CN106850431B - Multi-attribute optimal routing method applied to low-orbit information network - Google Patents

Abstract

The invention discloses a multi-attribute optimal routing method applied to a low-orbit information network, and belongs to the field of satellite communication. Firstly, selecting N attributes which can represent the network quality; then, calculating an evaluation function of each attribute value of all edges in the whole network topological structure; next, a Choquet fuzzy integral value A of each edge in the topology of the whole network is calculated with respect to the evaluation function of the N attributes representing the network quality_mM is 1 to M, and finally, the fuzzy integral value A is set_mAnd M is 1-M, which is used as a measurement weight of the edge in the network, and the optimal route is calculated by utilizing a Dijkstra algorithm. The method takes a fuzzy integral of Chquet based on fuzzy measure as an integrated operator, converts multi-attribute parameters into single comprehensive attribute evaluation parameters, and then utilizes Dijkstra algorithm to select a route. This method has the following advantages: (1) the optimal route under the condition of multi-attribute routing can be obtained; (2) the routing algorithm of the node has low complexity; (3) the parameter adjustment is flexible and the adaptability is strong.

Description

Multi-attribute optimal routing method applied to low-orbit information network

Technical Field

The invention relates to a multi-attribute optimal routing method applied to a low-orbit information network, belonging to the technical field of space networks.

Background

The low-orbit information network communication system is based on low-orbit satellite constellation, and uses the excellent coverage characteristic of the satellite to realize the information access and transmission to the mobile users in a larger area (or the world). The low-orbit information network communication system has the characteristics of wide coverage, flexible and rapid networking, flexible information access and the like, and becomes an important component of the next generation mobile communication (5G) and the national information highway. The satellite nodes in the low-orbit information network have the characteristics of expensive equipment, large information transmission delay, limited bandwidth and the like, so that the optimal routing selection for transmitting data in the low-orbit information network needs to comprehensively consider the combined action of a plurality of attributes. Aiming at the particularity of routing selection of the low-orbit information network, the method for researching the optimal routing under the action of multiple attributes has important practical significance.

The optimal routing problem under the action of multiple attributes can be substantially reduced to a multiple attribute decision analysis problem. The traditional multi-attribute decision analysis problem mostly adopts arithmetic mean, weighted mean and other linear integration operators based on classical additive measurement to fuse the evaluation information of each attribute, and the assumption is that each attribute is independent and independent. But this premise is difficult to satisfy in real-life decision problems. The three attributes of path cost, time delay and bandwidth in the satellite network are correlated, so that the additive property of the property measurement is damaged.

The japanese scholars sugeno first proposed replacing the additivity set function with weaker monotonicity and continuity and became the fuzzy measure. The lambda measurement is a basic fuzzy measurement, and is characterized in that the parameters to be determined are few, and the measurement value of each single-point attribute set is determined only under the condition of satisfying the lambda law. The k-plus-blur measure proposed by Grabisch has strong representation capability, and the larger the k value is, the stronger the representation capability is, but has higher complexity relative to the lambda measure.

The multi-attribute decision method using fuzzy integration based on fuzzy measure as an integration operator not only fully considers the relative weight among the attributes, but also considers the interaction among the attributes. Three fuzzy integrals are commonly used, sugeno fuzzy integral, Zhenyuan fuzzy integral, Choquet fuzzy integral. The calculation complexity of the Sugeno fuzzy integral is the minimum of the three fuzzy integrals, but much information is ignored, and the evaluation effect is poor; the computational complexity of the Zhenyuan fuzzy integral is the maximum, and the integral value of the Zhenyuan fuzzy integral represents the maximum potential among various combinations of the attribute set; the computational complexity of the Choquet fuzzy integral is slightly higher than that of Sugeno integral, so that the properties of monotonicity, idempotency, boundedness and the like are met, and the Choquet fuzzy integral is widely applied to multi-attribute object evaluation and information fusion.

Disclosure of Invention

The technical problem to be solved by the invention is as follows: the method overcomes the defects of the prior art, and provides the multi-attribute optimal routing method which has low algorithm complexity, flexible parameter adjustment and strong adaptability and is applied to the low-orbit information network.

The technical solution of the invention is as follows: a multi-attribute optimal routing method applied to a low-orbit information network is based on a known full-network topological structure and comprises the following steps:

(1) selecting N attributes which can represent the network quality;

(2) calculating an evaluation function f of each attribute value of all edges in the topology structure of the whole network_m,k(x) K is 1 to N, M is 1 to M, and M is the total number of edges of the topology structure of the whole network;

(3) calculating each edge in the topology structure of the whole network_NChoquet fuzzy integral value A of evaluation function of attribute value_m；

(4) And (4) calculating the fuzzy integral value A calculated in the step (3)_mAnd as the measurement weight of the edge in the network, calculating by utilizing a Dijkstra algorithm to obtain the optimal route.

The evaluation function f of the attribute value in the step (2)_m,k(x) The specific calculation formula of (A) is as follows:

in which x is the number_mK-th attribute value, p, of an edge_{k_best}Represents the maximum value, p, of the kth attribute of all edges in the full-network topology_{k_worst}Is the minimum value of the kth attribute of all edges in the full-network topology.

Step (3) calculating Choquet fuzzy integral value A of evaluation functions of N attributes of each edge in the whole network topology structure and representing the network quality_mThe method comprises the following specific steps:

(3.1) abbreviation f_m,k(x)＝f_m,kAccording to f_m，kWith its density function mu (f) of fuzzy measure_m，k) Determining the fuzzy measure density mu (f) of N attributes of the m-th edge representing the network quality_m,k) Specific value of (a), abbreviated as μ (f)_m,k) Is mu_m,k；

(3.2) fuzzy measure density mu according to N attributes of all edges_m,kAnd calculating a fuzzy measure parameter lambda according to the following formula:

(3.3) the m-th edge determined according to step (3.1) represents the fuzzy measure mu (f) of the N attributes of the network quality_m,k) And (3) calculating the Choquet fuzzy integral value A according to the specific value and the fuzzy measure parameter lambda determined in the step (3.2)_m:

Will f is_m,kSorting from small to large to obtain f_m,i,i＝1～N：

In the formula, F_m,i＝{f_m,i,f_m，i+1，…f_m,NIs the m-th side after sorting_iSet of attributes to the Nth attribute evaluation function, μ (F)_m,i)＝μ(f_m,i+f_m,i+1+...+f_m,N-1+f_m,N) For any two attribute evaluation functions f with different sequence numbers in the sorted attribute evaluation function set_k'And f_j' the density value of the fuzzy measure satisfies the following relation:

μ(f_m,k'+f_m,j')＝μ(f_m,k')+μ(f_m,j')+λ_mμ(f_m,k')μ(f_m,j') K '≠ j', where μ (F)_N+1)＝0。

In step (3.1) f_m，kWith its density function mu (f) of fuzzy measure_m，k) The correspondence of (a) is obtained as follows: let mu (f)_m，k) Is f_m，kThird order function of, i.e. mu (f)_m，k)＝af_m，k+bf² _m，k+f³ _m，kWherein, for the kth attribute of all edges, the following condition is satisfied:

a+b＝1，

wherein p is_{k_mean}Is the average value of the k-th attribute of all edges in the full-network topology structure.

Compared with the prior art, the invention has the beneficial effects that:

(1) the routing method can obtain the optimal route under the condition of multi-attribute routing, and overcomes the limitation that the prior art only carries out routing decision according to single attribute;

(2) the relevance among the attributes is fully considered by the selected fuzzy measure, and the cost of each link can be evaluated in an all-around manner;

(3) the method adopts linear mapping to calculate the attribute value evaluation function, has small calculation complexity, is suitable for the calculation of the on-satellite router, and can truly and reasonably reflect the quality degree of the attribute parameters;

(4) the method adopts the third-order function to fit the fuzzy measure of the attributes, has simple and effective model, flexible parameter adjustment and strong adaptability, can select any number of attributes, and has no restriction on the specific selection of the attributes.

Drawings

FIG. 1 is a flow chart of a multi-attribute optimal routing method of the present invention;

fig. 2 is an abstract view of a satellite network according to an embodiment of the invention.

Detailed Description

The invention is described in detail below with reference to the drawings and the detailed description.

As shown in figure 1, the invention provides a multi-attribute optimal routing method applied to a low-orbit information network, which is based on a known full-network topological structure, replaces classical additive measurement with fuzzy measurement, takes Choquet fuzzy integral based on the fuzzy measurement as an integrated operator, converts multi-attribute parameters into single comprehensive attribute evaluation parameters, and then utilizes a shortest path algorithm to perform routing. The method comprises the following specific steps:

(1) selecting N attributes which can represent the network quality; such as bandwidth, cost, latency, etc.

(2) (2) calculating an evaluation function f of each attribute value of all edges in the whole network topology structure_m,k(x) K is 1 to N, M is 1 to M, and M is the total number of edges of the topology structure of the whole network; f. of_m,k(x) The specific calculation formula of (A) is as follows:

in which x is the number_mK-th attribute value, p, of an edge_{k_best}Represents the maximum value, p, of the kth attribute of all edges in the full-network topology_{k_worst}The minimum value of the k-th attribute of all edges in the full-network topological structure; p is a radical of_{k_best}And p_{k_worst}Obtained by counting the past data.

Evaluation function f of the above attribute values_m，k(x) For linear mapping, attribute optimum p_{k_best}Mapping to 1, attribute worst value p_{k_worst}The mapping is 0. Taking the cost attribute as an example, p_{k_best}The value of (1) is the minimum value of the path cost in all loop-free paths from the source point S to other nodes; p is a radical of_{k_worst}The maximum value of the path cost in all loop-free paths from the source point S to other nodes is the upper limit and the lower limit that the attribute of the path can reach in the routing process. The linear mapping is adopted because the calculation complexity is small, the linear mapping is suitable for the calculation of the on-satellite router, and the quality degree of the attribute parameters can be reflected really and reasonably.

(3) Calculating the network quality of each edge in the topology structure of the whole network_NChoquet fuzzy integral value A of evaluation function of each attribute_m；

(3.1) abbreviation f_m,k(x)＝f_m,kAccording to f_m，kWith its density function mu (f) of fuzzy measure_m，k) Determining the fuzzy measure mu (f) of N attributes of the mth edge representing the quality of the network_m,k) Specific value of (a), abbreviated as μ (f)_m,k) Is mu_m,k；

The fuzzy measure mu on the set of attributes is used to describe the importance of a certain attribute or several attributes. Can be determined by the following method:

f_m，kwith its density function mu (f) of fuzzy measure_m，k) The correspondence of (a) is obtained as follows: let mu (f)_m，k) Is f_m，kThird order function of, i.e. mu (f)_m，k)＝af_m，k+bf² _m，k+f³ _m，kWherein, for the kth attribute of all edges, the following condition is satisfied:

a+b＝1，

wherein p is_{k_mean}Is the average value of the k-th attribute of all edges in the full-network topology structure.

And can also be obtained by table look-up according to experience.

(3.2) fuzzy measure density mu according to N attributes of all edges_m,kAnd calculating a fuzzy measure parameter lambda according to the following formula:

(3.3) the m-th edge determined according to step (3.1) represents the fuzzy measure mu (f) of the N attributes of the network quality_m,k) And (3) calculating the Choquet fuzzy integral value A according to the specific value and the fuzzy measure parameter lambda determined in the step (3.2)_m:

Will f is_m,kSorting from small to large to obtain f_m,i,i＝1～N：

In the formula, F_m,i＝{f_m,i,f_m,i+1，…f_m,NIs the m-th side after sorting_iSet of attributes to the Nth attribute evaluation function, μ (F)_m,i)＝μ(f_m,i+f_m,i+1+...+f_m,N-1+f_m,N) For any two attribute evaluation functions f with different sequence numbers in the sorted attribute evaluation function set_k'And f_j' the density value of the fuzzy measure satisfies the following relation:

μ(f_m,k'+f_m,j')＝μ(f_m,k')+μ(f_m,j')+λ_mμ(f_m,k')μ(f_m,j') K '≠ j', where μ (F)_N+1)＝0。

(4) And (4) calculating the fuzzy integral value A calculated in the step (3)_mAnd as the measurement weight of the edge in the network, calculating by utilizing a Dijkstra algorithm to obtain the optimal route.

Example (b):

as shown in fig. 2, G is a satellite network abstraction graph (the graph is connected), and there are 5 nodes and 6 edges in the graph.

(1) Selecting two attributes x capable of indicating network quality₁,x₂Wherein x is₁As a tariff, x₂Is the bandwidth. The two attribute values are positive numbers;

(2) calculating an evaluation function f of each attribute value of all edges in the topology structure of the whole network_m,k(x) K is 1 to 2, and m is 1 to 6. Maximum value p of tariff in 6 edges_{1_best}Minimum value of p_{1_worst}(ii) a Maximum value p of band width in 6 edges_{2_best}Minimum value of p_{2_worst}(ii) a Attribute evaluation function f_m,k(x) The specific calculation formula is as follows:

from the above formula, it can be seen that: f. of_m,k(x)∈[0,1]For the tariff attribute, a tariff evaluation function f for each edge may be obtained_1,1(x₁),f_2,1(x₁),f_3,1(x₁),f_4,1(x₁),f_5,1(x₁),f_6,1(x₁) (ii) a The bandwidth evaluation function f of each edge can be obtained for the bandwidth property_1,2(x₂),f_2,2(x₂),f_3,2(x₂),f_4,2(x₂),f_5,2(x₂),f_6,2(x₂)。

(3) Computing Choquet fuzzy integral value A of each edge in the whole network topological structure relative to 2 attribute evaluation functions representing network quality_mM is 1-6; the method specifically comprises the following steps:

(3.1) determining f (x)) The corresponding fuzzy measure density function μ (f (x)). The specific method comprises the following steps: mu (f)_m，k) Is f_m，kThird order function of, i.e. mu (f)_m，k)＝af_m，k+bf² _m，k+f³ _m，kAnd determining the values of the coefficients a and b by using a waiting coefficient method. According to μ (f (x) ═ af (x) + bf²(x)+f³(x) Thus, it can be seen that:

μ(0)＝0

a+b＝1

applying this function to the 1 st attribute (tariff) of all edges of the whole network, then:

order to

Wherein p is_{1_mean}Expressed as the average of all edge tariff attributes in the full network topology.

According to the constraint conditions, coefficients a and b can be determined, so that a fuzzy measure density function mu (f) of the 6-edge tariff attribute can be determined_1,1(x₁)),μ(f_2,1(x₁)),μ(f_3,1(x₁)),μ(f_4,1(x₁)),μ(f_5,1(x₁)),μ(f_6,1(x₁)),。

Applying this function to the 2 nd attribute (bandwidth) of all edges of the whole network, then there are:

order to

Wherein p is_{2_mean}Expressed as the average of all the sideband wide attributes in the full network topology.

Similarly, a fuzzy measure density function μ (f) of the bandwidth attribute of the full network may be determined_1,2(x₂)),μ(f_2,2(x₂)),μ(f_3,2(x₂)),μ(f_4,2(x₂)),μ(f_5,2(x₂)),μ(f_6,2(x₂))。

The fuzzy measure density function can also be obtained empirically by table lookup.

And (3.2) selecting a lambda fuzzy measure and evaluating the relevance among the 2 attributes. Density mu of fuzzy measure according to N attributes of all edges_m,kM is 1 to M, k is 1 to N, and the following formula calculates a fuzzy measure parameter λ:

the density values of the 12 fuzzy measures calculated in (3.1) are formed into a set [ mu (f)_1,1(x)),μ(f_2,1(x₁)),μ(f_3,1(x₁)),μ(f_4,1(x₁)),μ(f_5,1(x₁)),μ(f_6,1(x₁)),μ(f_1,2(x₂)),μ(f_2,2(x₂)),μ(f_3,2(x₂)),μ(f_4,2(x₂)),μ(f_5,2(x₂)),μ(f_6,2(x₂))}

For a set of 12 elements:

in the formula, mu_i'Is a set

The inner elements.

Solving the above equation can obtain a fuzzy measure parameter lambda.

(3.3) computing the Choquet fuzzy integral of each side.

The specific calculation of the Choquet fuzzy integral will be described in detail by taking the calculation of the Choquet fuzzy integral of the first side as an example.

First, will { f_1,1(x₁)，f_1,2(x₂) Sorting according to size, and noting f_1,1(x₁) Is f₁，f_1,2(x₂) Is f₂Let f be₁<f₂To obtain a new set { f₁，f₂}. Choquet fuzzy integral value A₁：

Shorthand F_1,iIs F_i，F_1,i+1Is F_i+1The above formula can be simplified as:

wherein, F_i＝{f_i,f_i+1,…f_NAnd the evaluation function is a set of the ith attribute evaluation function to the Nth attribute evaluation function which are ordered from small to large. A. the₁Can be further developed into:

wherein, F_N+1＝φ μ(φ)＝0。

Similarly, the fuzzy integral value A of other edges can be obtained_m,m＝2～M。

(4) And (4) calculating the fuzzy integral value A calculated in the step (3)_mAnd M is 1-M, which is used as a measurement weight of the edge in the network, and the optimal route is calculated by utilizing a Dijkstra algorithm.

Those skilled in the art will appreciate that those matters not described in detail in the present specification are well known in the art.

Claims (4)

1. A multi-attribute optimal routing method applied to a low-orbit information network is based on a known full-network topological structure and is characterized by comprising the following steps:

(1) selecting N attributes which can represent the network quality;

(2) calculating an evaluation function f of each attribute value of all edges in the topology structure of the whole network_m,k(x) K is 1 to N, M is 1 to M, and M is the total number of edges of the topology structure of the whole network;

(3) calculating the Choquet fuzzy integral value A of the evaluation function of N attribute values of each edge in the whole network topological structure_m；

(4) And (4) calculating the fuzzy integral value A calculated in the step (3)_mAnd as the measurement weight of the edge in the network, calculating by utilizing a Dijkstra algorithm to obtain the optimal route.

2. The method according to claim 1, wherein the evaluation function f of the attribute values in step (2) is_m,k(x) The specific calculation formula of (A) is as follows:

wherein x is the k attribute value of the mth edge, p_{k_best}Represents the maximum value, p, of the kth attribute of all edges in the full-network topology_{k_worst}Is the minimum value of the kth attribute of all edges in the full-network topology.

3. The method according to claim 1, wherein the step (3) calculates Choquet fuzzy integral A of evaluation function of N attributes representing network quality of each edge in the topology of the whole network_mThe method comprises the following specific steps:

(3.1) abbreviation f_m,k(x)＝f_m,kAccording to f_m,kWith its density function mu (f) of fuzzy measure_m,k) Determining the fuzzy measure density mu (f) of N attributes of the m-th edge representing the network quality_m,k) Specific value of (a), abbreviated as μ (f)_m,k) Is mu_m,k；

(3.2) fuzzy measure density mu according to N attributes of all edges_m,kAnd calculating a fuzzy measure parameter lambda according to the following formula:

(3.3) the m-th edge determined according to step (3.1) represents the fuzzy measure mu (f) of the N attributes of the network quality_m,k) In particularTaking values and the fuzzy measure parameter lambda determined in the step (3.2), and calculating a Choquet fuzzy integral value A_m:

Will f is_m,kSorting from small to large to obtain f_m,i,i＝1～N：

In the formula, F_m,i＝{f_m,i,f_m,i+1,…f_m,NIs the m-th side after sorting_iSet of attributes to the Nth attribute evaluation function, U (F)_m,i)＝U(f_m,i+f_m,i+1+...+f_m,N-1+f_m,N) For any two attribute evaluation functions f with different sequence numbers in the sorted attribute evaluation function set_k'And f_j'The density value of the fuzzy measure meets the following relational expression:

U(f_m,k'+f_m,j')＝μ(f_m,k')+μ(f_m,j')+λ_mμ(f_m,k')μ(f_m,j') K '≠ j', where U (F)_N+1)＝0。

4. The method of claim 3, wherein f is the step (3.1)_m,kWith its density function mu (f) of fuzzy measure_m,k) The correspondence of (a) is obtained as follows: let mu (f)_m,k) Is f_m,kThird order function of, i.e. mu (f)_m,k)＝af_m,k+bf² _m,k+f³ _m,kWherein, for the kth attribute of all edges, the following condition is satisfied:

a+b＝1，

wherein p is_{k_mean}Is the average value of the k-th attribute of all edges in the full-network topology structure.

Images