LoS is:
whereind i,BS To representiThe distance between the BS and the BS,andis a parameter that is related to the air conditions,ithe path loss with the BS can be expressed as:
wherein the content of the first and second substances,it is indicated that the path loss exponent,it is indicated that the carrier frequency is,andthe additional path loss under both LoS and NLoS conditions is shown separately.
4.3 Energy loss from the cluster head
The node energy loss mainly comes from the signal transmission and reception process, and the part adopts a free space multipath fading channel model based on the distance d between a transmitter and a receiver. And transmitting data of l bits with a distance threshold ofd th Then the energy consumption can be expressed as:
wherein the content of the first and second substances,representing a parameter of the electronic energy consumption, depending on the energy consumed per bit between the transmitter and the receiver.、Representing the energy of the amplifier in two cases, the magnitude of which depends ond. The energy consumption for receiving data of l bits is:
assume an average number of clusters in an entire ammunition ad hoc network isN p The average energy consumed per cluster isE Np The average number of nodes per cluster isN/N p WhereinNIs the total number of nodes. The average energy consumption of a round of communication is:
representing a quality parameter of the communication channel, the average energy consumption per cluster is:
wherein the content of the first and second substances,E CH indicating the energy consumption of the cluster head CH,、the energy consumed by the CM node in the free space and multipath amplification models respectively,、the scores of the member nodes transmitted using the free space model and the multipath model, respectively.E CH Can be expressed as:
、representing the average distance from CM to its corresponding CH in the free space and multipath models, respectively. For in formula (56)In the case of (60), (61), (62), we can obtain:
substituting (63) for equation (58) has:
4.4 determination of optimal clustering quantity
Typically the total energy consumption of an ad hoc network is determined by the number of clusters in the network. Determining the optimal clustering number can reduce energy consumption and achieve the effect of balancing the network. Next, a mathematical model of the optimal number of clusters will be constructed, which determines the optimal number of clusters. The volume of the cubic network area in the air is:
m represents the length of the cubic network region, and the average volume occupied by each cluster is:
the clustered regions can be represented asLet the position of CH be the centroid of the clustering region, and the spatial estimation value of the distance from the node to CH can be obtained by the following calculation
It is assumed that the cluster region is spherical,is a constant for the cluster region, the radius of the sphere and the node density can be expressed as:
then, equation (67) may be expressed as:
by substituting equation (70) into equation (64), the optimal number of clusters in the ad hoc network can be obtained:
the probability of selecting the best number of cluster heads is:
4.5 clustering algorithm
In this section, an ammunition ad hoc network is mainly clustered by adopting an HCOGW (hybrid control over open gas) optimization algorithm, and the purpose is to divide N sub-ammunition nodes into preset or optimal sub-ammunition nodesN a opt The number of clusters. The association degree of the wolf cluster and the ammunition cluster is shown in table 1
TABLE 1
Suburb wolf group | Ammunition cluster |
Suburb wolf | Ammunition node |
Food or game location | Objective function |
Prey position unknown by suburb wolf | Nodes without knowledge of cost function |
Optimum value of fitness function | Best node, i.e. cluster head |
The number of the average wolfs of each suburb wolf group is 8-12 | The ideal average number of nodes per ammunition cluster is 5-12 |
Location of wolf | Location of ammunition |
In the clustering process, the nodes with the closest Euclidean distance are allocated to the same cluster. This ensures a lower data transmission distance and reduces power consumption. However, the positions of the nodes are difficult to determine under the high dynamic topological condition of ammunition networking, and in order to solve the problem, the HCOGW algorithm is adopted to measure the distances of the nodes. First, we group ammunition by minimizing (74) the sum of the squared errors, thereby determining their geographic location.
WhereinIs the uniqueness of the association of the wolf with the cluster,is a clusterThe set of associated wolves may be,is a wolfIn the dimensionhOf (c) is used. If the new location of the suburb wolf (i.e., the location of the ammunition node) is beyond the search range, (75) the wolf pack is directed toward the beyond boundary. This process can be expressed as:
whereinAndrespectively the upper and lower borders of the search area,is [0,1]]The random number of (2). In the HCOGW clustering method, a cluster is formed,GPthe position of the suburb is the center of mass of the cluster. The clustering scheme based on HCOGW is described in Algorithm 4.
Clustering algorithm based on HCOGW:
Output Each ClusterC g G =1,2 \8230;, M
4. repeat
8. Updating wolf and cluster relationships via Table 1
9. end for
10. end for
14. The position of each wolf is updated by equation (12).
15. The latest position of the w-th wolf is calculated by (75).
17. Optimal wolf and median wolf in the calculation set
18. Using (44), (45) to calculate the intra-component growth process
19. According to a sine crossing strategy, the growth of the wolf group is partially carried out by adopting a Gauss-oriented optimal growth operator, and partially carried out by adopting a simplified wolf optimization algorithm
20. Calculating the social adaptation capability of each suburb, greedy selecting to reserve the optimal suburb and updating the globally optimal suburbGP
21. And (3) randomly selecting the father suburb and the variation value of the environmental influence to generate the new suburb according to the formula (21), calculating the social adaptability of the new suburb, and judging the life and death of the new suburb.
22. Updating the age of the wolf and the globally optimal wolfGP
23. end for
24. end for
26.end while
27.Return GP
4.7 selection of Cluster heads
This section mainly takes the HGWO to select the cluster head CH in the network, the selection process of which is shown in algorithm 7. In the HGWO algorithm, the best solution is. In the first round, the position of the wolf is updated according to (46) - (48). And then, updating the position of the wolf according to the value of the fitness. The procedure for measuring the fitness value is as follows:
case 1:
therefore:
Case 2:
case 3:
Case 4:
for cluster headCH S The choice of (a) is usually made taking into account several fitness functions, the main objective being to maximize the fitness function. In the fitness model, an intra-cluster distance function-ICD, an adjacent node number function-NON and a residual power function-RP of an ammunition node are mainly considered.
WhereinAndthe three coefficients are taken to be 0.2, 0.3 and 0.5 based on priority, respectively. The average distance between the suburb wolf node and the initial cluster head CH is a fitness function ICD, which is defined as:
wherein the content of the first and second substances,showing suburb nodes to cluster headsCH g The distance of (a) to (b),N g is shown inCH m The number of sub-ammunition nodes in the cluster that is the cluster head. The expression of the fitness function NoN is as follows:
wherein, the first and the second end of the pipe are connected with each other,Rthe radius is indicated as such and,N a opt representing the optimal number of clusters. The expression for the residual energy is:
wherein the content of the first and second substances,E R CH g indicating current cluster headCH g The remaining energy level of (a) is,. A detailed cluster head selection procedure such as algorithm 6.
Output cluster headCH g ,g=1,2……,M
Initialization parameterse、CR
3. Calculating fitness value of each wolf (fit)
6. for each search agent, do
7. The location of the current search unit is updated.
8. if round =0 then
9. Updating the location by (46) - (48)
10. else
11. The position is updated through (76) - (88).
12. end if
13. end for
14. Updatinge、CRValue of (2)
15. Calculating fitness values of all search units: (fit).
18. end while
19. Return CH S
5. performance evaluation
A simulation experiment is carried out on the SIC clustering algorithm in MATLAB to evaluate the performance of the algorithm under different scenes. First, we studied the performance of SIL and compared it with that of Selective 3-Anchor DV-Hop, DCDV-Hop, ND-DV-Hop, GWO-LPWSN and DV-Hop. The performance of SIC was then evaluated and compared to the performance of cbladr, ACO, EALC, GA, and SOCS.
5.1 simulation Environment
TABLE 2
5.2 Clustering simulation results and analysis
The performance of the clustering algorithm is compared and analyzed, the clustering algorithm based on HCOGW optimization is mainly compared with the common clustering algorithms based on swarm intelligence optimization, such as CBLADSR, ACO, EALC, GA and SOCS, and the result analysis is carried out.
(1) Number of clusters
In the clustering algorithm, the number of clusters largely determines the consumption of network energy, and the optimal number of clusters can effectively reduce the consumption of energy. As shown in fig. 1, the relationship between the number of clusters and the total number of nodes is mainly shown, and it can be seen from the figure that the performance of the SIC algorithm of the present invention is obviously superior to that of other clustering algorithms, and the number of clusters can be controlled to be small. The modeling herein analyzes the optimal number of clusters to ensure that energy consumption in the network is minimized. If the number of nodes in the cluster is small, the number of clusters is increased; if the number of nodes in the cluster is large, the data transmission amount in the cluster is increased. This means that once the optimal cluster number is not obtained, part of nodes are not added into the cluster, resulting in that a single node is clustered independently, which may greatly increase the cluster number. To optimize the number of clusters, the SIC clustering scheme proposed herein can prevent single nodes from clustering and optimize the network topology.
(2) Aspect of clustering time
The clustering time refers to the total time required by the clustering algorithm in CH selection and clustering formation, which affects the computational complexity of the clustering algorithm. In the clustering process, when each cycle starts, each CH node broadcasts a message and declares itself to be a cluster head in the network, and each node in the cluster broadcasts the position and the residual energy of each node. In SIC, the number of transmission control packets depends on the selection of CH and the distance between the node and the BS. Longer clustering times consume a lot of energy and shorten network life. Since the battery mounted on the flying patrol is limited in energy, the required clustering time is short. As shown in fig. 2, as the number of network nodes increases, the SIC algorithm clustering time is the shortest compared with the other five algorithms, mainly because in the HCOGW algorithm, an improved suburb algorithm is introduced to avoid the wolf group from falling into local optimum, so that CH and relay nodes are obtained, which makes the SIC more energy-saving. Therefore, the SIC algorithm has shorter clustering time, and simultaneously reduces the routing delay.
(3) Number of surviving nodes
After each cycle, some of the missiles lose their ability to fight due to the exhaustion of the battery power, as shown in fig. 3, which shows the variation curve of the node survival rate and the network life cycle. As can be seen from the figure, the number of surviving nodes of SIC is greater than other algorithms under the same cycle number, indicating that the network is more stable. Meanwhile, the SIC algorithm proposed herein can be demonstrated to consume the least energy among six algorithms.
(4) Aspect of total energy consumption
As can be seen from fig. 4, compared with the other five algorithms, the total energy consumption of the SIC algorithm is obviously better than that of the other algorithms, mainly because the optimal clustering and the optimal cluster head number are adopted in the algorithm.
The foregoing is illustrative of the best mode of the invention and details not described herein are within the common general knowledge of a person of ordinary skill in the art. The protection scope of the present invention is subject to the content of the claims, and any equivalent changes based on the technical teaching of the present invention are also within the protection scope of the present invention.
Claims (4)
1. A high-efficiency energy-saving clustering method based on a biological heuristic algorithm is characterized by comprising the following steps;
step 1, setting the clustering numberN a opt Inputting the positions of ammunition nodes to be clustered;
step 2, according to the number of clustersN a opt Clustering ammunition nodes to be clustered, and distributing the nodes with the closest Euclidean distance to the same cluster to obtain an initial cluster;
step 3, adjusting the initial clusters and determining the centroid position according to a clustering algorithm based on HCOWG to obtain a plurality of final clusters;
the clustering algorithm based on HCOWG in the step 3 comprises the following steps:
step 3.1, taking each node position in the initial cluster as a suburb wolf, taking the suburb wolf in the same initial cluster as an initial wolf group, and obtaining a plurality of initial wolf groups;
3.2, calculating a new position of each suburb under the guidance of 3 optimal wolfs in the wolf group by using a wolf optimization algorithm to obtain the first updated suburb;
step 3.3, calculating coefficientse=2-2t/MaxDT, sine function adaptive control crossover probabilityCR=0.5×(sin(2π× 0.25×t+π)×(t/MaxDT) + 1), whereintThe current first iteration time is MaxDT which is a preset first maximum iteration time;
step 3.4, randomly grouping the suburbs in each wolf group, and calculating the global optimal suburb of the wolf group by using a suburb optimization algorithmGPThe best wolf in each groupalphaAnd medium value suburb wolfcult;
Step 3.5, updating all the suburbs for the second time, wherein the method for updating each suburb for the second time comprises the steps of calculating the value of each dimension of the current suburb, obtaining the updated suburb after all the dimensions are calculated, and calculating the value of each dimension by using a random probability function to obtain a random probabilityrandIf at allrand<CRIf so, acquiring the value of the current dimension by adopting an SGWO searching mode; otherwise, adopting a Gaussian global optimization growth operator to obtain the value of the current dimension;
in step 3.5, the current dimension of the current wolf is obtained by adopting a Gaussian global approach growth operatorThe formula of (1) is as follows:in the formula (I), the reaction is carried out,=GP-soc cr1 ,= cult-soc cr2 ,GPfor the current value of the global optimum suburb,soc cr1 andsoc cr2 respectively represents the randomly selected suburb wolfs in the group where the current suburb wolfs are locatedcr1Wen and suburbcr2The value of (a) is,rn 1 andrn 2 is a random number generated by a gaussian/normal distribution with a mean value of 0 and a variance of 1;
in step 3.5, the formula of the SGWO search mode is:
wherein the content of the first and second substances,newcis the value of the current dimension of the current suburb wolf,temp c indicating the state of the current suburb,A 1 、A 2 andA 3 is a coefficient vectorAIn the global optimum suburbGPOptimum suburb wolf in groupalphaAnd medium value suburb wolfcultThe corresponding values under different conditions are taken,NX 1 、NX 2 andNX 3 indicating that the current suburbs are respectively in the global optimal suburbsGPOptimum suburb wolf in groupalphaAnd medium value suburb wolfcultTo obtain the growth condition under the guidance of (2);
step 3.6, calculating the social adaptation capacity of each suburb before and after the second updating of each suburb, selecting and reserving the better suburb by greedy, screening all suburbs, and finally obtaining all screened suburbs;
step 3.7, updating the global optimal suburb wolf according to all screened suburb wolfsGP;
Step 3.8, the global optimum suburb wolfGPAs the mass center of the corresponding wolf group, obtaining the mass center positions of all wolf groups, calculating the Euclidean distance between the screened suburb wolf and each mass center in the step 3.6, dividing the screened suburb wolf into the wolf group corresponding to the mass center closest to the selected suburb, and obtaining the iterative updated wolf group;
step 3.9, adding 1 to the first iteration times, judging the first iteration times, if the first iteration times is smaller than a preset first maximum iteration times, turning to step 3.1, updating the initial wolf group into the wolf group updated in the current iteration in step 3.8, and otherwise, turning to the next step;
step 4.0, taking each wolf pack after the last iteration update as a final cluster to obtain a plurality of final clusters;
step 4, determining a cluster head for each final cluster according to a cluster head selection algorithm based on HCOWG, and finishing clustering;
the cluster head selection algorithm based on HCOWG comprises the following steps:
step 4.1, taking each cluster in the final clustering result as an initial wolf cluster, and taking each node in the cluster as an initial suburb in the wolf cluster;
step 4.2, calculating the fitness value of each suburb;
step 4.3, randomly grouping the suburbs in each wolf group, and calculating the global optimal suburb of each wolf group by using a suburb optimization algorithmGPThe best wolf in each groupalphaAnd medium value suburb wolfcult;
4.4, updating all the suburbs according to the SGWO searching mode to obtain the suburbs to be processed;
step 4.5, calculating the globally optimal suburbs of the suburbs to be processedGPOptimum suburb wolf in groupalphaAnd medium value suburb wolfcultNew state under guidance of fitness value、Andand find out、Andthe average value of the suburb wolfs is obtained, the new position of the current suburb wolf is updated again, and the iteration suburb wolf is obtained;
step 4.6, adding 1 to the second iteration number, judging the second iteration number, if the second iteration number is smaller than a preset second maximum iteration number, turning to step 4.4, updating the suburb wolf to be processed into the current iteration suburb wolf in step 4.5, and meanwhile, updating the global optimal suburb wolfGPThe best wolf in each groupalphaAnd medium value suburb wolfcultUpdating, otherwise, turning to the next step;
and 4.7, taking the node corresponding to the global optimal suburb wolf of each wolf cluster obtained by the last iteration as a cluster head of the corresponding cluster of the wolf cluster.
2. The method for energy-efficient clustering based on the biological heuristic algorithm of claim 1, characterized in that the number of clusters isN a opt The calculation formula of (2) is as follows:
wherein the content of the first and second substances,Nas to the number of ammunition nodes to be clustered,is a set of edges that are to be considered,is the carrier frequency of the carrier wave,
3. The energy-efficient clustering method based on the biological heuristic algorithm of claim 1, characterized in that in the step 4.5:
wherein the content of the first and second substances,、andrespectively as global optimum suburb wolfGPOptimum suburb wolf in groupalphaAnd medium value suburb wolfcultThe value of the fitness of (a) is,is the current fitness value of the current suburb wolf,A 1 、A 2 andA 3 is a coefficient vectorAIn globally optimal suburbGPAnd the best suburb in groupalphaAnd medium value suburb wolfcultAnd (4) corresponding values under different conditions.
4. The energy-efficient clustering method based on biological heuristic algorithm of claim 1, wherein in step 3.2, if any one of the obtained locations corresponding to the suburbs after the first update exceeds the search boundary, the suburb location after the first update is adjusted, the adjusting method is:
wherein, the first and the second end of the pipe are connected with each other,in order to adjust the position of the suburb wolf,is the suburb wolf after the first update,is the suburb wolf before the first update,andrespectively the upper and lower borders of the search area,is [0,1]]The random number of (2).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211223207.6A CN115297497B (en) | 2022-10-08 | 2022-10-08 | High-efficiency energy-saving clustering method based on biological heuristic algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211223207.6A CN115297497B (en) | 2022-10-08 | 2022-10-08 | High-efficiency energy-saving clustering method based on biological heuristic algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115297497A CN115297497A (en) | 2022-11-04 |
CN115297497B true CN115297497B (en) | 2023-02-03 |
Family
ID=83834834
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211223207.6A Active CN115297497B (en) | 2022-10-08 | 2022-10-08 | High-efficiency energy-saving clustering method based on biological heuristic algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115297497B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116074991A (en) * | 2023-01-10 | 2023-05-05 | 中国人民解放军海军工程大学 | Intelligent cluster clustering method based on K-means++ clustering algorithm |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101873606A (en) * | 2010-06-09 | 2010-10-27 | 中国人民解放军海军航空工程学院 | Network clustering method of wireless sensor based on fixed cluster heads |
CN110062390A (en) * | 2019-04-19 | 2019-07-26 | 江西理工大学 | Based on the wireless sensor network node Optimization deployment method for improving wolf pack algorithm |
CN111263325A (en) * | 2020-01-17 | 2020-06-09 | 南华大学 | Routing method, device and medium for dynamic clustering heterogeneous wireless sensor network |
CN111368891A (en) * | 2020-02-27 | 2020-07-03 | 大连大学 | K-Means text classification method based on immune clone wolf optimization algorithm |
CN111698705A (en) * | 2020-05-29 | 2020-09-22 | 华南理工大学 | Non-uniform clustering routing method of wireless sensor network based on energy optimization |
CN111784211A (en) * | 2020-08-04 | 2020-10-16 | 中国人民解放军国防科技大学 | Cluster-based group multitask allocation method and storage medium |
CN112085147A (en) * | 2020-09-16 | 2020-12-15 | 北京邮电大学 | Feature selection method based on improved suburb optimization algorithm |
CN114783215A (en) * | 2022-04-18 | 2022-07-22 | 中国人民解放军战略支援部队信息工程大学 | Unmanned aerial vehicle clustering method and device and electronic equipment |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7539155B1 (en) * | 2000-08-15 | 2009-05-26 | Michael Holloway | Centralized feature platform in a packetized network |
US10893420B2 (en) * | 2018-02-16 | 2021-01-12 | Telefonaktiebolaget Lm Ericsson (Publ) | System and method of automatic physical cell ID allocation to reduce collision |
CN108520272B (en) * | 2018-03-22 | 2020-09-04 | 江南大学 | Semi-supervised intrusion detection method for improving Cantonese algorithm |
-
2022
- 2022-10-08 CN CN202211223207.6A patent/CN115297497B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101873606A (en) * | 2010-06-09 | 2010-10-27 | 中国人民解放军海军航空工程学院 | Network clustering method of wireless sensor based on fixed cluster heads |
CN110062390A (en) * | 2019-04-19 | 2019-07-26 | 江西理工大学 | Based on the wireless sensor network node Optimization deployment method for improving wolf pack algorithm |
CN111263325A (en) * | 2020-01-17 | 2020-06-09 | 南华大学 | Routing method, device and medium for dynamic clustering heterogeneous wireless sensor network |
CN111368891A (en) * | 2020-02-27 | 2020-07-03 | 大连大学 | K-Means text classification method based on immune clone wolf optimization algorithm |
CN111698705A (en) * | 2020-05-29 | 2020-09-22 | 华南理工大学 | Non-uniform clustering routing method of wireless sensor network based on energy optimization |
CN111784211A (en) * | 2020-08-04 | 2020-10-16 | 中国人民解放军国防科技大学 | Cluster-based group multitask allocation method and storage medium |
CN112085147A (en) * | 2020-09-16 | 2020-12-15 | 北京邮电大学 | Feature selection method based on improved suburb optimization algorithm |
CN114783215A (en) * | 2022-04-18 | 2022-07-22 | 中国人民解放军战略支援部队信息工程大学 | Unmanned aerial vehicle clustering method and device and electronic equipment |
Non-Patent Citations (3)
Title |
---|
Classification of Microarray Gene Expression Data;Amika Achom;《IEEE》;20191212;全文 * |
基于RSSI的灰狼优化差分修正质心定位算法;高栋;《长春工业大学学报》;20210831;全文 * |
基于改进灰狼优化算法的无线传感网络;周文详;《硕士毕业论文电子期刊》;20210731;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN115297497A (en) | 2022-11-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | A communication strategy of proactive nodes based on loop theorem in wireless sensor networks | |
CN112822752B (en) | Route establishing method and system for unmanned aerial vehicle self-organizing network | |
Bharany et al. | Energy-efficient clustering protocol for underwater wireless sensor networks using optimized glowworm swarm optimization | |
Fei et al. | Energy-efficient clustering algorithm in underwater sensor networks based on fuzzy C means and Moth-flame optimization method | |
CN115297497B (en) | High-efficiency energy-saving clustering method based on biological heuristic algorithm | |
CN113573271A (en) | DV-hop positioning method based on Levy flight strategy and three-dimensional wolf algorithm optimization | |
CN110225569B (en) | WSNs clustering multi-hop routing protocol method based on improved particle swarm optimization | |
Sefati et al. | Cluster selection for load balancing in flying ad hoc networks using an optimal low-energy adaptive clustering hierarchy based on optimization approach | |
Gupta et al. | Multi-UAV deployment for NOMA-enabled wireless networks based on IMOGWO algorithm | |
Khedr et al. | MWCRSF: Mobility-based weighted cluster routing scheme for FANETs | |
CN109348518B (en) | Method for searching route by using artificial fish swarm algorithm in underwater acoustic cooperative communication network | |
Pan et al. | The phasmatodea population evolution algorithm and its application in 5g heterogeneous network downlink power allocation problem | |
Moussaoui et al. | Clustered chain founded on ant colony optimization energy efficient routing scheme for under-water wireless sensor networks | |
Karunkuzhali et al. | An adaptive fuzzy c means with seagull optimization algorithm for analysis of WSNs in agricultural field with IoT | |
Reddy et al. | Secure Cluster based Routing Using Multiobjective Trust Centric Reptile Search Algorithm for WSN. | |
Maharajan et al. | Energy Efficient QoS Aware Cluster Based Multihop Routing Protocol for WSN. | |
CN115333990B (en) | Efficient transmission routing method based on suburb and gray wolf hybrid optimization | |
Fadhel | Optimization of head cluster selection in WSN by human-based optimization techniques | |
Razooqi et al. | Enhanced Ant Colony Optimization for Routing in WSNs An Energy Aware Approach. | |
Hao et al. | Swarm intelligence based clustering and localizing methods for loitering munitions in a satellite denied environment | |
Kaur et al. | Acoustic monitoring in underwater wireless sensor networks using energy-efficient artificial fish swarm-based clustering protocol (EAFSCP) | |
Bahl et al. | Probabilistic Multi-Tiered Grey Wolf Optimizer-Based Routing for Sustainable Sensor Networks | |
Khedr et al. | A hybrid MGO-JAYA based clustered routing for FANETs | |
Khasim et al. | A Hybrid Bird Swarm-Differential Search Methods, Based On Wsn For Trusted Cluster-Based Energy And Lifetime Aware Routing | |
Komala et al. | Multi-UAV computing enabling efficient clustering-based IoT for energy reduction and data transmission |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |