JP4909060B2 - Network topology design method and design system using AHP - Google Patents

Description

  In the network topology design problem, the present invention uses an AHP that obtains an optimal network topology by simultaneously considering a plurality of evaluation scales having different units by evaluating with an AHP (Analytic Hierarchy Process). The present invention relates to a network topology design method and a network topology design system.

  Physical topology design of how to configure physical network topology for telecommunications carriers and some ISPs (Internet Service Providers) that manage and operate physical resources of networks such as optical fibers and routers Is an important issue. The physical topology is a factor that determines the facility costs of nodes and links, network management / maintenance costs, reliability against failures, and the like. In addition, if a logical path such as LSP (Label Switched Path) or wavelength path cannot be installed and the route to which the packet is transferred cannot be set explicitly, the physical topology directly affects the link load or the packet route. This is a factor that determines user quality such as transmission delay quality and throughput.

  On the other hand, when a logical path is installed on a physical network such as an MPLS (Multi Protocol Label Switching) network or a wavelength path network, and a path path between grounds can be explicitly designed, the physical topology is designed by logical topology design. The influence on the user quality can be eliminated to some extent. In this case, in addition to the physical topology design, it is necessary to appropriately design the logical topology. Thus, network topology design problems can be broadly divided into physical topology design problems and logical topology design problems. The outline of each will be described below.

(Physical topology design problem)
As generalized components of the network, consider three types: edge nodes, core nodes, and links. The edge node is a node for switching traffic and accommodates end hosts either directly or via a lower network. On the other hand, the core node is a node having only a relay function and does not accommodate users. The link plays a role of connecting edge nodes and core nodes. It is assumed that an AC traffic matrix between arbitrary edge nodes is given. Telecom operators already have civil infrastructure such as pipelines and stations that are arranged in consideration of geographical characteristics, and it is not realistic to redesign them. Therefore, the position of the edge node and the position where the core node or link can be installed are given.

  It is a practical problem to select these existing pipes and stations according to changes in the requirements of the network and to examine where to place equipment such as optical fibers and routers. Therefore, here, the physical topology design is defined as “an act of selecting a candidate position to be actually installed for a given core node and link installation candidate position”. However, edge nodes are installed at all edge node installation positions.

(Logical topology design problem)
When an MPLS router or OXC (Optical Cross Connect) is used, a virtual path (LSP or wavelength path) can be installed on the physical network, and a route through which traffic between each ground flows can be explicitly designed. The physical topology is difficult to change once constructed, and the design cycle ranges from one year to several years, but the logical topology can be easily changed, and several years according to changes in AC traffic It is possible to perform flexible operations such as dynamically changing paths at intervals of several hours. The logical topology design is what kind of route a virtual path is set up for a given physical topology and an AC traffic matrix between edge nodes. Since the design is based on physical network resources, it is usually necessary to design in consideration of various constraints such as link capacity.

When the logical path can be installed on the physical topology, the logical topology can be designed at the same time when the physical topology is designed. Since the design period of the logical topology is much shorter than that of the physical topology, only the logical topology is designed after the operation is started.
It is desirable for the user to avoid congestion at the transit node and to shorten the transmission delay so that the total distance of the route is short. From the viewpoint of the utilization efficiency and reliability of transmission resources, it is desirable to avoid the concentration of traffic on a specific link and to distribute it evenly over all links as much as possible. However, if the number of installed nodes and the number of links are reduced in order to reduce the equipment cost and management / maintenance cost, the flexibility of path setting is reduced, and it becomes difficult to appropriately distribute traffic and suppress the path length. As described above, when designing the network topology, it is necessary to simultaneously consider a plurality of evaluation measures having different units such as the degree of traffic distribution, the average path length, and the network cost, which are mutually contradictory.

A parade optimum concept used in the field of microeconomics is a typical method for considering a plurality of evaluation measures at the same time, and examples of application to logical topology design are also seen. If there are _M evaluation scales V _{I to} V _{M and} the m-th evaluation scale value of the candidate x is V _{m, x} , then V _{m, x} ≦ V _{m, y} for any m, and , V _{m, x} <V _{m, y} is present, m is said to dominate candidate y in a Pareto manner (provided that all evaluation measures are smaller and better). Then, it is assumed that all candidates for which there are no other controlled candidates are Pareto optimal.

As another method for simultaneously considering a plurality of evaluation scales, DEA (Data Environment Analysis) is known. DEA is an evaluation method that has been developed mainly as a method for evaluating investee entities, and it is possible to obtain efficiency from the attitude of evaluating each business entity's strength (evaluation scale). Is also being applied to network topology design.
In addition, a method of calculating ranks in all candidate sets of each evaluation scale for each candidate topology and evaluating the superiority or inferiority of the candidate topologies using an average value in all the evaluation scales has been studied.
The above technique is described in Non-Patent Document 1 (“Network Topology Design Considering Multiple Evaluation Scales” (The Institute of Electronics, Information and Communication Engineers, IEICE Technical Report, vol. 106, no. 118, IN 2006-31, pp. 85-90). , June 2006)

“Network Topology Design Considering Multiple Evaluation Scales” (The Institute of Electronics, Information and Communication Engineers, IEICE Technical Report, vol. 106, no. 118, IN 2006-31, pp. 85-90, June 2006)

When network design is performed by Pareto optimization, there are many candidates that are determined to be optimal, and it is difficult to effectively narrow down candidates.
Moreover, DEA has an intuitively difficult concept of efficiency, and it is difficult to determine the optimality of the obtained solution. In addition, the efficiency calculation needs to solve a linear programming problem, and there is a problem in scalability.
In addition, the method using the average rank has a problem that it cannot make a difference in the degree of emphasis among the evaluation scales, and the shape of distribution among all candidates of each evaluation scale cannot be considered.

(the purpose)
The object of the present invention is to give a difference in the degree of emphasis between the evaluation scales, reflect the difference in distribution form of the evaluation scales in the result, and simultaneously consider a plurality of evaluation scales with different units that can be intuitively understood. It is an object of the present invention to provide a network topology design method and design system using AHP that can obtain an optimum network topology.

The network topology design method using AHP according to the present invention provides a set of edge nodes that accommodate users, a set of candidate core nodes having only a relay function, and a set of link candidates connecting them. The method is characterized by selecting a set of core nodes and links to be used while simultaneously considering a plurality of evaluation measures.
Further, in the above network topology design method, a pairwise comparison between evaluation scales is performed according to the AHP method to calculate the weight of each evaluation scale, and the evaluation scale is used as a score value for each evaluation scale of each candidate topology. It is also characterized in that the superiority or inferiority of each candidate is determined based on the score value of each candidate that is finally calculated using a value obtained by normalizing the value obtained by subtracting the minimum value among all candidates.
Further, in the above network topology design method, as examples of evaluation measures to be considered at the same time, the total number of nodes, the total link length, the product of the length of each path and the traffic amount for all paths, the traffic of the maximum load link It is also characterized by the use of four quantities.

  The network topology design system using AHP according to the present invention includes a topology information / demand matrix input device to which an edge node position, a core node installation position, and a link installation position are input, and an initial topology for generating an initial topology. A generation apparatus, a topology candidate generation apparatus that generates a plurality of candidate topologies with some links and core nodes added to the initial topology, an AHP application apparatus that evaluates superiority or inferiority between candidate topologies, and the obtained optimal topology And an optimum topology output device for outputting the set.

  According to the present invention, when a location where a core node or a link can be installed and an AC traffic matrix between edge nodes and edge nodes are given, an optimal network topology is considered in consideration of a plurality of evaluation measures. It has the effect that it is calculated | required by evaluating by.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a system configuration diagram of a network topology design apparatus according to an embodiment of the present invention.
In FIG. 1, 101 is a topology information / demand matrix input device, 102 is an initial topology generation device, 103 is a topology candidate generation device, 104 is an AHP application device, and 105 is an optimum topology output device. All of these devices operate under the control of a computer.
The topology information / demand matrix input device 101 inputs the edge node position, the core node installation position, and the link installation position. An initial topology is generated by the initial topology generation apparatus 102. A plurality of candidate topologies obtained by adding several links and core nodes to the initial topology are generated by the topology candidate generation apparatus 103, and the superiority or inferiority of the candidate topologies is evaluated by the AHP application apparatus 104.
Then, the obtained optimal topology set is output by the optimal topology output device 105.

Next, a network topology design method using AHP according to the embodiment of the present invention will be described in detail.
As an example, the study will proceed with the physical topology design, but it is assumed that the logical path is installed at the upper level, and the logical path is also installed at the same time as the physical topology design (however, constraints such as link capacity are not considered) .
However, the network topology design method of the present invention focuses on how to consider a plurality of evaluation measures at the same time, and the constraint conditions are highly versatile. It can also be applied to topology design.

(Evaluation scale)
Consider the evaluation measures that should be considered when designing the physical topology. Considering a backbone network as a design object, the physical topology requires connectivity between all edge nodes and redundancy that can maintain connectivity even when a failure occurs. Therefore, here, as constraints for physical topology design, (a) connectivity between all edge nodes and (b) connectivity between all edge nodes is maintained by a detour path even in the event of a link failure. think of.
Various evaluation scales can be considered. Here, the following four cases are considered as examples (the i-th evaluation scale is expressed as V _i ).

(1) Total nodes: the total number of the combined installation edge nodes and core nodes, the number of edge nodes N _e, When the N _c installation core node number,

とする。Ｎ_ｃは全ての候補トポロジで同一であるため、Ｎ_ｃの差異がそのままＶ_１の差異となる。Ｖ_１は小さい方が望ましい。

And Since N _c is the same in all candidate topologies, the difference in N _c becomes the difference in V ₁ as it is. A smaller V ₁ is desirable.

(2) total link distance: is the sum of the distance of the installed link, the distance of link d _l l, when the set of installed links E _a,

とする。Ｖ_２も小さい方が望ましい。

And It is desirable that V _{2 is} also small.

(3) Sum obtained by weighting path distance by traffic amount: It is desirable that the path length is short from the viewpoint of transmission delay, and sum ζ = Σ _{s, dεN e, s} , wherein path length is weighted by path traffic amount. _{≠ d} D _sd T _sd is considered as an evaluation scale. However, the path through which the traffic directed from the edge node s to d _{P sd,} path _{P sd} traffic amount _{T sd,} path _{P sd} set the [Phi _sd links through which the path _P the length of the _{sd D sd} ( D _sd = Σ _{lεΦ sd} ^dl ). If t _l is the total traffic amount of link l, ζ matches the sum of the products of link distance d _l and traffic amount t _l ,
ζ = Σ _le E _a ^{dl × tl} . Therefore,

とする。Ｖ_３も小さい方が望ましい。

And It is desirable that V _{3 is} also small.

(4) Maximum link load: When traffic is concentrated on a specific link, the transmission quality of the path passing through the bottleneck link is greatly deteriorated. Also, since the transmission bandwidth of the link is increased in discrete units such as optical fibers and wavelengths, it is desirable to suppress and disperse the link load in order to effectively use the transmission resources. Therefore, the traffic volume of the maximum load link is considered as an evaluation measure,

とする。Ｖ_４も小さい方が望ましい。

And It is desirable that V _{4 is} also small.

(Create candidate topology)
When links and core nodes are arranged freely at a given installable position, the obtained topology candidates include those that do not satisfy the constraint conditions. Therefore, a physical topology satisfying the constraint conditions is first created as an initial solution, and a set of physical topologies created by additionally deploying arbitrary core nodes and links to the initial solution is set as a solution candidate. As a result, since the constraint condition is already satisfied at the stage of creating the initial solution, it is possible to adopt an arbitrary candidate topology.

Let z be the number of link installable positions remaining excluding the links arranged in the initial topology. Any installable results established the link to the position, the total number of the resulting topology is the 2 ^z-number. In each topology created in this way, the core node is placed at the core node installable position where the order is 3 or more, but the exchange function is not necessary at the core node installable position where the order is 2. Directly connect two placement links without installing a core node. In addition, since the topology in which the position where the core node can be placed having the order of 1 exists is an inappropriate topology, it is excluded from the candidates for examination.

Next, for each topology created as a candidate, a logical path is set between the edge nodes using a given AC traffic matrix. Here, only one path Psd is installed for the traffic flowing from the node s to d, and the path Psd is installed by a construction method (the installed path is not changed) in descending order of the requested traffic amount Tsd. . When determining the route of each path, the route that minimizes D _sd = Σ _{lεΦ sd} ^{dl × t′l} is selected. However, the traffic amount of the link l at the time of installing the corresponding path is t ′ _l . Link capacity constraints are not taken into account when setting up paths. As a result of setting all the paths, the topology in which there is a link through which no path passes is still an inappropriate topology and is excluded from the candidates for examination.

(Topology evaluation by AHP)
FIG. 2 is a block diagram of an AHP application model in the present invention.
When AHP is applied to the network topology design, as shown in FIG. 2, the problem P to be solved (in this case, “optimal network selection”) is the central hierarchy (hierarchy 1). rating scale V _i is in is, and candidate topology lined the bottom (Tier 2). Here, the weight for the element j in the hierarchy c-1 of the element i in the hierarchy c (c = 1, 2) is expressed as w ^c _ij . Consider the derivation of the final score S _i for each candidate i.
In FIG. 2, V _{1 to} V ₄ (for example, the total number of nodes, the total link distance, the sum obtained by weighting the path distance by the traffic amount, and the maximum link load) are shown as the evaluation scale Vi, and the candidate topology Ai is A _{1 to} A ₈₃₈₆₈ are shown.

FIG. 3 is a diagram showing a paired comparison value and its meaning in the present invention.
Since the superiority or inferiority of the hierarchy 1 (evaluation scale) depends on the intention of the network designer, the weight for the problem P needs to be quantified by some method. In AHP, the importance level a _ij of the evaluation scale j with respect to the evaluation scale i is set to the value shown in FIG. 3 by performing pairwise comparison. The vector having the element with the weight w ^c _ij matches the eigenvector corresponding to the maximum eigenvalue λ _max of the matrix A having the element a _ij . Here, the degree of consistency C.I. I. (Consistency index) is used. C. I. = (Λ _max −N _c ) / (N _c −1) (N _c is the number of elements in layer c). C. I. If it is 0.1 or less, it can be usually judged that the consistency is acceptable.
In FIG. 3, as an example, when the paired comparison value as the importance level a _ij is 1, a certain evaluation scale is as important as the other evaluation scales, and when 3, the former is slightly more important than the latter. Indicates that the former is more important than the latter, 7 indicates that the former is much more important than the latter, and 9 indicates that the former is absolutely more important than the latter. Note that 2, 3, 6, and 8 are the importance levels located in the middle of 1, 3, 5, 7, and 9, respectively. In addition, 1-1 / 9, which is the reciprocal of 1-9, is used when the former is viewed from the latter.

The score S _i ^c for the problem P of the element V _i ^c is obtained by the following equation using the scores S _i ^c-1 of all elements related to V _i ^c of the hierarchy c-1.

ただし、Φ_ｉ ^ｃはＶ_ｉ ^ｃが関係する要素の集合である。階層１においては、Ｐに対する重みがそのままスコアとなる。以下、ｃ＝２についてスコアが算出できる。

However, Φ _i ^c is a set of elements related to V _i ^c . In the hierarchy 1, the weight for P becomes the score as it is. Hereinafter, a score can be calculated for c = 2.

The value of the evaluation scale j of the candidate i is expressed as V _ij . As the weight of the candidate topology i in the evaluation scale j, since V _ij has a quantitative value, normally, in AHP, a value X _ij obtained by simply normalizing V _ij among all candidates is used. Therefore, in the case of obtaining the score _S ^{i c} is the value _{X ij} may be assigned to _{omega ij} ^c in the formula above [equation 5].

しかし、網トポロジの候補は、通常、非常に多数となるため、このままでは上式中の分母が大きくなり、候補間のＸ_ｉｊの値の差異が僅かとなる。その結果、単に重視された評価尺度が良好な候補を選択する傾向が強くなる。

However, since there are usually a large number of network topology candidates, the denominator in the above equation becomes large as it is, and the difference in the value of X _ij between the candidates becomes small. As a result, there is a strong tendency to select candidates with a good evaluation scale that is simply emphasized.

In the present invention, to solve this problem, Z _ij defined by the following equation is used as a weight.

ただし、ｍｉｎ｛Ｖ_ｊ｝は全候補中のＶ_ｉｊの最小値である。この値Ｚ_ｉｊも正規化されているので、スコアＳ_ｉ ^ｃを得る場合には、この値Ｚ_ｉｊを上記〔数５〕の式中のω_ｉｊ ^ｃに代入すればよい。

However, min {V _j } is the minimum value of V _ij among all candidates. This value Z _ij is also normalized, in the case of obtaining the score S _i ^c is the value Z _ij may be assigned to omega _ij ^c in the formula above [equation 5].

As described above, the first feature of the network topology design method according to the present invention is that a set of edge nodes that accommodate users, a set of candidate core nodes having only a relay function, and a set of link candidates connecting them. When given, it is to select a set of core nodes and links to be used while simultaneously considering a plurality of evaluation measures (see paragraph [0016] above and FIG. 1).
The second feature is that, in the network topology design method described above, a pairwise comparison between evaluation measures (see FIG. 3) is performed according to the AHP method, whereby the weight of each evaluation measure is calculated, and each evaluation of each candidate topology is further performed. As a score value for the scale, a value obtained by subtracting the minimum value among all candidates from the value of the evaluation scale (refer to each formula of [Formula 6] and [Formula 7]) is finally calculated. In other words, the superiority or inferiority of each candidate is determined by the score value of each candidate (see the equation of [Formula 5]).
Furthermore, the third feature is that the total number of nodes, the total link length, the product of the length of each path and the traffic amount are added to all the paths as an example of an evaluation measure to be considered simultaneously in the above network topology design method. The maximum load link traffic volume (see paragraphs [0018] (evaluation scale) (1) to (4)) described above.

Other evaluation methods include the following topology evaluation.
(Topology evaluation by average rank)
For each candidate topology i, the rank in the N candidate set of each evaluation scale is calculated. Since all the evaluation scales are preferably as small as possible, the rank can be calculated simply by sorting the evaluation scales V _{k, i} in ascending order (k = 1, 2, 3, 4).
Using the rank r (k, i) obtained for each evaluation scale V _{k, i} of topology i in this way, the average rank (AR) of topology i is calculated by the following equation (8), and the average is simply calculated. A candidate topology having a small rank (AR) value is evaluated as an excellent topology. That is, among all N candidate topologies, the topology with the smallest average rank (AR) value is determined as the optimum network topology.

通常、単位の異なる複数の評価尺度を単純に足し合わせて評価することはできない。しかし、各評価尺度を全候補中の順位という相互比較が可能な同一の尺度に置き換えることにより、単純にそれらの平均値で最終的な優劣が評価できる。
平均順位法は、任意数Ｍ個の評価尺度を同時に考慮することができ、その場合、トポロジｉの平均順位（ＡＲ）は下記（９）式となる。

Usually, it is not possible to simply add a plurality of evaluation scales having different units for evaluation. However, by replacing each evaluation scale with the same scale that can be compared with each other in terms of rank among all candidates, it is possible to simply evaluate the final superiority or inferiority based on their average value.
In the average rank method, an arbitrary number M of evaluation scales can be considered at the same time. In this case, the average rank (AR) of topology i is expressed by the following equation (9).

評価値である平均味位の計算は非常に簡易である。各評価尺度の順位を求めるのにソート処理が必要であるが、ＭｅｒｇｅソートやＱｕｉｃｋソートといったソートアルゴリズムを用いれば０（ＮｌｏｇＮ）の処理量で済むことから、ＤＥＡと比較して遥かに短い時間で網トポロジ設計が可能である。

Calculation of the average taste, which is an evaluation value, is very simple. Sort processing is required to determine the rank of each evaluation scale. However, if a sort algorithm such as Merge sort or Quick sort is used, a processing amount of 0 (NlogN) is sufficient, so it takes much shorter time than DEA. Network topology design is possible.

It is a block diagram of the network topology design system using AHP which concerns on one Embodiment of this invention. It is a figure which shows the AHP application model in this invention. It is a figure which shows the pair comparison value and its meaning in this invention.

Explanation of symbols

101 Topology Information / Demand Matrix Input Device 102 Initial Topology Generation Device 103 Topology Candidate Generation Device 104 AHP Application Device 105 Optimal Topology Output Device

Claims (3)

A topology design method for applying AHP to topology design considering a plurality of evaluation measures by computer control,
The topology information demand matrix input device gives a set of edge nodes that accommodate users, a set of candidate core nodes having only a relay function, and a set of candidate links that connect the edge node set and the core node set,
An initial topology generating device generates an initial topology from the plurality of sets;
The topology candidate generation device generates a plurality of candidate topologies by adding some links and core nodes to the generated initial topology,
The AHP application apparatus calculates the weight of each evaluation measure by performing pairwise comparison between evaluation measures according to the AHP method for the plurality of candidate topologies, and further, as a score value for each evaluation measure of each candidate topology , Using the value obtained by subtracting the minimum value of all candidates from the value of the evaluation scale, and determining the superiority or inferiority of each candidate by the score value of each candidate finally calculated,
A network topology design method using AHP, wherein an optimum topology output device outputs a selected optimum topology. The network topology design method according to claim 1 ,
As examples of the evaluation measures to be considered simultaneously by the AHP application apparatus, there are four types: total node number, total link length, product of the length of each path and traffic amount for all paths, and traffic amount of maximum load link. A network topology design method using AHP, characterized in that A topology design apparatus that applies AHP to topology design considering a plurality of evaluation measures by computer control,
Topology information / demand matrix input device for inputting the position of the edge node, the installable position of the core node, and the installable position of the link;
An initial topology generator for generating an initial topology;
A topology candidate generator for generating a plurality of candidate topologies by adding several links and core nodes to the initial topology;
According to the AHP method, a pairwise comparison between evaluation scales is performed to calculate the weight of each evaluation scale. Further, as a score value for each evaluation scale of each candidate topology, the minimum value among all candidates is calculated from the value of the evaluation scale. An AHP application device that evaluates the superiority or inferiority of candidate topologies by judging the superiority or inferiority of each candidate using the score value of each candidate that is finally calculated, using what is subtracted ,
A network topology design system using AHP, comprising: an optimum topology output device that outputs a set of obtained optimum topologies.

Images