CN105306372A - Comprehensive topological optimization method - Google Patents

Abstract

The invention relates to a comprehensive topological optimization method. According to the method, network topological optimization is divided into two phases, firstly, the running state of a current system is determined; before running of the network topology of the system, the invention provides a topological optimization algorithm based on node criticality, which is used for quantitatively computing the criticality of each node in the topology, and keeping node variance in the whole topology within a certain range, so that the impact caused by invalidation of each node is effectively reduced, and the robustness of the network topology is improved; after the network topology of the system runs, a topological optimization method based on credibility and weighting load flow is provided to perform quantitative computation on the credibility of each node and the weighting load flow of each link, and keep the credibility of the node and the weighting load flow of each link at a certain level, so that the topology can be well optimized and can find a certain node or the fault of a certain link rapidly.

Description

A kind of comprehensive Topology Optimization Method

Technical field

The present invention designs a kind of comprehensive Topology Optimization Method, and particularly a kind of comprehensive Topology Optimization Method based on node key degree, confidence level and weighted load flow, belongs to network topology structure technical field.

Background technology

In the last few years, topological optimization was applied in a network and was more and more received publicity.The origin main cause of topology optimization is as follows: each node in network, reduces energy expense during node communication, effectively can extend network lifetime, can also effectively process the phenomenons such as the network congestion caused due to node failure simultaneously.Topological optimization controlling mechanism is the basis realizing a good network energy efficiency, and as the operation platform of routing layer agreement and MAC protocol, network topology is obviously realize the highest necessary condition of network connectivty.Therefore the power of communications adopting the in good time knot modification of rational topological optimization mechanism is only had, network could the square relation of plane such as balanced topology connective, openness, energy response, interference characteristic effectively, is ensureing connective basis to build the topological structure with high reliability, high connectivity.

In addition, scholar is had to inquire into the relation between network configuration robustness and network configuration feature from the ability aspect that network resists destruction, namely in network, some node, after being attacked, still can keep the ability be communicated with between remaining node, and residue Connectivity is more weak, illustrates that this node is more important.Scholar is also had to recognize that alone network configuration is not enough to portray pitch point importance, flow on network and corresponding expense are also very important key factors, to propose using flow as weight to the weighted network research mode of network node weighted to obtain the more network information for this reason.A lot of researcher is using the important means of redundancy scheme as raising network reliability.Main from node degree, network topology is optimized, to the node number of degrees higher carry out redundancy backup.Some scholar is the reliable data transmission ensureing network, provides a kind of mechanism promptly finding by scanning and search node route list under node sudden failure state and set up new route path.There is scholar according to the relation between the existence of hubs in complex networks and network vulnerability, by controlling the formation of DCS control and changing topology of networks to a certain extent, thus the ability of the anti-concerted attack of Scale-free Network can be strengthened.Also propose the topological key point elimination algorithm under non-structural P 2 P network in some scholar, namely topological key point is converted into general node.In literary composition, labor is carried out to two kinds of existing elimination algorithms (linearly connected elimination algorithm and band string ring elimination algorithm), and propose a kind of topological key point elimination algorithm connected based on focus.

Above generally only node degree is considered to the critical evaluation of node, do not consider other attributes of node, and these algorithms are be optimized after independent topological network runs or before running, and do not have comprehensive optimal design-aside that each stage has.

So, in order to ensure the high availability of application system, the network topology structure built has good elasticity, avoid because certain node failure causes the obvious decline of system crash and service quality, find the weak link in network, strengthening targetedly and optimize, is the reliable and stable important means of maintenance system.Network connectivty is the prerequisite be optimized network topology structure, in order to ensure being interconnected of each node in network, strengthens the reliability of network topology, robustness and survivability.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, and provide a kind of comprehensive Topology Optimization Method, the topology key degree that topology runs the front each unit of quantitative analysis network chooses suitable Topology Optimization Method; Topology carries out topological optimization by the confidence level and weighted load flow of collecting each node after running.Solve current generally with the deficiency of the degree of the node unification optimized algorithm that is benchmark, the reliability of the operating network topology of better guarantee and load balancing, the criticality of each node can be effectively reduced before operation, operationally can from the unexpected damaged condition of topological operation angle processing node and link, effectively avoid the impact on whole topological network in real time, improve reliability and the survivability of whole topology.

Technical solution of the present invention is: a kind of comprehensive Topology Optimization Method, for being optimized the network topology in navigational satellite system, comprises the following steps:

(1) whether interpretation current system is in running status, if current system is in running status, then enters step (4); Otherwise, enter step (2);

(2) the crucial degree of each node of computing system, i.e. the gray scale relating value of each node;

(3) whether the variance of node key degree that determining step (2) calculates is less than the threshold value preset, the threshold value preset if be less than, then enter step (4); Otherwise, adopt the method for looped network equivalence to be optimized the node that degree crucial in network is maximum, after having optimized, return step (2);

(4) optimize after judging whether to carry out system cloud gray model, if so, then enter step (5); Otherwise topological optimization algorithm terminates;

(5) carry out the topological optimization based on node credibility, enter step (6), carry out the topological optimization based on weighted load flow simultaneously, enter step (8);

(6) confidence level between each node and adjacent node is calculated in real time, specifically by formula:

$K_{ij}=K(i,j)=2\frac{v+1}{v+f+2}-1$

Provide, wherein K _ijfor node i is to the confidence level of j, v is that between node i with j, j communicates successful number of times, and f is the number of times of communication failure between node i and j;

(7) if the confidence level between certain node and adjacent node is less than threshold value given in advance; Then judge whether there is backup path between these two nodes, if there is backup path, then increase the path between two nodes and the flow balanced between two nodes; If there is not backup path, then cannot be optimized this node, terminate optimized algorithm;

If the confidence level between certain node and adjacent node is more than or equal to threshold value given in advance, then present node is without the need to optimizing, and returns step (6), recalculates the confidence level between this node and adjacent node;

(8) the weighted load flow F of each bar link is calculated in real time, specifically by formula:

$F=\left(\begin{array}{cccc}{F}_{11}& F_{12}& \mathrm{...}& F_{1j}\\ F_{21}& \mathrm{...}& \mathrm{...}& F_{2j}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ F_{i1}& F_{i2}& \mathrm{...}& F_{ij}\end{array}\right),i=1,2,\mathrm{...},n;j=1,2,\mathrm{...},n$

Provide, wherein n is node total number, F _ijfor being flowed to the weighted load stream of node j by node i;

(9) if the weighted load flow of certain link is greater than threshold value given in advance; Then judging whether there is backup path between these two nodes, if there is backup path, then increasing path between two nodes for balancing the flow between two nodes, and return step (8) and recalculate weighted load flow F; If there is not backup path, then cannot be optimized this node, terminate optimized algorithm; If the weighted load flow of certain link is less than threshold value given in advance, then returns step (8) and recalculate weighted load flow F.

The crucial degree of described each node of computing system, i.e. the gray scale relating value of each node; Concrete steps are:

(2-1) degree of selected node, the degree of approach of node, the betweenness of node, the network efficiency of the core degree sum node of node is as the parameter of node key degree;

(2-2) calculation procedure (2-1) choose each refer to target value;

(2-3) parameter is normalized;

(2-4) weight of each index is calculated;

(2-5) the selected normalized ideal value b' of each index _j, calculate the gray scale relating value of each node, be the crucial degree of each node.

The degree D of described node j _jfor the number of direct other nodes be connected with node j;

The degree of approach of node j refers to the inverse of the distance sum of other all nodes in node j to network, by formula:

Provide, wherein CC _jfor the degree of approach of node j, d _ifor the beeline of each node in node j and other nodes of network topology, N is total number of network topology interior joint;

The betweenness of node j to refer in network through the ratio of node j in all shortest paths, by formula:

$B_j=\underset{i,k\∈v}{\Σ}\frac{n_{ik}\left(j\right)}{n_{ik}}$

Provide, wherein B _jfor the betweenness of node j, n _ikfor the number of shortest path between node i and node k, n _ik(j) in shortest path between node i and node k through the number of node j;

The core degree of node j is long-pending refer to that node j is isolated after, the number of links in the shortest path total length that in network, all nodes intercom mutually and network, by formula:

$T_j=(S_{G-V_j}-S_G)\×(l_G-l_{G-V_j})$

Provide, wherein T _jfor node j is amassed by the core degree after isolating, S _gfor the shortest path length summation that two nodes any in network topology structure G intercom mutually, for node j is isolated the shortest path length summation that in rear network topology structure G, any two nodes intercom mutually, l _gfor the number of links in network topology structure G, for node j is by the number of links isolated in rear network topology structure G;

The network efficiency η of node j _jfor node j is isolated the mean value of the inverse of any two euclidean distance between node pair in rear network topology structure G, by formula:

η _j＝MEAN(1/d _ij),1≤i,j≤N&&i≠j

Provide, wherein η _jfor the network efficiency of node j, MEAN carries out averaging the function of computing, d _ijrepresent that node j is isolated any two internodal distances in rear network topology structure G, N is total number of network topology structure G interior joint.

Described parameter to be normalized; Specifically by formula:

$x_{i,j}^{\′}=(x_{i,j}-\underset{i=1,2,...,n}{\min }{x}_{i,j})/(\underset{i=1,2,...,n}{\max }{x}_{i,j}-\underset{i=1,2,...,n}{\min }{x}_{i,j}),j=1,2,...,p$

Provide, wherein x _i,jrepresent the desired value of a jth input pointer of the i-th node in network topology structure G, x ' _i,jto index x _i,jvalue after normalization, n represents total number of network topology structure G interior joint; P is the total number of index.

The method that described employing entropy power calculates calculates the weight of each index; Be specially:

(2-4-1) probability that the desired value of each node is shared in whole index is calculated, specifically by formula:

$P_{i,j}=(x_{i,j}^{\′}+\mathrm{\ζ})/\left(\underset{i=1}{\overset{n}{\Σ}}\right(x_{i,j}^{\′}+\mathrm{\ζ})),(j=1,2,...,p;i=1,2,...,n)$

Provide, wherein P _i,jfor the probability that the desired value of a jth input pointer of the i-th node in network topology structure G is shared in the desired value of all indexs of all nodes, x ' _i,jfor the normalization desired value of each node; P is the total number of index selected; N represents total number of whole network topology interior joint; ζ is adjustment parameter, and span is 10 ^-8-10 ^-6;

(2-4-2) comentropy of each index is calculated according to the result of step (2-4-1); Specifically by formula:

$H_j=-\frac{1}{\ln n}\[\underset{i=1}{\overset{n}{\Σ}}{P}_{i,j}*{\mathrm{lnP}}_{i,j}\]$

Provide, wherein H _jfor the information entropy of index j;

(2-4-3) the entropy weight of each index is calculated, specifically by formula:

$w_j=\[1-H_j\]/\underset{i=1}{\overset{n}{\Σ}}\[1-H_i\]$

Provide, wherein w _jfor the entropy weight of index j.

The gray scale relating value of described each node is specifically by formula:

$I=\underset{i=1}{\overset{n}{\Σ}}{\mathrm{\α}}_{i,j}{w}_j,j=1,2,...,p$

Provide, wherein α _i,jby formula:

α _i,j＝(Δ _min+βΔ _max)/(|b' _j-x′ _i,j|+βΔ _max)i＝1,2,...,n；j＝1,2,...,p

Provide, wherein ${\mathrm{\Δ}}_{min}=\underset{i=1,2,...,n}{\min }\underset{j=1,2,\mathrm{...},p}{\min }\left|b_j^{\′}-x_{i,j}^{\′}\right|,{\mathrm{\Δ}}_{\max }=\underset{i=1,2,\mathrm{...},n}{\max }\underset{j=1,2,\mathrm{...},p}{\max }\left|b_j^{\′}-x_{i,j}^{\′}\right|,$ β is constant given in advance; w _jfor the entropy weight of index j.

Described F _ijspecifically by formula:

F _ij＝∑γ _kd _kk＝1,2,...,K

Provide, wherein, d _kfor the data rate of kth category information in this load flow, unit is bit/s; γ _kfor the weight of kth category information in this load flow, specifically by formula:

${\mathrm{\γ}}_k={\mathrm{\λ}}_k\left(\frac{1}{K-1}-\frac{D_k}{(K-1){\mathrm{\ΣD}}_k}\right)$

Provide, wherein λ _kfor the factor of influence of kth category information given in advance, λ _kfor constant and ∑ λ _k=1, D _kfor the generation speed of kth category information, unit is bit/s, K is information category number.

Flow between described step (7) and the middle balance of step (9) two nodes is specifically by formula:

$\frac{{\mathrm{Link}}_1}{{\mathrm{Link}}_2}=\frac{C_1}{C_2}$

Provide, wherein Link ₁and Link ₂be respectively the flow of two links, C ₁and C ₂be respectively the channel capacity of two links.

The present invention's usefulness is compared with prior art:

(1) the present invention is that a kind of network topology that spreads all over runs Topology Optimization Method that is front and two stages when running, and the cycle of operation different from system topological carries out omnibearing optimization, like this, not only compensate for the deficiency that may run in design in early stage topology, with in later stage topology running, certain node that can not resist or link exceed the situation of load, the effective whole topological feasibility of design improved and the reliability ensureing topology.

(2) before topology is run, the present invention proposes a kind of comprehensive evaluation method to the key of node, the method considers degree, the degree of approach, betweenness, the core degree sum network efficiency of node, the key characteristic of evaluation node is carried out from multiple angle, the method of each node is evaluated with conventional method single employing node degree, more can effectively distinguish each node, some can be looked from node degree the node that importance is identical, branch away from the angular area of crucial degree, thus effectively can make up the deficiency of node degree, more targeted to the optimization of topology.

(3) after the network operation, the present invention in turn gives real-time control based on node and link and optimized algorithm, these two kinds of algorithms are the fault that occurs in real time of processing node and link effectively, effectively enable backup, and also have the good alert capability to link and node failure, add the connectedness of network, guarantee being interconnected of each node in network, strengthen the reliability of network topology, robustness and survivability.

Accompanying drawing explanation

Fig. 1 is method flow diagram of the present invention;

Fig. 2 is topological optimization structure chart method schematic diagram;

Fig. 3 be based on the optimization of the topological optimization of node key degree before topology diagram;

Fig. 4 be based on the optimization of the topological optimization of node key degree after topology diagram;

Fig. 5 be based on the optimization of the topological optimization of node key degree before node key degree schematic diagram;

Fig. 6 be based on the optimization of the topological optimization of node key degree after node key degree schematic diagram;

Fig. 7 is the topology diagram of the topological optimization based on node credibility;

Fig. 8 is the algorithm simulating result figure of the topological optimization based on node credibility;

Fig. 9 is the algorithm simulating result figure of the topological optimization based on weighted load stream.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail.

Topology Optimization Method in the present invention is used for being optimized the network topology in navigational satellite system, described network topology structure refers to the physical layout of the various equipment that to interconnect with transmission medium, and main topological structure has bus-type topology, Star topology, ring topology, tree topology and their mixed type.

Be illustrated in figure 1 flow chart of the present invention, as can be seen from Figure 1, the comprehensive Topology Optimization Method of one that the present invention proposes, comprises the following steps:

(1) whether be in running status mark (definition running status is Running) interpretation current system according to current system and whether be in running status (Running=1), if current system is in running status, then enter step (4); If current system is not in running status (Running=0), enter step (2);

(2) the crucial degree of each node of computing system; Concrete steps are:

(2-1) as shown in Figure 2, the degree of selected node, the degree of approach of node, the betweenness of node, the network efficiency of the core degree sum node of node is as the parameter of node key degree;

(2-2) calculation procedure (2-1) choose each refer to target value, be specially:

The degree D of node j _jfor the number of direct other nodes be connected with node j;

The degree of approach of node j refers to the inverse of the distance sum of other all nodes in node j to network, by formula:

Provide, wherein CC _jfor the degree of approach of node j, d _ifor the beeline of each node in node j and other nodes of network topology, N is total number of network topology interior joint;

The betweenness of node j to refer in network through the ratio of node j in all shortest paths, by formula:

$B_j=\underset{i,k\∈v}{\Σ}\frac{n_{ik}\left(j\right)}{n_{ik}}$

Provide, wherein B _jfor the betweenness of node j, n _ikfor the number of shortest path between node i and node k, n _ik(j) in shortest path between node i and node k through the number of node j;

The core degree of node j is long-pending refer to that node j is isolated after, the number of links in the shortest path total length that in network, all nodes intercom mutually and network, by formula:

$T_j=(S_{G-V_j}-S_G)\×(l_G-l_{G-V_j})$

Provide, wherein T _jfor node j is amassed by the core degree after isolating, S _gfor the shortest path length summation that two nodes any in network topology structure G intercom mutually, for node j is isolated the shortest path length summation that in rear network topology structure G, any two nodes intercom mutually, l _gfor the number of links in network topology structure G, for node j is by the number of links isolated in rear network topology structure G;

The network efficiency η of node j _jfor node j is isolated the mean value of the inverse of any two euclidean distance between node pair in rear network topology structure G, by formula:

η _j＝MEAN(1/d _ij),1≤i,j≤N&&i≠j

Provide, wherein η _jfor the network efficiency of node j, MEAN carries out averaging the function of computing, d _ijrepresent that node j is isolated any two internodal distances in rear network topology structure G, N is total number of network topology structure G interior joint;

The network efficiency of the degree of above computing node, the degree of approach of node, the betweenness of node, the core degree of amassing of node and node is canonical algorithm, the basis of this algorithm is improved canonical algorithm or pro forma change also can be used for calculating in the present invention the value of each network topology structure parameter.

(2-3) parameter is normalized; Specifically by formula:

$x_{i,j}^{\′}=(x_{i,j}-\underset{i=1,2,...,n}{\min }{x}_{i,j})/(\underset{i=1,2,...,n}{\max }{x}_{i,j}-\underset{i=1,2,...,n}{\min }{x}_{i,j}),j=1,2,...,p$

Provide, wherein x _i,jrepresent the desired value of a jth input pointer of the i-th node in network topology structure G, x ' _i,jto index x _i,jvalue after normalization, n represents total number of network topology structure G interior joint; P is the total number of index selected; For those skilled in the art, other normalization processing method can be adopted to obtain the normalized value of input pointer value.

(2-4) weight of each index is calculated; Calculating for weight can adopt many indexes as the parameter of weight, the absolute value of such as index, the comentropy etc. of index, the method that the present invention adopts entropy to weigh to calculate, to calculate the weight of each index, is specially:

(2-4-1) probability that the desired value of each node is shared in whole index is calculated, specifically by formula:

$P_{i,j}=(x_{i,j}^{\′}+\mathrm{\ζ})/\left(\underset{i=1}{\overset{n}{\Σ}}\right(x_{i,j}^{\′}+\mathrm{\ζ})),(j=1,2,...,p;i=1,2,...,n)$

Provide, wherein P _i,jfor the probability that the desired value of a jth input pointer of the i-th node in network topology structure G is shared in the desired value of all indexs of all nodes, x ' _i,jfor the normalization desired value of each node; P is the total number of index selected; N represents total number of whole network topology interior joint; ζ is adjustment parameter, and span is 10 ^-8-10 ^-6;

(2-4-2) comentropy of each index is calculated according to the result of step (2-4-1); Specifically by formula:

$H_j=-\frac{1}{\ln n}\[\underset{i=1}{\overset{n}{\Σ}}{P}_{i,j}*{\mathrm{lnP}}_{i,j}\]$

Provide, wherein H _jfor the information entropy of index j;

(2-4-3) the entropy weight of each index is calculated, specifically by formula:

$w_j=\[1-H_j\]/\underset{i=1}{\overset{n}{\Σ}}\[1-H_i\]$

Provide, wherein w _jfor the entropy weight of index j;

Adopt entropy assessment agriculture products weight can avoid the evaluation deviation caused because subjectivity determines weight, calculate generation insignificant value introducing adjustment parameter ζ (being in order to avoid when taking the logarithm operation, occurring the situation of ln0) here for avoiding entropy to weigh simultaneously;

(2-5) the selected normalized ideal value b' of each index _j, calculate the gray scale relating value of each node, be the crucial degree of each node, specifically by formula:

$I=\underset{i=1}{\overset{n}{\Σ}}{\mathrm{\α}}_{i,j}{w}_j,j=1,2,...,p$

Provide, wherein α _i,jby formula:

α _i,j＝(Δ _min+βΔ _max)/(|b' _j-x′ _i,j|+βΔ _max)i＝1,2,...,n；j＝1,2,...,p

Provide, wherein ${\mathrm{\Δ}}_{min}=\underset{i=1,2,...,n}{\min }\underset{j=1,2,\mathrm{...},p}{\min }\left|b_j^{\′}-x_{i,j}^{\′}\right|,{\mathrm{\Δ}}_{\max }=\underset{i=1,2,\mathrm{...},n}{\max }\underset{j=1,2,\mathrm{...},p}{\max }\left|b_j^{\′}-x_{i,j}^{\′}\right|,$ β is constant given in advance, usual 1; Described normalization ideal value is: if index is the bigger the better, then ideal value is 1, otherwise ideal value is 0;

(3) whether the variance of node key degree that determining step (2) calculates is less than the threshold value preset threshold value described here is the requirement required for design system, the threshold value preset if be less than, then enter step (4); Otherwise, adopt the method for looped network equivalence (when the task that a certain key node carries outbalance is in again between multiple subnet, can consider to be linked to be ring-type to replace this node with multiple node, its workload is shared or backups on multiple node, and can transmit via the node that loop two kinds is other when certain node has problems) to network in the maximum node of crucial degree be optimized, return step (2) after having optimized;

(4) optimize after judging whether to carry out system cloud gray model, if so, then enter step (5); Otherwise topological optimization algorithm terminates;

(5) carry out the topological optimization based on node credibility, enter step (6), carry out the topological optimization based on weighted load flow simultaneously, enter step (8);

(6) confidence level between each node and adjacent node is calculated in real time, specifically by formula:

$K_{ij}=K(i,j)=2\frac{v+1}{v+f+2}-1$

Provide, wherein K _ijfor node i is to the confidence level of j, v is that between node i with j, j communicates successful number of times, and f is the number of times of communication failure between node i and j;

Wherein, communicate successfully: refer to from adjacent node and receive routing packets, and have passed the communication of pre-detection, or in official hour, have received the HELLO message from adjacent node, then think and receive a successfully communication.If not by pre-detection, then this node is resorted to as malicious node, does not need the calculating carrying out confidence level again.Proposed pre-detection, refers to and carries out some restrictions, as the number of requests etc. sent in the unit interval according to trusting degree to the behavior of node.

Communication failure: refer to the error message received due to link breakdown.The link breakdown in so-called path, refer to when through HELLO week after date, a node is finding that there is disruption in the path between adjacent node, an error message can be sent to other node broadcasts, or within the cycle time of a network reception message, do not receive the message forwarded from adjacent node, be regarded as communication failure yet.

(7) if the confidence level between certain node and adjacent node is less than threshold value given in advance; Then judge whether there is backup path between these two nodes, if there is backup path, then increase the path between two nodes and the flow balanced between two nodes; If there is not backup path, then cannot be optimized this node, terminate optimized algorithm;

If the confidence level between certain node and adjacent node is more than or equal to threshold value given in advance, then present node is without the need to optimizing, and returns step (6), recalculates the confidence level between this node and adjacent node;

(8) the weighted load flow F of each bar link is calculated in real time, specifically by formula:

$F=\left(\begin{array}{cccc}{F}_{11}& F_{12}& \mathrm{...}& F_{1j}\\ F_{21}& \mathrm{...}& \mathrm{...}& F_{2j}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ F_{i1}& F_{i2}& \mathrm{..}& F_{ij}\end{array}\right),i=1,2,...,n;j=1,2,...,n$

Provide, wherein n is node total number, F _ijfor being flowed to the weighted load stream of node j by node i, specifically by formula:

F _ij＝∑γ _kd _kk＝1,2,...,K

Provide, wherein, d _kfor the data rate of kth category information in this load flow, unit is bit/s; γ _kfor the weight of kth category information in this load flow, specifically by formula:

${\mathrm{\γ}}_k={\mathrm{\λ}}_k\left(\frac{1}{K-1}-\frac{D_k}{(K-1){\mathrm{\ΣD}}_k}\right)$

Provide, wherein λ _kfor the factor of influence of kth category information given in advance, λ _kfor constant and ∑ λ _k=1, D _kfor the generation speed of kth category information, the kind of described information is as shown in table 1, is divided into I class, II class and III class; Unit is bit/s, K is information category number;

(9) if the weighted load flow of certain link is greater than threshold value F given in advance _thread; Then judging whether there is backup path between these two nodes, if there is backup path, then increasing path between two nodes for balancing the flow between two nodes, and return step (8) and recalculate weighted load flow F; If there is not backup path, then cannot be optimized this node, terminate optimized algorithm; If the weighted load flow of certain link is less than threshold value given in advance, then returns step (8) and recalculate weighted load flow F.

Flow between step (7) and the middle balance of step (9) two nodes is specifically by formula:

$\frac{{\mathrm{Link}}_1}{{\mathrm{Link}}_2}=\frac{C_1}{C_2}$

Provide, wherein Link ₁and Link ₂be respectively the flow of two links, C ₁and C ₂be respectively the channel capacity of two links.

Certainly, except above flow equilibrium method, other method also can be adopted to carry out flow equilibrium, and the residual capacity of such as link, the reliability etc. of link, concrete balance criterion changes according to the difference of stressed factor.

Specific embodiment

As shown in Figure 3, be the test topology that the present invention selects, first, system topological is not in running status (Running=0), calculates the crucial degree of each node, and the crucial degree normalization variance calculating each node is σ ²=0.1332, the annexation of analysis node, checks whether the crucial degree of node is greater than the threshold value of our regulation therefore we will adopt the Topology Optimization Method based on node, and we adopt the method for looped network equivalence here, and node 5 equivalence is become four node looped networks, and node 5a, 5b, 5c, 5d, obtain result as Fig. 4.Recalculate the crucial degree of now each node, the crucial degree normalization variance that we obtain the new topology calculated is σ ²=0.0746, be less than our threshold value Thread=0.1, so optimized algorithm terminates, by optimum results Fig. 5 and Fig. 6, and the display of variance result of calculation, and our effect of optimization or effectively.

System topological is arranged to running status (Running=1) by us, and in order to more convenient differentiation, topology renames by we, as shown in Figure 7, is the network topology carrying out optimizing after topology is run.After starting based on the optimized algorithm of confidence level, the confidence level of each node calculate and neighbor node, then the threshold value whether confidence level of certain neighbor node requires lower than us is judged, here we arrange this threshold value is 0.8, we find the with a low credibility in our requirement of node 10, so we increase the link between 2, and balance the flow of two links.Return and recalculate confidence level, confidence level 0.87 now; If can not increase link, send the alarm signal that reliability cannot meet, algorithm terminates.The operational effect figure of this algorithm is referring to Fig. 8.

Based on the topological optimization algorithm of weighted load flow, calculate the load flowing matrix F of whole topology, and the limit finding load maximum, its load F _i.Wherein, different data traffics is referring to table 1.If F _ibe less than the threshold value of our setting, return to computational load traffic matrix.Here the threshold value arranged is F _thread=8Mbps.Result link 7 has been greater than threshold value, and we will judge whether to increase link, if can increase link, link between increasing at 2, and balance the flow of two links, return recalculate load flowing matrix obtain new result, now the weighted load flow of link 7 is less than 7; If can not increase link, send the alarm signal of excessive flow load, algorithm terminates.The operational effect figure of this algorithm is referring to Fig. 9.

Table 1

Simulation result display of the present invention: as can be seen from Fig. 5 and 6, tradition optimized algorithm effectively can not distinguish the criticality of each node, and the method for use looped network in this paper equivalence carry out topology run before optimized algorithm, effectively distinguish the crucial degree of each node, this topological optimization algorithm simultaneously, operation result can effectively reduce the criticality of each node, mainly embody from variance aspect, the node so just can effectively effectively avoiding those crucial degree high from topological angle damages the impact on whole topological network, improve reliability and the survivability of whole topology, as can be seen from Figure 8, the postrun optimized algorithm of topology based on confidence level in this paper, the requirement of whole network to confidence level can be controlled, so just can effectively from the unexpected damaged condition of topological operation angle processing node, effectively avoid the impact on whole topological network in real time, improve reliability and the survivability of whole topology, as can be seen from Figure 9, the postrun optimized algorithm of topology based on flow in this paper, whole network can be controlled to the requirement of each link to weighted load flow, so just can effectively from the unexpected damaged condition of topological operation angle handle link, impact with effectively avoiding in real time whole topological network, improves reliability and the survivability of whole topology.

The content be not described in detail in specification of the present invention belongs to the known technology of professional and technical personnel in the field.

Claims (8)

1. a comprehensive Topology Optimization Method, for being optimized the network topology in navigational satellite system, is characterized in that comprising the following steps:

(1) whether interpretation current system is in running status, if current system is in running status, then enters step (4); Otherwise, enter step (2);

(2) the crucial degree of each node of computing system, i.e. the gray scale relating value of each node;

(3) whether the variance of node key degree that determining step (2) calculates is less than the threshold value preset, the threshold value preset if be less than, then enter step (4); Otherwise, adopt the method for looped network equivalence to be optimized the node that degree crucial in network is maximum, after having optimized, return step (2);

(4) optimize after judging whether to carry out system cloud gray model, if so, then enter step (5); Otherwise topological optimization algorithm terminates;

(5) carry out the topological optimization based on node credibility, enter step (6), carry out the topological optimization based on weighted load flow simultaneously, enter step (8);

(6) confidence level between each node and adjacent node is calculated in real time, specifically by formula:

$K_{ij}=K(i,j)=2\frac{v+1}{v+f+2}-1$

Provide, wherein K _ijfor node i is to the confidence level of j, v is that between node i with j, j communicates successful number of times, and f is the number of times of communication failure between node i and j;

(7) if the confidence level between certain node and adjacent node is less than threshold value given in advance; Then judge whether there is backup path between these two nodes, if there is backup path, then increase the path between two nodes and the flow balanced between two nodes; If there is not backup path, then cannot be optimized this node, terminate optimized algorithm;

If the confidence level between certain node and adjacent node is more than or equal to threshold value given in advance, then present node is without the need to optimizing, and returns step (6), recalculates the confidence level between this node and adjacent node;

(8) the weighted load flow F of each bar link is calculated in real time, specifically by formula:

$F=\left(\begin{array}{cccc}{F}_{11}& F_{12}& ...& F_{1j}\\ F_{21}& ...& ...& F_{2j}\\ ...& ...& ...& ...\\ F_{i1}& F_{i2}& ...& F_{ij}\end{array}\right)i=1,2,...,n;j=1,2,...,n$

Provide, wherein n is node total number, F _ijfor being flowed to the weighted load stream of node j by node i;

(9) if the weighted load flow of certain link is greater than threshold value given in advance; Then judging whether there is backup path between these two nodes, if there is backup path, then increasing path between two nodes for balancing the flow between two nodes, and return step (8) and recalculate weighted load flow F; If there is not backup path, then cannot be optimized this node, terminate optimized algorithm; If the weighted load flow of certain link is less than threshold value given in advance, then returns step (8) and recalculate weighted load flow F.

2. the comprehensive Topology Optimization Method of one according to claim 1, is characterized in that: the crucial degree of described each node of computing system, i.e. the gray scale relating value of each node; Concrete steps are:

(2-1) degree of selected node, the degree of approach of node, the betweenness of node, the network efficiency of the core degree sum node of node is as the parameter of node key degree;

(2-2) calculation procedure (2-1) choose each refer to target value;

(2-3) parameter is normalized;

(2-4) weight of each index is calculated;

(2-5) the selected normalized ideal value b' of each index _j, calculate the gray scale relating value of each node, be the crucial degree of each node.

3. the comprehensive Topology Optimization Method of one according to claim 2, is characterized in that:

The degree D of described node j _jfor the number of direct other nodes be connected with node j;

The degree of approach of node j refers to the inverse of the distance sum of other all nodes in node j to network, by formula:

Provide, wherein CC _jfor the degree of approach of node j, d _ifor the beeline of each node in node j and other nodes of network topology, N is total number of network topology interior joint;

The betweenness of node j to refer in network through the ratio of node j in all shortest paths, by formula:

$B_j=\underset{i,k\∈v}{\Σ}\frac{n_{ik}\left(j\right)}{n_{ik}}$

Provide, wherein B _jfor the betweenness of node j, n _ikfor the number of shortest path between node i and node k, n _ik(j) in shortest path between node i and node k through the number of node j;

The core degree of node j is long-pending refer to that node j is isolated after, the number of links in the shortest path total length that in network, all nodes intercom mutually and network, by formula:

$T_j=(S_{G-V_j}-S_G)\×(l_G-l_{G-V_j})$

Provide, wherein T _jfor node j is amassed by the core degree after isolating, S _gfor the shortest path length summation that two nodes any in network topology structure G intercom mutually, for node j is isolated the shortest path length summation that in rear network topology structure G, any two nodes intercom mutually, l _gfor the number of links in network topology structure G, for node j is by the number of links isolated in rear network topology structure G;

The network efficiency η of node j _jfor node j is isolated the mean value of the inverse of any two euclidean distance between node pair in rear network topology structure G, by formula:

η _j＝MEAN(1/d _ij),1≤i,j≤N&&i≠j

Provide, wherein η _jfor the network efficiency of node j, MEAN carries out averaging the function of computing, d _ijrepresent that node j is isolated any two internodal distances in rear network topology structure G, N is total number of network topology structure G interior joint.

4. the comprehensive Topology Optimization Method of one according to claim 2, is characterized in that: be describedly normalized parameter; Specifically by formula:

$x_{i,j}^{\′}=(x_{i,j}-\underset{i=1,2,...,n}{min}{x}_{i,j})/(\underset{i=1,2,...,n}{\max }{x}_{i,j}-\underset{i=1,2,...,n}{min}{x}_{i,j}),j=1,2,...,p$

Provide, wherein x _i,jrepresent the desired value of a jth input pointer of the i-th node in network topology structure G, x ' _i,jto index x _i,jvalue after normalization, n represents total number of network topology structure G interior joint; P is the total number of index.

5. the comprehensive Topology Optimization Method of one according to claim 2, is characterized in that: the method that described employing entropy power calculates is to calculate the weight of each index; Be specially:

(2-4-1) probability that the desired value of each node is shared in whole index is calculated, specifically by formula:

$P_{i,j}=(x_{i,j}^{\′}+\mathrm{\ζ})/\left(\underset{i=1}{\overset{n}{\Σ}}\right(x_{i,j}^{\′}+\mathrm{\ζ}\left)\right),(j=1,2,...,p;i=1,2,...,n)$

Provide, wherein P _i,jfor the probability that the desired value of a jth input pointer of the i-th node in network topology structure G is shared in the desired value of all indexs of all nodes, x ' _i,jfor the normalization desired value of each node; P is the total number of index selected; N represents total number of whole network topology interior joint; ζ is adjustment parameter, and span is 10 ^-8-10 ^-6;

(2-4-2) comentropy of each index is calculated according to the result of step (2-4-1); Specifically by formula:

$H_j=-\frac{1}{lnn}\[\underset{i=1}{\overset{n}{\Σ}}{P}_{i,j}*{\mathrm{lnP}}_{i,j}\]$

Provide, wherein H _jfor the information entropy of index j;

(2-4-3) the entropy weight of each index is calculated, specifically by formula:

$w_j=\[1-H_j\]/\underset{i=1}{\overset{n}{\Σ}}\[1-H_i\]$

Provide, wherein w _jfor the entropy weight of index j.

6. the comprehensive Topology Optimization Method of one according to claim 2, is characterized in that: the gray scale relating value of described each node is specifically by formula:

$I=\underset{i=1}{\overset{n}{\Σ}}{\mathrm{\α}}_{i,j}{w}_j,j=1,2,...,p$

Provide, wherein α _i,jby formula:

α _i,j＝(Δ _min+βΔ _max)/(|b' _j-x′ _i,j|+βΔ _max)i＝1,2,...,n；j＝1,2,...,p

Provide, wherein ${\mathrm{\Δ}}_{min}=\underset{i=1,2,...,n}{min}\underset{j=1,2,...,p}{min}|b_j^{\′}-x_{i,j}^{\′}|,{\mathrm{\Δ}}_{\max }=\underset{i=1,2,...,n}{\max }\underset{j=1,2,...,p}{\max }|b_j^{\′}-x_{i,j}^{\′}|,$ β is constant given in advance; w _jfor the entropy weight of index j.

7. the comprehensive Topology Optimization Method of one according to claim 1, is characterized in that: described F _ijspecifically by formula:

F _ij＝∑γ _kd _kk＝1,2,...,K

Provide, wherein, d _kfor the data rate of kth category information in this load flow, unit is bit/s; γ _kfor the weight of kth category information in this load flow, specifically by formula:

${\mathrm{\γ}}_k={\mathrm{\λ}}_k(\frac{1}{K-1}-\frac{D_k}{(K-1){\mathrm{\ΣD}}_k})$

Provide, wherein λ _kfor the factor of influence of kth category information given in advance, λ _kfor constant and ∑ λ _k=1, D _kfor the generation speed of kth category information, unit is bit/s, K is information category number.

8. the comprehensive Topology Optimization Method of one according to claim 1, is characterized in that: the flow between described step (7) and the middle balance of step (9) two nodes is specifically by formula:

$\frac{{\mathrm{Link}}_1}{{\mathrm{Link}}_2}=\frac{C_1}{C_2}$

Provide, wherein Link ₁and Link ₂be respectively the flow of two links, C ₁and C ₂be respectively the channel capacity of two links.