CN103702276B - A kind of complex task collaborative service method in wireless sensor network based on sub-clustering - Google Patents

Abstract

The present invention relates to a kind of complex task collaborative service method in wireless sensor network based on sub-clustering, its step includes: 1. complex task layering, according to task DAG figure, it is carried out layering and divides, and the most successively distributes during task distribution；2. center cluster bunch head is by member node in task distributes to center cluster in selected layer；If 3. center cluster bunch energy expenditure degree exceedes setting threshold value, center cluster request periphery one jumps bunch assist process, bunch energy expenditure degree not less than threshold value bunch as the cooperative cluster of center cluster.4. untreated task is distributed to cooperative cluster by center cluster, it is ensured that cooperative cluster energy expenditure degree difference minimum.5. in cooperative cluster assigns the task to bunch, member node processes.With it, the complex task sub-clustering in network can be processed, task treatment effeciency, balance network load can be improved, extend network life.

Description

A kind of complex task collaborative service method in wireless sensor network based on sub-clustering

Technical field

The invention belongs to wireless multimedia sensor network field, specifically the present invention relates to a kind of based on sub-clustering wireless The method of complex task collaborative service in sensor network, processes the complex task sub-clustering in network, can improve task and process Efficiency, balance network load, extend network life.

Background technology

Along with wireless sensor network application requirement of real-time is more and more higher, task complexity is increasing, due to sensing Device individual node resource-constrained, if considerable task is distributed to single node and completed, easily causes this node energy and exhausts, therefore The problem of node energy consumption balance becomes a key issue of wireless sensor network.One given complex task, resolves into Multiple subtasks, and distribute to different sensor nodes go process, multinode cooperates, and jointly completes task, Ke Yijie About node energy, raising systematic function.

Task distribution is the important research problem of wireless sensor network, and different task allocative decisions causes tasks carrying The required traffic is different with amount of calculation, thus affects the energy expenditure of tasks carrying, for reducing energy expenditure, extends network raw The life cycle needs to design efficient task allocative decision.

Pertinent literature is as follows:

1,2005, Y.Yu et al. was at " Energy-Balanced Task Allocation for Collaborative Processing in Wireless Sensor Networks " propose the task allocation algorithms of a kind of balancing energy, this calculation Method meet minimize energy expenditure while reach the purpose of balancing energy.The communication issue of multiple wireless channels is built by they Mould, and propose to utilize heuritic approach to provide a feasible solution.

2,2009, Sekhar et al. was at " A state-space search approach for optimizing Reliability and cost of execution in distributed sensor networks " a kind of base of middle proposition In A* algorithm, this algorithm assigns the task to a large amount of sensor node, simultaneously takes account of the energy constraint problem of node, it is proposed that Greedy A* algorithm (Greedy A*) reduces the complexity of A* algorithm optimal distributing scheme.By optimization problem is formulated And the value arranging specific parameter μ and λ is analyzed.

But, the distribution of above task is all completed by same node with communicating, and so easily causes individual node energy Too much consume；At task allocated phase, it is not easy to ensure that the parallelization of task processes, do not embody the real-time of application.

3,2009, Li Zhigang et al. formed at isomorphism node in " the effective task allocation algorithms of the sensor network energy " Proposing a kind of effective task allocation algorithms of the energy in network, assessment of tasks is perception task collection and process task by this algorithm Collection, therefore task distribution is also divided into perception task distribution and process task two stages of distribution, and process task is assigned as secondary One-zero programming problem, and distributed hierarchical optimization allocation algorithm is proposed, it is achieved that it is energy-optimised that task processes, but does not beg for How opinion shortens task processes the time.

Summary of the invention

The technical problem to be solved is: the complex task sub-clustering in network processed, and processes for improving task Efficiency, balance network load, extend network life.The present invention is achieved by the following technical solutions, the concrete step of the present invention Rapid as follows:

Step one: task is layered, carries out layering according to task DAG figure to it and divides, during task distribution the most successively Distribution, every layer is distributed some tasks；

Step 2: after the event of task occurs, incident point place bunch as center cluster, bunch head of center cluster is by layer Task is allocated；

Step 3: when bunch dump energy of center cluster is than less than setting threshold value, then jump a bunch transmission solicited message to periphery one, If periphery one is jumped bunch dump energy and is not less than threshold value than, then accept request, as the cooperative cluster of center cluster；Described center cluster with Periphery one jumps a bunch formation cooperation cluster, and bunch head of periphery one jumping bunch accepts the arrangement of center cluster bunch head；

Step 4: untreated task is distributed to cooperative cluster by center cluster, ensures the balancing energy of each cooperative cluster during distribution, Dump energy is than difference minimum；

Step 5: in cooperative cluster assigns the task to bunch, member node processes.

In above-mentioned steps one, that task carries out the process that layering divides is as follows for DAG figure according to task: entrance task divides For ground floor, if a certain subtask forerunner's maximum layer is k, then this subtask is divided into k+1 layer.

Above-mentioned steps three judging, center cluster accepts after new task dump energy than whether less than the calculating setting threshold value Journey is as follows:

(3a) calculate bunch present energy and consume C_cluster(i): bunch present energy consumption refers to certain bunch of current existing task Energy expenditure sum；

(3b) bunch total surplus energy is calculatedBunch total surplus energy be bunch in the dump energy sum of all nodes；

(3c) bunch total primary power is calculatedBunch total primary power be bunch in the primary power sum of all nodes；

(3d) calculate bunch dump energy and compare R_cluster(i): a bunch dump energy ratio is that total surplus energy deducts current energy in this bunch Amount consumes the ratio with bunch total primary power, $R_{\mathrm{cluster}\left(i\right)}=\frac{E_{\mathrm{residual}}^{\mathrm{cluster}\left(i\right)}-C_{\mathrm{cluster}\left(i\right)}}{E_{\mathrm{initrial}}^{\mathrm{cluster}\left(i\right)}}\×100\%.$

In above-mentioned steps five bunch in the evaluation procedure of member node service ability as follows:

(4a) degree of belief reliability of node processing task is calculated_i,Represent that node i processes the success of task Number of times, usesRepresent the total degree of node i accumulated process task, then the degree of belief of node i process task can be expressed as ${\mathrm{reliability}}_i=\frac{t_{\mathrm{reliability}}^i}{t_{\mathrm{total}}^i},{\mathrm{reliability}}_i\∈\left[\mathrm{0,1}\right];$

(4b) calculate residue energy of node and compare residual_i,Represent the dump energy of node i,Represent node The primary power of i, node i dump energy compares residual_iIt it is exactly the dump energy ratio with the primary power of node i of node i Value, it may be assumed that

(4c) service ability of node is calculated: node i service ability can be expressed as by equation below ${\mathrm{capacity}}_i={\mathrm{\αreliability}}_i+{\mathrm{\βresidual}}_i=\mathrm{\α}\frac{t_{\mathrm{reliacility}}^i}{t_{\mathrm{total}}^i}+\mathrm{\β}\frac{e_{\mathrm{residual}}^i}{e_{\mathrm{initial}}^i},$ Wherein alpha+beta=1, α >=0, β >=0.

In above-mentioned steps five bunch in member node categorizing process as follows:

(5a) in general bunch, member node is divided into three classifications, and a category node is to have service ability, and service ability is strong Node set, two category nodes are to have service ability, and the node set that service ability is general, and three category nodes are temporarily without service energy The node set of power；

(5b) node classification initializes: if residue energy of node ratio is less than setting threshold value r, then it is assumed that node is temporarily without clothes Business ability, this category node returns the node that is divided three classes；In bunch, in addition to three category nodes, remaining node is considered there is service ability node, Calculating has service ability node serve ability and sorts, and having the node division of a% before in service ability node is a category node, a ∈ [0,100], the node division outside a category node, three category nodes is two category nodes；If there being service ability node number to be less than Or equal to two, then will have service ability node universal formulation is a category node；

(5c) node classification dynamically adjusts: after each layer of task has processed, will to bunch in member node classification weight New division, calculates residue energy of node ratio, is saved as three classes less than the node division setting threshold value r by residue energy of node ratio Point, calculates and has service ability node serve ability and sort, be divided into respective classes according to its service ability.

In above-mentioned steps three, the evaluation procedure of the service ability size of periphery one jumping bunch is as follows:

(6a) bunch service ability and a category node, the quantity of two category nodes, a category node, the service ability of two category nodes, Bunch dump energy ratio is relevant.

(6b) assume that the service ability of bunch cluster (k) interior joint i is expressed as capacity_i, one type Node number is N₁, two category node numbers are N₂, then a bunch capability list is shown as ${\mathrm{Capacity}}_{\mathrm{cluster}\left(k\right)}=({\mathrm{\ω}}_1\×N_1\×\underset{i\∈G1}{\mathrm{\Σ}}{\mathrm{capacity}}_i+{\mathrm{\ω}}_2\×N_2\×\underset{i\∈G2}{\mathrm{\Σ}}{\mathrm{capacity}}_i)\×\frac{R_{\mathrm{cluster}\left(k\right)}-r}{r}$ Wherein, G1 represents one Category node, G2 represents two category nodes, ω₁+ω₂=1, ω₁>=0, ω₂>=0, R_cluster(k)Represent the residual energy of bunch cluster (k) Amount ratio, bunch dump energy that r sets compares threshold value.

Concretely comprising the following steps of composition cooperative cluster in above-mentioned steps three:

(7a) ground floor task distributes to center cluster, while ground floor task processes, carries out second layer task distribution；

If (7b) dump energy of center cluster is than less than setting threshold value, then in being no longer allocated to task unallocated in layer The heart bunch, center cluster jumps a bunch request cooperation to periphery one, and periphery one is jumped bunch according to self situation, bunch provides centered by choosing whether Service, if periphery one jumps bunch dump energy, ratio is less than or equal to set threshold value, then refusal provides service, periphery one jumping bunch residue Energy than more than setting threshold value, then can provide service, and self bunch service ability value is returned to center cluster；

If being (7c) n with layer unallocated task number, in periphery one is jumped bunch, select bunch service ability bunch making at front n For alternative cooperative cluster, n task is assigned in n alternative cooperative cluster, it is ensured that the dump energy of each bunch is than difference as far as possible Little, and assign to task number on each bunch less than bunch in the number of a category node；

If (7d) number of center cluster periphery one jumping bunch is all made less than unallocated task number, the jumping bunch of the most all peripheries one For alternative cooperative cluster；

If (7e) center cluster periphery one jumping bunch all refuses to provide service, then center cluster periphery two jumps a bunch transmission message, asks Ask periphery two to jump bunch assistance to complete；

(7f) after this layer of task is assigned to bunch, bunch head in task being assigned to bunch, member node performs, and carries out one simultaneously Layer task distribution, until all tasks are assigned.

In above-mentioned steps four, untreated task is distributed to the specific requirement of cooperative cluster by center cluster: have n task task₁, task₂,…,task_n, m cooperative cluster clouster₁,clouster₂,…,clouster_mComplete this n task, kth bunch The energy expenditure completing this n task is respectively C_k1,C_k2,…,C_kn, and the dump energy of m cooperative cluster is than respectively R₁, R₂,…,R_m, the primary power of each bunch is respectivelyN task is distributed to m cooperative cluster Complete so that it is less, to reach the energy balance than difference that each cooperative cluster completes the dump energy after task；

Assuming that each task is only completed by a cooperative cluster, a cooperative cluster can complete multiple task；

x_ij=1 represents that jth task is completed by i-th cooperative cluster, x_ij=0 represents that jth task is not complete by i-th bunch Become, C_ijRepresent that i-th cooperative cluster completes the energy expenditure of jth task, i=1,2 ..., m, j=1,2 ..., n, make X= (x_ij)_m×nMatrix for logger task distribution condition；

Then completing this n task, the energy expenditure of cooperative cluster is

$\left(\begin{array}{c}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j}\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j}\\ \·\\ \·\\ \·\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}}\end{array}\right),$

The dump energy of cooperative cluster is

$\left(\begin{array}{c}{R}_1{E}_{\mathrm{initial}}^1\\ R_2{E}_{\mathrm{inntial}}^2\\ \·\\ \·\\ \·\\ R_m{E}_{\mathrm{initial}}^m\end{array}\right)-\left(\begin{array}{c}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j}\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j}\\ \·\\ \·\\ \·\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}}\end{array}\right)=\left(\begin{array}{c}{R}_1{E}_{\mathrm{initial}}^1-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j}\\ R_2{E}_{\mathrm{initial}}^2-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j}\\ \·\\ \·\\ \·\\ R_m{E}_{\mathrm{initial}}^m-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}}\end{array}\right),$

Bunch dump energy ratio be

$\left(\begin{array}{c}(R_1{E}_{\mathrm{initial}}^1-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j})/E_{\mathrm{initial}}^1\\ (R_2{E}_{\mathrm{initial}}^2-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j})/E_{\mathrm{initial}}^2\\ \·\\ \·\\ \·\\ (R_m{E}_{\mathrm{initial}}^m-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}})/E_{\mathrm{initial}}^m\end{array}\right),$

The average residual energy ratio of m cooperative cluster is approximately

$R_0=(\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{R}_i{E}_{\mathrm{initial}}^i-\underset{i=1}{\overset{m}{\mathrm{\Σ}}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{C}_{\mathrm{ij}}/m)/\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{E}_{\mathrm{initial}}^i$

Seek x_ijMake

The least, wherein $b_i=R_i-R_0,a_{\mathrm{ij}}=C_{\mathrm{ij}}/E_{\mathrm{initial}}^i;$

I.e. solve following one-zero programming problem

$\underset{x_{\mathrm{ij}}}{\min }\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{[b_i-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{a}_{\mathrm{ij}}{x}_{\mathrm{ij}}]}^2$

$s.t.\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{x}_{\mathrm{ij}}=1,j=\mathrm{1,2},...,n,$

x_ij∈{0,1}

By the solution of above-mentioned optimization problem, can obtain distributing to n task the task distribution side that m cooperative cluster completes Case, and the dump energy after making each cooperative cluster complete task differs the least than as far as possible, thus ensure balancing energy.

In above-mentioned steps two, what task in selected layer was allocated by center cluster bunch head concretely comprises the following steps:

(9a) subtask is allocated by the order that center cluster bunch head arrives according to task, is sent by mission bit stream to be allocated To a category node；

(9b) after a category node receives message, check task description, decide whether to process according to self-energy consumption degree This task, if energy expenditure degree exceedes setting threshold value, then refusal provides service, if energy expenditure degree is not less than setting threshold Value, then provide service, and calculated this task estimation time, estimated energy consumption, these two information are returned bunch heads；

(9c) bunch head is according to a category node return information, selects the node of energy expenditure minimum as tasks carrying node；

If (9d) category node all refuses to provide service, then bunch head sends message to two category nodes, in two category nodes Select service available node.

The invention have the benefit that compared with prior art, the invention provides a kind of wireless sensing based on sub-clustering Complex task collaborative service method in device network, single bunch when cannot complete complex task, to periphery bunch request cooperation, composition association Work bunch, completes complex task jointly, whether bunch dump energy is used for participating in the threshold of cooperation, thus balances each bunch of energy, Extend network lifecycle；Dump energy ratio and execution Mission Success rate according to bunch interior nodes calculate node serve ability, right Node is classified, thus the service ability obtained bunch, by the performance quantified bunch, improves task allocative efficiency and completes matter Amount.

Accompanying drawing explanation

Fig. 1: bunch interior nodes classification process figure

Fig. 2: task is assigned to cooperative cluster flow chart

Fig. 3: a bunch head assigns the task to a bunch interior nodes flow chart

Detailed description of the invention

Below in conjunction with drawings and Examples, the present invention is described in further detail.

As shown in Figure 1, 2, 3, in a kind of sensor network based on bunch complex task decomposition method, its step includes:

Step one: task is layered, carries out layering according to task DAG figure to it and divides, during task distribution the most successively Distribution, every layer is distributed some bunches；

Step 2: after the event of task occurs, incident point place bunch as center cluster, bunch head of center cluster is by layer Task is allocated；

Step 3: when bunch dump energy of center cluster is than less than setting threshold value, then jump a bunch transmission solicited message to periphery one, If periphery one is jumped bunch dump energy and is not less than threshold value than, then accept request, as the cooperative cluster of center cluster；Described center cluster with Periphery one jumps a bunch formation cooperation cluster, and bunch head of periphery one jumping bunch accepts the arrangement of center cluster bunch head；

Step 4: untreated task is distributed to cooperative cluster by center cluster, ensures the balancing energy of each cooperative cluster during distribution, Dump energy is than difference minimum；

Step 5: in cooperative cluster assigns the task to bunch, member node processes.

In above-mentioned steps one, that task carries out the process that layering divides is as follows for DAG figure according to task: entrance task divides For ground floor, if a certain subtask forerunner's maximum layer is k, then this subtask is divided into k+1 layer.

Above-mentioned steps three judging, center cluster accepts after new task dump energy than whether less than the calculating setting threshold value Journey is as follows:

(3a) calculate bunch present energy and consume C_cluster(i): bunch present energy consumption refers to certain bunch of current existing task Energy expenditure sum；

(3b) bunch total surplus energy is calculatedBunch total surplus energy be bunch in the dump energy sum of all nodes；

(3c) bunch total primary power is calculatedBunch total primary power be bunch in the primary power sum of all nodes；

(3d) calculate bunch dump energy and compare R_cluster(i): a bunch dump energy ratio is that total surplus energy deducts current energy in this bunch Amount consumes the ratio with bunch total primary power, $R_{\mathrm{cluster}\left(i\right)}=\frac{E_{\mathrm{residual}}^{\mathrm{cluster}\left(i\right)}-C_{\mathrm{cluster}\left(i\right)}}{E_{\mathrm{initrial}}^{\mathrm{cluster}\left(i\right)}}\×100\%.$

In above-mentioned steps five bunch in the evaluation procedure of member node service ability as follows:

(4a) degree of belief reliability of node processing task is calculated_i,Represent that node i processes the one-tenth of task Merit number of times, usesRepresent the total degree of node i accumulated process task, then the degree of belief of node i process task can be expressed as ${\mathrm{reliability}}_i=\frac{t_{\mathrm{reliability}}^i}{t_{\mathrm{total}}^i},{\mathrm{reliability}}_i\∈\left[\mathrm{0,1}\right];$

(4b) calculate residue energy of node and compare residual_i,Represent the dump energy of node i,Represent joint The primary power of some i, node i dump energy compares residual_iIt it is exactly the dump energy primary power with node i of node i Ratio, it may be assumed that

(4c) service ability of node is calculated: node i service ability can be expressed as by equation below ${\mathrm{capacity}}_i={\mathrm{\αreliability}}_i+{\mathrm{\βresidual}}_i=\mathrm{\α}\frac{t_{\mathrm{reliacility}}^i}{t_{\mathrm{total}}^i}+\mathrm{\β}\frac{e_{\mathrm{residual}}^i}{e_{\mathrm{initial}}^i},$ Wherein alpha+beta=1, α >=0, β >=0.

In above-mentioned steps five bunch in member node categorizing process as follows:

(5a) in general bunch, member node is divided into three classifications, and a category node is to have service ability, and service ability is strong Node set, two category nodes are to have service ability, and the node set that service ability is general, and three category nodes are temporarily without service energy The node set of power；

(5b) node classification initializes: if residue energy of node ratio is less than setting threshold value r, then it is assumed that node is temporarily without clothes Business ability, this category node returns the node that is divided three classes；In bunch, in addition to three category nodes, remaining node is considered there is service ability node, Calculating has service ability node serve ability and sorts, and having the node division of a% before in service ability node is a category node, a ∈ [0,100], the node division outside a category node, three category nodes is two category nodes；If there being service ability node number to be less than Or equal to two, then will have service ability node universal formulation is a category node.

(5c) node classification dynamically adjusts: after each layer of task has processed, will to bunch in member node classification weight New division, calculates residue energy of node ratio, is saved as three classes less than the node division setting threshold value r by residue energy of node ratio Point, calculates and has service ability node serve ability and sort, be divided into respective classes according to its service ability.

In above-mentioned steps three, the evaluation procedure of the service ability size of periphery one jumping bunch is as follows:

(6a) bunch service ability and a category node, the quantity of two category nodes, a category node, the service ability of two category nodes, Bunch dump energy ratio is relevant.

(6b) assume that the service ability of bunch cluster (k) interior joint i is expressed as capacity_i, one type node Number is N₁, two category node numbers are N₂, then a bunch capability list is shown as ${\mathrm{Capacity}}_{\mathrm{cluster}\left(k\right)}=({\mathrm{\ω}}_1\×N_1\×\underset{i\∈G1}{\mathrm{\Σ}}{\mathrm{capacity}}_i+{\mathrm{\ω}}_2\×N_2\×\underset{i\∈G2}{\mathrm{\Σ}}{\mathrm{capacity}}_i)\×\frac{R_{\mathrm{cluster}\left(k\right)}-r}{r}$ Wherein, G1 represents one Category node, G2 represents two category nodes, ω₁+ω₂=1, ω₁>=0, ω₂>=0, R_cluster(k)Represent the residual energy of bunch cluster (k) Amount ratio, bunch dump energy that r sets compares threshold value.

Concretely comprising the following steps of composition cooperative cluster in above-mentioned steps three:

(7a) ground floor task distributes to center cluster, while ground floor task processes, carries out second layer task distribution；

If (7b) dump energy of center cluster is than less than setting threshold value, then in being no longer allocated to task unallocated in layer The heart bunch, center cluster jumps a bunch request cooperation to periphery one, and periphery one is jumped bunch according to self situation, bunch provides centered by choosing whether Service, if periphery one jumps bunch dump energy, ratio is less than or equal to set threshold value, then refusal provides service, periphery one jumping bunch residue Energy than more than setting threshold value, then can provide service, and self bunch service ability value is returned to center cluster；

If being (7c) n with layer unallocated task number, in periphery one is jumped bunch, select bunch service ability bunch making at front n For alternative cooperative cluster, n task is assigned in n alternative cooperative cluster, it is ensured that the dump energy of each bunch is than difference as far as possible Little, and assign to task number on each bunch less than bunch in the number of a category node；

If (7d) number of center cluster periphery one jumping bunch is all made less than unallocated task number, the jumping bunch of the most all peripheries one For alternative cooperative cluster；

If (7e) center cluster periphery one jumping bunch all refuses to provide service, then center cluster periphery two jumps a bunch transmission message, asks Ask periphery two to jump bunch assistance to complete；

(7f) after this layer of task is assigned to bunch, bunch head in task being assigned to bunch, member node performs, and carries out one simultaneously Layer task distribution, until all tasks are assigned.

In above-mentioned steps four, untreated task is distributed to the specific requirement of cooperative cluster by center cluster: have n task task₁, task₂,…,task_n, m cooperative cluster clouster₁,clouster₂,…,clouster_mComplete this n task, kth bunch The energy expenditure completing this n task is respectively C_k1,C_k2,…,C_kn, and the dump energy of m cooperative cluster is than respectively R₁, R₂,…,R_m, the primary power of each bunch is respectivelyN task is distributed to m cooperative cluster Complete so that it is less, to reach the energy balance than difference that each cooperative cluster completes the dump energy after task；

Assuming that each task is only completed by a cooperative cluster, a cooperative cluster can complete multiple task；

x_ij=1 represents that jth task is completed by i-th cooperative cluster, x_ij=0 represents that jth task is not complete by i-th bunch Become, C_ijRepresent that i-th cooperative cluster completes the energy expenditure of jth task, i=1,2 ..., m, j=1,2 ..., n, make X= (x_ij)_m×nMatrix for logger task distribution condition；

Then completing this n task, the energy expenditure of cooperative cluster is

$\left(\begin{array}{c}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j}\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j}\\ \·\\ \·\\ \·\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}}\end{array}\right),$

The dump energy of cooperative cluster is

$\left(\begin{array}{c}{R}_1{E}_{\mathrm{initial}}^1\\ R_2{E}_{\mathrm{inntial}}^2\\ \·\\ \·\\ \·\\ R_m{E}_{\mathrm{initial}}^m\end{array}\right)-\left(\begin{array}{c}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j}\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j}\\ \·\\ \·\\ \·\\ \underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}}\end{array}\right)=\left(\begin{array}{c}{R}_1{E}_{\mathrm{initial}}^1-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j}\\ R_2{E}_{\mathrm{initial}}^2-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j}\\ \·\\ \·\\ \·\\ R_m{E}_{\mathrm{initial}}^m-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}}\end{array}\right),$

Bunch dump energy ratio be

$\left(\begin{array}{c}(R_1{E}_{\mathrm{initial}}^1-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{1j}{C}_{1j})/E_{\mathrm{initial}}^1\\ (R_2{E}_{\mathrm{initial}}^2-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{2j}{C}_{2j})/E_{\mathrm{initial}}^2\\ \·\\ \·\\ \·\\ (R_m{E}_{\mathrm{initial}}^m-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{mj}}{C}_{\mathrm{mj}})/E_{\mathrm{initial}}^m\end{array}\right),$

The average residual energy ratio of m cooperative cluster is approximately

$R_0=(\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{R}_i{E}_{\mathrm{initial}}^i-\underset{i=1}{\overset{m}{\mathrm{\Σ}}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{C}_{\mathrm{ij}}/m)/\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{E}_{\mathrm{initial}}^i$

Seek x_ijMake

The least, wherein $b_i=R_i-R_0,a_{\mathrm{ij}}=C_{\mathrm{ij}}/E_{\mathrm{initial}}^i;$

I.e. solve following one-zero programming problem

$\underset{x_{\mathrm{ij}}}{\min }\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{[b_i-\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{a}_{\mathrm{ij}}{x}_{\mathrm{ij}}]}^2$

$s.t.\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{x}_{\mathrm{ij}}=1,j=\mathrm{1,2},...,n,$

x_ij∈{0,1}

By the solution of above-mentioned optimization problem, can obtain distributing to n task the task distribution side that m cooperative cluster completes Case, and the dump energy after making each cooperative cluster complete task differs the least than as far as possible, thus ensure balancing energy.

In above-mentioned steps two, what task in selected layer was allocated by center cluster bunch head concretely comprises the following steps:

(9a) subtask is allocated by the order that center cluster bunch head arrives according to task, is sent by mission bit stream to be allocated To a category node；

(9b) after a category node receives message, check task description, decide whether to process according to self-energy consumption degree This task, if energy expenditure degree exceedes setting threshold value, then refusal provides service, if energy expenditure degree is not less than setting threshold Value, then provide service, and calculated this task estimation time, estimated energy consumption, these two information are returned bunch heads；

(9c) bunch head is according to a category node return information, selects the node of energy expenditure minimum as tasks carrying node；

If (9d) category node all refuses to provide service, then bunch head sends message to two category nodes, in two category nodes Select service available node.

The above, be only presently preferred embodiments of the present invention, and the present invention not makees any pro forma restriction, though So the present invention is disclosed above with preferred embodiment, but is not limited to the present invention, any technology people being familiar with this specialty Member, in the range of without departing from technical solution of the present invention, when the technology contents of available the disclosure above makes a little change or modification For the Equivalent embodiments of equivalent variations, as long as being the content without departing from technical solution of the present invention, the technical spirit of the foundation present invention Any simple modification, equivalent variations and the modification being made above example, all still falls within the range of technical solution of the present invention.

Claims (9)

1. complex task collaborative service method in a wireless sensor network based on sub-clustering, it is characterised in that: its step is such as Under:

Step one: task is layered, carries out layering according to task DAG figure to it and divides, and the most successively distributes during task distribution, Every layer is distributed some tasks；

Step 2: after the event of task occurs, incident point place bunch as center cluster, bunch head of center cluster is by task in layer It is allocated；

Step 3: when bunch dump energy of center cluster is than less than setting threshold value, then jump a bunch transmission solicited message to periphery one, if Periphery one is jumped bunch dump energy and is not less than threshold value than, then accept request, as the cooperative cluster of center cluster；Described center cluster and periphery One jumps a bunch formation cooperation cluster, and bunch head of periphery one jumping bunch accepts the arrangement of center cluster bunch head；

Step 4: untreated task is distributed to cooperative cluster by center cluster, ensures the balancing energy of each cooperative cluster during distribution, residue Energy is than difference minimum；

Step 5: in cooperative cluster assigns the task to bunch, member node processes.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: in described step one, that task carries out the process that layering divides is as follows for DAG figure according to task: entrance task is drawn Being divided into ground floor, if a certain subtask forerunner's maximum layer is k, then this subtask is divided into k+1 layer.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: described step 3 judging, center cluster accepts after new task dump energy than whether less than the calculating setting threshold value Process is as follows:

(3a) calculate bunch present energy and consume C_cluster(i): bunch present energy consumption refers to the energy of certain bunch of current existing task Amount consumes sum；

(3b) bunch total surplus energy is calculatedBunch total surplus energy be bunch in the dump energy sum of all nodes；

(3c) bunch total primary power is calculatedBunch total primary power be bunch in the primary power sum of all nodes；

(3d) calculate bunch dump energy and compare R_cluster(i): a bunch dump energy ratio is that total surplus energy deducts present energy and disappears in this bunch Consume the ratio with bunch total primary power,。

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: in described step 5 bunch in the evaluation procedure of member node service ability as follows:

(4a) degree of belief reliability of node processing task is calculated_i,Represent that node i processes the success time of task Number, usesRepresent the total degree of node i accumulated process task, then the degree of belief of node i process task is expressed asreliability_i∈[0,1]；

(4b) calculate residue energy of node and compare residual_i,Represent the dump energy of node i,Represent at the beginning of node i Beginning energy, node i dump energy compares residual_iIt is exactly the dump energy ratio with the primary power of node i of node i, it may be assumed that

(4c) service ability of node is calculated: node i service ability equation below is expressed asWherein alpha+beta=1, α >=0, β >=0.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: in described step 5 bunch in member node categorizing process as follows:

(5a) in general bunch, member node is divided into three classifications, and a category node is to have service ability, and the node that service ability is strong Set, two category nodes are to have service ability, and the node set that service ability is general, and three category nodes are temporarily without service ability Node set；

(5b) node classification initializes: if residue energy of node ratio is less than setting threshold value r, then it is assumed that node is temporarily without service energy Power, this category node returns the node that is divided three classes；In bunch, in addition to three category nodes, remaining node is considered there is service ability node, calculates Having service ability node serve ability and sort, having the node division of a% before in service ability node is a category node, a ∈ [0, 100], the node division outside a category node, three category nodes is two category nodes；If have service ability node number less than or etc. In two, then will have service ability node universal formulation is a category node；

(5c) node classification dynamically adjusts: after each layer of task has processed, will to bunch in member node classification again draw Point, calculate residue energy of node ratio, by residue energy of node ratio less than setting the node division of threshold value r as three category nodes, meter Calculation has service ability node serve ability and sorts, and is divided into respective classes according to its service ability.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: in described step 3, the evaluation procedure of the service ability size of periphery one jumping bunch is as follows:

(6a) bunch service ability and a category node, the quantity of two category nodes, a category node, the service ability of two category nodes, bunch surplus Complementary energy ratio is relevant；

(6b) assume that the service ability of bunch cluster (k) interior joint i is expressed as capacity_i, one type saves Point number is N₁, two category node numbers are N₂, then a bunch capability list is shown asWherein, G1 represents one Category node, G2 represents two category nodes, ω₁+ω₂=1, ω₁>=0, ω₂>=0, R_cluster(k)Represent the residual energy of bunch cluster (k) Amount ratio, r is that bunch dump energy set compares threshold value.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: concretely comprising the following steps of the composition cooperative cluster in described step 3:

(7a) ground floor task distributes to center cluster, while ground floor task processes, carries out second layer task distribution；

If (7b) dump energy of center cluster is than less than setting threshold value, then it is no longer allocated to center with task unallocated in layer Bunch, center cluster is to periphery one jumping bunch request cooperation, and periphery one jumping bunch is according to the situation of self, and centered by choosing whether, a bunch offer takes Business, if periphery one jumps bunch dump energy, ratio is less than or equal to set threshold value, then refusal provides service, periphery one jumping bunch residual energy Amount ratio more than setting threshold value, then provides service, and self bunch service ability value is returned to center cluster；

If being (7c) n with layer unallocated task number, in periphery one is jumped bunch, select bunch conduct at front n of bunch service ability Alternative cooperative cluster, is assigned to n task in n alternative cooperative cluster, it is ensured that the dump energy of each bunch is less than difference, And assign to task number on each bunch less than bunch in the number of a category node；

If (7d) number of center cluster periphery one jumping bunch is less than unallocated task number, the jumping bunch of the most all peripheries one is all as standby Select cooperative cluster；

If (7e) center cluster periphery one jumping bunch all refuses to provide service, then center cluster periphery two jumps a bunch transmission message, asks week Limit two is jumped bunch assistance and is completed；

(7f) after this layer of task is assigned to bunch, bunch head in task being assigned to bunch, member node performs, and carries out next layer simultaneously Task is distributed, until all tasks are assigned.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 7, It is characterized in that: untreated task is distributed to the specific requirement of cooperative cluster by the center cluster in described step 4: have n task task₁,task₂,…,task_n, m cooperative cluster clouster₁,clouster₂,…,clouster_mComplete this n task, I-th bunch completes the energy expenditure of this n task and is respectively C_i1,C_i2,…,C_in, and the dump energy score of m cooperative cluster Wei R₁,R₂,…,R_m, the primary power of each bunch is respectivelyN task is distributed to m association Work bunch completes so that it is less, to reach the energy balance than difference that each cooperative cluster completes the dump energy after task；

Assuming that each task is only completed by a cooperative cluster, a cooperative cluster completes multiple task；

x_ij=1 represents that jth task is completed by i-th cooperative cluster, x_ij=0 represents that jth task is not completed by i-th bunch, C_ij Represent that i-th cooperative cluster completes the energy expenditure of jth task, i=1,2 ..., m, j=1,2 ..., n, make X=(x_ij)_m×nFor The matrix of logger task distribution condition；Then completing this n task, the energy expenditure of cooperative cluster is

The dump energy of cooperative cluster is

Bunch dump energy ratio be

The average residual energy ratio of m cooperative cluster is approximately

Seek x_ijMake

The least, wherein b_i=R_i-R₀,

I.e. solve following one-zero programming problem

x_ij∈{0,1}

By the solution of above-mentioned optimization problem, obtain distributing to n task the task allocative decision that m cooperative cluster completes, and make Obtain the dump energy after each cooperative cluster completes task and differ the least than as far as possible, thus ensure balancing energy.

Complex task collaborative service method in a kind of wireless sensor network based on sub-clustering the most according to claim 1, It is characterized in that: in described step 2, what task in selected layer was allocated by center cluster bunch head concretely comprises the following steps:

(9a) subtask is allocated by the order that center cluster bunch head arrives according to task, and mission bit stream to be allocated is sent to one Category node；

(9b) after a category node receives message, check task description, decide whether to process this according to self-energy consumption degree Business, if energy expenditure degree exceedes setting threshold value, then refusal provides service, if energy expenditure degree is not less than setting threshold value, then Service is provided, and has calculated this task estimation time, estimated energy consumption, these two information are returned bunch heads；

(9c) bunch head is according to a category node return information, selects the node of energy expenditure minimum as tasks carrying node；

If (9d) category node all refuses to provide service, then bunch head sends message to two category nodes, selects in two category nodes The node of service is provided.