JP5595342B2 - Multiple path search method and apparatus - Google Patents

Description

    The present invention relates to a multi-path search method and apparatus, and in particular, sets a plurality of paths between a start point and an end point in a router / routing server or the like to perform load distribution / failure avoidance, or Video On Demand (VOD). The present invention relates to a multipath search method and apparatus for setting a multipath that secures a band from a stream server to a plurality of edge routers in a service or the like.

  The existing technology will be described with reference to FIG. An existing technique is a k shortest path algorithm that generates k paths in order from the lowest path cost between one desired start node and one end node (for example, see Non-Patent Document 1). The route cost here is the total of all link costs constituting the route.

  Step 1) A node is selected (at the beginning of the algorithm, the starting point node, and then the end node of the route candidate selected in the immediately preceding step 4).

  Step 2) Select a link from the node selected in Step 1. At this time, a plurality of links may be selected.

  Step 3) A new route candidate is created by connecting each link selected in Step 2 to the tail of the route candidate selected in the previous Step 4, and the route candidate list corresponding to the terminal node of the route candidate is created. Add. At that time, a loop route having a plurality of the same nodes on the same route is not a route candidate.

  Step 4) Select the route candidate having the lowest route cost among all route candidates on the network, and delete it from the route candidate list including the route candidate.

  Step 5) It is determined whether or not the terminal node of the selected route candidate is a desired terminal node. If it is not the desired terminal node, the process of Step 1 is performed. If it is a desired terminal node, the process proceeds to step 6.

  Step 6) The route candidate is added to the route list.

  Step 7) It is determined whether or not there are k routes in the route list. If k routes exist, the algorithm ends. If it is less than k, step 4 is performed.

IEICE technical report IN2010-107 Hiroshi Matsuura "Proposal of k shortest simple path algorithm to minimize computational complexity".

The existing k shortest path algorithm has both the number of start nodes and the number of end nodes, and if k paths to multiple end nodes are to be found, the k shortest path algorithm must be run for the number of end nodes. . In this case, there is a problem that the processing time is increased.
The k shortest path algorithm calculates k routes in order from the route with the lowest route cost, but these routes may not be optimal from the viewpoint of traffic distribution. In other words, there is a high possibility that a plurality of routes use the same link, and the use bandwidth of the link becomes large, resulting in a problem of creating a link that becomes a bottleneck.

  The present invention has been made in view of the above points, and can reduce a plurality of route search processing times, and a small number of fixed links with a small number of vias from a start node by a plurality of routes often seen as a result of a k shortest path algorithm. An object of the present invention is to provide a multi-path search method and apparatus capable of avoiding excessive use.

In order to solve the above-described problem, the present invention (Claim 1) is configured such that a plurality of nodes are connected to each other by a link to which a cost is given, and a desired end node (a plurality of end nodes) is formed from a desired start point node. A multi-route search method that searches for a plurality of routes up to
A route candidate creation unit creates a route candidate by selecting all links connectable from the start point node, and stores it in a route candidate list corresponding to the end node or end point of each link in the storage unit. Route candidate creation step,
The route candidate list control means searches for one route candidate having the lowest route cost among all the route candidate lists in the storage means, deletes it from the route candidate list , and terminates the route candidate end node. Is a desired end node, a minimum cost route deletion step of storing the route candidate in a route list corresponding to the end node ; and
The route candidate list control means further selects all links connectable from the end nodes of the searched route candidates, creates new route candidates by adding the selected links to the route candidates, A second route candidate creation step of storing in a route candidate list corresponding to the end node or the end point of the link;
The necessary number determination means continues the processing from the first route candidate creation step to the second route candidate creation step until the necessary number of route candidates are stored in the route list corresponding to all desired terminal nodes. Repeating steps,
The intends line.

Further, the present invention (Claim 2) sets, in claim 2, an upper limit number of route candidates including the number of route candidates already deleted as the minimum route in each route candidate list in the storage unit,
In the first route candidate creation step and the second route candidate creation step,
A route candidate storage determining step of storing route candidates in the route candidate list within a range not exceeding the upper limit of the number of route candidates.

Further, the present invention (Claim 3), in the route candidate storage determining step of claim 2,
When determining whether to store the new route candidate in the route candidate list corresponding to the terminal node / link terminal point of the route candidate list, the number of route candidates deleted as the minimum route in the route candidate list When the upper limit value of the number of route candidates in the route candidate list is reached, the new route candidate is not stored in the route candidate list.

Further, the present invention (Claim 4) is the route candidate storage determining step according to Claim 3,
When determining whether to store a new route candidate in the route candidate list corresponding to the terminal node / link terminal point of the route candidate list, (1) the route candidate deleted as the minimum route in the route candidate list The number of route candidates in the route candidate list does not reach the upper limit value of the route candidate list, and (2) the sum of the number of route candidates deleted as the minimum route in the route candidate list and the number of route candidates in the route candidate list If the route cost of the new route candidate is smaller than the maximum route cost, the route cost of the new route candidate is compared with the route candidate having the maximum route cost in the route candidate list. A route candidate having a cost is deleted from the route candidate list, and a new route candidate is newly stored in the route candidate list.

Further, the present invention (Claim 5), Oite before SL minimum cost path deletion steps of claim 1,
The route cost of the route candidate is not the sum of the costs of the links constituting the route, but a route with a large number of links from the start node is given a higher weight on the link cost so as to increase the route cost . including path cost adjustment steps.
Further, the present invention (Claim 6) is the route candidate storage determining step according to Claim 4,
The route cost of the route candidate is not the sum of the costs of the links constituting the route, but the route cost is increased so that the route cost is increased so that the route cost of the route having a large number of links from the start node increases. Route cost adjustment step.

Further, the present invention (invention 7 ) is characterized in that in the first route cost adjustment step of claim 5,
The path candidate with the minimum path cost in all the path candidate lists is set to path (s, u), the link from the end node is set to E (u, v), and the path candidate and the link are newly connected. If the path candidate is path (s, v), the path cost of the path (s, v) | path (s, v) |
| path (s, v) | = | path (s, u) | + | E (u, v) | + β | hop (path (s, v)) |
Where | E (u, v) | is the link cost, | hop (path (s, v)) | is the number of links through which path (s, v) passes, and β is a positive number.
The present invention (Claim 8) is characterized in that, in the second route cost adjustment step of Claim 6,
The path candidate with the minimum path cost in all the path candidate lists is set to path (s, u), the link from the end node is set to E (u, v), and the path candidate and the link are newly connected. If the path candidate is path (s, v), the path cost of the path (s, v) | path (s, v) |
| path (s, v) | = | path (s, u) | + | E (u, v) | + β | hop (path (s, v)) |
Where | E (u, v) | is the link cost, | hop (path (s, v)) | is the number of links through which path (s, v) passes, and β is a positive number.

The present invention (Claim 9 ) provides a route from a desired start point node to a desired end node (s) on a network formed by connecting a plurality of nodes to each other by links to which costs are given. A multi-path search device for searching a plurality of lines,
Storage means for storing a route candidate list;
First route candidate creation means for selecting all links connectable from the start point node to create route candidates and storing them in a route candidate list corresponding to the end node or end point of each link of the storage means; ,
If all the route candidate lists in the storage means search for one route candidate with the lowest route cost, delete it from the route candidate list, and if the end node of the route candidate is the desired end node A minimum cost route deleting means for storing the route candidate in the route list corresponding to the terminal node;
Further select all the links that can be connected from the end nodes of the searched route candidates, create new route candidates by adding the selected links to the route candidates, and set them as the end node or end point of each link. Second route candidate creation means for storing in a corresponding route candidate list;
The first route candidate creating unit, the minimum cost route deleting unit, and the second route candidate until the necessary number determining unit stores the necessary number of route candidates in the route list corresponding to all desired end nodes. A necessary number determining means for repeating each process of the creating means;
Have

The present invention (Claim 10 ) is a multi-path search program for causing a computer to function as each means of the multi-path search apparatus according to claim 9 .

  As described above, the present invention searches for one route candidate having the smallest route cost among all route candidate lists on the network, and if the terminal node of the route candidate is a desired terminal node, The k shortest path algorithm is stored by storing the route candidate in the route list corresponding to the terminal node and repeating the above processing until the required number of route candidates are stored in the route list corresponding to the desired terminal node. Does not need to be executed several times at the terminal node, and the processing time can be shortened.

It is a conventional multi-path search algorithm flow. It is a multiple path search method for a start node and an end node created in the present invention. It is the data memory structure in PCE in one embodiment of this invention. It is a data memory structure in case the path | route candidate list | wrist in one embodiment of this invention respond | corresponds to each link termination point. 1 is a configuration diagram of a multi-path search device in a first embodiment of the present invention. 5 is a flowchart of multi-path search in the first embodiment of the present invention. It is a figure for demonstrating the data structure of the route candidate list | wrist and new route acceptance determination in the 2nd Embodiment of this invention.

  Embodiments of the present invention will be described below with reference to the drawings.

  First, the outline of the present invention will be described.

  FIG. 2 shows an application example of the multipath search method of the start node and end node created in the present invention to the network. In the figure, a PCE (path computation element) is used as a multi-path search device.

  In the case where a plurality of paths are created by a PCE (path computation element) 1, in the figure, it is assumed that a plurality of paths are created between the start node 2 and the end node 7 between the nodes 2 to 7. However, the number of terminal nodes may be plural. In this example, it is assumed that after four paths are determined between the start node 2 and the end node 7 by the PCE 1, the paths are set in the switching table of each network node.

  FIG. 3 shows the structure of the data memory in the PCE when a plurality of paths are obtained from the start node 11 held in the PCE 1 to the three end nodes 12, 15 and 18 on the network holding 10 nodes. The nodes 11-20 and links L1-L14 held in the PCE are not nodes or links existing in the real network, but hold the configuration / information of the real network as they are, and each link L1-L14 has its direction. Have two different link costs.

  The route candidate list 121-201 held corresponding to each node other than the start point node holds the route candidates from the start point node to the node sorted by a binary search tree in the order of route cost. As indicated by 21, the route candidate holds a node permutation from the start node to the end node of the route candidate. In addition, the route candidate with the smallest route cost in each route candidate list is Fibonacci heap ("ML Fredman and RE Tarjan," Fibonacci heaps and their uses in improved network optimization algorithms, "Journal of the ACM (JACM), vol. 34, no. 3, pp. 596-615, 1987)), and the smallest node in this Fibonacci heap is a route candidate having the smallest route cost in all route candidate lists. Become.

  The three multipath end nodes 12, 15, 18 held in the PCE 1 hold the route lists 122, 152, 182 respectively, and are selected from the corresponding route candidate lists 121, 151, 181 in the route list. In addition, route candidates having the minimum route cost among all route candidate lists are stored in order.

  In FIG. 3, the route candidate list is held corresponding to each node. However, as shown in FIG. 4, the route candidate list can be held for each end point of the link. FIG. 4 shows an example in which a route candidate list is held for each link termination point in the node 14 in FIG. Since there are three link termination points in the node 14, three route candidate lists 142, 143, and 144 correspond to each termination point.

  FIG. 5 shows an apparatus configuration according to the first embodiment of the present invention.

  The multi-route search apparatus shown in FIG. 1 includes a node selection unit 10, a link selection unit 20, a route candidate creation unit 30, and a route list control unit 40.

  The operation in the above configuration will be described below along the flow of FIG.

  Step 201) When node information is input, the node selection unit 10 selects a node. At this time, the starting point node is selected at first, and thereafter, it becomes the terminal node of the route candidate selected in step 204.

  Step 202) The link selection unit 20 selects all links coming out of the node selected in Step 201.

  Step 203) The route candidate creation unit 30 creates a new route candidate in which each link selected in Step 202 is connected to the tail of the route candidate selected in the previous step 204, and outputs it to the route list control unit 40. The route list control unit 40 adds a route (node permutation) from the start point node to the node having the link end selected above in the route candidate list of the route list storage unit 50 corresponding to the end node / final link end point. to add. At this time, route candidates that pass through the same node twice on the route candidates are excluded. If the upper limit number of route candidates is set in advance, the number of route candidates equal to or less than the upper limit number is maintained (this process will be described in detail with reference to FIG. 7).

Incidentally, the path cost of the new path candidates is made in the form of adding the link cost that is added to the path cost of the route candidate selected in step 2 04 immediately before, when the number of links through which the route passes is large route cost It is also possible to give a specific gravity so as to increase. For example, the path cost | path (s, v) |
| path (s, v) | = | path (s, u) | + | E (u, v) | + β | hop (path (s, v)) |
It is also possible to define Here, when β is a positive number, the specific gravity is set according to the number of links, but when β is set to 0, the path cost of the conventional method is the same. By performing this processing, the cost of a link with a small number of vias from the start node is lower than the cost of a link with a large number of vias, and as a result, a link with a small number of vias searches for multiple routes equally. Will be selected. This avoids excessive use of fixed links by multiple paths that are often seen as a result of the k shortest path algorithm.

  Step 204) The route list control unit 40 uses the Fibonacci heap 22 of FIG. 3 to select the smallest route candidate from all the route candidate lists, extracts the route candidate, and deletes it from the route candidate list.

  Step 205) It is determined whether or not the terminal node of the route candidate selected in Step 204 is a desired terminal node. If it is a desired terminal node, the process proceeds to Step 206. If not, the process proceeds to Step 201. .

  Step 206) The route list control unit 40 stores the route candidate in the route list corresponding to the terminal node of the route list storage unit 50.

  Step 207) The route list control unit 40 determines whether or not a desired number of routes are stored in the route list corresponding to all the desired end nodes. If all the routes are stored, the processing is terminated, and the route list is still stored. If not, the process returns to step 201.

[Second Embodiment]
In the present embodiment, a method for managing the number of route candidates will be described.

  FIG. 7 is a diagram for explaining the data structure of the route candidate list and the new route candidate acceptance determination in the second embodiment of the present invention.

  In the present embodiment, in the route candidate list storage unit 50, the number of route candidates including the route candidate condition management memory 41, the route candidate number management memory 42 deleted as the minimum route, and the number of route candidates deleted as the minimum route. A management memory 43 is included.

  The route candidate number upper limit management memory 41 manages the upper limit of the number of route candidates in the route candidate list including the number of route candidates already deleted. Here, the upper limit value is A.

  The route candidate number management memory 42 deleted as the minimum route manages the number of routes deleted as the minimum route in the route candidate list. Here, the value is B.

  The route candidate number management memory 43 including the number of route candidates deleted as the minimum route manages the number of route candidates in the route candidate list including the number of route candidates already deleted as the minimum route. Here, the value is C.

  The route candidate 44 having the smallest route cost in the route candidate list shown in the route candidate list in FIG. 6 is linked to the node of the Fibonacci heap 22. The route candidate 45 is a route candidate having the largest route cost in the route candidate list.

  It is assumed that the route list control unit 40 performs the following control for the values A, B, and C held in the respective management memories.

  (1) In the case of C <A, since the route candidates are not stored in the route candidate list up to the upper limit of the number of route candidates, the new route candidates generated in step 203 are stored in the route candidate list.

  (2) When B = A, route candidates are not stored in the route candidate list.

  (3) When C = A∧B ≠ A, the new route candidate 46 is compared with the route cost of the route candidate 45, and only when the route cost of the new route candidate is smaller than the route cost of the route candidate 45, After the candidate 45 is deleted from the route candidate list, the new route candidate 46 is stored in the route candidate list.

  By performing the control as described above, it is possible to limit the number of route candidates to be handled in the processing shown in FIG. 6, and further shorten the processing time. In addition, since the upper limit of the number of route candidates in the route candidate list corresponding to each node and link termination point can be set, adjustments such as setting a higher upper limit value for nodes and links with larger capacities are possible.

  Note that it is possible to construct the operation of the constituent elements of the multi-path search device shown in FIG. 5 as a program and install it on a computer used as the device to execute it, or distribute it via a network. .

  The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

1 Multi-path search device (PCE: Path computing element)
2 Start point nodes 3 to 6 Nodes 7, 12, 15, 18 Multipath end node 10 Node selection unit 11 Start point nodes 13, 14, 16, 17, 19, 20 Network node 20 Link selection unit 21 Path candidate 22 Fibonacci heap 30 Path Candidate creation unit 40 Route list control unit 41 Path candidate number upper limit management memory 42 Route candidate number management memory 43 deleted as the minimum route Route candidate number management memory 50 including the number of route candidates deleted as the minimum route Route list storage unit 122,152,182 Route list 121,131,141,151,161,171,181,191,201 Route candidate list

Claims (10)

A multi-path search method for searching for a plurality of paths from a desired start node to a desired end node (s) on a network formed by connecting a plurality of nodes to each other by links to which costs are given. There,
A route candidate creation unit creates a route candidate by selecting all links connectable from the start point node, and stores it in a route candidate list corresponding to the end node or end point of each link in the storage unit. Route candidate creation step,
The route candidate list control means searches for one route candidate having the lowest route cost among all the route candidate lists in the storage means, deletes it from the route candidate list, and terminates the route candidate end node. Is a desired end node, a minimum cost route deletion step of storing the route candidate in a route list corresponding to the end node; and
The route candidate list control means further selects all links connectable from the end nodes of the searched route candidates, creates new route candidates by adding the selected links to the route candidates, A second route candidate creation step of storing in a route candidate list corresponding to the end node or the end point of the link;
The necessary number determination means continues the processing from the first route candidate creation step to the second route candidate creation step until the necessary number of route candidates are stored in the route list corresponding to all desired terminal nodes. Repeating steps,
A method for searching for a plurality of routes, characterized by: In the storage means, in each route candidate list, set a route candidate number upper limit including the number of route candidates already deleted as the minimum route,
In the first route candidate creation step and the second route candidate creation step,
The multiple route search method according to claim 1, further comprising a route candidate storage determination step of storing route candidates in a route candidate list within a range not exceeding the upper limit of the number of route candidates. In the route candidate storage determining step,
When determining whether to store the new route candidate in the route candidate list corresponding to the terminal node / link terminal point of the route candidate list, the number of route candidates deleted as the minimum route in the route candidate list The multiple route search method according to claim 2, wherein the new route candidate is not stored in the route candidate list when the upper limit value of the number of route candidates in the route candidate list is reached. In the route candidate storage determining step,
When determining whether to store a new route candidate in the route candidate list corresponding to the terminal node / link terminal point of the route candidate list, (1) the route candidate deleted as the minimum route in the route candidate list The number of route candidates in the route candidate list does not reach the upper limit value of the route candidate list, and (2) the sum of the number of route candidates deleted as the minimum route in the route candidate list and the number of route candidates in the route candidate list If the route cost of the new route candidate is smaller than the maximum route cost, the route cost of the new route candidate is compared with the route candidate having the maximum route cost in the route candidate list. 4. The multiple route search method according to claim 3, wherein a route candidate having a cost is deleted from the route candidate list, and a new route candidate is newly stored in the route candidate list. Oite to the minimum cost path deletion steps,
The route cost of the route candidate is not the sum of the costs of the links constituting the route, but a route with a large number of links from the start node is given a higher weight on the link cost so as to increase the route cost . multiple route searching method according to claim 1 Symbol mounting including path cost adjustment steps. In the route candidate storage determining step,
The route cost of the route candidate is not the sum of the costs of the links constituting the route, but the route cost is increased so that the route cost is increased so that the route cost of the route having a large number of links from the start node increases. Includes route cost adjustment steps
The multi-path search method according to claim 4. In the first route cost adjustment step,
The path candidate with the minimum path cost in all the path candidate lists is set to path (s, u), the link from the end node is set to E (u, v), and the path candidate and the link are newly connected. If the path candidate is path (s, v), the path cost of the path (s, v) | path (s, v) |
| path (s, v) | = | path (s, u) | + | E (u, v) | + β | hop (path (s, v)) |
Where | E (u, v) | is the link cost, | hop (path (s, v)) | is the number of links that path (s, v) passes through, and β is a positive number
The multi-path search method according to claim 5. In the second route cost adjustment step,
The path candidate with the minimum path cost in all the path candidate lists is set to path (s, u), the link from the end node is set to E (u, v), and the path candidate and the link are newly connected. If the path candidate is path (s, v), the path cost of the path (s, v) | path (s, v) |
| path (s, v) | = | path (s, u) | + | E (u, v) | + β | hop (path (s, v)) |
Where | E (u, v) | is the link cost, | hop (path (s, v)) | is the number of links that path (s, v) passes through, and β is a positive number
The multiple route search method according to claim 6. A multi-path search device that searches for a plurality of paths from a desired start node to a desired end node (s) on a network formed by connecting a plurality of nodes to each other by links to which costs are given. There,
Storage means for storing a route candidate list;
First route candidate creation means for selecting all links connectable from the start point node to create route candidates and storing them in a route candidate list corresponding to the end node or end point of each link of the storage means; ,
If all the route candidate lists in the storage means search for one route candidate with the lowest route cost, delete it from the route candidate list, and if the end node of the route candidate is the desired end node A minimum cost route deleting means for storing the route candidate in the route list corresponding to the terminal node;
Further select all the links that can be connected from the end nodes of the searched route candidates, create new route candidates by adding the selected links to the route candidates, and set them as the end node or end point of each link. Second route candidate creation means for storing in a corresponding route candidate list;
The first route candidate creating unit, the minimum cost route deleting unit, and the second route candidate until the necessary number determining unit stores the necessary number of route candidates in the route list corresponding to all desired end nodes. A necessary number determining means for repeating each process of the creating means;
A multi-path search device characterized by comprising: Computer
A multi-route search program for functioning as each unit of the multi-route search device according to claim 9 .

Images