CN103095577A - Context-dependent non-uniform clustering routing algorithm - Google Patents

Abstract

The invention discloses a context-dependent non-uniform clustering routing algorithm. According to the context-dependent non-uniform clustering routing algorithm, the defect that clustering and route are isolated from each other by a traditional hierarchical routing model is overcome, information at a clustering stage is applied to a routing formation stage, set-up time of a network is largely shortened, and energy consumption of the network is reduced. Based on a clustering algorithm of a vote mechanism, the context-dependent non-uniform clustering routing algorithm optimizes energy consumption of an area, a time-sharing reversed flooding algorithm is used, and energy consumed by route discovery is largely saved. The context-dependent non-uniform clustering routing algorithm can be effectively applied to large-scale area monitoring, energy load balancing of the context-dependent non-uniform clustering routing algorithm is confirmed, node behavior is stable, and network lifetime is much longer.

Description

Context-sensitive Uneven Cluster routing algorithm

Technical field

The present invention relates to a kind of Uneven Cluster routing algorithm, relate in particular to a kind of context-sensitive Uneven Cluster routing algorithm.

Background technology

In the world to the showing great attention to of " Internet of Things ", " Internet of Things " was described as the information technology revolution for the third time after computer and the Internet, became the commanding elevation of future science in recent years.The upsurge that has started research and used, 2006, China WSN as great special project, first develop theme, Disciplinary Frontiers is put into " national medium-term and long-term science and technology development plan (2006-2020) ".The U.S. classified Internet of Things and new forms of energy as revitalize the economy two large emphasis in 2009, and " the wisdom earth " this theory proposed first, the same year, China proposed " perception China ", and formally classify Internet of Things one of as country's five large emerging strategic industries, and write " government work report ", cause the very big concern of the whole society.

Compare with traditional the Internet, Internet of Things has cognition technology, the foundation distinct features such as Ubiquitous Network, Intelligent treatment and control in real time on the internet of extensive use.Its objective is and realize thing and thing, thing and people, all article and the interconnection of network, convenient identification, management and control.And wireless sensor network is the senior application of " Internet of Things ", is again wherein indispensable key technology.

Wireless sensor network (WSN) combines sensor technology, microelectric technique, embedded technology, wireless communication technology and distributed information processing, the information of Real Time Monitoring, perception and collection monitored target synergistically, by data fusion with after processing, send the user who needs these information to, thereby realize remote real-time monitoring.Wireless sensor network has following characteristics: limited node energy, wireless self-organization network, low power consumption and low cost are disposed, data-centered and application relativity.Therefore, for different application demands, develop focus and difficult point that corresponding communication protocol becomes current research.

The life cycle of utilizing the existing energy to extend network is one of wireless sensor network key technology.The form of node organization cluster can be reduced effectively the energy consumption of network.Cluster algorithm is to implement the important method that the layering route adopts in wireless sensor network.WCA has proposed a kind of Weighted Clustering Algorithm of considering the many factors combination for Ad Hoc network.But this algorithm is for AD Hoc network, and Selecting All Parameters is more difficult.In the cluster algorithm of wireless sensor network, the most representative is low power consumption adaptive clustering route protocol (LEACH).Although it is compared with general plane multi-hop Routing Protocol, network lifecycle can be extended 15%,, the ratio that the network cluster head of expectation accounts for all nodes has determined every sub-clustering number of taking turns.In actual applications, determining of this optimal proportion is very difficult, and this is relevant with network size and node density, usually gets fixed number 5% or 10%.In the HEED cluster algorithm, two parameters of selection Main Basis primary and secondary of bunch head, principal parameter is dump energy, the node that has than multi-energy becomes a bunch head with larger probability; Subparameter be bunch in communication cost.Experiment shows, the sub-clustering speed of HEED is fast, can produce distribute substantially uniform bunch head and rational network topology.But each node energy consumption of this algorithm is unbalanced, and the routing policy energy is optimized not.

Due to layering route in wireless network, can bear more transfer of data near the node of base station, therefore, energy consumption is faster, thereby causes faster near the node death in the zone of base station, and we become this phenomenon is the energy cavity.Divide even Clustering Routing (EEUCR) to propose a kind of clustering algorithm based on the competition radius for this phenomenon, alleviated this phenomenon.But the cluster algorithm of EEUCR still uses the method for random selection, does not consider local message, and this makes sub-clustering efficient optimize not.The probe response pattern that its routing algorithm uses, this is not obviously a kind of routing algorithm of stability and high efficiency.

Therefore need a kind of Clustering Routing of not only can the consumption of balancing network energy but also can extend network lifetime to solve above-mentioned problem.

Denomination of invention: based on the wireless sensor network topology control method of non-uniform sections, application number: 200810035214, the sub-clustering stage of this patent, although used Uneven Cluster, the not explanation of specifying information exchange details patent, and to choose a small bundle of straw, etc. for silkworms to spin cocoons on head be only according to the position but not with reference to energy; The route stage, use reverse route, send explore frame and find route, do like this and can not determine effective broadcast distances, can not determine to send the not response later on of this information, in this patent, the stability of disclosed Routing Protocol is inadequate.And this algorithm has only been done a minute evenly sub-clustering processing for the empty problem of energy, the route aspect does not process, so be very fast for a small bundle of straw, etc. for silkworms to spin cocoons on head energy consumption near the base station, and, the replacing of a small bundle of straw, etc. for silkworms to spin cocoons on head is just according to positional information, balancing energy is inadequate like this, can not guarantee the connectedness of whole route.

Denomination of invention: a kind of wireless sensor network routing method, application number: 200910039496, in this patent: 1. produce random number when sub-clustering and threshold value T relatively determines whether becoming a small bundle of straw, etc. for silkworms to spin cocoons on head, but the formation of a small bundle of straw, etc. for silkworms to spin cocoons on head is completely random, therefore there is drawback in cluster-dividing method, the informixs such as energy and position can not be got up consider; 2. the obvious energy consumption of routing algorithm is inhomogeneous,, all data to the base station all will be passed through main a small bundle of straw, etc. for silkworms to spin cocoons on head, and just use this standard, the link of use that repeats is easy to cause the too early death of part of nodes, but also needs to sound out energy consumption.

Denomination of invention: a kind of wireless sensor network Uneven Cluster method for routing of high energy efficiency, application number: 201110131502, cluster algorithm in this patent is not considered energy and the resource that embedded node is limited, use this cluster algorithm, the information of sub-clustering is not fully utilized, but sends again broadcast message when route, and if reach so complicated control logic, need a large amount of Information Statistics and node to calculate, these all need a large amount of energy consumption undoubtedly.The agreement of this application only is concerned about analog result, finds a balance point between complexity and energy consumption, and the control of the complex logic in this patent, is that race is obstructed in actual conditions.

Summary of the invention

Purpose of the present invention is exactly in order to address the above problem, and a kind of context-sensitive Uneven Cluster routing algorithm is provided, and it has realizes simply efficiently effectively saving energy, extends network lifetime, saves manpower financial capacity's advantage.

To achieve these goals, the present invention adopts following technical scheme:

A kind of context-sensitive Uneven Cluster routing algorithm comprises sub-clustering stage and route stage, and concrete steps are:

Step 1: all nodes are with d ₀For radius is broadcasted once, d ₀Be radio energy consumption threshold value; After each node is received respective broadcast information, calculate this node weights and to the ballot value of neighbor node; The ballot value of each node broadcasts oneself, neighbor node are received the final gained vote of ballot value calculating later on oneself;

Step 2: if the ballot value of node is maximum in subrange, to select oneself be a bunch head to node so, and with oneself information, comprise position, dump energy, the competition radius of oneself, broadcasts away with the competition radius of oneself by control frame;

Step 3: if receive the node bunch head of control frame, this bunch head joins corresponding information in the middle of higher level or subordinate's list; If receiving the node of control frame is ordinary node, select an optimum bunch head to add according to fitness function and distance, and the information that election is withdrawed from transmission has comprised a bunch position to a bunch head, dump energy and competition radius;

Step 4: receive and withdraw from election information but the node that does not enter bunch, the information of the respective cluster head that record is received, comprise a bunch position, dump energy, the last neighbor node ballot value of these node utilizations increases the node weights of oneself, makes to have larger priority become a bunch head in the election of next round;

Step 5: repeating step one is to step 6, until all nodes become bunch head or bunch member;

Step 6: whether the inspection oneself of all bunches head has higher level's list, if higher level's list is empty, so just with n competition radius broadcasting doubly once, n is the real number greater than 1;

Step 7: with base station distance less than bunch head of threshold value DIS_DIRECTTO_BASE directly and base station communication, own check is indicated be set to 1, be set to 1 to the jumping figure of base station, calculate the energy of base station; Then select the member of a maximum apart from broadcasting once by control frame as the broadcasting radius from the subordinate's list of oneself, control frame comprised the position, dump energy of bunch head, to the energy of base station needs, to the jumping figure of base station, threshold value DIS_DIRECTTO_BASE is that bunch head is to the distance of base station;

Step 8: the state of receiving bunch head inspection oneself of the control frame in step 9;

step 9: bunch head in step 10 checks the higher level's list of oneself, if all higher level's lists had all sent control information, this bunch head arranges the check flag bit 1 of oneself so, and broadcasting ROUTE_FORM_MSG control frame, after all higher level's lists of this a small bundle of straw, etc. for silkworms to spin cocoons on head that and if only if having all sent ROUTE_FORM_MSG, bunch head is revised the inspection position of oneself, then send the ROUTE_FORM_MSG control frame, the ROUTE_FORM_MSG control frame comprises the sign of this bunch head, the position of this bunch head, the dump energy of this bunch head, this bunch head transmits data to the energy that the base station consumes, this bunch head is to the needed jumping figure in base station,

Step 10: repeating step eight is to step 11 until the end of the time period of route;

Step 11: if bunch head finishes still there is no higher level's list to the route stage, this bunch head directly and base station communication so.

In described step 2, the competition radius is:

$R_i=(1-c\×\frac{d_{\max }-d(v_i,\sin k)}{d_{\max }-d_{\min }})R_{\max }$

Wherein, d (V _i, Sink) be V _iTo the distance of base station Sink, d _minThat wireless sensor node is apart from the minimum distance of base station Sink, d _maxThe maximum distance of nodal point separation base station Sink, R _maxOptimal value between 1/2d ₀– 2/3d ₀Between, the optimal value of c is between 0.2-0.4.

In described step 1, the step of node weights is:

(1-1) determine node v _iThrowing is to its a certain neighbor node v _jBallot paper v (v _i, v _j),

$v(v_i,v_j)=\left\{\begin{array}{cc}\frac{e_j}{\underset{d_{\mathrm{ik}}\≤R_i}{\mathrm{\Σ}}{e}_k}& d_{\mathrm{ij}}\≤R_i\\ 0& d_{\mathrm{ij}}>R_i\end{array}\right.$

Wherein, d _ikBe node v _iTo v _kDistance, v _kBe node v _iAll neighbours, R _iBe node v _iThe competition radius, e _jBe node v _jDump energy, e _kBe one of them neighbor node energy, d _ijFor node i arrives node j distance;

(1-2) determine node v _iNumber of ballot papers, node v _iNumber of ballot papers be all neighbours' ballot papers and be:

$\mathrm{vote}\left(v_i\right)=\underset{d_{\mathrm{ij}}\≤R_i}{\mathrm{\Σ}}v(v_j,v_i);$

(1-3) v _iCombining weights I _iFor:

$I_i=\frac{\mathrm{vote}\left(v_i\right)}{P_i}$

Wherein, P _iNode v _iTo the average distance of its all neighbor nodes, if this node without neighbours, weight I so _iAssignment is 1.

If in described step 6, bunch head checks that oneself is empty to the higher level's list of oneself, so just broadcasts once with the competition radius of twice.

The concrete steps of described step 8 are:

If (8-1) check be 0 and the source node of this frame be higher level's list member of this bunch head, so just upgrade corresponding higher level's list entry, dump energy is the dump energy in control frame, being that jumping figure in control frame adds 1 to the jumping figure of base station, is to send data to the energy that source bunch head adds to be needed to the base station in this control frame to the energy of base station;

If (8-2) check is 0, and the source node of this frame creates a list item so not in higher level's list;

If (8-3) check is 1, directly ignore this control frame.

The technical solution used in the present invention is: by to evenly spreading 300 points in a 500*500, repeatedly sampling, make it approach extensive monitoring system in reality (comprising to field-crop forest monitoring etc.), base station (sink) is positioned at the edge of monitoring range.For each sample (suppose carry out be periodic transfer of data), add up respectively bunch number that each is taken turns, the average load of each bunch head, the energy consumption that each is taken turns, wheel number when first node is dead, the energy curves of node, and come the quality of critic network model by foregoing.Specifically comprise following model:

(1). energy consumption model

The present invention adopts single order wireless transmission model (First Order Radio Model), and radio transmitters sends the energy consumption of kbits packet:

$E_{\mathrm{Tx}}(k,d)=E_{\mathrm{Tx}-\mathrm{elec}}\left(k\right)+E_{\mathrm{Tx}-\mathrm{amp}}(k,d)$

$=\left\{\begin{array}{cc}{\mathrm{kE}}_{\mathrm{elec}}+{\mathrm{k\&epsiv;}}_{\mathrm{fs}}{d}^2,& d<d_0\\ {\mathrm{kE}}_{\mathrm{elec}}+{\mathrm{k\&epsiv;}}_{\mathrm{mp}}{d}^4,& d\&GreaterEqual;d_0\end{array}\right.$

ETx represents that data send the energy that consumes, by starting ENERGY E _Tx-elecWith the transmission ENERGY E _Tx-amp(k, d) forms, and ETx-elec only depends on the size of transmitted data amount and ETx-amp (k, d) depends on the size of transmitted data amount and a long and short two factor of transmitting range.Wherein, Eelec is the energy consumption coefficient of wireless transceiver operation circuit, and k represents the size of the data volume that transmits, and d represents the distance that transmits, d ₀The expression radio energy consumption threshold value and ε _fsAnd ε _mpThe power amplification coefficient power amplification ratio that represents respectively two kinds of channel energy consumption models.The implication of above-mentioned formula is for working as transmitting range less than d ₀The time, the transmitter energy consumption model adopts free space (Free Space) model, transmission energy consumption and d ²Be directly proportional; When transmitting range is not less than d ₀The time, model adopts multichannel decay (Multi-path Fading) model, transmission energy consumption and d ⁴Be directly proportional.

Wireless receiver receives the energy consumption of kbits packet:

E _Rx(k)＝E _Rx-elec(k)=kE _elec

ERx represents the energy that data receiver consumes, and only depends on the size of transmitted data amount.

Energy consumption model is under perfect condition, establishes

Eelec=50nJ/bit，ε _fs=10pJ/(bit·m2)，ε _mp=0.0013pJ/(bit·m4)。

(2). the energy consumption model analysis of multi-hop transmission

Long-range Data Transmission, namely d is larger, E _TxSize mainly by E _Tx-amp(k, d) decides, and energy consumption is mainly derived from data and sends.

E _Tx(k,d)＝E _Tx-elec(k)+E _Tx-amp(k,d)=kE _elec+kε _mpd ⁴(d≥d ₀)

Suppose d ₀=50m, data transmitting range d=100m, radio transmitters sends the packet of 1bit, gets E according to energy consumption model _Tx-amp(1,100)=1300nJ, E _Tx-elec=50nJ.Transmission range is larger, d ⁴* E _Tx-ampEffect is more obvious, E _Tx-elecCan be left in the basket.In true environment, corresponding to remote wireless transmission, when data send, energy consumption is:

$E_{\mathrm{Tx}}(k,d)=E_{\mathrm{Tx}-\mathrm{elec}}\left(k\right)+E_{\mathrm{Tx}-\mathrm{amp}}(k,d)\&ap;E_{\mathrm{Tx}-\mathrm{amp}}(k,d)$

$=\left\{\begin{array}{cc}{\mathrm{k\&epsiv;}}_{\mathrm{fs}}{d}^2,& d<d_o\\ {\mathrm{k\&epsiv;}}_{\mathrm{mp}}{d}^4,& d\&GreaterEqual;d_o\end{array}\right.$

For the near radio transmission, multi-hop transmission is not necessarily saved energy, because multi-hop transmission need repeatedly transmit and receive data.Yet for remote wireless transmission, multi-hop transmission can be saved energy, improves energy efficiency.

When an a relaying bunch C and formed auxiliary angle were arranged between a bunch A and BS greater than 90 °, multi-hop transmission can be saved energy, and main energy consumption is distributed in 2 leader cluster nodes.

Beneficial effect of the present invention: the present invention uses the cluster algorithm based on voting apparatus system, has optimized regional energy consumption.Preserve the effective information of sub-clustering, reduced the information exchange that route forms.Use the reverse flooding of timesharing, greatly saved the energy that route discovery consumes.Can effectively be applied to large-scale area monitoring, and confirmed that its energy load is balanced, and nodes ' behavior is stable, and network lifetime is longer, the agreement that the present invention proposes is a kind of Data Transport Protocol remote, stationary nodes that is applicable to, and can be applied to medical monitoring and environmental monitoring.

Description of drawings

Fig. 1 is demonstrating model of the present invention;

Fig. 2 routing algorithm flow chart of the present invention.

Fig. 3 is the schematic diagram of route stage routing algorithm in the present invention;

Embodiment

The invention will be further described below in conjunction with accompanying drawing and embodiment.

Any wireless sensor network protocols all has application relativity, has concrete application background, the CRUCR agreement (Context Related Unequal-Clustering Routing protocol for WSNs) that the present invention proposes is a kind of Data Transport Protocol remote, stationary nodes that is applicable to, and can be applied to medical monitoring and environmental monitoring.The applicable elements of New Deal CRUCR is:

(a) supposition BS energy can replenish at any time; Each node is known the position of BS, and the position of node can obtain (such as GPS, RSSI etc.) by the base station by some algorithm.

(b) in network, all the sensors node physical configuration parameter is consistent, and primary power is identical, and cloth postpone physical location is constant.The data volume that sensor node need transmit is less but transfer ratio is more frequent, and each node has unique number and knows geographical position separately by navigation systems such as GPS, with two-dimensional coordinate＜X, and Y〉expression.Numbering, coordinate information are stored in the RAM of node;

(c) each transmission node position is that fix and far away with the communication distance of BS, transmission situation more complicated, and as complicated in geographical situation, obstacle is more;

(d) New Deal belongs to network layer protocol, and in agreement discussion and experimental situation, hypothesis has desirable physical layer and data link layer support.

The below will provide detailed implementation of the present invention, be divided into two parts and specifically describe:

(1) the Uneven Cluster algorithm of optimizing

In a square surveyed area, N sensor node of random evenly deployment, each sensor node calculates respectively itself and the distance between neighbor node on every side, and calculation base station Sink and each internodal distance, after evaluation work is completed, base station Sink comprises base station Sink to the longest distance d of nodes to whole Web broadcast a piece of news in message _maxWith beeline d _minThese two distance measurements are used for calculating the size of relevant bunch in the Uneven Cluster algorithm.

Comprise a bunch of head and some general sensor nodes for information gathering in each bunch, less the closer to bunch radius of base station Sink, bunch member is fewer, otherwise can have larger bunch radius and bunch member.Wherein the criterion of neighbor node depends on bunch a competition radius R, and the nodes that node covers in the circle take R as radius is exactly the neighbor node number of this node, so the quality of R value directly relatively has influence on the quality of election bunch head.The selective dependency of R is in the threshold values d of network area size and radio energy consumption model ₀, R＜d is arranged usually ₀

For with the network Uneven Cluster, need to control the competition radius that the nodes competition becomes bunch head.If R _maxBe the maximum contention radius that presets in the network design front nodal point, node V _iThe competition radius R _iSize is a function of distance between itself and base station Sink:

$R_i=(1-c\&times;\frac{d_{\max }-d(v_i,\sin k)}{d_{\max }-d_{\min }})R_{\max }$

Wherein, d (V _i, Sink) be V _iTo the distance of base station Sink, c is a constant coefficient of 0 to 1, d _minThat wireless sensor node is apart from the minimum distance of base station Sink, d _maxIt is the maximum distance of nodal point separation base station Sink.Can find out from formula, the competition radius of node is from (1-c) R _maxTo R _maxThis is traditional Uneven Cluster algorithm, but this radius calculation mode is independently with the routing algorithm of back, for cluster algorithm can be combined better with routing algorithm, after the size that we adjust Rmax made last sub-clustering, each bunch head was at d ₀Scope in can find at least one adjacent bunch head, so just avoided with d ⁴Carry out transfer of data, thus the energy that between having reduced greatly bunch, transmission is used.Evidence, R _maxOptimal value between 1/2d ₀-2/3d ₀Between, the optimal value of c is between 0.2-0.4.

Node competes whether become a bunch head according to its weight separately.In the election process of bunch head, topological structure, dump energy and through-put power become three principal elements.We define a ballot paper function vote and represent sensor node topological structure on every side and the dump energy of itself, and a distance function P represents transmitting energy.

The importance of node should reflect by its all neighbours, and its ballot paper of neighbours is separately collected and calculated to node.The ballot paper of node accumulation is more, illustrates that it is more important in this localized network.In the cluster stage, the number of ballot papers that internodal competition is received based on them.A node calculates ballot paper with following rule: (a) node is 1 to its all neighbours' (comprising node itself) number of ballot papers; (b) neighbor node that has more dump energy obtains more ballot paper than the neighbor node that has less energy.Based on above-mentioned rule, node v _iThrowing is to its a certain neighbor node v _jBallot paper v (v _i, v _j) be:

$v(v_i,v_j)=\left\{\begin{array}{cc}\frac{e_j}{\underset{d_{\mathrm{ik}}\&le;R_i}{\mathrm{\&Sigma;}}{e}_k}& d_{\mathrm{ij}}\&le;R_i\\ 0& d_{\mathrm{ij}}>R_i\end{array}\right.$

D in formula _ikBe node v _iTo v _kDistance, v _kBe node v _iAll neighbours, R _iBe node v _iThe competition radius, e _jBe node v _jDump energy.

Node v _iNumber of ballot papers be all neighbours' ballot papers and:

$\mathrm{vote}\left(v_i\right)=\underset{d_{\mathrm{ij}}\&le;R_i}{\mathrm{\&Sigma;}}v(v_j,v_i)$

Computing node v _iCombining weights I _iFor:

$I_i=\frac{\mathrm{vote}\left(v_i\right)}{P_i}$

P in formula _iNode v _iAverage distance to its all neighbor nodes.If this node is without neighbours, weight weight assignment is 1 so.A node just can obtain these neighbours' position by the broadcast message that receives a certain neighbor node.Select the node of weight maximum to become a bunch head.

Number of members in the energy consumption of bunch head depends on bunch, for balanced energy consumption, node should be near a high-octane bunch of head.Define a fitness function fitness(v _i) this character of bunch head described:

$\mathrm{fitness}\left(v_i\right)=\frac{e_i}{{\mathrm{degree}}_i}$

Degree in formula _iBe node v _iDegree, e _iBe the dump energy of node i, when a node was in bunch radius of a plurality of bunches of heads, it selected the highest bunch head of fitness function value to add.When a bunch dump energy was identical, a bunch neighbour were more, and its fitness function value is less.

In order to calculate each neighbour's combining weights, node need to know its all neighbour dump energy and and neighbours between distance.Node broadcasts in network address ID separately, position and dump energy information.After receiving these information, each sensor node upgrades and broadcasts its all neighbour's weight.Fig. 2 has provided route flow chart of the present invention, if a node has a bunch head to exist at its sub-clustering radius with interior, we claim this node to be capped node so, otherwise are called not capped node.Clustering process is comprised of a series of iterative process.In iteration each time, if a node finds that it is the maximum node of number of votes obtained in all neighbor nodes, its selects own to bunch head so.After a node became bunch head, it can send one bunch broadcast message to its all neighbours.Remaining node can be waited for t ₂Be confirmed whether to also have other bunch head second in accepting scope.If a node has been capped, it can send a message that withdraws from election so, inform bunch head and neighbor node on every side, this message has comprised leader cluster node and position that node will add, so receive that bunch head of this message can know bunch member's number of oneself.If other bunches head is received this message, other bunches head just can be determined to have bunch head around oneself according to this message so, and corresponding information is joined in corresponding higher level or subordinate's list.Those have received that but the node that withdraws from election of cluster head message do not have to be capped can be with the information of bunch head, record, improved the ballot value of these nodes, when this makes the next round iteration, this node can more likely become a bunch head, and this has just guaranteed to exist at least two bunches of heads in certain scope.By this continuous iteration, during all nodes all can add bunch at last.

Although multi-hop transmission has effectively reduced the energy consumption of whole network, but due to the forwarding that can repeat near bunch head of base station from the information of neighbours' bunch head to the base station, so, node near the base station is understood the faster of death, this phenomenon is referred to as energy cavity (energy hole) usually, many researchs are verified, and the life span of network is shortened in energy cavity meeting.the common way that solves this phenomenon is generally to allow the node of close base station have more energy or to allow near the node density of base station larger, large-scale area monitoring due to what the present invention is directed to, node evenly distributes, therefore the present invention uses and increases the node that is connected with the base station, namely judge with a threshold value DIS_DIRECTTO_BASE node that directly is connected with the base station, as a rule threshold value DIS_DIRECTTO_BASE is larger, the node that directly is connected with the base station is more, the effect in energy cavity is more not obvious, but also to consider simultaneously and the base station between communication cost, so DIS_DIRECTTO_BASE is not too large, usually between d0-2d0.

Due to the sub-clustering stage control size of sub-clustering radius, so can prove at 2d0 (node is survived more than 50%) and must exist adjacent two bunches of heads proof as follows with interior:

Because in the time of sub-clustering, the sub-clustering radius of maximum is less than d0, suppose that a bunch of head forms, so in next round cluster process, receive that the node that withdraws from sub-clustering information and do not enter bunch has higher priority and becomes a bunch head, bunch head of next round must be in the scope of an epicycle bunch 2d0 so, iteration, therefore having guaranteed must have two bunches of heads to exist in the scope of 2d0, and we can adjust the sub-clustering radius and make it between 1/2d0-d0, have further guaranteed the correctness of this inference.

In the starting stage that forms connected graph, do not have bunch head of higher level's list once to broadcast with the radius of 2d0 in order, will form like this a connected graph, because we are optimized the sub-clustering radius, therefore each bunch head should be received one or more broadcast message, these broadcast messages have comprised the ID of bunch head, and bunch head is to the distance of base station, and the dump energy of bunch head.each bunch head can be safeguarded higher level's list and subordinate's list herein, these two lists can be configured according to practical application, but should comprise some minimum information, can represent higher level's list (ID with one polynary group, Dis, EngToBase, CurEng, Hop), ID represents the ID of superior node, Dis represents the distance of higher level's bunch head, EngToBase represents to arrive the needed least energy in base station through this higher level bunch head, CurEng represents this a small bundle of straw, etc. for silkworms to spin cocoons on head current remaining, Hop represents to arrive through this higher level bunch head the jumping figure of base station, these tuples can consist of higher level's list.Comparatively speaking, subordinate's list is simpler, as long as safeguard that (ID, Dis) is just passable, its implication is similar to higher level's list.Each node can be safeguarded a check sign, and after receiving all higher levels bunch head broadcasting, namely higher level's list is not after changing, and the check sign is set up, and this bunch head just can send the control message of connected graph to subordinate's list.Fig. 3 is the schematic diagram of routing procedure.

The time-sharing multiplex mechanism that the route formation stages uses, the below will provide detailed implementation procedure

1. all bunches head can check whether oneself has higher level's list, if do not have higher level's list so this bunch head can keep listening state.Between two bunches of heads according to judging to the distance of base station with to the vertical range of base station who is higher level's bunch head, for example A has received the broadcast control information of B, A can compare the distance to the base station of B, if B arrives the distance of base station and arrives the distance of base station less than A, and A arrives the vertical range of base station to the vertical range of base station greater than B, B is exactly higher level's bunch head of A so, and A can add B to come in higher level's list.B receives that the broadcast control information of A will join it in the middle of subordinate's list simultaneously.Through the broadcasting in sub-clustering stage, all nodes should be able to have one or more higher level bunch head (may have bunch head that there is no higher level's list, this situation is rarely found) and subordinate's bunch head (zero or a plurality of) basically.

2. bunch head can check that oneself is to the distance of base station, if distance is less than threshold value DIS_DIRECTTO_BASE, this bunch head just directly is connected with the base station so, the check sign of oneself is set, and the base station is joined higher level's list, calculate the needed energy in base station, jumping figure to the base station is 1, bunch head can check the subordinate's list of oneself, finds the distance list member of subordinate farthest, and sends control information as broadcasting radius take the distance to this member.Information comprises the ID of oneself, and oneself is to the needed energy in base station, and oneself is to the jumping figure of base station, and the dump energy of bunch head.An its subordinate bunch head can calculate and upgrade a series of data after receiving this broadcasting

(1). calculate and directly be sent to the needed energy in base station

(2). arrive the needed energy in base station by current higher level bunch head

(3). through the jumping figure of current higher level bunch head to the base station

Best higher level's bunch head can be selected according to time-delay or energy or residue energy of node, the present invention arrives base station institute energy requirement according to minimum and is used as best higher level's, when all superior nodes of this bunch head have all sent data, namely check all is set up, this bunch head just can arrange the check position so, and sends control information to subordinate's list.

Calculate the energy of base station:

(1) be to use a Communication Model of jumping in the scope of threshold value Dis_DIRECTo_BASE,

$E_{\mathrm{Tx}}(k,d)=E_{\mathrm{Tx}-\mathrm{elec}}\left(k\right)+E_{\mathrm{Tx}-\mathrm{amp}}(k,d)$

$=\left\{\begin{array}{cc}{\mathrm{kE}}_{\mathrm{elec}}+{\mathrm{k\&epsiv;}}_{\mathrm{fs}}{d}^2,& d<d_0\\ {\mathrm{kE}}_{\mathrm{elec}}+{\mathrm{k\&epsiv;}}_{\mathrm{mp}}{d}^4,& d\&GreaterEqual;d_0\end{array}\right.$

Energy consumption model is under perfect condition, establishes E _Elec=50nJ/bit, ε _fs=10pJ/ (bitm ²), ε _mp=0.0013pJ/ (bitm ⁴), if to the base station, being updated to the energy of base station so, multi-hop use following model to upgrade,

$\mathrm{EngToSink}\left(i\right)=\mathrm{EngToSink}\left(j\right)+{\mathrm{ke}}_{\mathrm{rx}}\left(j\right)+{\mathrm{ke}}_{\mathrm{rx}}\left(i\right)+\left\{\begin{array}{cc}{\mathrm{k\&epsiv;}}_{\mathrm{fs}}d{(i,j)}^2& d\&le;d_0\\ {\mathrm{k\&epsiv;}}_{\mathrm{mp}}d{(i,j)}^4& d>d_0\end{array}\right.$

EngToSink (j) is that a bunch j transfers data to the needed energy in base station, ke _rx(j) be that a bunch j accepts the needed energy of these data, ${\mathrm{ke}}_{\mathrm{tx}}\left(i\right)+\left\{\begin{array}{cc}{\mathrm{k\&epsiv;}}_{\mathrm{fs}}d{(i,j)}^2,& d(i,j)\&GreaterEqual;d_0\\ {\mathrm{k\&epsiv;}}_{\mathrm{mp}}d{(i,j)}^4,& d(i,j)<d_0\end{array}\right.$ Data are passed to bunch needed energy of a j.

3. repeating step 2, until all bunch heads all obtain best higher level.

4. if still have higher level's list to exist for empty bunch head when routing procedure finishes, this bunch head directly and base station communication so.

Be illustrated in figure 1 as demonstrating model of the present invention.

Although above-mentionedly by reference to the accompanying drawings the specific embodiment of the present invention is described; but be not limiting the scope of the invention; one of ordinary skill in the art should be understood that; on the basis of technical scheme of the present invention, those skilled in the art do not need to pay various modifications that creative work can make or distortion still in protection scope of the present invention.

Claims (5)

1. a context-sensitive Uneven Cluster routing algorithm, comprise sub-clustering stage and route stage, it is characterized in that, concrete steps are:

Step 1: all nodes are with d ₀For radius is broadcasted once, d ₀Be radio energy consumption threshold value; After each node is received respective broadcast information, calculate this node weights and to the ballot value of neighbor node; The ballot value of each node broadcasts oneself, neighbor node are received the final gained vote of ballot value calculating later on oneself;

Step 2: if the ballot value of node is maximum in subrange, to select oneself be a bunch head to node so, and with oneself information, comprise position, dump energy, the competition radius of oneself, broadcasts away with the competition radius of oneself by control frame;

Step 3: if receive the node bunch head of control frame, this bunch head joins corresponding information in the middle of higher level or subordinate's list; If receiving the node of control frame is ordinary node, select an optimum bunch head to add according to fitness function and distance, and the information that election is withdrawed from transmission has comprised a bunch position to a bunch head, dump energy and competition radius;

Step 4: receive and withdraw from election information but the node that does not enter bunch, the information of the respective cluster head that record is received, comprise a bunch position, dump energy, the last neighbor node ballot value of these node utilizations increases the node weights of oneself, makes to have larger priority become a bunch head in the election of next round;

Step 5: repeating step one is to step 6, until all nodes become bunch head or bunch member;

Step 6: whether the inspection oneself of all bunches head has higher level's list, if higher level's list is empty, so just with n competition radius broadcasting doubly once, n is the real number greater than 1;

Step 7: with base station distance less than bunch head of threshold value DIS_DIRECTTO_BASE directly and base station communication, own check is indicated be set to 1, be set to 1 to the jumping figure of base station, calculate the energy of base station; Then select the member of a maximum apart from broadcasting once by control frame as the broadcasting radius from the subordinate's list of oneself, control frame comprised the position, dump energy of bunch head, to the energy of base station needs, to the jumping figure of base station, threshold value DIS_DIRECTTO_BASE is that bunch head is to the distance of base station;

Step 8: the state of receiving bunch head inspection oneself of the control frame in step 9;

step 9: bunch head in step 10 checks the higher level's list of oneself, if all higher level's lists had all sent control information, this bunch head arranges the check flag bit 1 of oneself so, and broadcasting ROUTE_FORM_MSG control frame, after all higher level's lists of this a small bundle of straw, etc. for silkworms to spin cocoons on head that and if only if having all sent ROUTE_FORM_MSG, bunch head is revised the inspection position of oneself, then send the ROUTE_FORM_MSG control frame, the ROUTE_FORM_MSG control frame comprises the sign of this bunch head, the position of this bunch head, the dump energy of this bunch head, this bunch head transmits data to the energy that the base station consumes, this bunch head is to the needed jumping figure in base station,

Step 10: repeating step eight is to step 11 until the end of the time period of route;

Step 11: if bunch head finishes still there is no higher level's list to the route stage, this bunch head directly and base station communication so.

2. a kind of context-sensitive Uneven Cluster routing algorithm as claimed in claim 1, is characterized in that, in described step 1, the step of node weights is:

(1-1) determine node v _iThrowing is to its a certain neighbor node v _jBallot paper v (v _i, v _j),

$v(v_i,v_j)=\left\{\begin{array}{cc}\frac{e_j}{\underset{d_{\mathrm{ik}}\&le;R_i}{\mathrm{\&Sigma;}}{e}_k}& d_{\mathrm{ij}}\&le;R_i\\ 0& d_{\mathrm{ij}}>R_i\end{array}\right.$

Wherein, d _ikBe node v _iTo node v _kDistance, node v _kBe node v _iAll neighbours, R _iBe node v _iThe competition radius, e _jBe node v _jDump energy, e _kEnergy for one of them neighbour;

(1-2) determine node v _iNumber of ballot papers, node v _iNumber of ballot papers be all neighbours' ballot papers and be:

$\mathrm{vote}\left(v_i\right)=\underset{d_{\mathrm{ij}}\&le;R_i}{\mathrm{\&Sigma;}}v(v_j,v_i);$

(1-3) v _iCombining weights I _iFor:

$I_i=\frac{\mathrm{vote}\left(v_i\right)}{P_i}$

Wherein, P _iNode v _iTo the average distance of its all neighbor nodes, if this node without neighbours, weight I so _iAssignment is 1.

3. a kind of context-sensitive Uneven Cluster routing algorithm as claimed in claim 1, is characterized in that, in described step 2, competition is partly:

$R_i=(1-c\&times;\frac{d_{\max }-d(v_i,\sin k)}{d_{\max }-d_{\min }})R_{\max }$

Wherein, d (V _i, Sink) be V _iTo the distance of base station Sink, d _minThat wireless sensor node is apart from the minimum distance of base station Sink, d _maxThe maximum distance of nodal point separation base station Sink, R _maxOptimal value between 1/2d ₀-2/3d ₀Between, the optimal value of c is between 0.2-0.4.

4. a kind of context-sensitive Uneven Cluster routing algorithm as claimed in claim 1, is characterized in that, if in described step 6, bunch head checks to be sky to own higher level's list, so just to broadcast once with the competition radius of twice own.

5. a kind of context-sensitive Uneven Cluster routing algorithm as claimed in claim 1, is characterized in that, the concrete steps of described step 8 are:

If (8-1) check be 0 and the source node of this frame be higher level's list member of this bunch head, so just upgrade corresponding higher level's list entry, dump energy is the dump energy in control frame, being that jumping figure in control frame adds 1 to the jumping figure of base station, is to send data to the energy that source bunch head adds to be needed to the base station in this control frame to the energy of base station;

If (8-2) check is 0, and the source node of this frame creates a list item so not in higher level's list; If (8-3) check is 1, directly ignore this control frame.

Images