KR100444339B1 - Method for efficiently determining the capacity of a ring communication network - Google Patents

Abstract

The present invention discloses a method for calculating the capacity of an optical communication ring having the effect of reducing the computation time in the problem of distributing traffic in SONET network to reduce the overall network resource usage. The method of the present invention obtains information of expected communication demands for each link and source and destination node pairs of the ring network from the memory device; Calculating the clockwise maximum link capacity and the counterclockwise maximum link capacity in advance for the node pair having the first rerouting order among the node pairs connected to the source node; Under the condition that the maximum link capacity in the clockwise direction is greater than the maximum link capacity in the counterclockwise direction, the value obtained by dividing the maximum link capacity in the clockwise direction minus the maximum link capacity in the counterclockwise direction by half and the link capacity of the node pair. Compare the values and record the small value as the value of the destination node of the node pair; Update the clockwise maximum link capacity and the counterclockwise maximum link capacity with reference to the calculated value of the destination node; Recalculating each link capacity of the ring network based on the rerouting start node pair 10 and the rerouting end node pair 20 after the end of the rerouting calculation process.

Description

METHOOD FOR EFFICIENTLY DETERMINING THE CAPACITY OF A RING COMMUNICATION NETWORK}

The present invention relates to a method for calculating the capacity of an optical communication ring. More particularly, the present invention relates to a capacity of an optical communication ring having an effect of reducing the calculation time in a problem of distributing traffic in a SONET network to reduce overall network resource usage. It is about a calculation method.

In general, the capacity determination problem of an optical communication ring network is an optical transmission ring necessary to satisfy a given communication demand in an optical transmission network constructed by a synchronous optical transmission (SONET) and a synchronous digital hierarchy (SDH) method. This applies to the Ring Loading Problem (RLP).

This ring loading problem (RLP) is defined as follows. Set of nodes Set of links with Undirectional ring-shaped graph made up of Assume that is given. link And link Is defined as referring to the same link. Just link This link Anything marked with is recognized as an exception. Is the set of indices of two nodes, or node pairs. Each node pair about, Communication demand exists, Wow ( ) Is defined as representing the supply and demand nodes of node pair k, respectively. Wow The flow between can be sent in both clockwise and counterclockwise directions. In other words, the flow Moving in the direction, the flow is called clockwise Moving in the direction is called counterclockwise. When the ring is formed, the same capacity link must be installed in all the links, so the capacity required for the ring configuration is determined by the flow rate of the link through which the most flow of the link passes. Since the amount of flow per link in a ring is determined by how to distribute and route the communication demand between each pair of nodes, the capacity determination problem of the ring becomes a problem to find a routing method that minimizes the capacity of the ring.

In a ring capable of routing in both directions, it is divided into a case in which the demand between two nodes can be divided and transmitted in both directions at the same time, and a case in which the transmission must be performed in either the clockwise or counterclockwise direction. Under these conditions, the capacity determination problem of the ring can be divided into a case where split routing is possible and a case where it is not possible. In the former case, that is, a capacity determination problem in a ring where split routing is allowed will be referred to as a ring loading problem with demand splitting (RPLW). In some cases, even when partitioning is allowed, transmission must be an integer unit. This is because the case where the capacity of the link can be installed only as an integral multiple of the unit capacity is considered. Even for cases where partitioning is not allowed and for partitioning where only integers are allowed, RLPW's solution is also useful for solving the remaining problems because of the nature of relaxation.

In other words, the capacity determination problem of the ring is applied to the problem of minimizing the capacity of the SONET ring required to satisfy a given communication demand in a synchronous optical network (SONET) operating in both directions. In SONET ring, the optical cable connecting ADM (Add-Drop Multiplexer), which is an optical transmission device, is connected to a node as a link, and there are two types of transmissions in one direction only and two directions in clockwise and counterclockwise directions. . SONET rings are subject to technical constraints that the links that make up them all must have the same capacity. Therefore, in the case of a SONET ring that transmits in both directions, the required ring capacity varies depending on how to route the communication demand between each node pair in the clockwise and counterclockwise directions. Therefore, an optimal routing method for minimizing the capacity of the ring is determined. Determining capacity of the ring is an important issue. Particularly, in SONET ring, the unit of demand is represented by the number of channels having a certain capacity. Therefore, the problem that is allowed in integer unit even if partitioning is not allowed or allowed in the capacity determination problem of ring.

As such, the capacity determination problem of the ring has been studied because of its practical importance. Of these, RLPW's solution, which finds a routing method that minimizes the capacity of the ring when segmentation of demand is allowed, is directly applied to the capacity determination of synchronous optical networks (SONET), and even if segmentation of demand is not allowed or allowed, Even problems that can only be used in units can be usefully used as solutions to mitigation problems.

Due to this applicability, Myung et al. (Y.-S. Myung, H. -G. Kim and D. -W. Tcha, "Optimal load balancing on SONET bidirectional rings," Operations Research, 45 1997), pp. 148-152) have been proposed simultaneously.

Therefore, in the present invention, the theoretical calculation time of the solutions developed so far By improving the rerouting process of the fastest Myung et al solution The aim is to propose a new solution which has been improved.

In other words, the present invention The purpose of this paper is to propose a solution that can find the optimal solution of RLPW within the calculation time. This is the quickest solution to RLPW than any solution ever published.

1 is an exemplary view for explaining an example of a method for calculating the capacity of a conventional optical communication ring;

2 is an exemplary diagram illustrating a relationship between a node pair k = {i, j} and k + 1 = {i, h} of an optical communication ring;

3A to 3F are exemplary views for explaining a method for calculating the capacity of the optical communication ring of the present invention.

Explanation of symbols on the main parts of the drawings

10: Rerouting start node pair

20: Rerouting end node pair

The present invention for achieving the above object is to provide a processor and a memory device to determine the routing direction to minimize the capacity required for communication demand between the pair of source and destination nodes on a bidirectional ring network having a plurality of nodes. CLAIMS 1. A method of calculating an optical communication ring capacity for driving a computer system equipped with: (a) obtaining information of expected communication demand for each link and source and destination node pairs of a ring network from a memory device; (b) calculating a clockwise maximum link capacity and a counterclockwise maximum link capacity for the node pair having the first rerouting order among the node pairs connected to the source node; (c) Under the condition that the maximum link capacity in the clockwise direction is greater than the maximum link capacity in the counterclockwise direction, the maximum link capacity in the clockwise direction minus the maximum link capacity in the counterclockwise direction is divided by half and the corresponding node pair. Compare the link capacity values of and record the small value as the value of the destination node of the corresponding node pair; (d) update the maximum link capacity in the clockwise direction by subtracting the value of the recorded destination node from the maximum clock capacity in the clockwise direction, and add the value of the recorded destination node to the maximum link capacity in the counterclockwise direction; Compare a value obtained by subtracting the value of the recorded destination node from the link capacity corresponding to the destination node of the node pair of the next rerouting order, and update the large value to the maximum link capacity in the counterclockwise direction; (e) Terminate the rerouting calculation process under the condition that the updated maximum clock capacity in the clockwise direction is equal to the maximum maximum clock capacity in the updated clockwise direction, based on the rerouting start node pair and the rerouting end node pair. And recalculating each link capacity of the optical communication ring.

In the present invention described above, step (c) and step (d) is characterized in that it is repeated sequentially under the condition that the maximum link capacity in the clockwise direction is greater than the maximum link capacity in the counterclockwise direction.

In the above-described present invention, the step of recalculating each link capacity of the ring network based on the rerouting start node pair and the rerouting end node pair of the step (e) is half of the rerouting start node pair. The capacity of the links belonging to the clockwise direction is calculated by adding a recording value of the destination node of the rerouting end node pair, and the capacity of the links belonging to the clockwise direction of the rerouting end node pair is calculated from the destination node of the rerouting end node pair. It is calculated by subtracting the record value, and the capacity of the remaining links is calculated by subtracting twice the record value of the corresponding object node of each link and adding the record value of the object node of the rerouting end node pair.

In addition, the method for calculating the capacity of the optical communication ring of the present invention is prepared by a program of a predetermined type, and can be used in various computer systems including a processor and a memory device and calculating the capacity of the optical communication ring.

Hereinafter, the present invention will be described in more detail.

For example, node pairs in a SONET ring Wow Communication demand That exists in Of course, communication demand in Telecommunication demand Means. That is, it is assumed that the same communication demand exists in both directions. In the ring capacity problem of the present invention, since the demand is assumed to exist in only one direction, it can be considered that the SONET ring is different from the capacity decision problem. However, in SONET ring, because two or four optical cables are overlapped, the routing method that minimizes the capacity of the problem addressed in this paper coincides with the solution that minimizes the capacity of SONET ring.

In the present invention We propose a solution to find the optimal solution for RLPW within the calculation time. The solution of the present invention is an improvement of the rerouting method (meaning to change the routing method) in the solution of Myung et al., So that the optimal solution of the RLPW can be found faster than any solution ever published. Also, if Myung's solution to solve the problem that the division is allowed only in integer units with higher applicability in the design of SONET ring, the solution presented here instead of Myung et al. We can solve the problem that partitioning is allowed only in integer units.

First, the symbols used in the present specification are defined.

Each node pair Supply node for Demand node from Set of links that exist on the clockwise path to The set of links that exist on the path in the counterclockwise direction. It is defined as In other words, ego to be. Thus link Is always all about Included in Also all links Link to Is a set of node pairs that Is a set of node pairs that are included in the counterclockwise path. It is defined as In other words, ego to be. therefore It is easy to see that. Node pair Supply and demand nodes Wow Node pairs when To Instead of Also referred to as.

Each node pair Variable for Using Wow Let's display the amount of flow routed clockwise among the demands. And If you define Becomes a vector corresponding to a routing method that satisfies a given demand. Given routing method About,

If you define silver Links when routing communications demand It means the sum of the flow passing through. Each link Has capacity for Suppose is given by then the link The sum of the flows at can not exceed the capacity of the link, so all links about This must be satisfied.

RLPW is formulated as a problem of determining the capacity of a ring to determine the routing method that minimizes the maximum of the capacity of the link needed to meet a given demand.

.

(RLPW) can also be expressed as a linear planning model:

Next, the calculation time according to the solution of the present invention will be described by comparing with the solution of Myung et al.

In this comparative example, the solution according to the present invention This paper proposes a solution that can find the optimal solution of (RLPW) within the calculation time. The solution according to the present invention improves the efficiency of calculation time by improving the rerouting method, which is the most important element of the solution of Myung et al. (Hereinafter referred to as MKT solution), and therefore, first introduces the MKT solution and introduces the improved part. The solution according to the present invention will be described.

The outline of the conventional MKT solution is described first. The MKT solution first arranges pairs of nodes belonging to K in a certain order and then routes the demand between all pairs of nodes clockwise. In order to minimize the capacity of the ring, rerouting is performed. Pairs of nodes in the order in which they are arranged for Change some or all of the demands of the routing in the counterclockwise direction so that the maximum capacity of the ring is reduced. Changing the routing is as simple as: Current Routing Method For which node pairs are the most flowing links That belongs to the clockwise path of

If, the node pair currently being routed clockwise If you reroute the demand counterclockwise, you can reduce the ring capacity. Node pairs in the rerouting phase Demand is rerouted to minimize the capacity of the ring. The amount of demand for which the routing is changed may be total demand or only part of the demand may be rerouted. In the latter case, the routing method after rerouting about,

This happens when only part of the demand is rerouted until. The MKT solution can be expressed as

The MKT solution is

1. (ordering of node pairs) To Belong to The second node pair Back side Make it happen, If is Back side Be sure to

2. (Decision of initial solution) Each node pair About Shall be.

3. (rerouting) each node pair Do the following in the order in which they are arranged for. In other words,

After calculating If,

And put it as follows Readjust:

An example of application of the MKT solution is shown in FIG. 1. 1 shows an example in which three node pairs have communication demands in a ring composed of five nodes. 1 shows a ring composed of five nodes and links, and a thick straight line connecting the node pairs represents node pairs {1,4}, {3,4}, and {3,5} where communication demand exists. The numbers on the thick straight lines represent the communication demands of these node pairs.

Rerouting { i, j } Link capacity g ( x , l ) l = 1 2 3 4 5 {1,4} {3,5} {3,4} 4 4 18 4 0 min {(18-4) / 2, 4) = 4min {(14-4) / 2, 4) = 4min {(10-4) / 2, 10) = 1 0 0 14 8 4 4 4 10 4 8 5 5 9 5 9

Table 1 shows the process of sending all demand clockwise initially and reducing the capacity of the link through rerouting during the rerouting phase. In this case, the rerouting order in the MKT solution is {1,4}, {3,5}, and {3,4}, which are the arrangement order of the node pairs. This rerouting order first consists of two node pairs. About Back side end Ahead, then Back side end It is chosen to go ahead.

In the table of Table 1, 4, 4, 18, 4, and 0 in the first column represent the sum of the flow amounts of each link by the initial allocation, that is, the required link capacity. In step 3, {1,4} is rerouted first by the rerouting order. Links 1, 2, and 3 included in the clockwise direction of the node pair {1,4} are indicated by bold numbers. And counterclockwise links 4 and 5 are usually numbers of thickness. The maximum value at the current capacity of the links included in the clockwise direction is , The maximum value in the counterclockwise direction is to be.

therefore This will cause rerouting. The amount of demand sent counterclockwise Because of Becomes Therefore, 4 is subtracted from the clockwise links (links 1, 2 and 3), and added to the counterclockwise links (links 4 and 5), and then passed to the next rerouting node pair {3,5}. If the same process is repeated again for node pairs {3,5} and {3,4}, the link capacity is the fourth column, 5, 5, 9, 5, 9, and the maximum link capacity is 9 Is the optimum capacity of the ring.

Rearrangement of stage 1 It is possible by calculation and setting of initial solution of step 2 Calculation is possible. Repeat step 3 for any node pair Once for It takes time, so step 3 full Is performed in time. Myung et al. Have shown that the MKT solution can find an optimal solution that minimizes link capacity. The MKT solution is the best in the theoretical calculation time of RLPW's solutions to date, but if demand exists at every node pair MKT solution It can be seen that the time required for.

As the analysis above, the time required for the MKT solution depends on the calculation of the rerouting step. Therefore, if the rerouting phase can be calculated more efficiently, the overall time can be improved. In the MKT solution, the rerouting of one node pair The reason it takes time All links in the calculation of about I need information from According to the change of To adjust the value of Because time is needed.

The new method according to the invention (called MK solution) proceeds basically the same as the MKT solution but at the rerouting stage. It is designed to reduce the total calculation time by making the calculation more efficient.

The following symbols are further defined to illustrate the new MK solution according to the present invention. Set of node pairs Set of supply nodes for Define. And each For the supply node Set of pairs of nodes It is defined as The MK solution newly proposed in the present invention is About Reroute node pairs belonging to By completing it in time Enable it within In other words, the MK solution of the present invention is different from the MKT solution. Iterative process Not times You will complete the rerouting which repeats only once. At this time So the rerouting in the MK solution It is completed within the calculation time of. Later, the entire process of the MK solution, including the rerouting process, We will show that this is done at the time of calculation, but first we will introduce the rerouting process.

As mentioned above, the MK solution of the present invention about Rerouting takes place for each pair of nodes belonging to The node pairs of are arranged as in the MKT solution. In the MK solution Reroute node pairs belonging to To complete in time Adjust the value After rerouting is complete for all node pairs belonging to. In this case all links For Without information of the value I need to be able to calculate This is possible using the structural characteristics of the supply and demand nodes of the pair of nodes belonging to.

As in Figure 2 Wow this Suppose that two consecutive pairs of nodes belong to. Because of the way the node pair to be. Node pair The rooting method obtained just before the rerouting Say,

Wow

Define. In other words Is a node pair Routing method obtained just before rerouting Node pairs for The maximum capacity of a link in the clockwise direction of, Is the maximum capacity of the link in the counterclockwise direction. therefore By the definition of Wow If you know Can be obtained. Use this to pair nodes The rooting method obtained after the rerouting Let's say If the and All links to save For If we adjust the value of, the method will proceed in the same way as the MKT solution.

The new solution of the present invention Wow If you know only some links Even if you know the value of and It comes from the idea that you can save. first Can be expressed by definition as:

As also shown in Figure 2,

Considering the above relationship, the above relationship can be easily seen. therefore To save Wow For some links besides Information about the value of,

If only you can know Can be obtained.

Meanwhile If you know Can be easily derived from the following facts.

(Theorem 1)

Two consecutive pairs of nodes belonging to Wow The following holds true for:

The necessary condition for ego,

Also Back side to be.

(proof)

By the nature of the MKT solution Always holds true. Node pair The rooting method obtained just before the rerouting Let's say now Of links in Consider this maximum link. Of course, there may be a plurality of such links. One of these links Consider each case and the case where it does not. Of links in Of these largest links Link to Suppose that exists. In other words,

and

to be. then, and By the definition of

This holds true. If the Of links in While this is maximum Suppose no link belongs to. In this case This holds true. therefore, Or Situation always occurs only in the latter case, Since this holds true, theorem 1 holds.

Node pairs by the above theorem Rerouting for Only needed when I can get it If you are aware of this, you can perform the next step of rerouting. In other words When rerouting each node pair belonging to the Wow If you know that not all links Information about the value of, To determine whether rerouting is necessary in the next step, and even if it is necessary and Can be obtained.

From now on, based on the above facts Rerouting for all node pairs in Let's show that it can be completed in time. For this purpose about Consider the algorithm REROUTING_SUB, which performs the rerouting on all node pairs in. REROUTING_SUB is the current routing method And all corresponding links For Receive information from Routing method created as a result of rerouting after rerouting for all node pairs belonging to And all corresponding links For Algorithm to output information. Thus each If REROUTING_SUB is executed repeatedly for, it is to perform the rerouting step of MKT solution.

<Algorithm REROUTING_SUB >

(For convenience of expression Is assumed)

1. (initialization stage)

, , .

2. (Repeat Step) Each Node Pair Do the following:

( Always in the nature of ).

If the Then go to step 3:

, .

3. ( Readjusted)

In step 2 As explained earlier This is the maximum value of capacity for additional links needed to calculate. Of step 2 Meaning and phase of To help you understand how to readjust From the first node pair of Table 2 shows the process of rerouting up to the first node pair. In Table 2, changes in the flow rate of each link in the rerouting phase are shown as '+' and '-'. The flow rate of the link in the clockwise direction of each node pair decreases and the flow rate of the link included in the counterclockwise direction increases. The shaded portion of the table shows the links included in the counterclockwise direction.

Node pair Link capacity change l = 1 … i- 1 i d ( k ) d ( k ' +1) d ( k ' ) d ( k ' -1) d (1) … n One + … + - … - - … - - + … + k'-1 + … + - … … … - - + … + + k'` + … + - … … - - + … + … + + k '+ 1 + … + - … … - + … … + k + … + - - +

Referring to Table 2, in step 2 When calculating After the first rerouting Satisfying links field How to get the rescaled value of All links after rerouting until For Understand how to find the value of.

Now let's analyze the calculation time of REROUTING_SUB. In initialization It will be time consuming and in step 2 one link will be Since the calculation never overlaps All the iterations for each node pair of The calculation Possible in time. In step 2 The remaining calculations only take constant time for one node pair. Because of Possible in time. So the total calculation time for step 2 is Complete in time. In step 3 The readjustment of the link between equations (1) and (2) Is calculated once, it takes constant time for each link. If the readjustment for the link in equation (3) Node pair about In Link When setting the value of Part As increases This can be done in constant time by further reducing. So the whole process of step 3 Possible in time.

As mentioned above, the MK solution according to the present invention proceeds substantially the same as the MKT solution. However, unlike the MKT solution, the MK solution of the present invention uses an initial solution for each node pair. About For the early years Find the value of, and then Repeat REROUTING_SUB for. In addition, the MK solution of the present invention is guaranteed to find the optimal solution because the process is the same as the conventional MKT solution.

Now, the calculation time of the MK solution is analyzed and explained. Finding the initial solution Available in time and for early years The value of will be explained later in Step 3 of REROUTING_SUB. In a similar way to rebalancing You can finish in time. Then REROUTING_SUB will be Can be done in time, so the MK solution You can finish in time.

In conclusion Since the MK solution of the present invention The optimal solution of the RLPW can be found within the calculation time. The process outlined below is for the initial year. To get the value of. Algorithm INIT is in step 3 of REROUTING_SUB Like if you readjust It is readily apparent that this is possible in time.

<Initial year ( For) Decision Process>

1. (Reset)

2. (Repeat Step) Each about Do this.

algorithm

(For convenience of expression Assume)

Each node pair For sequential steps:

Next, an application example of the MK solution according to the present invention will be described with reference to FIGS. 3A to 3F. As described above, the rerouting order according to the present invention is substantially the same as the conventional MKT rerouting order, and the link recalculation process of the link is different.

The MK solution of the present invention first calculates the maximum link capacities in clockwise and counterclockwise directions for the node pair (2.6) having the first rerouting order among the node pairs connected to the source node 2 as shown in FIG. Start by setting each to a and b.

Here, if the clockwise maximum link capacity a set for the node pairs 2 and 6 is larger than the counterclockwise maximum link capacity b, that is, a> b, the rerouting calculation process continues.

In this example, since a = 20 and b = 3, as shown in FIG. 3A, the condition of a> b is satisfied. Therefore, as shown in FIG. 3B, the value obtained by subtracting the maximum link capacity a in the clockwise direction minus the maximum link capacity b in the clockwise direction and dividing it in half again is compared with the capacity of the node pairs (2, 6). Record the small value as the value of node [6], which is the destination node of node pair (2,6). That is the formula The value obtained by the calculation of Record as.

Then, the maximum link capacity a in the clockwise direction and the maximum link capacity b in the counterclockwise direction are updated. At this time, the updated maximum link capacity a in the clockwise direction is updated by subtracting the previously recorded value of the destination node. In addition, the maximum counter capacity in the clockwise direction (b) is updated by adding the value of the previously recorded target node and the value of the recorded target node in the capacity of the corresponding node pair (here, node pair (2, 6)). The subtracted value is compared and updated to a larger value. In this example, a = a-3 = 17 Is updated.

Next, confirm that the clockwise maximum link capacity a set in advance for the rerouting calculation of the node pairs 2 and 5 is larger than the counterclockwise maximum link capacity b as shown in FIG. 3C. Continue the rerouting calculation process for the node pair (2,5) in the routing order.

As described above, the updated a = 17 and b = 6 satisfy the condition of a> b. Therefore, the destination node before the small value obtained by comparing the capacity of the node pair (2,5) with the value obtained by subtracting the maximum link capacity (a) in the clockwise direction minus the maximum link capacity (b) in the counterclockwise direction. The value of the sum of the recording values is recorded as the value of node [5] which is the target node of the node pair (2, 5). In other words, the formula The value obtained by the calculation of is added to the node [5] by adding the recorded value of the target node. In other words Record as.

Then, the maximum link capacity a in the clockwise direction and the maximum link capacity b in the counterclockwise direction are updated. At this time, the updated maximum link capacity a in the clockwise direction is updated by subtracting the previously recorded value of the destination node. In addition, the maximum counter capacity in the clockwise direction (b) is updated by adding the value of the previously recorded destination node [6] and the recorded destination node in the capacity of the corresponding node pair (here, node pair (2, 5)). The value is updated by subtracting the value of. In this example, a = a-2 = 15 Is updated.

Next, as shown in FIG. 3D, it is confirmed that the clockwise maximum link capacity a set in advance for the rerouting calculation of the node pairs 2 and 4 is larger than the counterclockwise maximum link capacity b. Continue the rerouting calculation process for the node pair (2,4) in the routing order.

As described above, the updated a = 15 and b = 10 satisfy the condition of a> b. Therefore, the destination node before the small value obtained by comparing the capacity of the node pairs (2,4) with the value obtained by subtracting the maximum link capacity (a) in the clockwise direction minus the maximum link capacity (b) in the counterclockwise direction. The value obtained by adding the recording value of [5] is recorded as the value of node [4] which is the target node of node pair (2, 4). In other words, the formula The value obtained by the calculation of is added to the node [4] by adding the recorded value of the target node. In other words Record as.

Then, the maximum link capacity a in the clockwise direction and the maximum link capacity b in the counterclockwise direction are updated. At this time, the updated maximum link capacity a in the clockwise direction is updated by subtracting the previously recorded value of the destination node. In addition, the maximum counter capacity in the clockwise direction (b) is updated by adding the value of the previously recorded destination node [5] and the capacity of the corresponding node pair (here, node pairs 2 and 4). The value is updated by subtracting the value of. In this example, a = a-2.5 = 12.5 Is updated.

Next, as shown in FIG. 3E, for the rerouting calculation of the node pairs (2, 3), it is first confirmed that the clockwise maximum link capacity (a) set above is larger than the counterclockwise maximum link capacity (b). . In this case, since a = b, the rerouting calculation process for the node pair (2,3) does not proceed.

In addition, since the remaining node pairs connected to the source node 2 do not need to be rerouted anymore, the rerouting calculation process is terminated.

Then, each link capacity of the ring network is recalculated. Specifically, referring to FIG. 3F, the links (links [6] and [1]) in the counterclockwise direction of the rerouting start node pair 10 are added by adding the recording values of the node [4] to the two link capacities, respectively. Recalculate link capacity. The links in the clockwise direction of the rerouting end node pair 20 (links [2] and [3]) recalculate the corresponding link capacities by subtracting the recording values of the node [4] from the two link capacities, respectively. The other links (links [4] and [5]) subtract the recording value of each node from the link capacity and add the recording value of node [4], which is the end node of the rerouting, to restore the link capacity. Calculate

As described above, the conventional calculation scheme is to recalculate the capacity of all links for rerouting node pairs, but the calculation scheme according to the present invention is to recalculate the capacity of a link by rerouting node pairs connected to a specific node. Are doing it all at once. That is, for the rerouting of the order (2,6), (2,5), (2,4), (2,3), the capacity recalculation of the link is performed after the rerouting of these node pairs. Therefore, the present invention can shorten the calculation time required to calculate the capacity of the ring network.

As described above, according to the present invention, the theoretical calculation time among the solutions for calculating the capacity of the ring network developed so far is By improving the rerouting process of the fastest Myung et al solution The new solution is improved.

In other words, according to the present invention, it is possible to contribute to the problem of reducing the overall network resource usage by distributing traffic in SONET network by presenting a solution that can obtain an optimal solution of RLPW faster than any solution ever published. .

In addition, the present invention is advantageously applied to a computer system that determines the routing direction so as to minimize the capacity of the communication demand required for the communication demand between a pair of source and destination nodes on a bidirectional ring network having a plurality of nodes. There is this.

In addition, according to the present invention, if it is used instead of the conventional solution to the problem that the partitioning is allowed only in integer units of higher applicability in the design of SONET ring It is advantageous to solve the problem that partitioning is allowed only in integer units.

In addition, if the MK solution of the invention is used in the same way, the problem that only division by integer is allowed There is an advantage that this becomes possible.

Claims (4)

Calculation of optical communication ring capacity that drives a computer system with a processor and a memory device to determine the routing direction to minimize the required capacity for communication demand between a pair of source and destination nodes on a bidirectional ring network with multiple nodes In the method, (a) retrieve information from the memory device of expected communication demand for each link and source and destination node pairs in the ring network; (b) calculating a clockwise maximum link capacity and a counterclockwise maximum link capacity for the node pair having the first rerouting order among the node pairs connected to the source node; (c) Under the condition that the maximum link capacity in the clockwise direction is greater than the maximum link capacity in the counterclockwise direction, the maximum link capacity in the clockwise direction minus the maximum link capacity in the counterclockwise direction is divided by half and the corresponding node pair. Compare the link capacity values of and record the small value as the value of the destination node of the corresponding node pair; (d) update the maximum link capacity in the clockwise direction by subtracting the value of the recorded destination node from the maximum clock capacity in the clockwise direction, and add the value of the recorded destination node to the maximum link capacity in the counterclockwise direction; Compare a value obtained by subtracting the value of the recorded destination node from the link capacity corresponding to the destination node of the node pair of the next rerouting order, and update the large value to the maximum link capacity in the counterclockwise direction; (e) End the rerouting calculation process under the condition that the updated maximum clock capacity in the clockwise direction equals the updated maximum clock capacity in the counterclockwise direction, and the rerouting start node pair 10 and the rerouting end node pair 20 And recalculating each link capacity of the ring network based on the reference. The method of claim 1, And (c) and (d) are repeated sequentially under the condition that the maximum link capacity in the clockwise direction is greater than the maximum link capacity in the counterclockwise direction. The method according to claim 1 or 2, Recalculating each link capacity of the ring network based on the rerouting start node pair 10 and the rerouting end node pair 20 of step (e), The capacity of the links belonging to the counterclockwise direction of the rerouting start node pair 10 is calculated by adding a recording value of the destination node of the rerouting end node pair 20, The capacity of the links in the clockwise direction of the rerouting end node pair 20 is calculated by subtracting the recording value of the destination node of the rerouting end node pair 20, The capacity of the remaining links is calculated by subtracting twice the recording value of the corresponding target node of each link and adding the recording value of the target node of the rerouting end node pair 20. How to. The method of claim 1, The rerouting order is for two node pairs (i, x), (j, y) first if i> j (i, x) precedes (j, y) and then x> y (i , x) is ahead of (j, y).