JP6737747B2 - MPP network, MPP network construction method, MPP network design apparatus, and MPP network design method - Google Patents

Description

The present disclosure discloses an MPP network in which a structure of non-blocking (to realize an arbitrary connection request) is constructed by using a plurality of mechanical patch panels (MPPs), a construction method thereof, a design device and a design thereof. Regarding the method.

Fiber optic cabling is a tedious task, but it is an essential foundation for fast and robust network services. The operating costs of these tasks in the network office building are high. Because these tasks usually require work procedure preparation, dispatch of two workers to connect NW equipment at both ends of the cable, cable communication and power level testing, and control of these tasks. This is because In addition, human operators may misroute cables, which may cause degradation of quality of service and/or additional work to correct these work mistakes. If the wiring work can be performed automatically and remotely without human intervention, the cost required for these works can be reduced. Furthermore, if this is realized, it would be possible to dynamically change the network topology in accordance with, for example, traffic demand. In this way, automation of wiring work not only reduces costs but also opens the possibility of new network operations.

The mechanical patch panel (MPP) (Non-Patent Document 1) may realize automation of wiring work in a network (NW) station building. Like a normal patch panel, the mechanical patch panel has N optical fiber ports on the front surface and the same number of optical fiber cables are connected to the rear surface (FIG. 1). In a normal patch panel, the front port and the back fiber are internally connected one-to-one. On the other hand, in the mechanical patch panel, the correspondence between the front port and the back fiber can be set arbitrarily (FIGS. 1 and 2). This is achieved by a mechanical arm incorporated inside.

All wiring would need to be accommodated to automate the wiring work in the NW building, which would require the use of multiple MPPs. This is because it is difficult to construct a huge mechanical patch panel due to physical constraints and economical cost constraints. Actually, the size of the current patch panel is about 1000, which is insufficient to accommodate all the connections in the NW building. The size of the patch panel is the number of optical fiber connection ports formed on the panel surface.

Therefore, in order to realize the automation of the wiring work, it is necessary to construct a patch panel network in which a plurality of mechanical patch panels are connected and to realize any connection pattern. A network that can satisfy any such connection pattern is called non-blocking.

A mechanical patch panel network such as that of FIG. 3 cannot satisfy any connection pattern. For example, x _i and z _j (i, jε{1,..., 6}) cannot be connected. The number of connections between x _i and y _j is at most 3.

Even if MPP is added as shown in FIG. 4, it does not become non-blocking. For example, the connection between x _i and y _j is at most 2, and x _i and x _j cannot be connected.

A network structure called a non-blocking network solves such a problem. The classical Clos network (Non-Patent Document 2) and the like are famous as non-blocking networks. The Clos network is mainly used when constructing a large non-blocking network (a switch having a large number of input/output) from a small non-blocking network (a switch having a large number of input/output).

In fact, the intermediate folded structure of the Clos net can be applied to a mechanical patch panel net to meet any connection requirements. For example, an arbitrary connection request can be satisfied by using the folded structure of the three-stage Clos network shown in FIG. It is possible to connect x _i and y _j , x _i and z _j , x _i and w _j , and the maximum number of connections is 6, and it is also possible to connect x _i and x _j . However, it may be necessary to relocate existing connections. This means that there are cases where it is necessary to modify the intermediate mechanical patch panel used by an existing connection in order to implement the new connection. By the way, such relocation of the existing connection may not be permitted depending on the nature of the service provided by the wiring or the service contract. In such a case, the arbitrary connection pattern cannot be satisfied in FIG. However, it is known that by increasing the number of intermediate mechanical patch panels, it is possible to satisfy an arbitrary connection request even when relocation of existing wiring is not permitted (Non-Patent Documents 2 and 3). ..

A. S. Kewitsch, "Large scale, all-fiber optical cross-connect switches for automated patch-panels," Journal of Lightwave Technology. 27, no. 15, pp. 3107-3115, Aug 2009. C. Clos, "A study of non-blocking switching networks", Bell System Technical Journal, vol. 32, no. 2, pp. 406-424, 1953. F. Hwang, The mathematical theory of nonblocking switching networks. World Scientific, 2004, vol. 15.

The non-blocking MPP network has a problem that a large number of MPPs are required. Therefore, in order to solve the above-mentioned problems, the present invention provides an MPP network having a network structure having a higher accommodation efficiency than an MPP network composed of a conventional Clos network, a construction method thereof, a design apparatus thereof, and a design method thereof. The purpose is to Note that good accommodation efficiency means that a small number of MPPs are required to achieve the same number of connections without being blocked.

In order to achieve the above object, the MPP net according to the present invention is a mechanical patch panel (MPP) net having a structure in which a three-stage Clos net is folded, and both sides of the outer (first and third stage) MPP are An input/output port and an intermediate (second stage) connection port for connecting to the MPP were installed.

Specifically, the MPP network according to the present invention is a mechanical patch panel (Mechanical) having a size N (N is a natural number of 2 or more) that is the number of optical fiber ports on the first surface and the number of optical fiber ports on the second surface. An MPP network including a plurality of (P+1) Patch Panels (kPl),
Of the MPPs, k are input/output MPPs, l are intermediate MPPs,
Of the plurality of optical fiber ports on the first surface (11) of the input/output MPP, n optical fiber ports are ports for connecting to the outside, and m×l optical fiber ports are the second surface of the intermediate MPP. The optical fiber connection port for connecting the m optical fiber ports in (22) with the optical fiber,
Out of the plurality of optical fiber ports on the second surface (12) of the input/output MPP, n optical fiber ports are ports for external connection, and m×l optical fiber ports are the first surface of the intermediate MPP. It is characterized in that it is used as an optical fiber connection port for connecting with m optical fiber ports in (21) by an optical fiber.
However, k, l, m, and n are natural numbers,
n+m·l≦N,
m·k≦N, and 2 floor((n−1)/m)+1≦l
Meet

The MPP network construction method according to the present invention is a mechanical patch panel (Mechanical) having a size N (N is a natural number of 2 or more) that is the number of optical fiber ports on the first surface and the number of optical fiber ports on the second surface. A method for constructing an MPP network having a plurality of (P+1) Patch Panels (kPl),
Of the MPPs, k are input/output MPPs and 1 is an intermediate MPP,
Of the plurality of optical fiber ports on the first surface (11) of the input/output MPP, n optical fiber ports are ports for connecting to the outside, and m×l optical fiber ports are the second surface of the intermediate MPP. The optical fiber connection port for connecting the m optical fiber ports in (22) with the optical fiber,
Out of the plurality of optical fiber ports on the second surface (12) of the input/output MPP, n optical fiber ports are ports for external connection, and m×l optical fiber ports are the first surface of the intermediate MPP. It is characterized in that it is an optical fiber connection port for connecting with m optical fiber ports in (21) by an optical fiber.
However, k, l, m, and n are natural numbers,
n+m·l≦N,
m·k≦N, and 2 floor((n−1)/m)+1≦l
Meet

Further, the MPP network designing apparatus according to the present invention is
A parameter input unit into which the required number r of connection ports and the size N of the MPP are input,
An optimization calculation unit that uses r and N input to the parameter input unit and performs a full search for k, l, m, and n that minimizes the number of MPPs (k+1) while satisfying condition 1.
An optimal structure output unit that outputs k, l, m, and n that the optimization calculation unit has performed a full search;
Equipped with.
However, condition 1 is
n+m·l≦N,
m·k≦N,
2floor((n-1)/m)+1≦l, and n·k≧r
Is.

Further, the MPP network design method according to the present invention is
A parameter input procedure for inputting the required number r of connection ports and the size N of the MPP;
An optimization calculation procedure for using k and l, m, and n that minimizes the number of MPPs (k+1) while satisfying the condition 1 using r and N input in the parameter input procedure;
An optimal structure output procedure that outputs k, l, m, and n that have been fully searched in the optimization calculation procedure;
How to design.
However, condition 1 is
n+m·l≦N,
m·k≦N,
2floor((n-1)/m)+1≦l, and n·k≧r
Is.

As will be described later, the present invention can provide an MPP network corresponding to a service in which connection relocation is not possible. As will be described later, if the number of connection requests that can be accommodated is the same, the number of MPPs is greater than that in the conventional Clos structure. Can be reduced. Therefore, the present invention can provide an MPP network having a network structure with better accommodation efficiency than an MPP network configured by a conventional Clos network, a construction method thereof, a designing apparatus thereof, and a designing method thereof.

On the other hand, if the following is done, it is possible to provide an MPP network compatible with services in which connections can be rearranged.

Specifically, in the MPP network according to the present invention, the intermediate MPP is divided into two groups, one of the intermediate MPPs of the group is used as a service path of connection relocation prohibition, and the other intermediate MPP of the group is used. It is characterized in that the MPP is used as a service path for which the connection can be relocated.

Further, in the method for constructing an MPP network according to the present invention, the intermediate MPP is divided into two groups, one of the intermediate MPPs of the group is used as a service route of connection relocation prohibition, and the other intermediate MPP of the group is used. It is characterized in that the MPP is used as a service path for which the connection can be relocated.

Further, the MPP network designing apparatus according to the present invention is
A parameter input unit into which the required number r of connection ports and the size N of the MPP are input,
Using r and N input to the parameter input unit, k, l ₁ , l ₂ , m ₁ , m ₂ , n that minimizes the number of MPPs (k+l ₁ +l ₂ ) while satisfying the condition 2 An optimization calculation unit that searches all ₁ and n ₂ ;
An optimal structure output unit that outputs k, l ₁ , l ₂ , m ₁ , m ₂ , n ₁ and n ₂ that the optimization calculation unit has searched in full;
Equipped with.
However,
l=l ₁ +l ₂ ,
m=m ₁ +m ₂ ,
n=n ₁ +n ₂
And condition 2 is
n ₁ +n ₂ +m ₁ ·l ₁ +m ₂ ·l ₂ ≦N,
m ₁ ·k≦N,
m ₂ ·k≦N,
2floor((n ₁ −1)/m ₁ )+1≦l ₁ ,
n ₁ +n ₂ ≦m ₂ ·l ₂ and (n ₁ +n ₂ )k≧r
Is.

Further, the MPP network design method according to the present invention is
A parameter input procedure for inputting the required number r of connection ports and the size N of the MPP;
Using r and N input in the parameter input procedure, k, l ₁ , l ₂ , m ₁ , m ₂ , n that minimizes the number of MPPs (k+l ₁ +l ₂ ) while satisfying the condition 2 An optimization calculation procedure for searching all ₁ and n ₂ ;
An optimal structure output procedure that outputs k, l ₁ , l ₂ , m ₁ , m ₂ , n ₁ and n ₂ that have undergone a full search in the optimization calculation procedure;
I do.
However,
l=l ₁ +l ₂ ,
m=m ₁ +m ₂ ,
n=n ₁ +n ₂
And condition 2 is
n ₁ +n ₂ +m ₁ ·l ₁ +m ₂ ·l ₂ ≦N,
m ₁ ·k≦N,
m ₂ ·k≦N,
2floor((n ₁ −1)/m ₁ )+1≦l ₁ ,
n ₁ +n ₂ ≦m ₂ ·l ₂ and (n ₁ +n ₂ )k≧r
Is.

INDUSTRIAL APPLICABILITY The present invention can provide an MPP network having a network structure having a higher accommodation efficiency than an MPP network configured by a conventional Clos network, a construction method thereof, a design apparatus thereof, and a design method thereof.
If a network of non-blocking patch panels consisting of mechanical patch panels can be constructed, there is a possibility that wiring work in the network building can be automated.

(A) It is a figure explaining MPP. (B) The MPP can arbitrarily change the correspondence between the front and back ports by changing the internal connection settings. It is a figure explaining MPP. By changing the internal settings of the mechanical patch panel, arbitrary x _i and y _j (i, jε{1, 2, 3, 4}) can be connected. It is a figure explaining the MPP network which is not non-blocking. In this example, x _i and z _j (i, jε{1,..., 6}) cannot be connected. It is a figure explaining the MPP network which is not non-blocking. For example, x _i and x _j cannot be connected. It is a figure explaining a non-blocking MPP network. However, relocation of existing connections is allowed. This structure is obtained by folding a 3-stage Clos network with an intermediate layer. Here, the MPP size N is 8 or more. It is a figure explaining a non-blocking MPP network. Folded Clos structure composed of k+1 MPPs. This structure can meet a maximum of nk connection requirements. It is a figure explaining the MPP network which concerns on this invention. Proposed structure using k+1 MPPs. Both sides of the input/output MPP are used for connection with the intermediate MPP and for device connection. This MPP network can satisfy a maximum of nk connection requests. It is a figure explaining the MPP network which concerns on this invention. Proposed structure using k+l ₁ +l ₂ MPPs. Intermediate MPPs for non-relocatable request and intermediate MPPs for relocatable type request are prepared as l ₁ and l ₂ , respectively. This MPP network can satisfy a maximum of (n ₁ +n ₂ ) k connection requests. Solid lines are connection wirings for request that cannot be rearranged, and dotted lines are connection wirings for request that can be rearranged. It is a figure explaining the effect of the MPP network which concerns on this invention. The relationship (N=1000) between the number of used MPPs and the number of requests that can be accommodated when all connection requests cannot be relocated is shown. It is a figure explaining the effect of the MPP network which concerns on this invention. The relationship between the number of used MPPs and the number of requests that can be accommodated (N=1000) when non-relocatable requests and relocatable requests coexist is shown. It is a figure explaining the design apparatus which designs the MPP network which concerns on this invention.

Embodiments of the present invention will be described with reference to the accompanying drawings. The embodiments described below are examples of the present invention, and the present invention is not limited to the following embodiments. In addition, in the present specification and the drawings, components having the same reference numerals indicate the same components.

First, the MPP construction problem and the existing structure will be described in order to explain the structure of the MPP network according to the present invention. Next, the structure when all connection requests cannot be relocated will be described. Finally, the structure in the case where non-relocatable requests and relocatable requests are mixed will be described.

1. Introduction In order to explain the MPP network structure according to the present invention, the non-blocking property of the MPP network and the existing structure will be described first. For that purpose, an MPP network and a connection request for it are defined. Moreover, the conditions under which the existing structure is not blocked will be summarized.

1.1 MPP network The MPP network is assumed to have a plurality of MPPs connected as shown in FIG. It is assumed that all the fibers used in the NW station building are of the same type and any fiber can be accommodated in the MPP. The size of the front surface and the back surface of the MPP is the same, which is written as N. By changing the internal connection settings, it is possible to realize an arbitrary connection relationship between the front port and the back port. Since the front surface and the back surface have equivalent functions, they will not be distinguished hereinafter unless otherwise specified.

The MPP network is defined as a network in which a plurality of MPPs are connected by a fiber. In FIG. 5, MPPs {P ₁ , P ₂ , P ₃ , P ₄ } on the left side are connected to MPPs {P ₅ , P ₆ , P ₇ } on the right side. Of the ports not used for MPP connection, some are used as device connection ports. In FIG. 5, the left side surface of the left MPP {P ₁ , P ₂ , P ₃ , P ₄ } is used as a device connection port. Here, it is assumed that it is not necessary to use all free ports to satisfy an arbitrary connection pattern, that is, non-blocking property. The MPP network of FIG. 5 is not non-blocking when the connection request cannot be relocated. However, it becomes non-blocking by restricting not to use some free ports as described later.

1.2 Connection Request There are the following two connection requests to the MPP network. It is a request to create a new connection or a request to terminate an existing connection. For the new connection request, two unused ports that are not used in the MPP network are designated. It is required to change the internal settings of MPP so that they are connected. Two connection requests arrive as one pair. This is because an optical fiber normally uses two cores as a pair and is used for one connection, one for transmission and the other for reception. For example, two pairs of connection x ₁ and y ₁ and connection x ₂ and y ₂ arrive as one connection request. A request for creating such a connection is written as {(x ₁ , y ₁ ), (x ₂ , y ₂ )}. The termination request is cancellation of the connection request, and similarly, two pairs arrive as one. For example, it is requested as cancellation of the two pairs of connections of x ₁ and y ₁ and x ₂ and y ₂ that have been requested in the past. These requests sequentially arrive, and future requests are unknown.

Among the connection requests, those that cannot change the connection path are called non-relocation, and those that can be changed are called relocation. The connection path is a wiring for MPP connection used when connecting input/output ports (for example, x ₁ and y ₁ ). For example, assuming that x ₁ and y ₁ are connected to P ₁ , P ₅ , and P _2, and if this connection can be rearranged (if there is an empty wiring), connect as P ₁ , P ₆ , and P ₂ . It is also possible. A connection request that cannot be relocated cannot have such a route change.

The request only includes creating and canceling connections, but a switch request can also be expressed as: Here, the switching request is, for example, changing (x ₁ , y ₁ ) to (x ₁ , z ₁ ) (for simplicity of description, one of the two pairs is omitted for description). .. The switching request can be expressed as cancellation of (x ₁ , y ₁ ) and creation of (x ₁ , z ₁ ). Also, switching back can be expressed. Switching back is, for example, canceling the switching because a communication problem has occurred after switching the connection. Using the previous example, this can be expressed as canceling the new connection (x ₁ , z ₁ ) and creating the old connection (x ₁ , y ₁ ).

1.3 Non-blocking MPP network is non-blocking means that the MPP internal connection can be set to satisfy any connection request that arrives, provided that the connection request is to an unused device connection port. Suppose it is requested. For example, it is assumed that the connection creation request (x ₁ , y ₁ ) arrives and the connection creation request (x ₁ , z ₂ ) does not arrive before the connection cancellation request arrives. Existing connection requests that can be relocated upon request arrival are allowed to change connection routes.

For example, the MPP network of FIG. 4 is (x ₁ , y ₁ ), (x ₂ , y ₂ ), (x ₃ , y ₃ ), (x ₄ , y ₄ ), (x ₅ , y ₅ ), (x ₆ ,y ₆ ) cannot be satisfied, so it is not non-blocking. Note that the MPP network in FIG. 4 is not non-blocking regardless of whether the request can be relocated or not.

Further, the MPP network of FIG. 5 is not non-blocking when the request cannot be relocated. In fact, it is possible to construct a connection request that cannot be satisfied as follows.

Consider the case where the next six pairs of connection creation requests arrive as connection requests.
{( _X2i-1 , _y2i-1 ), ( _x2i , _y2i )} _i=1 , ₂ , ₃ and {( _z2i-1 , _w2i-1 ), ( _z2i , _w2i )} _i=1 , ₂ , ₃ and uses the intermediate MPPs {P _{j_i} , P _{j′_i} } _i=1 , ₂ , ₃ and {P _{k_i} , P _{k′_i} } _i=1 , ₂ , _{3 respectively} . To establish a connection. Note that all wiring is now in use. At this time, a certain integer pair i, i′ε{1, 2, 3} exists, and {j _i , j′ _i }≠{k _i′ , k′ _i′ } holds. This is because the number of connections between the input/output MPPs {P ₁ , P ₂ , P ₃ , P ₄ } and the intermediate MPPs {P ₅ , P ₆ , P ₇ } is 2. Therefore, {j ₁ , j′ ₁ }={5, 6} and {k ₂ , k′ ₂ }={6, 7} are set without loss of generality. At this time, after two connection cancellation requests {(x ₁ , y ₁ ), (x ₂ , y ₂ )} and {(z ₃ , w ₃ ), (z ₄ , w ₄ )} arrive, a new connection is made. When the creation request {(x ₁ , z ₃ ), (x ₂ , z ₄ )} arrives, this connection cannot be established. This is because it is necessary to use the intermediate MPP {P ₆ } to connect x ₁ and z ₃ or x ₂ and z ₄ . However, at this time, the other wiring is in use by the existing connection, and there is only one wiring between P ₁ and P ₆ .

1.4 Non-blocking existing structure The Clos network used when constructing a non-blocking MPP network and the theorem related to it are introduced. The Clos network is generally composed of an input layer, a plurality of intermediate layers, and an output layer. Since the MPP network does not distinguish between the input layer and the output layer, consider a configuration folded in the intermediate layer. This structure is called a folded Clos structure. FIG. 6 illustrates a structure in which a Clos net having one intermediate layer is folded. K MPPs {P _{1 to} P _k } are used for input/output, and ₁ MPPs {P _{k+1 to} P _k+1 } are used for the intermediate layer. The second surface 12 of the input/output MPP {P _{1 to} P _k } is formed by the optical fibers 51 of m-core and the respective surfaces (the first surface 21 or the second surface 22) of the respective intermediate layers MPP {P _{k+1 to} P _k+1 }. Connecting. The 2n-core optical fiber 52 connected to the first surface 11 of the input/output MPP {P _{1 to} P _k } is used for device connection. The n cores of the optical fiber 52 are used for transmission, and the remaining n cores are used for reception. That is, the MPP network of FIG. 6 has a total of 2 nk core device connection ports (connection ports to the outside). By setting k=4, l=3, m=1, and n=3 in FIG. 6, the MPP structure of FIG. 5 is obtained. Further, a multi-stage network structure having two or more intermediate layers MPP is also conceivable, but since the power of light is attenuated and noise is present, the first step in the present invention is to have only one intermediate layer. To do. Some ports may be unused in order to realize non-blocking property.

一方、接続要求が全て再配置不可能ある場合は、折り畳みＣｌｏｓ網が非閉塞であるための必要十分条件は以下であることが分かっている（非特許文献２、３）。

したがって、図５のＭＰＰ網において各入出力で使用する機器接続用ポートを６から４に制限すれば、つまり送信２ポート、受信２ポートに制限すれば、接続要求が再配置不可能である場合でも非閉塞となる。 Two conditions for the folded Clos network to be non-occluded are introduced. It has been found that the necessary and sufficient conditions for the folded Clos network to be non-blocking when all connection requests are relocatable (Non-Patent Document 3).

On the other hand, when all the connection requests cannot be relocated, it is known that the necessary and sufficient conditions for the folded Clos network to be non-blocking are as follows (Non-Patent Documents 2 and 3).

Therefore, if the device connection ports used for each input/output in the MPP network of FIG. 5 are limited to 6 to 4, that is, transmission 2 ports and reception 2 ports, the connection request cannot be rearranged. But it is non-blocking.

1.5 Minimization of the number of used MPPs In the present invention, a necessary connection port number r is given, and a problem of searching for an MPP network that minimizes the number of used MPPs in a non-blocking MPP network that achieves that Think This problem is called minimization of the number of used MPPs.

When all connection requests are non-relocatable and the folding Clos structure is used, the number of connections m, n ∈ (natural number) and the number of MPPs k, l ∈ (natural number) of each layer that minimizes the number of used MPPs are optimized by the following optimization. It can be calculated by solving the problem (for example, since all variables are N or less, it can be calculated by full search). FIG. 11 shows a device configuration diagram for calculating the optimum structure.

この最適値をｆ（ｒ；Ｎ）と書く。以降でこの値を改善するＭＰＰ構造を提案する。

This optimum value is written as f(r;N). The MPP structure that improves this value is proposed below.

2. Proposed structure when relocation is not possible 2.1 Basic idea First, we show that there is room for improvement in the folded Clos network. The number N of ports on one side of each MPP {P _{1 to} P _k , P _{k+1 to} P _k+l }, which will be specifically described using the folded Clos network of FIG. 5, is set to 8. As described in the previous section, when the request cannot be rearranged, it is necessary to satisfy the conditional expression (2) to realize the non-blocking property. When only the number n of device connection ports is changed and the conditional expression (2) is satisfied, it is necessary to reduce n from 3 to 2. At this time, less than half the number of available ports is used on the left side of the left MPP. If this unused port can be used, the accommodation efficiency can be expected to improve. Therefore, in the present invention, this unused port is used for connection with the MPP of the intermediate layer to improve the accommodation efficiency.

2.2 MPP Network Structure of the Present Invention The MPP network 301 of the present invention changes the connection method so that both sides of the input/output MPPs {P _{1 to} P _k } can be used as device connection ports as shown in FIG. 7. Specifically, the MPP network 301 includes a plurality of (k+1) MPPs having a size N (N is a natural number of 2 or more) that is the number of optical fiber ports on the first surface 11 and the number of optical fiber ports on the second surface 12. ) An MPP network provided,
Of the MPPs, k are input/output MPPs {P _{1 to} P _k }, l are intermediate MPPs {P _{k+1 to} P _k+1 },
Of the plurality of optical fiber ports on the first surface 11 of the input/output MPPs {P _{1 to} P _k }, n optical fiber ports are external connection ports and m×l optical fiber ports are intermediate MPPs. P _{k+1 to} P _k+1 } are optical fiber connection ports that connect the m optical fiber ports on the second surface 22 of the optical fibers 51 to each other,
Of the plurality of optical fiber ports on the second surface 12 of the input/output MPPs {P _{1 to} P _k }, n optical fiber ports are external connection ports and m×l optical fiber ports are intermediate MPPs { The optical fiber connection ports connect m optical fiber ports on the first surface 21 of P _{k+1 to} P _k+1 } with the optical fiber 51.
Where k, l, m, and n are natural numbers,
n+m·l≦N,
m·k≦N, and 2 floor((n−1)/m)+1≦l
Meet

As shown in FIG. 7, n cores of device reception ports are prepared on the first surface 11 of each input/output MPP {P _{1 to} P _k }. Furthermore, the remaining ports of the first surface 11 and the second surface 22 of each intermediate MPP {P _{k+1 to} P _k+1 } are connected by an m-core optical fiber 51. Similarly, n cores of equipment transmission ports are prepared on the second surface 12 of each input/output MPP {P _{1 to} P _k }. Furthermore, the remaining ports on the second surface 12 and the first surface 21 of each intermediate MPP {P _{k+1 to} P _k+1 } are connected by an m-core optical fiber 51. Since the MPP network 301 can install the input/output port and the intermediate MPP connection port on both sides of the input/output MPP {P _{1 to} P _k }, it is possible to avoid the inefficiency caused by the folding Clos.

2.3 Minimum MPP Number of Proposed Structure The minimum value of the MPP number minimization problem in the MPP network 301 is written as f _P (r;N). Similar to the folded Clos structure, the number of connections m, nε (natural number) and the number of MPPs k, lε (natural number) that minimize the number of used MPPs can be calculated by solving the following optimization problem. Since all variables are N or less, they can be calculated by full search.

FIG. 11 is a configuration diagram illustrating a design device for calculating the optimum structure of the MPP network 301. The design device is
A parameter input unit 71 into which the required number r of connection ports and the size N of the MPP are input,
Optimization using r and N input to the parameter input unit 71 to fully search k, l, m, and n while minimizing the number of MPPs (k+1) while satisfying the condition 1 shown in Expression 4. A calculation unit 72,
An optimal structure output unit 73 that outputs k, l, m, and n that the optimization calculation unit 72 has performed a full search;
Equipped with.

が成立する。 It can be shown that the MPP network 301 has better accommodation efficiency than the folded Clos structure of FIG. 6 as follows.
(Theorem 1) The MPP network 301 uses a larger number of MPPs than the folded Clos structure. That is

Is established.

(Proof) It is assumed that a set of positive integers (k, 1, m, n) satisfies the constraint conditions (4) to (8) of the use MPP number minimization problem of the folded Clos structure. At this time, it is sufficient to show that this set of integers (k, l, m, n) satisfies the constraints (10) to (13) of the proposed structure. However, this is self-evident because Equation 6 holds.

2.4 Effects Numerical evaluation shows to what extent the MPP network 301 improves the accommodation efficiency when the MPP size N is 1000.
FIG. 9 illustrates the relationship between the number of MPPs used in a situation where all connection requests cannot be relocated and the number of requests that can be accommodated. Here, the single-sided size N of MPP was calculated as 1000. It can be seen that the MPP network 302 has a smaller number of used MPPs and higher accommodation efficiency than any of the existing structures (FIG. 6) for any number of accommodable connection requests. For example, when constructing a non-blocking MPP network capable of accommodating 167,000 connection requests, 1167 MPPs are required in the existing structure, but 833 is sufficient for the proposed method. The same number of connection requests can be accommodated with about 29% less MPP.

3. Mixed structure In this section, we consider the case where connection requests that cannot be relocated and requests that can be relocated are mixed. As described above, the existing structure assumes that all connection requests can be relocated or all connection requests cannot be relocated. However, since various services are accommodated in the fiber/NW device in the NW station building, the present invention considers a situation in which relocatable connection requests and impossible connection requests coexist. Even if they are mixed, it is possible to construct a non-blocking network by constructing an MPP network, assuming that all requests cannot be rearranged, but this does not utilize the characteristics of relocatable connections. .. We propose a structure that improves the accommodation efficiency by explicitly considering the mixed situation.

3.1 Ratio of connection requests that cannot be rearranged First, the situation of mixed requests will be described. Each MPP has only n pairs of connection ports, that is, n ports are for transmission and n ports are for reception. At this time, it is assumed that there exists a positive real number aε[0,1], and when attention is paid to each MPP, a connection request that cannot be rearranged arrives at the pair shown in Formula 7 at most. This a is called a rearrangement impossible rate. When the non-relocation ratio a is 1, all connection requests may be non-rearrangeable, so the situation is the same as in the previous section 2.

3.2 Basic ideas Before explaining the mixed-type structure, we introduce some basic ideas. If all requests can be rearranged, only conditional expression (1) needs to be satisfied. A non-blocking network can be realized by using approximately half of the intermediate MPPs as compared with the case where the rearrangement that needs to satisfy the conditional expression (2) is impossible. Now consider the case where a small number of requests are non-relocatable and the rest are relocatable. In such a case, for example, one or two dedicated intermediate MPPs may be prepared to accommodate a request that cannot be relocated, and the number of connections may be adjusted so that the constraint condition (2) is satisfied. In this case as well, a non-blocking network can be constructed with about half the number of intermediate MPPs.

3.3 MPP network structure of the present invention By advancing the idea in the previous section 3.2, the intermediate MPPs are classified into MPPs that accommodate non-relocatable types and MPPs that accommodate relocatable types. FIG. 8 is a diagram for explaining the structure of the MPP network 302 of the present invention, which can handle the case where the requests are mixed. The MPP network 302 divides the intermediate MPPs { _{Pk+1 to Pk+l} } into two groups with respect to the MPP network 301 of FIG. 7, and prohibits the rearrangement of connection of the intermediate MPPs { _{Pk+1 to Pk+l1} } of one of the groups. of the path of the service, characterized in that as the other intermediate _{MPP {P k + l1 + 1} ~P k + l1 + l2} route relocatable services connection of the group _{(l = l 1 + l 2} ). Hereinafter, this structure is referred to as a mixed structure.

In FIG. 8, for convenience, the upper side {P _{k+1 to} P _k+l1 } of the intermediate MPP is a non-relocatable type MPP, and the lower side {P _{k+l1 to} P _k+l1+l2 } is a relocatable type MPP. The number of connections between the MPP and the input/output MPP for accommodating the non-relocatable type is m ₁ , and the number of connections between the relocatable MPP and the input/output MPP is m ₂ . In the input/output MPP {P _{1 to} P _k }, n ₁ port of the input n ports is reserved for the relocation impossible request. The remaining n ₂ is reserved for relocatable type requests. Here, since a maximum of n relocatable requests arrive, it is necessary to use n ₁ requests for requests exceeding n ₂ . On the other hand, in the non-relocatable type, if n ₁ ≧an is satisfied, at most n ₁ will arrive, so there is no need to use n ₂ .

The condition that the mixed structure is non-blocking is that the following two conditions are satisfied (provided that n ₁ ≧an).

The correctness of this condition is shown by the following discussion. Consider when a new connection request arrives. First, if the request is a connection cancellation, it is obvious that it can be met. Also, in the case of a relocatable connection creation request, the connection can be created using the relocatable intermediate MPP from the latter condition. In the case of a non-relocatable type request, since there is a vacancy in the non-relocatable intermediate MPP from the former condition, a connection can be created by utilizing it.

3.4 Minimum Number of MPPs Used in Mixed Structure The minimum value of the number of used MPPs minimization problem in the MPP network 302 is written as f _H (r;N). The number of connections m ₁ , m ₂ , n ₁ , n ₂ ε (natural number) that minimizes the number of used MPPs and the number k, l ₁ , l ₂ ε (natural number) of MPPs in each layer should solve the following optimization problem. Can be calculated by Since all variables are N or less, they can be calculated by full search. FIG. 11 shows a device configuration diagram for calculating the optimum structure.

FIG. 11 is also a configuration diagram illustrating a design device for calculating the optimum structure of the MPP network 302. The design device is
A parameter input unit 71 into which the required number r of connection ports and the size N of the MPP are input,
Using r and N input to the parameter input unit 71, k, l ₁ , l ₂ , m ₁ , which minimizes the number of MPPs (k+l ₁ +l ₂ ) while satisfying the condition 2 shown in Expression 9, an optimization calculator 72 that searches all m ₂ , n ₁ and n ₂ ;
An optimal structure output unit 73 that outputs k, l ₁ , l ₂ , m ₁ , m ₂ , n ₁ and n ₂ that the optimization calculation unit 72 has searched in full;
Equipped with.

が成立する。 It can be shown that the MPP network 302 has better accommodation efficiency than the MPP network 301 of FIG. 7 as follows.
(Theorem 2) It is assumed that r>qN and a≦(q−1)/{2(q+1)} hold for a positive integer q of 2 or more. At this time, the mixed structure has a smaller number of used MPPs than the proposed structure in Section 2. That is

Is established.

とおき、混在構造の構成に関するパラメータ（ｋ’、ｌ_１、ｌ_２、ｍ_１、ｍ_２、ｎ_１、ｎ_２）を次のように設定する。

このパラメータが制約条件（１８）〜（２４）を満たし、かつｌ_１＋ｌ_２＜ｌであることを示せば証明は完了する。 (Proof) It is assumed that the set of positive integers (k, 1, m, n) satisfies the constraint conditions (10) to (13) of the use MPP number minimization problem of the proposed structure. At this time,

Then, the parameters (k′, l ₁ , l ₂ , m ₁ , m ₂ , n ₁ , n ₂ ) relating to the structure of the mixed structure are set as follows.

The proof is complete if this parameter satisfies the constraint conditions (18) to (24) and shows that l ₁ +l ₂ <l.

となり矛盾。よってｑ≦ｔであることが分かる。 As for the constraint condition, since it can be derived from the definition of the parameter, l ₁ +l ₂ <l is shown. First, it is shown that q≦t. Use the absurd method. Assume q>t. Then

It becomes a contradiction. Therefore, it can be seen that q≦t.

よって、

となる。 Finally, l ₁ <l-l ₂ is shown. Using integers r, r′ε{0, 1,..., M−1} and positive integer q′, n and ceil(an) in Expression 7 can be written as n−1=tm+r and Expression 14, respectively.

Therefore,

Becomes

3.5 Effects Numerical evaluation shows how much the MPP network 302 improves the accommodation efficiency when the MPP size N is 1000.
FIG. 10 illustrates the number of MPPs used and the number of requests that can be accommodated in a situation where non-relocatable connection requests and relocatable connection requests coexist. Again, the single-sided size N of the MPP is 1000. It can be seen that the MPP network 302 has a smaller number of used MPPs and better accommodation efficiency than any of the existing structures (FIG. 6) and the MPP network 301 with regard to the number of accommodable connection requests. For example, when constructing a non-blocking MPP network capable of accommodating 167,000 connection requests, 1167 MPPs are required in the existing structure, but 758 MPPs are sufficient when the relocation impossible rate a is 1/3. It can be seen that 693 MPPs are sufficient when a is 1/5, and 640 MPPs are sufficient when a is 1/10. When the rearrangement impossible rate a is 1/10, the mixed structure can accommodate the same number of connection requests with an MPP that is about 45% less than the existing structure.

4. The point of the invention The MPP network according to the present invention improves the accommodation efficiency of the existing structure by utilizing the following two characteristics of the fiber network of the network station.

The first is the possibility of relocating the connection request. The existing structure assumes that either all connection requests can be relocated or all connection requests cannot be relocated. However, since various services are accommodated in the fiber/NW equipment in the NW station building, the present invention considered a situation in which relocatable connection requests and impossible connection requests coexist. Even if they are mixed, it is possible to construct a non-blocking network by constructing an MPP network, assuming that all requests cannot be rearranged, but this does not utilize the characteristics of relocatable connections. .. In the MPP network according to the present invention, the accommodation efficiency is improved by explicitly assuming a mixed situation.

The second is how the fiber is used and the size of the patch panel. Usually, an optical fiber uses two cores in one connection, one for transmission and the other for reception. Further, since the patch panel is a device that physically connects the fibers, it is natural that the patch panel is manufactured with the same number of cores on the front surface and the back surface. In the existing method, the input/output port is installed only on one side, but in the MPP network according to the present invention, the installation efficiency is improved by installing it on both sides.

11, 21: first face 12 and 22: the second surface 51, 52: optical fiber _{P 1 ~P k, P k +} 1 ~P k + l1, P k + l1 + 1 ~P k + l1 + l2, P k + l: MPP
301, 302: MPP network

Claims (8)

A plurality (k+1) of mechanical patch panels (MPPs) of size N (N is a natural number of 2 or more), which is the number of optical fiber ports on the first surface and the number of optical fiber ports on the second surface, are provided. MPP network,
Of the MPPs, k are input/output MPPs, l are intermediate MPPs,
Of the plurality of optical fiber ports on the first surface (11) of the input/output MPP, n optical fiber ports are ports for connecting to the outside, and m×l optical fiber ports are the second surface of the intermediate MPP. The optical fiber connection port for connecting the m optical fiber ports in (22) with the optical fiber,
Out of the plurality of optical fiber ports on the second surface (12) of the input/output MPP, n optical fiber ports are ports for external connection, and m×l optical fiber ports are the first surface of the intermediate MPP. (21) An MPP network characterized by having m optical fiber ports in (21) as optical fiber connection ports connected by optical fibers.
However, k, l, m, and n are natural numbers,
n+m·l≦N,
m·k≦N, and 2 floor((n−1)/m)+1≦l
Meet The intermediate MPPs are divided into two groups, one of the intermediate MPPs of the group is used as a connection relocation-prohibited service path, and the other intermediate MPP of the group is used as a connection relocation-enabled service path. The MPP network according to claim 1, wherein: A plurality (k+1) of mechanical patch panels (MPPs) of size N (N is a natural number of 2 or more), which is the number of optical fiber ports on the first surface and the number of optical fiber ports on the second surface, are provided. A method for constructing an MPP network,
Of the MPPs, k are input/output MPPs and 1 is an intermediate MPP,
Of the plurality of optical fiber ports on the first surface (11) of the input/output MPP, n optical fiber ports are ports for connecting to the outside, and m×l optical fiber ports are the second surface of the intermediate MPP. The optical fiber connection port for connecting the m optical fiber ports in (22) with the optical fiber,
Out of the plurality of optical fiber ports on the second surface (12) of the input/output MPP, n optical fiber ports are ports for external connection, and m×l optical fiber ports are the first surface of the intermediate MPP. (21) A method for constructing an MPP network, characterized in that m optical fiber ports in (21) are used as optical fiber connection ports connected by optical fibers.
However, k, l, m, and n are natural numbers,
n+m·l≦N,
m·k≦N, and 2 floor((n−1)/m)+1≦l
Meet The intermediate MPPs are divided into two groups, one of the intermediate MPPs of the group is used as a connection relocation-prohibited service path, and the other intermediate MPP of the group is used as a connection relocation-enabled service path. 4. The method for constructing an MPP network according to claim 3, wherein: The MPP network design apparatus according to claim 1,
A parameter input unit into which the required number r of connection ports and the size N of the MPP are input,
An optimization calculation unit that uses r and N input to the parameter input unit and performs a full search for k, l, m, and n that minimizes the number of MPPs (k+1) while satisfying condition 1.
An optimal structure output unit that outputs k, l, m, and n that the optimization calculation unit has performed a full search;
Designing device.
However, condition 1 is
n+m·l≦N,
m·k≦N,
2floor((n-1)/m)+1≦l, and n·k≧r
Is. An apparatus for designing an MPP network according to claim 2, wherein
A parameter input unit into which the required number r of connection ports and the size N of the MPP are input,
Using r and N input to the parameter input unit, k, l ₁ , l ₂ , m ₁ , m ₂ , n that minimizes the number of MPPs (k+l ₁ +l ₂ ) while satisfying the condition 2 An optimization calculation unit that searches all ₁ and n ₂ ;
An optimal structure output unit that outputs k, l ₁ , l ₂ , m ₁ , m ₂ , n ₁ and n ₂ that the optimization calculation unit has searched in full;
Designing device.
However,
l=l ₁ +l ₂ ,
m=m ₁ +m ₂ ,
n=n ₁ +n ₂
And condition 2 is
n ₁ +n ₂ +m ₁ ·l ₁ +m ₂ ·l ₂ ≦N,
m ₁ ·k≦N,
m ₂ ·k≦N,
2floor((n ₁ −1)/m ₁ )+1≦l ₁ ,
n ₁ +n ₂ ≦m ₂ ·l ₂ and (n ₁ +n ₂ )k≧r
Is. A method for designing an MPP network according to claim 1, wherein
A parameter input procedure for inputting the required number r of connection ports and the size N of the MPP;
An optimization calculation procedure for using k and l, m, and n that minimizes the number of MPPs (k+1) while satisfying the condition 1 using r and N input in the parameter input procedure;
An optimal structure output procedure that outputs k, l, m, and n that have been fully searched in the optimization calculation procedure;
How to design.
However, condition 1 is
n+m·l≦N,
m·k≦N,
2floor((n-1)/m)+1≦l, and n·k≧r
Is. The method for designing an MPP network according to claim 2, wherein
A parameter input procedure for inputting the required number r of connection ports and the size N of the MPP;
Using r and N input in the parameter input procedure, k, l ₁ , l ₂ , m ₁ , m ₂ , n that minimizes the number of MPPs (k+l ₁ +l ₂ ) while satisfying the condition 2 An optimization calculation procedure for searching all ₁ and n ₂ ;
An optimal structure output procedure that outputs k, l ₁ , l ₂ , m ₁ , m ₂ , n ₁ and n ₂ that have undergone a full search in the optimization calculation procedure;
How to design.
However,
l=l ₁ +l ₂ ,
m=m ₁ +m ₂ ,
n=n ₁ +n ₂
And condition 2 is
n ₁ +n ₂ +m ₁ ·l ₁ +m ₂ ·l ₂ ≦N,
m ₁ ·k≦N,
m ₂ ·k≦N,
2floor((n ₁ −1)/m ₁ )+1≦l ₁ ,
n ₁ +n ₂ ≦m ₂ ·l ₂ and (n ₁ +n ₂ )k≧r
Is.

Images