JP2003198727A - Link cost calculating method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To calculate a link cost in an actual operation where standby paths for a current path for a priority call between different node pairs share a link resource and a non-priority call paths share the link resource with the priority call standby path in a line switching communication network. <P>SOLUTION: In step S-1, the state of use of resources by each link is observed so as to estimate a call type dependent call unit or the arrival rate of each communication path setting request. In steps S-3, S-4, S-5, only a call type link cost depending on the state of a link at that time is advertised to a network at a time interval shorter than an observation period while storing the call type link cost in each link state at least for one observation period. When the observation period elapses (S-5), the call type dependent call unit or the arrival rate of various communication path setting requests are newly estimated in each link so as to calculate the call type dependent link cost depending on the link state at that time. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a link cost calculation method, and more particularly, a protection path for a priority call working path between different node pairs shares link resources and a non-priority call path protects a priority call. The present invention relates to a link cost calculation method in a circuit switched communication network that shares a path and a link resource.

[0002]

2. Description of the Related Art A link such as an optical fiber and an optical cross-connect device (hereinafter referred to as a connect device) that performs a circuit switching function, etc., and selectively switch the connect device under the control of a management device. In a circuit-switched communication network that sets a communication path, searches for the minimum cost route when setting a communication path between node pairs by selecting a link through which wavelength-multiplexed optical information is switched by switching connect devices. However, it is common practice to set a communication path along the route. The cost of a communication path is the sum of the link costs of the links that make up that communication path, so it is necessary to estimate the link cost in some way when setting the communication path along the route with the lowest cost. is there.

As one of the methods for estimating the link cost, when a request for setting a communication path is received, for example, if it is assumed that all resources can be used, the request for setting the communication path is rejected and a loss occurs. There is a method to estimate the link cost so that the profit rate lost due to this is minimized, and a circuit-switched communication network in which there are calls with different communication speeds (required bandwidths) and profit rates obtained by connecting As a target, the number of simultaneous calls on each link is calculated for each communication speed, the call volume by each communication speed added to each link is estimated from this number of simultaneous connections, and this estimated value and the rate of return for calls by each communication speed are used. It is performed to estimate a link cost for a call according to communication speed in each state of each link.

[0004]

In the circuit switched communication network, in order to effectively use the link resource in the restoration mode, the protection path for the priority call working path between different node pairs shares the link resource. As a practical operation, it is considered that the non-priority call path shares the link resource with the priority call protection path. In such a circuit-switched communication network, there are the following three types of communication paths. Will be done. (1) Priority call working path (2) Priority call protection path (3) Non-priority call path Note that a single link failure and node failure are assumed here, and the priority call working path and priority The call backup path is set on a route that does not share nodes (relay nodes) on the way.

FIG. 1 and FIG. 2 are configuration diagrams showing an example of a state of a circuit switching type communication network having a restoration mode. This circuit-switched communication network includes links 1-1 to 1-8 such as optical fibers, connect devices AE that perform a circuit-switching function, and switches the connect devices AE under the control of the management device 2. And selectively link the wavelength-multiplexed optical information input to it
The communication path is set between the selected nodes among the users or the routers (hereinafter referred to as nodes) a to d by outputting to the nodes 1 to 1-7.

In FIG. 1, a solid line and a wavy line indicate a priority call working path and a priority call protection path, respectively. In this state, the connection device A to the link 1-1 to the connection device are connected between the node pairs a to c. A working path for priority call passing through D to link 1-2 to connect device C is set, and its backup path is set to connect device A to link 1-5 to connect device E to link 1-7 to connect device C. To be done. Further, the connection device B-link 1-is connected between the node pairs b-c.
3 to the connection device C, the working path for priority call is set, and the backup path is the connection device B to the link 1-8.
-Connect device E-Link 1-7-Connect device C is set. Therefore, the resources of the links 1-7 are shared by the protection path for the priority call working path between the node pairs a to c and the protection path for the priority call working path between the node pairs b to c. , If a failure occurs in the link 1-1, the management device 2 causes the connection device A,
By switching E and C, the working path for priority call between the node pairs a to c is switched to the protection path for priority call.

In FIG. 2, a solid line, a wavy line, and an alternate long and short dash line respectively indicate a priority call working path, a priority call protection path, and a non-priority call path.
Connect device A to link 1-1 to connect device D between c
~ A working path for priority call passing through the link 1-2 to the connection device C is set, and the backup path is the connection device A ~
The link 1-5 to the connection device E to the link 1-7 to the connection device C are set. Further, a non-priority call path that passes through the connect device B, the link 1-8, the connect device, the connect device E, the link 1-7, and the connect device C is set between the nodes b and c. Therefore, the resource of the link 1-7 is shared by the non-priority call path between the node pairs b to c and the priority call protection path between the node pairs a to c, and, for example, the link 1-1 fails. In this case, the connecting devices A, E, C are switched by the management device 2 and the working path for priority call between the node pairs a to c is switched to the protection path for priority call, resulting in interruption of the resources of the non-priority call path. . As a result, the service via the non-priority call path is stopped.

Although the conventional calculation of the link cost considers that the communication speed (requested bandwidth) and the rate of return of the call are different, the above-mentioned priority with interruption considered as a practical operation is considered. No consideration is given to the different calls.

According to the present invention, in a circuit-switched communication network, a protection path for a priority call working path between different node pairs shares a link resource, and a non-priority call path uses a priority call protection path and a link resource. It is an object of the present invention to provide a link cost calculation method capable of calculating the link cost in the practical operation of sharing.

[0010]

In order to solve the above problems, the present invention provides a priority call working path call for each link,
Based on the number of simultaneous connections of call types including backup call for priority call and pass call for non-priority call, the call type call volume added to each link or the arrival rate of each communication path setting request is estimated, and the estimated value and call The first feature is that the call type link cost in each state of each link is calculated based on the type profitability. A second feature of the present invention is that the number of simultaneous connections of the call type is obtained by observing the link status of each link. Furthermore, the present invention provides the number of simultaneous connections of the call type,
The third feature is that the points are obtained based on the observation result in one observation period and the average observation result in a plurality of periods.

According to the first and second characteristics, the protection path for the priority call working path between different node pairs shares the link resource, and the non-priority call path shares the priority call protection path and the link resource. It is possible to calculate the link cost in the practical operation of sharing, and it is possible to select the routes of the calls having different priorities so that the profit rate due to the loss of the call is minimized. Further, according to the third feature, there is no fear that the control of the network does not converge and becomes unstable.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION According to the present invention, a call type call added to each link from the number of simultaneous connections of call types including a priority call working path call, a priority call backup path call, and a non-priority call path call in each link. Estimate the amount or arrival rate of each communication path setting request,
The call type link cost in each state of each link is calculated based on the estimated value and the call type return rate.
First, the principle will be described. [Principle of the present invention]

Now, it is assumed that the communication path setting request and release are independently generated for each link, and the number of resources used by the priority call working path is h and the priority call is set as the link state of each link. I is the number of resources used only by the backup protection path, j is the number of resources shared by the priority call backup path and the non-priority call path, and k is the number of resources used only by the non-priority call path. And

Assuming that the communication path setup request and release follow the Markov process, given the arrival rate and release rate of the call type communication path setup request and the profit rate obtained from each communication path, the Markov decision theory Accordingly, the following equations (1-1) to (1-8) are derived in each link.
It should be noted that the profit obtained from each communication path is a profit loss due to not connecting each communication path. Therefore, the profit loss rate used in the following formulas (1-1) to (1-8) is , Is the same as the rate of return.

The link state (h, i, j, k) of each link is classified into eight cases, and equations (1-1) to (1-8) correspond to each of these cases. In addition, each equation represents the average return loss rate RewardLoss on the target link.

Case 1: Not all resources are used, some resources are used only by the priority call backup path, and some resources are used only by the non-priority call path. Link state (h RewardLoss = λ _restp (V (h + 1, i, j, k) -V (h, i, j) considering the state transition from + i + j + k <N, i ≠ 0, k ≠ 0) , k)) + λ _rests (V (h, i, j + 1, k-1) -V (h, i, j, k)) + λ _npr (V (h, i-1, j + 1, k) -V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + iμ _rests (V (h, i-1, j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j-1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) + kμ _npr (V (h, i, j, k-1) -V (h, i, j, k)) (1-1)

Case 2: All resources are not used, there is no resource used only by the priority call protection path, and there is a resource used only by the non-priority call path. RewardLoss = λ _restp (V (h + 1, i, j, k) -V (h, i, j) considering the state transition from + i + j + k <N, i = 0, k ≠ 0) , k)) + λ _rests (V (h, i, j + 1, k-1) -V (h, i, j, k)) + λ _npr (V (h, i, j, k + 1) -V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j-1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) + kμ _npr (V (h, i, j, k-1) -V (h, i, j, k)) (1-2)

Case 3: All resources are not used, and there is a resource used only by the priority call protection path, and there is no resource used only by the non-priority call path. + i + j + k <N, i ≠ 0, k = 0) Considering the state transition from RewardLoss = λ _restp (V (h + 1, i, j, k) -V (h, i, j , k)) + λ _rests (V (h, i + 1, j, k) -V (h, i, j, k)) + λ _npr (V (h, i-1, j + 1, k) -V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + iμ _rests (V (h, i- 1, j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j-1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) (1-3)

Case 4: A link state in which all resources are not used, no resources are used only by the priority call protection path, and no resources are used only by the non-priority call path (h + i + j + k <N, i = 0, k = 0) Considering the state transition from RewardLoss = λ _restp (V (h + 1, i, j, k) -V (h, i, j , k)) + λ _rests (V (h, i + 1, j, k) -V (h, i, j, k)) + λ _npr (V (h, i, j, k + 1) -V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j- 1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) ・ ・・ (1-4)

Case 5: All resources are used, some resources are used only by the priority call protection path, and some resources are used only by the non-priority call path. Link state (h + Considering the state transition from i + j + k = N, i ≠ 0, k ≠ 0), RewardLoss = q _restp + λ _rests (V (h, i, j + 1, k-1) -V (h , i, j, k)) + λ _npr (V (h, i-1, j + 1, k) -V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + iμ _rests (V (h, i-1, j, k) -V (h, i, j, k)) + jμ _rests (V ( h, i, j-1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j , k)) + kμ _npr (V (h, i, j, k-1) -V (h, i, j, k)) (1-5)

Case 6: Link state (h +) where all resources are used, there is no resource used only by the priority call protection path, and there is resource used only by the non-priority call path Considering the state transition from i + j + k = N, i = 0, k ≠ 0), RewardLoss = q _restp + λ _rests (V (h, i, j + 1, k-1) -V (h , i, j, k)) + q _npr + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j -1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) + kμ _npr (V (h, i, j, k-1) -V (h, i, j, k)) ・ ・ ・ (1-6)

Case 7: All resources are used, some resources are used only by the priority call protection path, and no resources are used only by the non-priority call path. Link state (h + RewardLoss = q _restp + q _rests + λ _npr (V (h, i-1, j + 1, k)-considering the state transition from i + j + k = N, i ≠ 0, k = 0) V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + iμ _rests (V (h, i-1 , j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j-1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) (1-7)

Case 8: A link state in which all resources are used, no resources are used only by the priority call protection path, and no resources are used only by the non-priority call path (h + RewardLoss = q _restp + q _rests + q _npr + hμ _restp (V (h-1, i, j, k) considering the state transition from i + j + k = N, i = 0, k = 0) -V (h, i, j, k)) + jμ _rests (V (h, i, j-1, k + 1) -V (h, i, j, k)) + jμ _npr (V (h, i + 1, j-1, k) -V (h, i, j, k)) (1-8)

Here, the meaning of each symbol is as follows. RewardLoss: Average profit loss rate, N: Total number of resources in the link V (h, j, j, k): Total profit loss when starting from the link state (h, i, j, k) for infinite time (Link status (h, i, j,
k) Value in) λ _restp: priority arrival rate of call for the working path setting request _λ rests: priority arrival rate of call protection path setup request for λ _npr: arrival rate of non-priority call for the path setting request _μ restp: for a priority call Working path release rate μ _rests : Priority call _protection path release rate μ _npr : Non-priority call path release rate q _restp : Revenue loss rate by not connecting priority call working path q _rests : Priority call _protection Revenue loss rate by not connecting a path q _npr : Revenue loss rate by not connecting a non-priority call path

The meanings of the above equations (1-1) to (1-8) are, for example, the case 6 of the equation (1-6) (all resources are used and only the backup path for priority call is used). Explaining the state transition from the link state in which there is no resource used and there is a resource used only by the non-priority call path, the first term on the right side of the equation (1-6) is It represents the revenue loss rate when a path call is added. Since all resources are used in this link state, the priority call working path call becomes a loss, and the revenue loss rate occurs because the priority call working path is not connected. To do. The second term on the right side represents the profit loss rate when the priority call backup path call is added. In this case, the resource used by the non-priority call path is also used as the resource for the priority call protection path call and enters the link state (h, i, j + 1, k-1). A revenue loss rate occurs due to state transition. The third term on the right side represents the revenue loss rate when a non-priority call is added. In this link state, all resources are used, and there is no resource used only by the priority call protection path, so non-priority The call becomes a loss, and a revenue loss rate occurs due to not connecting the non-priority call path. The fourth term on the right side corresponds to the case where the working path call for the priority call is released, the fifth term on the right side corresponds to the case where the protection path call for the priority call is released, and the sixth term on the right side corresponds to the priority call. Corresponding to the case where the non-priority call sharing the resource with the protection backup path call is released, the seventh item on the right side is the case where the non-priority call not sharing the resource with the protection call path for the priority call is released. It corresponds to. The above is the profit loss rate that occurs when all the state transitions in Case 6 are considered.

V (h, i, j, k) in the link state (h, i, j, k)
The value of k) is infinite from the viewpoint of meaning, but only the difference between the values in the two link states is a problem in calculating the link cost. Therefore, for example, by setting V (0,0,0,0) = 0, by selecting any two right sides of the above equations in each link state and solving the equation connected by the equal signs, The value can be calculated.

For example, link status (h, i, j, k) = (0,0,
0,0) corresponds to Case 4, Is given and the link state (h, i, j, k) = (0,0,0,1) corresponds to Case 2, Is given.

By combining the right sides of these equations, (λ _restp + λ _rests + λ _npr + μ _npr ) V ( _0,0,0,0 )-(λ _restp + λ _rests + 2λ _npr + μ _npr ) V (0,0,0,1) + λ _npr V (0,0,0,2) _{-λ restp} V (1,0,0,0) _{-λ rests} V (0,1,0,0) + λ _restp V ( _1,0,0,1 ) + λ _rests V (0,0,1,0) = 0 is obtained.

Here, If V ( _0,0,0,0 ) = 0 is set, (λ _restp + λ _rests + 2λ _npr + μ _npr ) V (0,0,0,1) _{-λ npr} V (0,0,0, _{2) + λ restp V (1,0,0,0} ) + λ rests V (0,1,0,0) - λ restp V (1,0,0,1) - λ rests V (0,0, A linear equation containing 6 unknowns is obtained.

Further, for example, the link status (h, i, j, k) (h +
When i + j + k <N, i> 1, k ≠ 0, it corresponds to Case 1 and RewardLoss = λ _restp (V (h + 1, i, j, k) -V (h, i, j, k)) + λ _rests (V (h, i, j + 1, k-1) -V (h, i, j, k)) + λ _npr (V (h, i-1, j + 1 , k) -V (h, i, j, k)) + hμ _restp (V (h-1, i, j, k) -V (h, i, j, k)) + iμ _rests (V (h , i-1, j, k) -V (h, i, j, k)) + jμ _rests (V (h, i, j-1, k + 1) -V (h, i, j, k) ) + _{Jμ npr} (V (h, i + 1, j-1, k) -V (h, i, j, k)) + kμ _npr (V (h, i, j, k-1) -V (h, i, j, k)) is given, and the link state (h, i-1, j + 1, k) (h + i + j + k <N, i
When> 1, k ≠ 0), it still corresponds to Case 1, and RewardLoss = λ _restp (V (h + 1, i-1, j + 1, k) -V (h, i-1, j + 1, k)) + λ _rests (V (h, i-1, j + 2, k-1) -V (h, i-1, j + 1, k)) + λ _npr (V (h, i -2, j + 2, k) -V (h, i-1, j + 1, k)) + hμ _restp (V (h-1, i-1, j + 1, k) -V (h, i-1, j + 1, k)) + (i-1) μ _rests (V (h, i-2, j + 1, k) -V (h, i-1, j + 1, k)) + (j + 1) μ _rests (V (h, i-1, j, k + 1) -V (h, i-1, j + 1, k)) + (j + 1) μ _npr (V ( h, i, j, k) -V (h, i-1, j + 1, k)) + kμ _npr (V (h, i-1, j + 1, k-1) -V (h, i-1, j + 1, k)) Is given.

Combine the right sides of these equations
As a result, the linear equation containing the following 16 unknowns is
can get. λ_restpV (h + 1, i, j, k)-λ_restpV (h + 1, i-1, j + 1, k)) + λ_restsV (h, i, j + 1, k-1)-λ_restsV (h, i-1, j + 2, k-1) -λ_nprV (h, i-2, j + 2, k) + hμ_restpV (h-1, i, j, k) -hμ_restpV (h-1, i-1, j + 1, k) + iμ_restsV (h, i-1, j, k) -(i-1) μ_restsV (h, i-2, j + 1, k) + jμ_restsV (h, i, j-1, k + 1) -(j + 1) μ_restsV (h, i-1, j, k + 1) + jμ_nprV (h, i + 1, j-1, k) + kμ_nprV (h, i, j, k-1)-kμ_nprV (h, i-1, j + 1, k-1) -(λ_restp+ λ_rests+ λ_npr+ hμ_restp+ iμ_rests+ jμ _rests + (2j + 1) μ_npr+ kμ_npr) V (h, i, j, k) + (λ_restp+ λ_rests+ 2λ_npr+ hμ_restp+ (i-1) μ_rests+ (j + 1) (μ_rests+ μ_npr) + kμ_npr) V (h, i-1, j + 1, k) = 0

Further, for example, the link status (h, i, j, k) (h +
When i + j + k = N, i = 0, k = 0), it corresponds to Case 8, and RewardLoss = q _restp + q _rests + q _npr + hμ _restp (V (h-1,0, j, 0) -V (h, 0, j, 0)) + jμ _rests (V (h, 0, j-1,1) -V (h, 0, j, 0)) + jμ _npr (V (h, 1, j-1,0) -V (h, 0, j, 0)) is given, and the link state (h, i, j-1, k + 1) (h + i + j + k = N, i =
0, k = 0) corresponds to case 6, and RewardLoss = q _restp + λ _rests (V (h, 0, j, 0) -V (h, 0, j-1,1)) + q _npr + hμ _restp (V (h-1,0, j-1,1) -V (h, 0, j-1,1)) + (j-1) μ _rests (V (h, 0, j- 2,2) -V (h, 0, j-1,1)) + (j-1) μ _npr (V (h, 1, j-2,1) -V (h, 0, j-1, 1)) + μ _npr (V (h, 0, j-1,0) -V (h, 0, j-1,1)) is given.

By combining the right sides of these equations, the following linear equations containing the unknowns can be obtained. (Λ _rests + hμ _restp + jμ _rests + jμ _npr ) V (h, 0, j, 0)-(λ _rests + hμ _restp + (2j-1) μ _rests + jμ _npr ) V (h, 0, j- 1, 1) + hμ _restp V (h-1,0, j-1,1)-hμ _restp (V (h-1,0, j, 0) + (j-1) μ _rests V (h, 0 , j-2,2) + (j-1) μ _npr (V (h, 1, j-2,1) + jμ _npr V (h, 1, j-1,0) + μ _npr V (h, 0, j-1,0) = q _rests

As described above, by selecting arbitrary two equations from the given number of states and combining the right sides with each other, the number of linear equations that is one less than the number of states can be derived. The value of V of each link state can be obtained by solving.

The revenue loss rate due to not connecting each communication path is expressed by the following equations (2-1) to (2-3) using the estimated arrival rate of each communication path setting request. q _restp = λ _restp × r _restp (2-1) q _rests = λ _rests × r _rests (2-2) q _npr = λ _npr × r _npr (2-3)

Here, the meaning of each symbol is as follows. r _restp : Return rate obtained from the priority call working path r _rests : Return rate obtained from the priority call _protection path r _npr : Return rate obtained from the non-priority call path

The rate of return obtained from each communication path in the above equations (2-1) to (2-3) can be set by either of the following two methods. [Method 1]
Distribute the earned profit equally to the working path and the backup path. That is, a method of setting by the following formulas (3-1) to (3-3). _{r restp = m × p rest /} [(m + 1) × μ restp] · · (3-1) r rests = m × p rest / [(m + 1) × μ rests] · · (3-2) r npr = p _npr / _{μnpr ···} (3-3) [Method 2] Allocate the profit to be lost when the communication path cannot be connected. That is, the following equations (4-1) to (4-
Method to set by 3). r _restp = p _rest / μ _restp ... (4-1) r _rests = m × p _rest / μ _rests ... (4-2) r _npr = p _npr / μ _npr ... (4-3) Here, the meaning of each symbol is as follows. m: number of active paths for priority call set between one node pair p _rest : priority call price p _npr : non-priority call price

By the above method, V in each link state
Find the value of. Generally, the call type link cost in the link state (h, i, j, k) is obtained by subtracting the V value in the link state before the state transition from the V value in the link state after the state transition due to the call connection request. It is the value obtained by dividing by the call release rate. Therefore, the call type link cost in the link state (h, j, j, k) is calculated using the value of V obtained above by the following equation (5-1-1).
It can be obtained by (5-8-3). Note that expression (7
-1) to (7-8) correspond to Case 1 to Case 8 described above.

Case 1: (h + i + j + k <N, i ≠ 0, k ≠ 0) C _restp = (V (h + 1, i, j, k) -V (h, i, j, k)) / μ _{restp ···} (5-1-1) C _rests = (V (h, i, j + 1, k-1) -V (h, i, j, k)) / μ _{rests ···} (5-1-2) C _npr = (V (h, i-1, j + 1, k) -V (h, i, j, k)) / μ _{npr ···} ( _5-1-3 )

Case 2: (h + i + j + k <N, i = 0, k ≠ 0) C _restp = (V (h + 1, i, j, k) -V (h, i, j, k)) / μ _{restp ···} (5-2-1) C _rests = (V (h, i, j + 1, k-1) -V (h, i, j, k)) / μ _{rests ···} (5-2-2) C _npr = (V (h, i, j, k + 1) -V (h, i, j, k)) / μ _{npr ···} ( _5-2-3 )

Case 3: (h + i + j + k <N, i ≠ 0, k = 0) C _restp = (V (h + 1, i, j, k) -V (h, i, j, k)) / μ _{restp ···} (5-3-1) C _rests = (V (h, i + 1, j, k) -V (h, i, j, k)) / μ _{rests ···} (5 -3-2) C _npr = (V (h, i-1, j + 1, k) -V (h, i, j, k)) / μ _{npr ···} ( _5-3-3 )

Case 4: (h + i + j + k <N, i = 0, k = 0) C _restp = (V (h + 1, i, j, k) -V (h, i, j, k)) / μ _{restp ···} (5-4-1) C _rests = (V (h, i + 1, j, k) -V (h, i, j, k)) / μ _{rests ···} (5 -4-2) C _npr = (V (h, i, j, k + 1) -V (h, i, j, k)) / μ _{npr ···} ( _5-4-3 )

Case 5: (h + i + j + k = N, i ≠ 0, k ≠ 0) C _restp = ∞ _··· (5-5-1) C _rests = (V (h, i, j + 1, k-1) -V (h, i, j, k)) / μ _{rests ···} (5-5-2) C _npr = (V (h, i-1, j + 1, k) -V (h, i, j, k)) / μ _{npr ···} (5-5-3)

Case 6: (h + i + j + k = N, i = 0, k ≠ 0) C _restp = ∞ (5-6-1) C _rests = (V (h, i, j + 1, k-1) -V (h, i, j, k)) / μ _{rests ···} (5-6-2) C _npr = ∞ ··· (5-6-3)

Case 7: (h + i + j + k = N, i ≠ 0, k = 0) C _restp = ∞ _··· (5-7-1) C _rests = ∞ ··· (5-7-2 ) C _npr = (V (h, i-1, j + 1, k) -V (h, i, j, k)) / μ _{npr ···} ( _5-7-3 )

Case 8: (h + i + j + k = N, i = 0, k = 0) C _restp = ∞ _··· (5-8-1) C _rests = ∞ ·· (5-8-2 ) C _npr = ∞ ··· (5-8-3)

Here, the meaning of each symbol is as follows. C _RESTP: link cost _{C Rests} for priority calls for the working path: link cost for the backup path for the priority call _{C npr:} link cost for non-priority call for the path

The above equations (5-1-1) to (5-8-3)
Means, for example, formulas (5-6-1) to (5-6-3)
In case 6, all resources are used, there is no resource used only by the priority call protection path, and there is a resource used only by the non-priority call path. This is a state transition from the link state. In such a link state, the working path call for priority call and the non-priority call cannot be connected any more. Therefore, the expressions (5-6-1) and (5-6- As shown in 3), the link cost for the priority call working path and the non-priority call path becomes infinite. On the other hand, since the priority call protection path call can be connected, the link cost for the priority call protection path is expressed in a general form as described above.

The present invention calculates the call type link cost based on the above principle, and one embodiment of the present invention will be described below with reference to the flowchart of FIG. [Estimation of call type call volume or arrival rate of each communication path setting request]

As is clear from the above-mentioned principle of the present invention, in order to calculate the call type link cost, first, the arrival rate and release rate of the call type communication path setting request must be given. In the present embodiment, first, the call type call volume or the arrival rate of each communication path setting request in each link is estimated based on the observation result obtained by observing the resource usage state in each link (S-1).

By observing the resource use state in each link, the average number of resources h used by the priority call working path, the average number of resources i used only by the priority call protection path during the observation period, and the priority call If the number of resources j shared by the backup protection path and the non-priority call path and the number of resources k used only by the non-priority call path are obtained, call type call volume in each link or each communication path setting The request arrival rate is estimated as follows.

First, using the Erlang B formula, the formula (6-
The estimated added call volume of all communication path calls can be obtained by 1) and (6-2). Next, using the obtained call volume and call type communication path release rate, equations (7-1)- (7-
According to 3), it is possible to estimate the call type call volume in each link or the arrival rate of each communication path setting request. In addition,
The release rate of the call type communication path corresponds to the reciprocal of the communication time, and can be obtained by observing the usage state of the resource in the link for a long time.

[Calculation of call type link cost] Using the call type call volume estimated as described above, the arrival rate and release rate of each communication path setting request, and the profit loss rate obtained from each communication path, In accordance with the principles of the present invention, the formula (1-
The call type link cost is calculated from 1) to (5-8-3) (S-2). Although the arrival rate of each communication path setting request is used in the above-mentioned principle of the present invention, the call type call volume in each link and the arrival rate of each communication path setting request are expressed by equation (7-
As shown in 1) to (7-3), if one of them is known, the other is also known. Therefore, instead of the estimated arrival rate of each communication path setting request, the estimated call type call on each link It is also possible to use the quantity to calculate the call category link cost.

Further, it is preferable to calculate the call type link cost periodically based on the observation result of the resource use state in each link, for example, at a time interval of about 10 times the average communication path hold time. At this time, if the calculation is performed using only the observation result within one observation period, the control of the network may not converge, that is, the feedback control for controlling the network may not converge due to the calculation result and may diverge. However, using the observation result in one observation period and the average observation result in a plurality of periods, for example, the call estimated by the following equations (8-1) to (8-3) If the call type link cost is calculated using the type call volume, there is no such fear. Note that the following formula is used to calculate the call type call volume, but the arrival rate of each communication path setting request can be further calculated using the release rate of each communication path.

Here, the meaning of each symbol is as follows, and 0 ≦ α ≦ 1. a _restp : Call volume added by working path calls for priority call estimated from average link state in multiple observation cycles a _rests : Call volume added by backup path call for priority call estimated from average link states in multiple observation cycles a _npr : Call volume added by non-priority call path calls estimated from the average link state in multiple observation periods

Further, when the priority call protection path is to be set, if the resource can be shared with the already existing priority call protection path, the link cost is regarded as zero. For this purpose, a table showing the relationship between each failure state and the number of restoration resources for each failure state may be held in each link. The technology of restoration using this table is `` BT Doshi, S.Dravida, P.
Harshavardhana, O.Hauser, andY.Wang, “Optical Networ
k design and restoration ”, Bell Labs. Tech.J., pp.58
-84, January March, 1999 ", detailed description thereof will be omitted. For example, in the case of a network in which a priority call working path and a priority call protection path are set as shown in FIG. In the table held by the link E-C, as shown in FIG. 5, the number of restoration resources for the failure state in the link B-C through which the two priority call working paths pass is two. The number of restoration resources for other failure states is one. By maintaining such a table, when setting a protection path for priority calls, from the link and node information used by the corresponding working path, when the working path fails, the link is used to restore It is possible to know whether or not it is not necessary to increase the resources for the backup path in the link even if the connection is performed, and if it is not necessary to increase the resource for the backup path, the link cost can be regarded as zero. The table showing the relationship between each failure state and the number of resources used for restoration at that time is updated at the time of setting and releasing the working path for priority call.

The calculated call type link cost C
_{restp, C rests, C np r} is, [Method 1]
_{Alternatively} , the profits r _restp , r _rests , r obtained from the communication path of each call type obtained by [Method 2]
_When it is larger than _npr , the link cost can be regarded as infinite so that the communication path of the call type cannot select the link. In this way, by retaining the unused resource for each call type, it is possible to reduce the profit loss in a long time interval.

The call type link cost in each link state is held for at least one observation period, and is changed depending on the link state at a period shorter than the observation period, for example, a time interval of about average communication path hold time. Only the call type link cost that was saved is advertised to the network (S-3, S-
4, S-5). When the observation period has elapsed (S-5), the call type call volume on each link or the arrival rate of each communication path setting request is newly estimated, and the call type link cost corresponding to the link state at that time is calculated.

Route selection can be performed by a well-known algorithm based on the calculated call type link cost in each link state. Basically, route selection is performed by selecting the route with the lowest cost, and m priority calls are used. When there is a request for setting the working path, the m paths are first selected, and then the backup path is selected. Also,
When selecting a route, it is possible to select a route based on cost from a plurality of route candidates that do not share a node with each other in advance, and it is also possible to select a route based on cost at all.

[0060]

As is apparent from the above description, according to the inventions of claims 1 and 2, the backup path for the working path for priority call between different node pairs shares the link resource, and the protection path for non-priority call is used. It is possible to calculate the link cost in a practical operation in which the path shares the link resource with the protection path for the priority call, and the route of the call with different priority so that the profit rate due to the call loss is minimized. Can be selected. According to the third aspect of the invention, it is possible to eliminate the risk that the control of the network does not converge and becomes unstable.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a state example of a circuit switched communication network in a restoration mode.

FIG. 2 is a configuration diagram showing another state example of the circuit switched communication network in the restoration mode.

FIG. 3 is a flowchart showing an embodiment of the present invention.

FIG. 4 is a configuration diagram showing still another state example of the circuit-switched communication network in the restoration mode.

FIG. 5 is an explanatory diagram of a table showing the relationship between each failure state in FIG. 4 and the number of restoration resources.

[Explanation of symbols]

1-1 to 1-8 ... Link, 2 ... Management device, AE ...
Optical cross-connect device, a to d ... Node

Continued front page    F-term (reference) 5K030 GA11 HA01 LA17 LB01 LC09                 5K051 AA05 BB01 BB02 CC00 DD09                       DD11 FF11 FF16 GG01 GG06

Claims (3)

[Claims] 1. A circuit switched communication network in which a protection path for a priority call working path between different node pairs shares a link resource, and a non-priority call path shares a priority call protection path and a link resource. In the link cost calculation method, the call type call volume added to each link or each call based on the number of simultaneous connections of call types including priority call working path call, priority call protection path call, and non-priority call path call in each link A link cost calculation method comprising estimating an arrival rate of a communication path setting request and calculating a call category link cost in each state of each link based on the estimated value and a call category profit rate. 2. The link cost calculation method according to claim 1, wherein the number of simultaneous connections of the call type is obtained by observing a link state of each link. 3. The link according to claim 2, wherein the number of simultaneous connections of the call type is obtained based on an observation result in one observation period and an average observation result in a plurality of periods. Cost calculation method.