CN103957544A - Method for improving survivability of wireless sensor network - Google Patents

Abstract

The invention discloses a method for improving survivability of a wireless sensor network. The method mainly includes the first step of dividing the whole WSN network into k subnets through a clustering algorithm and iteration combined method, the second step of calculating sink nodes in each subnet through an equilibrium point calculation method for the k subnets obtained through division in the first step, and the third step of putting forward to a multi-hopping multi-center network survivability detecting and evaluating method to evaluate the overall survivability of the network through combination with the results obtained in the second step. The method is suitable for the complex wireless sensor network which has the multi-hopping multi-center characteristic and is provided with the nodes obviously different from one another in the aspects of service types, data volumes and the like, and the survivability of the network of this type can be effectively improved.

Description

A kind of method that improves wireless sensor network survivability

Technical field

The invention belongs to wireless sensor network technology field, be specifically related to a kind of design of the method that improves wireless sensor network survivability.

Background technology

Wireless sensor network is a kind of special Ad-Hoc network, be the complication system that integrates data acquisition, data processing, transfer of data, formed by the sensor node with wireless communication module, data processing module, sensor assembly and power module of a lot of random distribution.Because sensor network is made up of with Ad hoc mode the huge node of quantity, and network condition is limited, node limited energy, layout environment is severe, and therefore robustness, reliability and the adaptability to environment etc. in sensor network design have become the problem that must consider in sensor network design.Along with the continuous expansion of WSN network size and the continuous expansion of application, the research of WSN Survivabilities of Networks comes into one's own day by day, wireless senser radio sensing network scale day by day huge and complicated, companion's and come be network fault also thereupon increase, the threat that network faces and attack are also more and more, once certain key node of network breaks down, will carry out significant impact to whole WSN network, when serious, can cause whole network paralysis.The reliability etc. that simultaneously generally depends on network topology structure and communication link capacity, distribution of traffic and network element due to the node characteristic in WSN, the importance degree of therefore studying network node is also very necessary.

In wireless sensor network, node is generally divided into aggregation node and general sensor nodes, these nodes form wireless network by Ad hoc mode, with specific information in the mode perception, acquisition and processing network's coverage area of cooperation, realize the collection at any time of anywhere information, process and analyze.Between sensing node, can intercom mutually, oneself be organized into net and be connected to aggregation node by the mode of multi-hop, aggregation node receives after data, completes and the connection of public Internet network the visible Fig. 1 of concrete model by gateway.The critical role of aggregation node in whole network as can be seen here, therefore aggregation node choose also most important.Aggregation node is data-collection nodes, also claims aggregation node, base-station node.

At present numerous to the various researchs of Survivabilities of Networks, " evaluation method of network survivability degree and node importance degree " (referring to document " Rao Yuping; Lin Jingyu; Zhou Dongfang. the evaluation method ［ J ］ of network survivability degree and node importance degree. computer engineering; 2009; 35 (6): 14-16. ") pointed out in communication network that between network node, the survivability of the more networks of shortest path is just stronger, weigh its survivability by the difference that compares other network self structure and full-mesh network, proposed the network survivability degree based on the average equivalent shortest path number of the whole network." a kind of novel WSN survivability evaluation method and the Realization of Simulation thereof " (referring to document " Lin Liwei; Xu Li; Ye Xiucai. a kind of novel WSN survivability evaluation method and the Realization of Simulation ［ J ］ thereof. computer system application; 2010,19 (4): 32-36. ") weighing on the basis of Survivabilities of Networks and considering the feature centered by Data Collection in wireless sensor network with network-in-dialing coefficient." the WSN survivability evaluation method based on shortest path number " (referring to document " Wang Xin; Li Bin. the WSN survivability evaluation method [J] based on shortest path number. electronics technology; 2012; 25 (11): 88-90. ") on the wireless sensor network model basis to there is worldlet characteristic, according to the characteristic of the higher aggregation of having of small-world network and less average distance, proposed WSN network based on effective covering bunch the survivability evaluation method of average equivalent shortest path, with bunch survivability weigh more exactly the survivability of whole network.

But above-mentioned the whole bag of tricks all do not take into account the importance degree of node, type of service and the data volume etc. of having ignored node affect the factor of WSN network entirety survivability.In addition, choose aggregation node in subnet, conventional method is to take to calculate the method that the physics barycenter of subnet is aggregation node.But, this method is ignored the actual conditions of radio sensing network, and under outdoor environment, between node, between pure geographical position relationship and node, network connectivty may not have strict correlation, although the computation complexity of this method is lower, not necessarily can obtain optimum solution.

Summary of the invention

The object of the invention is survivability in order to improve multi-hop multicentric network can and assess and a kind of method that improves wireless sensor network survivability of providing it.

Technical scheme of the present invention is: a kind of method that improves wireless sensor network survivability, specifically comprises:

Step 1: use the method for clustering algorithm in conjunction with iteration, overall WSN network is divided into k subnet;

Step 2: to dividing k subnet out in described step 1, by the mode of asking balance point, obtain respectively the aggregation node in each subnet;

Step 3: provide the result that multi-hop multicentric network survivability evaluating method obtains in conjunction with described step 2 and can evaluate the overall survivability of network.

Further, the concrete steps that in described step 1, subnet is divided are:

Step 1: the set of establishing the node in network is V (G) is chosen at random k node as initial cluster centre C from set V (G) ₀, $C_0\⋐V\left(G\right);$

Step 2: calculate in G except cluster centre C ₀all node v outside the node comprising _jto cluster centre C ₀in each node euclidean distance, if there is d (u _i, v _j)=min{d (u, v _j) | u ∈ V (C ₀), wherein wherein (x _u, y _u), (x _v, y _v) represent respectively node u and the coordinate position of node v in network, by node v _jbe divided into node u _icorresponding cluster; Repeat aforementioned account form, until all some v are divided into corresponding cluster;

Step 3: through after step 2, set of node V (G) is divided into k disjoint subset, and has V (G)=V' ₁∪ V' ₂... V' _k; Then each subset V' is asked to its central point u'(x', y' again based on plane coordinates), order have thereby can obtain new cluster centre C ₁;

Step 4: calculate variance criterion function

Step 5: repeating step 2,3,4, until variance criterion function convergence or number of repetition exceed the iterations of appointment;

Step 6: complete cluster, output cluster result.

Further, the concrete steps of asking for the aggregation node in subnet in described step 2 are:

Step 1: the set of establishing the node in subnet is V (G), to each the node u in each ready-portioned subnet, some u ∈ V (G), calculates other nodes v in this subnet _ito the weighted path length sum of this node u, described e (u) is the weighting irrelevance of u node, wherein, and m _irepresent v _ithe weights size of node, l (v _i, u) represent node v in this subnet _ito the linkage length of node u;

Step 2: the balance point that is this subnet by the node definition of weighting irrelevance minimum in subnet, that is the aggregation node of this subnet.

Further, the concrete steps of asking for the aggregation node in subnet in described step 2 are:

Step 1: the set of establishing the node in this subnet is V (G), to each the node u in V (G), the normalization weighting irrelevance of calculation level u in the whole network, computing formula is:

$\mathrm{\ξ}=\frac{\underset{i=1}{\overset{\left|V\right|}{\mathrm{\Σ}}}l(v_i,u)m_i}{\underset{i=1}{\overset{\left|V\right|}{\mathrm{\Σ}}}{m}_i}$

Wherein, m _irepresent v _ithe weights size of node, l (v _i, u) represent subnet V interior nodes v _ito the linkage length of node u;

Step 2: from u point, take breadth first search's algorithm, find the linkage length of all range points u to be less than or equal to the set V that forms of node _ξ;

Step 3: to V _ξin each some u calculate corresponding weighting irrelevance the balance point that reckling is in this subnet V is aggregation node.

Further, in described step 3, the specific implementation process of multi-hop multicentric network survivability evaluating method is: k the subnet of dividing out tried to achieve after corresponding aggregation node, and the available following anti-coefficient ρ that ruins assesses the survivability of network entirety:

$\mathrm{\ρ}=\frac{k}{\underset{i=1}{\overset{k}{\mathrm{\Σ}}}{e}^{\frac{\left|V_k\right|}{\left|V\right|}\underset{j=1}{\overset{\left|V_k\right|}{\mathrm{\Σ}}}l(v_j,c_k)m_j}};$

Wherein, k represents subnet number, | V| is whole network node number, | V _k| be k subnet interior nodes number, v _jrepresent a certain node in k subnet, c _krepresent the aggregation node in k subnet, m _jrepresent node v _jweights, l (v _j, c _k) expression subnet V interior nodes v _jto aggregation node c _klinkage length.

The invention has the beneficial effects as follows: the present invention is a kind of method that improves wireless sensor network survivability, by introducing the concept of node importance degree, give each node and given the weighted value corresponding to factors such as its type of service and data volumes, more objective and provided exactly the distribution characteristics of nodes, contribute to correctly choosing of aggregation node, and the concept of having introduced balance point is calculated the optimum position of aggregation node, provide more representational aggregation node more accurately, improved the survivability energy of network; Meanwhile, provide the algorithm that reduces computation complexity, in carrying out aggregation node location optimization, greatly reduced amount of calculation; There is the complicated wireless sensor network of significant difference in the present invention is applicable to possess multi-hop multicenter feature and node, can effectively improve the survivability energy of the type network at the aspect such as type of service and data volume.

Brief description of the drawings

Fig. 1 is the FB(flow block) of a kind of method that improves wireless sensor network survivability of the present invention;

Fig. 2 is the FB(flow block) that uses clustering algorithm subnet division in a kind of method that improves wireless sensor network survivability of the present invention;

Fig. 3 is the abstract model figure of multi-hop multicentric network;

Fig. 4 is the FB(flow block) of balance point computational process in step 2.

Embodiment

Below in conjunction with accompanying drawing and specific embodiment, the invention will be further elaborated.

The FB(flow block) that is illustrated in figure 1 a kind of method that improves wireless sensor network survivability of the present invention, specifically comprises:

Step 1: use the method for clustering algorithm in conjunction with iteration, overall WSN network is divided into k subnet;

Step 2: to dividing k subnet out in described step 1, by the mode of asking balance point, obtain respectively the aggregation node in each subnet;

Step 3: provide the result that multi-hop multicentric network survivability evaluating method obtains in conjunction with described step 2 and can evaluate the overall survivability of network.

The survivability the object of the invention is in order to improve multi-hop multicentric network can and can be evaluated the overall survivability of network by Survivabilities of Networks evaluating method; The method is mainly by carrying out the division of subnet to whole network, then each subnet of dividing gained is asked for respectively to corresponding aggregation node, finally utilize multi-hop multicentric network survivability evaluating method in conjunction with aggregation node location optimization result, can evaluate the overall survivability of network.Below by the method for dividing with regard to subnet respectively, the method for optimizing of aggregation node in each subnet, and the whole network survivability can appraisal procedure 3 aspects be introduced:

The concrete steps that in described step 1, subnet is divided are:

Because the node in whole WSN network is equally distributed at random, therefore can adopt clustering algorithm to carry out the division of subnet, the process that the set of physics or abstract object is divided into the multiple classes that are made up of similar object is called as cluster; Be illustrated in figure 2 the FB(flow block) that uses clustering algorithm subnet division, be specifically divided into following 6 steps:

Step 1: regard whole network as a figure, be illustrated in figure 3 the abstract model figure of multi-hop multicentric network, node in described network i.e. summit in figure, and its set is V (G), first from figure G, chooses at random k node as initial cluster centre C ₀, C ₀ì V (G);

Step 2: calculate in G except cluster centre C ₀all node v outside the node comprising _jto cluster centre C ₀in each node u (u ì C ₀) Euclidean distance, if there is d (u _i, v _j)=min{d (u, v _j) | u ∈ V (C ₀), wherein wherein (x _u, y _u), (x _v, y _v) represent respectively node u and the coordinate position of node v in network, by node v _jbe divided into node u _icorresponding cluster; Repeat aforementioned account form, until all some v are divided into corresponding cluster;

Step 3: through after step 2, set of node V (G) is divided into k disjoint subset, and has V (G)=V' ₁eV' ₂... V' _k, then each subset V' is asked to its central point again based on plane coordinates, i.e. average u'(x', y'), make v' ì V', have thereby can obtain new cluster centre C ₁;

Step 4: calculate variance criterion function

Step 5: repeating step 2,3,4, until variance criterion function convergence or number of repetition exceed the iterations of appointment;

Step 6: complete cluster, output cluster result.

In described step 2, ask for the aggregation node in subnet, traditionally main employing physically asks the mode of barycenter to calculate aggregation node position, but under outdoor environment, owing to having barrier or other orographic factor, network connectivty between node between pure geographical position relation and node does not probably have strict correlation, although therefore computational complexity is lower, its result, under mountain environment, may not be optimal solution.By the topological connection state in conjunction with network reality, calculate optimum aggregation node position by solving the method for balance point in a kind of network survivability performance evaluation method based on aggregation node of the present invention.

First define balance point, the node set of establishing in subnet is V (G), and G is the connected graph that represents this subnet, and some u ∈ V (G), claims for the weighting irrelevance of a u, wherein m _irepresent v _iweights size, l (v _i, u) represent this subnet interior nodes v _ito the linkage length of node u, if there is r (G)=min{e (u) | u ∈ V}, be called the weighting radius of figure G, for certain some u in figure G, if there is e (u)=r (G), claim that u is a balance point of figure G.

By the definition of balance point, first ask corresponding for each node u in whole subnet then from all points, choose satisfied point as balance point, be aggregation node; Because figure G is the figure of a cum rights, and may not be line graph or dendrogram, if therefore ask balance point, its basic theories complexity be O (N ²), by analysis and checking, the present invention can adopt another kind of algorithm to reduce the complexity of calculating, computation complexity can be reduced to O (lN) at this, the scale of l<<N, the flow process of its calculating as shown in Figure 4, is specifically divided into following 3 steps:

Step 1: optional 1 u in this subnet connected graph G, the normalization weighting irrelevance of calculation level u in this subnet, computing formula is: be the normalization weighting irrelevance that u is ordered, reality is exactly to weighting irrelevance in weights in whole subnet, done normalization (divided by whole subnet weights sum), therefore it is the average weighted distance of other joints point-to-point u; Described in it, calculate in the formula of the normalization weighting irrelevance that u orders m _irepresent v _ithe weights size of node, l (v _i, u) represent this subnet interior nodes v _ito the linkage length of node u;

Step 2: take breadth first search's algorithm, from a u, find the linkage length of all range points u to be less than or equal to the set V that forms of node _ξ; Wherein, the common practise that described breadth first search's algorithm is those skilled in the art is no longer described in detail in the present patent application scheme.

Step 3: to V _ξin each some u calculate corresponding weighting irrelevance get reckling and be the balance point in this subnet.

By the above-mentioned optimum position that has calculated the aggregation node in each subnet, thereby improve the survivability energy of network, in order to verify the survivability energy of asking for networks converge node select position and can improve overall network of the present invention, the present invention has provided Survivabilities of Networks assessment method so that the survivability of network entirety can be evaluated, and will be specifically described below:

In described step 3, the specific implementation process of multi-hop multicentric network survivability evaluating method is comprehensive above k subnet and the selected corresponding aggregation node of dividing, can can assess the survivability of multi-hop multicentric network with the following method, ρ is called the anti-coefficient of ruining of this network:

$\mathrm{\&rho;}=\frac{k}{\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{e}^{\frac{\left|V_k\right|}{\left|V\right|}\underset{j=1}{\overset{\left|V_k\right|}{\mathrm{\&Sigma;}}}l(v_j,c_k)m_j}}$

In this formula, each parameter declaration is as follows:

Represent the network topology of WSN with figure G=(V, E), G is the figure of UNICOM.Wherein, V is sensing node collection (being uniformly distributed in G), the wireless connections between E representation node, | V| represents to gather the number of node in V; Suppose two-way link symmetry between any two nodes, the limit in figure is all nonoriented edge, and figure G is divided into k subgraph, has figure G _k=(V _k, E _k), wherein subgraph G _krepresent k subnet, each subgraph G _kthere is a central point C _krepresent the aggregation node of corresponding k subnet; Each node v in net _jweight m _jrepresent, represent the importance degree size of node; L (v _j, c _k) represent node v in k subgraph _jto its corresponding aggregation node c _kpath, l (v _j, c _k) m _jbe called a node v _jto its corresponding central point C _kweighted path length.

For those skilled in the art can understand and implement technical solution of the present invention, describe below in conjunction with embodiment:

1. while adopting clustering algorithm to be network subnet division, need under target, carry out iteration repeatedly taking the convergence of variance criterion function (in the situation that can restraining), the concrete computational process of each iteration is as follows:

To figure G corresponding to overall network and former cluster centre C ₀, calculate in G except cluster centre C ₀all node v outside the node comprising _jdivide and be clipped to C ₀in the Euclidean distance of each node u , choose the node v that wherein minimum Eustachian distance is corresponding _jfor cluster corresponding to node u, with the method to C ₀in all node calculate one time, can draw cluster centre C ₀corresponding cluster, i.e. k disjoint subset, calculates and obtains its geometric center point u'(x', y' with how much each subset V'), computational methods are wherein v'(x _v', y _v') be the coordinate of the node in V', calculate thus to obtain the geometric center point of each subnet, i.e. new cluster centre C ₁, then, calculate variance criterion function carry out again new iteration once.

2. while adopting the mode that solves balance point to calculate aggregation node, as follows in conjunction with the concrete calculation procedure of innovative approach of the present invention:

If the set of the node in subnet is V (G), first optional 1 u in figure G, calculates its normalization weighting irrelevance be the average weighted distance of all the other joint point-to-point u in subnet, then take breadth first search's algorithm, from a u, find the linkage length of all range points u to be less than or equal to the set V that forms of node _ξ, finally to V _ξin each some u calculate corresponding weighting irrelevance get reckling, be the balance point of figure G; Need to be first a u be calculated N time by this algorithm, more right | ξ | in λ node carry out respectively N calculating, therefore computation complexity is O (λ N), has λ <<N here.

3. in the time that real network scale is relatively large, in the time asking for aggregation node by method described in previous step, can be for network size, suitably choose multiple u points, calculate corresponding ξ _u, obtain thus multiple V _ξ, then to these multiple V _ξintersection of sets collection V', reuses balance point method and asks aggregation node.

4. in the time that the balance point method that adopts normalization weighting irrelevance is asked for aggregation node, due to the aggregation node of finally the choosing larger node of inevitable more close those weights in theory, therefore choose at the very start the point of weight maximum in figure G and calculate ξ, can make as far as possible little, corresponding V _ξscale also can be less, thereby can make overall calculation amount further reduce.

5. the present invention is directed to multi-hop multicentric network, in the time adopting balance point method to solve aggregation node, use l (v _j, c _k) expression ordinary node v _jto aggregation node c _klinkage length, this need to consider under actual conditions, the l (v in multihop network _j, c _k) with c _kthe variation of position and changing.

6. from the anti-expression formula of ruining coefficient ρ of network in invention, in the time that k is fixing, each subnet size is more close, and each subnet interior nodes is better to the less capability to resist destruction of average weighted path sum of aggregation node.

Those of ordinary skill in the art will appreciate that, embodiment described here is in order to help reader understanding's principle of the present invention, should be understood to that protection scope of the present invention is not limited to such special statement and embodiment.Those of ordinary skill in the art can make various other various concrete distortion and combinations that do not depart from essence of the present invention according to these technology enlightenments disclosed by the invention, and these distortion and combination are still in protection scope of the present invention.

Claims (5)

1. a method that improves wireless sensor network survivability, is characterized in that, specifically comprises:

Step 1: use the method for clustering algorithm in conjunction with iteration, overall WSN network is divided into k subnet;

Step 2: to dividing k subnet out in described step 1, by the mode of asking balance point, obtain respectively the aggregation node in each subnet;

Step 3: provide the result that multi-hop multicentric network survivability evaluating method obtains in conjunction with described step 2 and can evaluate the overall survivability of network.

2. a kind of method that improves wireless sensor network survivability as claimed in claim 1, is characterized in that, the concrete steps that in described step 1, subnet is divided are:

Step 1: the set of establishing the node in network is V (G) is chosen at random k node as initial cluster centre C from set V (G) ₀, $C_0\&Subset;V\left(G\right);$

Step 2: calculate in G except cluster centre C ₀all node v outside the node comprising _jto cluster centre C ₀in each node euclidean distance, if there is d (u _i, v _j)=min{d (u, v _j) | u ∈ V (C ₀), wherein wherein (x _u, y _u), (x _v, y _v) represent respectively node u and the coordinate position of node v in network, by node v _jbe divided into node u _icorresponding cluster; Repeat aforementioned account form, until all some v are divided into corresponding cluster;

Step 3: through after step 2, set of node V (G) is divided into k disjoint subset, and has V (G)=V' ₁∪ V' ₂... V' _k; Then each subset V' is asked to its central point u'(x', y' again based on plane coordinates), order have thereby can obtain new cluster centre C ₁;

Step 4: calculate variance criterion function

Step 5: repeating step 2,3,4, until variance criterion function convergence or number of repetition exceed the iterations of appointment;

Step 6: complete cluster, output cluster result.

3. a kind of method that improves wireless sensor network survivability as claimed in claim 1, is characterized in that, the concrete steps of asking for the aggregation node in subnet in described step 2 are:

Step 1: the set of establishing the node in certain subnet is V (G), to each the node u in this subnet, some u ∈ V (G), calculates other nodes v in this subnet _ito the weighted path length sum of this node u, described e (u) is the weighting irrelevance of u node, wherein m _irepresent v _ithe weights size of node, l (v _i, u) represent node v in k subgraph _ito the linkage length of node u;

Step 2: the node u of weighting irrelevance minimum in the subnet obtaining in above-mentioned steps 1 is defined as to the balance point of this subnet, that is the aggregation node of this subnet.

4. a kind of method that improves wireless sensor network survivability as claimed in claim 1, is characterized in that, the concrete steps of asking for the aggregation node in subnet in described step 2 are:

Step 1: the set of establishing the node in subnet is V (G), to each the node u in each ready-portioned subnet, the normalization weighting irrelevance of calculation level u in this subnet, computing formula is:

$\mathrm{\&xi;}=\frac{\underset{i=1}{\overset{\left|V\right|}{\mathrm{\&Sigma;}}}l(v_i,u)m_i}{\underset{i=1}{\overset{\left|V\right|}{\mathrm{\&Sigma;}}}{m}_i};$

Wherein, m _irepresent v _ithe weights size of node, l (v _i, u) represent subnet V interior nodes v _ito the linkage length of node u;

Step 2: from u point, take breadth first search's algorithm in subnet, find the linkage length of all range points u to be less than or equal to the set V that forms of node _ξ;

Step 3: to V _ξin each some u calculate corresponding weighting irrelevance the balance point that reckling is in V is aggregation node.

5. a kind of method that improves wireless sensor network survivability as described in claim 1 to 4 any one, it is characterized in that, in described step S3, the specific implementation process of multi-hop multicentric network survivability evaluating method is: k the subnet of dividing out tried to achieve after corresponding aggregation node, and the available following anti-coefficient ρ that ruins assesses the survivability of network entirety:

$\mathrm{\&rho;}=\frac{k}{\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{e}^{\frac{\left|V_k\right|}{\left|V\right|}\underset{j=1}{\overset{\left|V_k\right|}{\mathrm{\&Sigma;}}}l(v_j,c_k)m_j}};$

Wherein, k represents subnet number, | V| is whole network node number, | V _k| be k subnet interior nodes number, v _jrepresent a certain node in k subnet, c _krepresent the aggregation node in k subnet, table m _jshow node v _jweights, l (v _j, c _k) expression node v _jto aggregation node c _klinkage length.