JP5459791B2 - Delivery route determination apparatus and method - Google Patents

Description

  The present invention relates to a distribution route determination apparatus and method for determining a distribution route for distributed distribution of content, and in particular, a highly secure distribution route using less computational resources that can be calculated even in a practical scale network. The present invention relates to a delivery route determination apparatus and method that can be calculated and determined.

  Divide one content into n pieces of encrypted partial content, and distribute the secret so that the original content cannot be restored unless k (≦ n) pieces of partial content are collected from the n pieces of partial content Non-patent document 1 discloses (k, n) threshold secret sharing. Also, Patent Documents 1 and 2 disclose techniques for secretly distributing and holding n partial contents divided by a plurality of nodes in order to distribute the risk of data retention.

  Such technology can be used for content distribution that makes wiretapping difficult, and if (k, n) threshold secret sharing is used, at least k partial contents are distributed and distributed to distribution source nodes on the network, Content distribution is realized by performing distribution from a distribution source node that holds each partial content to a distribution destination node that is to restore the original data.

  However, at this time, if the distribution route of each partial content passes through the same link or the same node, the original data is restored just by eavesdropping less than k links or less than k nodes, and safety is ensured. Decreases.

  Similarly, as a risk other than eavesdropping, if the distribution path of each partial content passes through the same link or the same node, the distribution destination node only has a failure in less than k links or less than k nodes. This makes it impossible to restore content.

  Therefore, in order to realize distribution that is resistant to risks such as eavesdropping and failure in content distribution using (k, n) threshold secret sharing etc., do not pass the same link or the same node as much as possible, Even when a situation where a common part occurs between different routes is unavoidable, it is required to select a route that ensures safety as much as possible.

  In addition, such route selection technology can be used in addition to content distribution that makes it difficult to eavesdrop. For example, when the same content is redundantly distributed by a plurality of routes, such route selection is applied to improve safety. .

  As a prior art of such route selection, Patent Document 3 discloses that all partial contents are transferred when transferring k pieces of partial contents secretly held by a plurality of nodes to a distribution destination node via a network. A technique for calculating a distribution route that minimizes the probability of eavesdropping in the middle is disclosed.

  Further, in Patent Document 3, when content held in a plurality of nodes is transferred to a delivery destination node using k routes, all the routes become a failure and the content cannot be transferred to the delivery destination node. A technique for calculating a delivery route that minimizes the probability is disclosed.

JP 2009-122731 A JP 2009-10531 A Japanese Patent Application No. 2009-297262

Shamir, Adi (1979), "How to share a secret", Communications of the ACM 22 (11): 612-613

  However, in the prior art disclosed in Patent Document 3, the content distribution route is calculated by solving the integer programming method. Therefore, when applied to a practical scale network, the distribution route is reduced in a short time due to the limitation of calculation resources. There is a problem that cannot be calculated.

  An object of the present invention is to solve the above-described problems of the prior art, and in a system that distributes content by a distribution form such as a plurality of partial contents that are threshold secret shared or the same content is made redundant at the same time, It is an object of the present invention to provide a delivery route determination apparatus and method capable of calculating a delivery route with high reliability against eavesdropping and failure using a small amount of computing resources.

  In order to solve the above-described problems of the prior art, the present invention provides a route determination apparatus that determines a plurality of content distribution routes for distributing and distributing content from a plurality of distribution source nodes to a distribution destination node on a network as follows. It is characterized by taking various measures.

  (1) Root node selection means for sequentially selecting root nodes as starting points for route search without overlapping, and for each adjacent node of the selected root node, reach any one of the distribution source nodes from the adjacent node. Content distribution by connecting route search means for searching for an upstream route as the shortest route, an upstream route searched for each adjacent node, and a downstream route connecting the adjacent node and the delivery destination node Route establishing means for establishing a route, wherein the root node selecting means selects the delivery destination node as an initial root node, and after the first step, the delivery destination node is selected as a root node. A second step of sequentially selecting a root node from a set of adjacent nodes searched for an upstream route, and after the second step, the number of established content distribution routes is a target Reach, and executes a third step of repeating the set of neighboring nodes upstream path is searched by sequentially selecting the root node to.

  (2) The root node selection means selects the next root node after the content distribution path is established with respect to all adjacent nodes of the selected root node, and selects the root node in the third step as the distribution destination node. Is selected in ascending order of hop count.

  (3) The route search means, for each node that is an adjacent node of the selected root node and is not a node constituting any of the downstream routes before the selection of the root node, An upstream route is searched for as a shortest route from an adjacent node to any one of the distribution source nodes.

  (4) The route search means is a node that is adjacent to the selected root node, and is selected from the adjacent nodes for each node that does not exist on the content distribution route already established by the route establishment means. The upstream route is searched as the shortest route to reach the distribution source node.

  (5) In the second step and the third step, the root node selecting means sequentially selects the root nodes in the order from the shortest distance of the upstream path from the searched set of adjacent nodes. It is characterized by.

  (6) Each time the content distribution route is established by the route establishment unit, the link distance of each link constituting the already established content distribution route group, and the route search unit sets the upstream route as the shortest route It is further characterized by further comprising a link distance correcting means for correcting a link distance to be referred to for searching.

  According to the present invention, the following effects are achieved.

  (1) By sequentially selecting a root node with the distribution destination node as the first root node and selecting its adjacent nodes, the downstream path preliminarily distributed on the downstream side where the distribution destination node exists and the shortest path calculation are obtained. Since the content distribution route is determined sequentially by connecting the upstream route, it is possible to determine a safe content distribution route that is distributed with a smaller calculation load than in the prior art.

  (2) The downstream path group is constructed by branching sequentially in order from the end point closer to the delivery destination node, so that it is constructed in advance and a safe content delivery path can be determined.

  (3) The downstream route group is constructed without duplication, and a safe content distribution route can be determined.

  According to the features (1) to (3), the downstream route group is constructed by distributing in advance according to the order of the node visits by the breadth-first search with the delivery destination node as the first root node, and safe content A delivery route can be determined.

  (4) Only one upstream path is defined from the end points of one downstream path, and a distributed safe content distribution path can be determined.

  (5) Since the root node that is the starting point of the route search is selected in order from the node closer to the delivery source node, a safe content delivery route with a short link distance as a whole can be determined.

  (6) Since the correction is made so that the link distance of the established content distribution route becomes long and the subsequent upstream route is searched, the upstream route is searched while suppressing the tendency of overlapping portions between the upstream routes, A safe content distribution route can be determined.

It is a figure which shows the network in which content delivery is performed by the delivery route determined by the delivery route determination apparatus. It is a figure for demonstrating the outline from which a delivery route determination apparatus selects an optimal route. FIG. 3 is a diagram showing that a delivery destination node is first selected as a root node in establishing the delivery route of FIG. 2 and an upstream route is searched from its adjacent node (first adjacent node). It is a figure which shows that a content delivery path | route is established as a path | route which connected the upstream path | route and downstream path | route by the search of FIG. Next to the state shown in FIG. 4, among the first adjacent nodes, a node through which the route having the shortest distance among the searched upstream routes is selected as a new root node, and the adjacent node (second adjacent node) is selected. It is a figure which shows that an upstream route is searched from. FIG. 6 is a diagram showing that a content distribution route is established as a route connecting the searched upstream route and downstream route next to the state of FIG. 5. Next to the state shown in FIG. 6, among the first adjacent nodes, the node through which the second route from the shortest distance passes among the searched upstream routes is selected as a new root node, and the adjacent node (first It is a figure which shows that an upstream path | route is searched from (2 adjacent nodes). FIG. 8 is a diagram showing that a content distribution route is established as a route connecting the searched upstream route and downstream route next to the state of FIG. 7. Similarly, after the state in FIG. 8, the first adjacent node is sequentially selected as the root node, the upstream path is searched from the adjacent node, and the distribution path is established by combining with the downstream path to reach the target number. FIG. 10 is a diagram showing a state in which a second adjacent node is sequentially selected as a root node in the same manner and a distribution route is established in the same manner. It is a functional block diagram of the delivery route determination apparatus in one basic embodiment. It is a functional block diagram of the delivery route determination apparatus in one detailed embodiment. It is a flowchart of the process in embodiment which traces a root node in the order which a breadth priority search determines, and determines a delivery route. It is a figure which shows the example in which a delivery route is decided on a network with the flow of FIG. It is a figure which shows the example of a network structure for demonstrating link distance correction | amendment. It is a figure which shows the example of a network structure for demonstrating link distance correction | amendment. It is a figure which shows the example of a network structure for demonstrating link distance correction | amendment. It is a figure which shows the example explaining the effect of link distance correction | amendment.

  Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 shows a network in which content distribution is performed by k transfer paths determined and established by the distribution path determination apparatus 1 of the present invention. As shown in the figure, there are a total of C distribution source nodes that hold contents, c1, c2, c3,..., CC. As shown in the figure, each of these distribution source nodes ci (i = 1 to C) exists on the network where the content distribution is performed, together with the distribution destination node d and the distribution route determination device 1.

  Assume that a given position on the network of a total of C distribution source nodes ci (i = 1 to C) and distribution destination node d is given by address information or the like. The delivery route determination device 1 calculates a predetermined number k of content delivery routes that start from one of the delivery source nodes ci and arrive at the delivery destination node d that is the end point. Further, the distribution route determination device 1 notifies the distribution result to each distribution source node ci so that the content distribution according to the calculation result can be performed.

  The content distribution is distribution for the purpose of redundancy of distribution, and the same single content may be distributed through k routes. In addition, the content distribution is for the purpose of preventing eavesdropping by applying (k, n) threshold secret sharing and the like, and any k pieces of partial content that can restore the original content among the n pieces of partial content, Distribution via k routes may be possible.

  As a result of the distribution route calculation, there may be a node that does not distribute the content among the distribution source nodes ci. When the predetermined number k is smaller than the distribution source node number C, there is always such a non-distributed node. Further, as a result of the calculation, there may be a node that distributes a plurality of contents among the distribution source nodes ci. When k> C, there is always such a node.

  Also, for example, in the case of performing distribution using (k, n) threshold secret sharing, each distribution source node holds in advance each partial content to which (k, n) threshold secret sharing is applied. . Then, it is assumed that the partial content determined to be the distribution source by the distribution route calculation result is distributed to the distribution destination node through the designated distribution route.

  Next, an outline of the flow in which the delivery route determination device 1 sequentially establishes the delivery route as described above, and basic concepts used when establishing the establishment (upstream side, downstream determined as the shortest arrival route) Side route, etc.). In the general description, for the sake of convenience of explanation, only the part of the flow of establishment of the distribution route will be described. This is a basic embodiment. And after the said general description, each detailed embodiment including the detail of calculation of the shortest arrival route, etc. is further demonstrated.

  As shown in FIG. 2, an optimum route for distributing content from a plurality of content distribution source nodes c1 to cC to a single content distribution destination node d through a plurality of routes for the purpose of redundancy and threshold distribution. Think about the decision.

  First, as shown in FIG. 3, the content distribution destination node d is selected as the root node that is the starting point of the route search, and the adjacent node (first adjacent node) N11 whose hop count from the root node is “1”. , N12, N13... Are searched for upstream paths Pupp (Pupp1, Pupp2, Pupp3...) That can reach any one of the distribution source nodes ci (i = 1 to C) in the shortest time.

  Next, each upstream path Pupp is routed from each of the first adjacent nodes N11, N12, N13... To the content distribution destination node d (here, the root node and the content distribution destination node d are the same) via the root node. By connecting to the downstream path Plow (Plow1, Plow2, Plow3,...), As shown in FIG. 4, each content distribution source node ci (i = 1 to C) is connected to the content distribution destination node d. A plurality of content distribution paths Pcd1, Pcd2, Pcd3,... (Pupp1 + Plow1, Pupp2 + Plow2, Pupp3 + Plow3,...) Are newly established. In other words, a route that becomes a member of the k routes that are actually used for content distribution as shown in FIG. 1 is newly determined at the time. Further, the distance of each upstream path Pupp is calculated. Here, the description will be continued assuming that the results of Pupp2, Pupp1, and Pupp3 are obtained in the order of the shortest distance.

  Next, as shown in FIG. 5, the first adjacent node N12 through which the upstream route Pupp2 with the shortest distance passes is selected as a root node, and the number of hops from the new root node N12 is “1”. From the set of nodes (second adjacent nodes) N121, N122, N123,..., The content distribution source node for each of the remaining second adjacent nodes N122, N123... Excluding the node N121 through which the already established content distribution path Pcd2 passes. An upstream route Pupp (Pupp4, Pupp5,...) that can reach any of ci (i = 1 to C) in the shortest time is searched.

  Next, each upstream path Pupp4, Pupp5... Is connected to each downstream path Plow (Plow4, Plow5...) From the second adjacent nodes N122, N123. Then, as shown in FIG. 6, the content distribution routes Pcd4, Pcd5,... (Pupp4 + Plow4, Pupp5 + Plow5) connecting any one of the content distribution source nodes ci (i = 1 to C) and the content distribution destination node d. ...) is established in addition to the already established Pcd2 or the like.

  Next, as shown in FIG. 7, in the first adjacent node set (N11, N12, N13...), The upstream path Pupp1 passes through the upstream path Pupp1 that is next to Pcd2 that passes through N12. One adjacent node N11 is selected as a root node, and a content distribution route Pcd1 already established from a set of adjacent nodes (second adjacent nodes) N111, N112, N113... With the number of hops from the root node N11 being “1”. Upstream path Pupp (Pupp6) that can reach at least one of the content distribution source nodes ci (i = 1 to C) for each of the remaining second adjacent nodes N111, N113. , Pupp7 ...) is searched. Then, the upstream paths Pupp6, Pupp7,... Are connected to the downstream paths Plow (Plow6, Plow7,...) From the second adjacent nodes N111, N113. Thus, as shown in FIG. 8, the content distribution paths Pcd6, Pcd7 (Pupp6 + Plow6, Pupp7 + Plow7...) Connecting any one of the content distribution source nodes ci (i = 1 to C) and the content distribution destination node d. ) Is established in addition.

  Similarly, the process of sequentially selecting the first adjacent node that has not been searched from the first adjacent node set as the root node in the shortest order of the upstream route and repeating the route search results in the total content distribution route ΣP being the target number k. Repeat until it reaches. Further, even if the search is completed for all of the first adjacent nodes, if the total number of paths ΣP still does not reach the target number k, as shown in FIG. Switch to the second adjacent node set. Then, in the same manner as described above, each adjacent node of the second adjacent node is set as the third adjacent node, the upstream route is searched from the third adjacent node, and the distribution route is determined by combining with the downstream route, and the target number k is reached. Repeat this until. If the target number k is not reached even if the delivery route is determined with all the second adjacent node sets as the root nodes, the root node setting target is sequentially repeated as the third adjacent node, the fourth adjacent node,.

  The basic embodiment of the present application in which the distribution route is sequentially established has been described above with reference to FIGS. FIG. 10A is a functional block diagram of the delivery route determination device 1 in the embodiment. As shown in FIG. 10A, the delivery route determination device 1 includes a root node selection unit 2, a route search unit 3, and a route establishment unit 4.

  In starting the process of sequentially establishing the content distribution paths, the distribution path determination device 1 grasps the network topology information from the management server as appropriate as shown in FIG. 10A. Each unit of the delivery route determination device 1 fulfills each function by referring to the topology information as necessary.

  As described with reference to FIGS. 2 to 9, the root node selection means 2 sequentially selects the root node that is the starting point of the route search. As described with reference to FIGS. 2 to 9, the route search means 3 searches for the shortest route among the routes from each adjacent node of the root node to any of the distribution source nodes, and determines it as the upstream route. As described with reference to FIGS. 2 to 9, the route establishment unit 4 determines the upstream route determined for each adjacent node by the route search unit 3 and the downstream route connecting the adjacent node and the delivery destination node. By connecting, a content distribution path is established.

  Next, a more detailed embodiment will be described. As shown in FIG. 10B, the distribution route determination device 1 in the present embodiment has a configuration in which a link distance correction unit 5 is added compared to the configuration in FIG. 10A in the basic embodiment. Also in the detailed embodiment, the processing by the root node selection unit 2, the route search unit 3, and the route establishment unit 4 is consistent with the basic processing described in FIGS. 2 to 9, but as described below. More detailed processing is performed.

  That is, in the detailed embodiment, the processing in the root node selection unit 2 will be described in more detail. With this processing, the processing flow of the distribution route determination device 1 is to manage the node list with the distribution destination node as the first root node in the well-known breadth-first search, and sequentially visit the adjacent nodes of the root node in the order determined by the breadth-first search. In addition, the search is performed by adding to the node list, and the distribution route is determined. In particular, the node addition management to the node list that has not been described in the basic embodiment will be described in detail in this embodiment.

  Furthermore, in the one embodiment, the link distance correction means 5 is used. The link distance correction unit 5 corrects the link distance that is referred to when the route search unit 3 searches for the upstream route as the shortest route. The link distance correction unit 5 corrects the link distance of the links constituting the already established distribution route to be increased immediately before the upstream route is searched by the route search unit 3. Due to the correction, the link constituting the already established distribution route is less likely to be searched by the route search means 3 as the link constituting the shortest route. As a result, there is an effect that the established content distribution route tends to be a highly secure route with few common link portions between the routes.

  Note that the corrected link distance is referred to only when the route search means 3 obtains the upstream route. In functional blocks other than the route search means 3, although the link distance (link distance obtained from the network topology information first) without correction is referred to, the corrected link distance is not referred to.

  That is, in the embodiment, the delivery route is sequentially determined by the node visit according to the order determined by the breadth-first search and the route confirmation and distance correction as additional processing (other than the breadth-first search) at each visited site. The downstream path is obtained by the breadth-first search portion. The upstream route is obtained by the additional processing part, and the delivery route is established by combining with the downstream route.

  FIG. 11 shows a flowchart of processing in the detailed embodiment. FIG. 12 shows an example in which content distribution paths are sequentially established on the network by the flow, and the flow of FIG. 11 will be described using the example of FIG.

  In starting the processing in step S0 in FIG. 11, the distribution route determining apparatus 1 first obtains network topology information. In the topology information, as shown in FIG. 12, the connection destination node d, the content holding nodes c1, c2,..., Ce,. Link distance is given. By referring to the information, for example, it is understood that there are m adjacent nodes (nodes one hop ahead) of the distribution destination node d, N1, N2,..., Nm as shown in FIG. To do. In addition, when starting the process in step S0, the node list in the breadth-first search is empty.

  Note that the node list is not a functional block in the distribution route determination device 1, but corresponds to a queue in the breadth-first search, and is added to the queue as each node on the network, and is called a node list. That is, as will be apparent from the following description, a node list is used to describe the order management process of sequentially selecting root nodes by the root node selecting means 2 in association with a known breadth-first search.

  The process is started from step S0 in the above state, and the process proceeds to step S1 to add the distribution destination node d to the node list. The node list is {d}. Proceeding to step S2, the first node is extracted from the node list and is set as the root node. Here, the state is root node = d, node list = {} (empty list).

  Proceeding to step S3, among the nodes located one hop upstream of the root node (currently, the delivery destination node d), a node existing on the already established content delivery path is processed in the loop L1 starting from the subsequent step S4. It is excluded from the target, that is, the relay node. At this stage, the 1-hop upstream nodes are N1, N2,..., Nm, but since no delivery route has been established, no nodes are excluded.

(Term Definition 1) One hop upstream node of root node In step S3, “one hop upstream node” for the root node is a node list by the root node selection means 2 in the flow among the adjacent nodes of the root node. Nodes that have not been added to, i.e., unvisited nodes in breadth-first search. Therefore, the one-hop upstream node with respect to the root node refers to the topology information for obtaining the adjacent node and the node list history for confirming whether or not the node has been added to the node list. Desired. Then, as will become clear from the following description of the flow, the distribution destination node as shown in FIG. 12 is determined by the continuation of the relationship that follows one hop upstream of the root node starting from the root node that is sequentially updated. The root tree structure is obtained.

  Further, as will become clear from the following description, the matter of whether or not it has been added to the above node list is a node that constitutes one of the downstream paths of the already established distribution paths. It is equivalent to the matter of whether or not. Therefore, the above description can be rephrased as follows. That is, when obtaining the 1-hop upstream node of the root node, it can be obtained by excluding the nodes constituting the downstream path of the distribution path already established at that time from the adjacent nodes of the root node. .

(Term Definition 2) One-hop upstream node on established route In addition, in each step of FIG. 11, the “one-hop upstream node” in the node on the already established delivery route is the node on the delivery route of the relevant node. An adjacent node and a node closer to the distribution source node.

  Proceeding to step S4, the node is one hop upstream (currently N1, N2,..., Nm) of the root node (currently the delivery destination node d), and the route is not established, that is, not excluded in step S3, and If there is a node that has not yet been established in the loop L1, the process proceeds to step S5, and the node is selected as a relay node for determining the upstream route. That is, in step S4, if a node that is one hop upstream of the root node and does not exist on the established route remains, the process proceeds to step S5 to select the node, and the route is determined by the subsequent processing in the loop L1. Let it be established.

  Proceeding to step S6, the link distance correcting means 5 performs correction so that the link distance of each link constituting the already established distribution route is increased. Details of the correction of the link distance (such as an algorithm) will be described later. As described above, the corrected link distance is used only when the route search means 3 obtains the upstream route as the shortest route.

  If no delivery route has been established yet, step S6 is skipped.

  Note that the distribution route is established in step S8, but when the distribution route is to be established by addition, the above-described step S6 is necessarily passed. Therefore, every time a delivery route is established in step S8, correction of all distances of the already established delivery route including the delivery route established in the immediately preceding step S8 is executed in the next step S6. The (However, if the established delivery route is the k-th delivery route, which is the target number, there is no need to find any further delivery routes, so there is no need to correct the distance, and the k-th delivery route. The distance is not corrected.)

  Proceeding to step S7, the shortest path is calculated among the paths from the relay node to any of the distribution source nodes ci (i = 1 to C). The calculation is performed by the route search unit 3 based on the topology information and the link distance referred to as a value corrected by the link distance correction unit 5. In one embodiment, the well-known Dijkstra method is used for the calculation. Then, the route search means 3 determines the calculated shortest route as the upstream route.

  Proceeding to step S8, the route establishment unit 4 combines the upstream route determined for the relay node and the downstream route connecting the relay node and the delivery destination node d, so that the content distribution route is newly set to 1. The book is established.

  Proceeding to step S9, if the calculated number of routes has reached the target predetermined number k, the process proceeds to step S11 and the flow ends. In order to continue the description of the flow, assuming that k> (number of adjacent nodes m of the distribution destination node d)> 1, since the current route calculation number 1 has not reached k, the process returns from step S9 to step S4. .

  A loop that proceeds from step S4 to steps S5, S6, S7, and S8 and returns to step S4 because the number of route calculations has not reached k in step S9 is referred to as a loop L1 as shown. Since the details of each processing step in the loop L1 have already been described, the following description will be made with emphasis on the flow in which the root node is selected in the order determined by the breadth-first search and the delivery route is determined. Therefore, the process of steps S0, S1, S2, S3 and the flow of the loop L1 that have been described above is performed to briefly explain again how the delivery route is determined in FIG. 12, and thereafter The flow will be described with reference to the loop L1.

  First, at the point from step S0 to step S3, the state is root node = d, node list = {} (empty node). The one-hop upstream node of the root node is N1, N2,..., Nm (N2 is not shown) shown as the node group R100 in FIG. Does not exist above. Therefore, these one-hop upstream nodes whose routes have not been determined are sequentially used as relay nodes for calculating the upstream route as the shortest route in the loop L1, and the content distribution routes are sequentially determined. When routes are determined for all of the one-hop upstream nodes, a distribution route group in which upstream routes are calculated from relay nodes that are one hop upstream as indicated by solid arrows is obtained.

  For example, a distribution route c1-N111-N11-N1-d calculated from the 1-hop upstream node N1 and a distribution route cC-Nmb1-Nmb-Nm-d calculated from the 1-hop upstream node Nm are obtained. Here, for example, the former route is obtained by connecting the downstream route d-N1 and the upstream route c1-N111-N11-N1 via the relay node N1. Then, the constituent links of the distribution route group that is established by connecting these upstream routes and connected to the downstream route are the link distances that are referred to in the subsequent shortest route calculation in step S7 in step S6 after the establishment. The value is corrected so that becomes larger.

  Note that the order of selecting the one-hop upstream node with respect to the root node in step S5 of the loop L1 is not limited to the case where the nodes N1, N2,. This is different from the case where the order is specified in step S10 described later.

  As described above, when the route is established for all N1, N2,..., Nm of the one-hop upstream nodes of the root node d and the number of established routes has not reached k, the process proceeds from step S4 to step S10. Here, among the upstream nodes N1, N2,..., Nm of the root node d, the distance from the distribution source node to each upstream node in the established distribution route, that is, the distance of the upstream route portion in each distribution route is small In turn, the root node selection means 2 adds these one-hop upstream nodes to the end of the node list.

  As described above, the order in which the distances are small is determined using the link distance in the first topology information regardless of the correction by the link distance correction unit 5.

  In step S10, the processing is also performed on “one hop upstream node with respect to the root node on the already established distribution route” as described in (Term definition 2). Therefore, not only the node processed as the relay node in the loop L1 before reaching step S10, but also the nodes excluded in step S3 before entering the loop L1 are compared.

  This is based on the following circumstances. Even if the node excluded in step S3 is a node that constitutes an already established route, there may be a node whose route has not been established in the 1-hop upstream node. By selecting such a path non-established node as a relay node, it is possible to construct a tree structure in which downstream path groups are distributed evenly. A specific example will be described later with reference to FIG. 12 (step S10 when the root node is N1).

  As described above, in step S10, by adding one hop upstream node to the node list in ascending order of the distance not considering correction, the flow is further continued and the distribution route is sequentially established. The root nodes are selected in ascending order of distance from the distribution source node. Therefore, the additional delivery route established with the root node as the starting point is more likely to have a smaller distance than when the order is not followed. Therefore, there is an effect that it is possible to increase the tendency to suppress the distance of the entire delivery route at the same time while obtaining the delivery route dispersed as much as possible by the correction distance.

  It is assumed that the distance from the 1-hop upstream node of the root node to the distribution source node on the established distribution route (distance not considering correction as described above) is N1, N2,. In the node list, N1 with the shortest link distance is added last to {N1}, then N2 is added last to {N1, N2}, and then added to Nm in order, and then the node list is sequentially added. The node list becomes {N1, N2, ..., Nm}.

  Returning to step S2, the root node selection means 2 takes out the first node, that is, N1, and sets it as the root node. The node list is {N2, ..., Nm}. In step S3, for the root node N1, the nodes on the already established distribution path among the one-hop upstream nodes are excluded from the processing targets of the next loop L1.

  As shown in FIG. 12, the one-hop upstream node of N1 is a node group R210 (N11, N12,..., N1a) (N12 is not shown), among which N11 has already been established with N1 as a relay node N11 is excluded from the processing target in the loop L1.

  When the root node is d, (term definition 1) is followed, and all adjacent nodes are 1-hop upstream nodes. After the root node becomes N1, not all adjacent nodes become 1-hop upstream nodes in accordance with (Term Definition 1). In other words, the distribution destination node d is included in the adjacent node of N1, but this is not included in the one-hop upstream node and is excluded. In addition, the adjacent node of N1 may include d one-hop upstream nodes (other than N1) N2,... Nm, but these are not included in the one-hop upstream node and are excluded.

  As is clear from this example, in step S3, the one-hop upstream node according to (Term Definition 1), that is, the node constituting the downstream path of the distribution path established at that time from the adjacent node of the root node is excluded. By sequentially obtaining the nodes, the newly constructed downstream path is always constructed by branching with respect to the previously constructed downstream path group, and is constructed so that it does not merge again at the branch destination. Is done. That is, there is an effect that a downstream route group having a tree structure is constructed, and as a result, distributed delivery routes can be obtained.

  Entering loop L1, N1 one-hop upstream nodes N12, ..., N1a excluding node N11 are used as relay nodes, distance correction is performed on the constituent links of the already established distribution route, and a new distribution route is established. The For example, ce-N1a1-N1a-N1-d is established as a delivery route using N1a as a relay node. This is calculated by calculating the shortest route (upstream side route) ce-N1a1-N1a and the downstream route N1a-N1- is established by connecting d at the relay node N1a.

  If the number of distribution routes does not reach k even if the processing of the loop L1 is performed for all the relay nodes N12, ..., N1a to be processed, at this time, the node is a one-hop upstream node of the root node N1 and Since there are no nodes for which no route has been established, the process proceeds to step S10.

  In step S10, unlike the case where N11 is excluded in step S3, the distance (link distance) from all N11, N12,..., N1a of the one-hop upstream nodes of the root node N1 to the distribution source node on each established distribution route No correction) is compared. If N11, N12,..., N1a are in ascending order, the node list is {N2, ..., Nm, N11, N12, ..., N1a}.

  Similarly, after repeating the steps (step S2 → S3) → (loop L1) → (step S10) until the number of distribution routes reaches k, the node list is sequentially updated. FIG. 12 shows a state consisting of only a node group two hops upstream from the distribution destination node d, indicated as R200. If the number of established paths does not reach k, the process continues in the same manner, and the upstream path is calculated by using each node 2 hops upstream as the root node and the node group R300 3 hops upstream from the distribution destination node d as a relay node. The delivery route is established.

  For example, a distribution route (ce-N11c-N11-N1-d, etc.) is additionally established with the 2-hop upstream node N11 as the root node and the node of the node group R310 3 hops upstream as the relay node. For the root node N1a of the 2-hop upstream node, a delivery route (c1-N1ad-N1a-N1-d, etc.) is additionally established with the node of the node group R3a0 upstream of 3 hops as the upstream node. Similarly, for example, a node group R2m0 (Nm1,..., Nmb) constituting a further one-hop upstream node of the one-hop upstream node Nm is set as a root node, and any one of the nodes Nm11,. ., Nmb1 is used as a relay node to establish a distribution route.

  As is clear from the above description, in this embodiment, each upstream node of the distribution destination node d is used as a relay node in the order of the number of hops from d, and distribution paths are sequentially established until a predetermined number k is reached. That is, the distribution route is obtained as indicated by the solid line arrow with the distribution destination node d as the root node and the one-hop upstream node shown in the node group R100 as the relay node. Subsequently, a distribution route as indicated by a broken line arrow is obtained with each 1-hop upstream node, that is, each of the node group R100 as a root node and a 2-hop upstream node in the node group R200 as a relay node.

  Similarly, each of the two hop upstream nodes, that is, each node in the node group R200 is set as a root node, and the three hops upstream node shown in the node group R300 is set as a relay node, and a distribution route indicated by a dotted arrow is obtained. . Hereinafter, the upstream node is continued as a relay node in the same manner until k nodes are reached. Then, among the nodes having the same number of upstream hops from the distribution destination node d, the order of the root nodes for searching for the relay node in establishing the distribution route is determined in step S10. Further, before a new distribution route is established, the link distance of the constituent links of the distribution route already established in step S6 is corrected.

  According to the embodiment described in FIGS. 11 and 12 as described above, the following effects can be obtained. That is, in this embodiment, there is an effect of reducing the probability that a risk that occurs in a plurality of links through which a content distribution route passes causes a risk that extends to all distribution routes as much as possible. The effect is that, in the embodiment, in order to avoid the risk of one link being a risk to many distribution routes, k distribution routes for transferring content are arranged on the same link as much as possible. Even if there is no choice but to overlap, the delivery route is established along the following (Policy 1) and (Policy 2) with the aim of ensuring safety as much as possible This is an effect.

  (Policy 1) Maximize the number of risk-occurring links (for example, the number of links necessary for eavesdropping all k partial contents) necessary to cause the risk to reach all the distribution routes. That is, the number of established delivery routes is as close as possible to k.

  (Policy 2) Minimize the number of combination patterns (for example, the number of link selection patterns that can eavesdrop on all k partial contents) of the risk-occurring links that span all distribution routes as described above.

  In this embodiment, the above policy is satisfied as follows. That is, in this embodiment, in the tree having the distribution destination node as the root node (the tree structure as shown in FIG. 12 is obtained by sequentially tracing the 1-hop upstream node from the distribution destination node), Then, the upstream node is selected as the root node, and the shortest path calculation with the content holding node is performed for each node one hop upstream of the selected root node based on the corrected link distance.

  Here, by selecting the root node in the form of breadth-first tree search and constructing the downstream path that is sequentially branched, it is possible to avoid overlapping of the paths on the same link as much as possible, and to reduce the risk of all paths. It is possible to increase the number of links that cause risk. That is, (Policy 1) is satisfied.

  In addition, since the process of step S10 in FIG. 11 exists, by selecting a node having a short distance from the content holding node as a root node preferentially between nodes located upstream from the distribution source node by the same hop number, The distance of the calculated route is shortened, and the number of combination patterns of risk-occurring links that cause a risk for all routes can be reduced. That is, (Policy 2) is satisfied.

  Here, the risk is a link failure when transferring a single content redundantly by k paths, and a link when transferring k partial contents by (k, n) threshold secret sharing. Is eavesdropping. That is, if a risk occurs in all of k distribution routes, content distribution is not achieved in the case of redundancy, and secret content is decrypted and leaked in the case of (k, n) threshold secret sharing. Even in other cases, a corresponding risk can be defined in each of the arbitrary distribution forms in which the route calculated and established according to the above policy is advantageous (for example, risk of falsification during distribution), and distribution The route determination device 1 can be used. That is, the distribution route determination device 1 plays the role of a distribution route determination server in each distribution form having the form as shown in FIG. 1, and performs route calculation in response to a request from the distribution destination node d.

  Next, the correction of the link distance in step S6 of FIG. 11 in the embodiment, that is, the correction by the link distance correction unit 5 of the link distance referred to when the route search unit 3 calculates the upstream route as the shortest route will be described. . Of course, the correction is also performed in accordance with the aforementioned (Policy 1) and (Policy 2).

  FIG. 13A shows a network configuration example for explaining the correction. In the figure, there are a distribution destination node d, distribution source nodes c1 to c4, and intermediate nodes n1 to n50, and P1 (c1−n30−n31−d) and P2 (c2−n40−n41) indicated by solid lines as distribution paths. -N1-d), P3 (c3-n50-n51-n1-d), and P4 (c4-n50-n51-n1-d) have already been established. It is a state that is. In this state, the root node is n1, and the upstream path node n2 is used as a relay node to establish the distribution path P. That is, among the routes from the relay node n2 to any one of the distribution source nodes c1 to c4, the shortest route is obtained as an upstream route (for example, as indicated by a dotted line c3-n2), and has already been determined as indicated by a solid line. The distribution path P is to be established by connecting to the downstream path portion n2-n1-d. That is, it is the stage where step S5 in the flow of FIG. 11 is finished. Then, in order to examine what correction is made in step S6, a preferable route as the upstream route is calculated as the shortest route in the subsequent step S8, various routes are virtually examined. Thus, the risk will be considered when each of them becomes an upstream route. And the correction algorithm based on the said risk consideration is given. The same applies to the description of FIGS. 13B and 13C in the second embodiment described later.

    The link distance correction policy (technical significance) is as follows. When the route P passes only through a link through which no already established route passes, the addition of the route P increases the number of risk-occurring links necessary to cause a risk covering all the routes (policy 1). ), Which is preferable in terms of safety. However, when the path P passes through a link through which an already established path passes, the number of risk-occurring links that cause a risk for all paths does not increase, which is contrary to (Policy 1) and is not preferable compared to the former.

  For example, in FIG. 13A, when the route P passes through links n30-n31, links n40-n41, or links n50-n51 that constitute the already established delivery route, when a risk occurs in these links, all routes The number of risk-occurring links that cause a risk of up to 1 does not increase, which is an undesirable route. Therefore, in order to calculate a route P that avoids such a link, the distance of the links that constitute the already established route is corrected to a sufficiently large value. Furthermore, in this amendment, even if it is necessary to select a link to be avoided as a part of the route because it violates (Policy 1), it follows from (Policy 2) and from among those links to be avoided. Then, the weight is corrected so as to select a link that reduces the risk level in all the distribution routes finally determined.

  As a first embodiment of the correction to a sufficiently large value, a sufficiently large predetermined value is set as M, and a value obtained by multiplying the number of already established distribution paths passing through a link whose distance is to be corrected by M is a corrected link. Distance. As a result, when adding a new route, it is possible to increase the number of risk-occurring links that cause risks for all routes according to (Policy 1), and as can be seen from the following considerations (Policy 2) Accordingly, there is an effect that the value can be corrected according to the risk of each link.

  When the algorithm of the first embodiment is applied to all established delivery routes, the link distance changes as a result of the correction only when the “number of already established delivery routes that pass” changes. It is. That is, the link distance is changed by the correction only in the distribution route established in step S8 immediately before step S6 in FIG. However, in terms of form, the application target of the algorithm of the first embodiment may be all established distribution paths, and correction in the present invention includes a case where the distance does not change.

  The effect of the first embodiment is clear from the following consideration. For example, in FIG. 13A, when the route P passes through links n30-n31, links n40-n41, or links n50-n51 constituting the already determined delivery route, when a risk occurs in these links, The number of risk-causing links that cause risk does not change. From this perspective, the risk remains the same regardless of which of the three links is passed.

  However, when these three links are further compared, it passes through links n50-n51, compared with the case of passing through links n30-n31 and links n40-n41 (there are only one established delivery path each passing through P1 and P2). (Two already established distribution routes that pass through are P3 and P4), so the number of routes that pass through the same link increases, and the risk that all the routes reach risk due to the occurrence of risk in that link. That is, there is a greater risk when the link n50-n51 through which two established paths are selected is selected than the links n30-n31 and links n40-n41 through which one established path passes.

  Therefore, in the first embodiment, in order to secure the maximum number of risk occurrence links that cause the risk to reach all routes, and further suppress the probability that the risk will reach all routes (that is, do not select n50-n51 in the above example). To correct the link distance as described above. That is, the link distance is corrected to a value proportional to the number of distribution routes that have already been determined that pass through the link.

  Here, the effect of correcting the link distance will be described with reference to FIG. 14 as a specific example. FIG. 4A shows that the shortest route from the relay nodes n10, n20, and n30, which is one hop upstream node of a root node (not shown), to any one of the distribution source nodes c10, c20, and c30 is determined. It shows the state to do. Nodes a, b, and c exist in the middle, and the link distance between the nodes is as shown in parentheses below the line indicating the link. If no correction is performed, three shortest paths as indicated by a solid line in FIG. In this case, the risk occurs in all routes due to the occurrence of risk in only one link between the links a-c20.

  Therefore, the process when the first embodiment of the link distance correction is applied and the correction is made 10 times as an example of the predetermined number of paths already established is shown in FIGS. First, the shortest path from n10 is obtained as n10-a-b-c-c20 (distance 9) based on the link distance in FIG. 60 [n10-a], 10 [a-b], 10 [b-c] and 10 [c-c20] shown in FIG.

  Next, based on the corrected link distance shown in FIG. 5A1, the shortest path from n20 is calculated to obtain n20−a−b−c30 (distance 21). As a result, the link distance is further corrected to 60 [n20−a], 20 [a−b], and 50 [b−c30] shown in FIG. Since the link a-b passes through two routes via n10 and n20, the link a-b is corrected to the link distance 20 by the initial link distance 1 × passage number 2 × predetermined multiple 10.

  Further, based on the corrected link distance shown in (A2) in the figure, the shortest route from n30 is finally calculated to obtain n30−b−c−c10 (distance 26), and the link distance is further corrected in the same manner. , 80 [n30-b], 20 [b-c], and 80 [c-c10], and the state shown in FIG. The distribution route of (A3) is clearly distributed as compared with the distribution route of (B). For example, two links a-b and b-c are required as links that give risks to the entire route. In the distribution route shown in FIG. Safety is increasing compared to the risk of all routes with only one of them.

  Although the first embodiment of link distance correction has been described with reference to FIG. 14, it is obvious that other embodiments that greatly correct the link distance have the same tendency.

  Here, referring back to FIG. 13A, the second embodiment, which is another embodiment in which the link distance is corrected to a sufficiently large value, will be further described. The policy of the second embodiment is as follows. That is, in accordance with (Policy 2), in order to suppress the number of combination patterns of risk-occurring links that cause a risk extending to all routes, the link distance is corrected based on the number of combination patterns.

  As a specific method, a link that has a ratio of link distances is a risk-occurring link that causes a risk to reach all routes even if the link is added by passing sequentially established routes. When the minimum number of numbers does not change (without increasing), that is, when a risk occurrence link appears in a form that does not match (Policy 1), the link distance is equal to the ratio of the number of combination patterns of risk occurrence links. Correct.

  Here, the “link distance ratio” is precisely the “ratio between values integrated with respect to the initial link distance of each link (to be referred to as a correction coefficient) to correct the link distance”. is there. However, in the second embodiment, for the sake of simplification of description, the latter is represented by the former term, particularly in the description by the specific example. The same applies to the “ratio of correction distance”.

  A method of obtaining the number of combination patterns of the risk occurrence link used for the correction and how the link distance of each link is corrected based on the number of patterns will be described through first and second examples described below.

  Also, in this policy, the increase in the number of combination patterns of risk-occurring links that cause risks for all routes is the already established part of the route that is trying to establish the delivery route, as in the specific example described below, That is, the calculation is performed excluding the downstream path portion. This is referred to as a “pattern number calculation policy”. In other words, the contribution to the security of the newly established distribution route is considered as the contribution of the newly generated route that branches off from other established routes, that is, the upstream route portion. . Therefore, in the state where the upstream route is to be obtained, the number of combination patterns is obtained by removing the already determined downstream route portion.

  In addition, in the second embodiment, in order to explain the technical significance of link distance correction, in the state where the upstream side route has not been established (the state at the time when the above-described step S5 is completed), various routes are set. Consider the risks if established. In the same configuration as that of FIG. 13A, the path P is being established, and the upstream path portion of the distribution path P temporarily passes two links Pa or n30-n31 or links n40-n41. FIG. 13B shows a case where Pb is determined as any one of Pb. First, correction according to the second embodiment will be described using this as a first example. As shown in the figure, the path candidate Pa is c3-n30-n31-n2, and the link n30-n31 on the way is on the already established path P1. The route candidate Pb is c3-n40-n41-n2, and the link n40-n41 on the way is on the already established route P2. It is assumed that the portions indicated by the remaining dotted lines in both Pa and Pb do not have an established distribution route and a common link.

  When the route candidate Pa is established as a delivery route, a risk occurs in any one of the links constituting the part from the node C2 to the node n1 of the route P2 and one link n30-n31. In this case, the number of risk-occurring links that cause a risk extending to the routes P1, P2, and P does not increase from two to three despite the addition of the route P, and remains two. Here, in accordance with the “pattern number calculation policy” described above, the number of risk occurrence links is examined by excluding the n1-d portion of the route P2 that overlaps the established portion of the route P. In this case, the combination pattern of the two risk occurrence links that cause the risk on the paths P1, P2, and P is the number of links constituting the part (c2-n40-n41-n1) from the nodes c2 to n1 of the path P2. It is. The number of links corresponds to the risk when the delivery route P is determined using the links n30-n31, that is, when Pa is selected.

  On the other hand, when the route candidate Pb is determined as the delivery route, the risk is determined by any one of the links constituting the part from the nodes c1 to d of the route P1 and one link n40-n41. However, the number of risk-occurring links that cause a risk on the routes P1, P2, and P does not increase from two to three, but remains two, regardless of the addition of the route P. In this case, the combination pattern of the two risk occurrence links that cause the risk on the paths P1, P2, and P is the number of links constituting the part (c1-n30-n31-d) from the nodes C1 to d of the path P1. It is. Since this portion does not include the portion constituting the downstream path of P, according to the “pattern number calculation policy” described above, this portion is used as it is for the examination of the number of risk occurrence links. The number of links corresponds to the risk when the link n40-n41 is selected and the delivery route P is determined, that is, when Pb is selected.

Here, the correction coefficient is obtained by applying the link distance correction method in the second embodiment. If the ratio of the correction coefficients of link n30-n31 and link n40-n41 is a1: a2,
a1: a2 = (distance of c2-n40-n41-n1): (distance of c1-n30-n31-d)
= (Distance between the established path passing through the link n40-n41, that is, the part of P2 that does not share the link with the downstream path part d-n1-n2 that is the established part of the path P): (passes through the link n30-n31) Established path, ie P1 distance)
It becomes.

  This is rephrased into a form suitable for a general correction method of the link distance of each link described later. That is, in the above, the correction coefficient (ratio a1) of link n30-n31 is given by the part related to link n40-n41, and the correction coefficient (ratio a2) of link n40-n41 is the part related to link n30-n31. Therefore, by taking the reciprocal, the correction coefficient ratio is given in the portion related to each link itself.

And the ratio a1: a2 of the correction coefficients of the link n30-n31 and the link n40-n41 in the second embodiment is
a1: a2 = (the reciprocal of the distance of the established path passing through link n30-n31, that is, path P1): (the established path passing through link n40-n41, ie, path P2, sharing the link with the downstream path portion of path P (Reciprocal of the distance of the part that is not)
It becomes.

  The first example particularly indicates that the ratio is obtained by obtaining the upstream route and excluding the downstream route which is the determined portion of the route P to be established by the above-described “pattern number calculation policy”. It is an example for. That is, in the above (Formula 1), the path P1 with the ratio a1 does not have a common part with the downstream path part of the path P, so the entire path P1 is considered, and the path P2 with the ratio a2 has the downstream path of the path P. Since there is a part in common with the part, this part is excluded and considered.

  Similarly, the correction of the link distance in the second embodiment will be further described using a second example. FIG. 13C shows a second example. The second example is an example for explaining a case where there are two or more established paths in the first example described above, in which there is only one established path passing through each distance correction target link. is there. FIG. 13C shows a state in which the path P is to be established with the same configuration as FIG. 13A, and the upstream part of the distribution path P temporarily passes through the link n30-n31 or the link n50-n51. The case where it is decided by either candidate Pa or Pc is shown.

  When the route candidate Pa is selected and the delivery route is established, when the risk occurs in the link from the node c3 or c4 to the node n1, the link where the routes P3 and P4 overlap and the link n30-n31, The number of risk-occurring links that cause risks extending to the routes P1, P3, P4, and P does not increase from two to three, but remains two, regardless of the addition of the route P. In this case, the combination pattern of the two risk-occurring links that cause the risk on the routes P1, P3, P4, and P is the number of links where the routes P3 and P4 overlap in the portion up to the node n1. This corresponds to the risk of selecting Pa. Although the overlap between the paths P3 and P4 exists between n1 and d, it is excluded by the “pattern number calculation policy” as described above.

  On the other hand, when the route candidate Pc is established as a delivery route, when a risk occurs in one of the links constituting the part from the nodes C1 to d of the route P1 and the links n50-n51, the route The number of risk-occurring links that cause risks extending to P1, P3, P4, and P does not increase from 2 to 3, despite the addition of path P, and remains unchanged at 2. In this case, the combination pattern of the two risk occurrence links that cause the risk on the routes P1, P3, P4, and P is the number of links constituting the part from the nodes C1 to d of the route P1, and the Pc is selected. It corresponds to the risk in the case.

Here, the correction method is obtained by applying the link distance correction method in the second embodiment. The ratio of the correction coefficients of link n30-n31 and link n50-n51 is a1: a3.
a1: a3 = (distance n50-n51) :( distance c1-n30-n31-d)
= (Distance between all established paths P3 and P4 that pass through links n50-n51, and the determined part of path P, that is, the part that downstream path part n2-d does not pass): (passes link n30-n31) Established route P1 distance)

Similar to the first example described above, the ratio is converted to the reciprocal so that the ratio of the correction coefficient is given to the portions related to the links n30-n31 and n50-n51 themselves. That is,
a1: a3 = (reciprocal of the distance of the established path P1 that passes through the link n30-n31): (all established paths P3 and P4 that pass through the link n50-n51 pass, and the determined part of the path P, that is, the downstream path The reciprocal of the distance of the portion n2−d that does not pass through) (Formula 2)
It becomes.

From the description of the first and second examples in FIGS. 13B and 13C as described above, in the second embodiment, the calculation method for correcting the link distance of each link in general is also apparent. It is as follows. That is, first, a set of links that pass through a link whose distance is to be corrected and that constitutes all already established distribution route groups in common is defined as a first link set. Further, the link set included in the downstream path portion determined by selecting the relay node in step S5 immediately before step S6 in FIG. That is, the second link set is a link set that constitutes a downstream path of the next newly established distribution path.

  A difference set obtained by subtracting the second link set from the first link set is defined as a third link set. In the second embodiment, a corrected link distance is obtained by adding a sufficiently large predetermined value M to a reciprocal of the number of links included in the third link set to the link distance given first. The correction target is all the established distribution routes.

  Although the above calculation method reproduces (Equation 1) and (Equation 2) obtained in the examples of FIGS. 13B and 13C and is a general correction calculation method, the first to third link sets are expressed as follows. In order to confirm with a specific example, it will be described below that a1, a2, and a3 similar to (Expression 1) and (Expression 2) are obtained by the above method.

[a1]
The distance correction target links are n30-n31. Since the already established distribution route group passing through the link is {P1} whose component distribution route consists of only one route P1, its common part matches the distribution route group itself and is {P1}. . Therefore, the first link set is a link set included in P1. The second link set is {n2-n1, n1-d}. Since these have no common part, the first link set minus the second link set is the first link set itself, and this is determined as the third link set. Therefore, the third link set is a link set included in P1,
a1 = (the reciprocal of the distance of the established path passing through the links n30-n31, that is, the path P1), which matches (Expression 1) (Expression 2)

[a2]
The distance correction target links are n40-n41. The already established distribution route group that passes through the link is {P2} in which the distribution route of the component consists of only one route P2. The common part coincides with itself and is {P2}. Therefore, the first link set is a link set included in P2. The second link set is {n2-n1, n1-d}. Taking the difference set, the common part {n1-d} is drawn, and the third link set becomes the part (c2-n40-n41-n1) of P2. Therefore,
a2 = (the reciprocal of the distance of the established route passing through the links n40-n41, that is, the route P2, that does not share the link with the downstream route portion of the route P), which matches (Equation 1).

[a3]
The distance correction target links are n50-n51. The already established distribution route group passing through the link is {P3, P4}. Therefore, the first link set is {P3∩P4} = {n50−n51, n1−d}. The second link set is {n2-n1, n1-d}. Taking the difference set, the common part {n1−d} is drawn and the third link set becomes {n50−n51}. Therefore,
a3 = (reciprocal of the distance of all established paths P3 and P4 that pass through links n50-n51 and the determined part of path P, that is, the downstream path part n2-d does not pass), (Equation 2) Matches.

  In the second embodiment, when the link distance is corrected in this way, the number of risk-occurring links that cause the risk for all routes is not changed in spite of adding a new route according to (Policy 1). If the number of risk occurrence links does not change as much as possible, an increase in the number of combination patterns of such risk occurrence links can be suppressed according to (Policy 2).

  The formal significance of the correction between the first embodiment and the second embodiment will be described again with reference to the flow of FIG. Each step Sn (n = 4 to 9) of the i-th time (i = 1 to k) in the loop L1 is expressed as Sn [i]. Consider the correction in step S6 [i + 1] after reaching the step S8 [i] and establishing the i-th distribution route.

  In the first embodiment, the correction of step S6 [i + 1] changes the distance of the i-th distribution route established among all the already established distribution routes to be corrected.

  In the second embodiment, in the correction in step S6 [i + 1], the first link set based on the i distribution routes already established up to the i-th delivery route, and then step S8. By using the second link set constituting the downstream route portion of the distribution route established in [i + 1], the distance correction of the entire established distribution route is performed.

  That is, in the second embodiment, the first link set is a set based on the entire distribution route already established in step S8 [j] (j = 1 to i), and the second link set is set in step S5 [i + 1] is a link set that constitutes a downstream route that constitutes a delivery route established in a later step S8 [i + 1], which is determined by the relay node selection in [1].

  In addition, in the present invention as a whole, when a downstream path is constructed, adjacent nodes one hop ahead are sequentially traced. The relationship between the number of hops and the link distance referred to when calculating the upstream route as the shortest route (especially, the link distance obtained from the network topology information first before correction) is as follows. . That is, one embodiment of the link distance is the number of hops. The link distance can be defined by other than the number of hops. And in the whole this invention, the adjacent node at the time of constructing | assembling a downstream path | route is good also as a node in the unit distance destination in link distances other than the number of hops.

  As described above, in the present invention, in accordance with (Policy 1), the number of risk-occurring links that cause a risk on all routes is increased, and in accordance with (Policy 2), the number of combination patterns of such risk-occurring links is reduced. Thus, the order of calculation routes is determined, the link distance is corrected, and the shortest route calculation is sequentially repeated. A Dijkstra method or the like is used for the shortest path calculation, which is the portion with the largest load in the calculation for executing the processing. Therefore, the content distribution route can be calculated as a sub-optimal solution using a sufficiently small number of calculation resources as compared with the prior art integer programming method in which an exact solution is obtained but the calculation is enormous.

  The present invention can calculate k paths for transferring arbitrary k partial contents of n partial contents secretly held by a plurality of nodes to a delivery destination node via a network. Applicable. The present invention maximizes the number of eavesdropping links required for eavesdropping all k partial contents when partial contents being transferred may be eavesdropped on the link. It is possible to calculate a safe partial content distribution route that minimizes the number of combination patterns.

  In addition, the present invention can be applied to calculation of k routes for transferring k pieces of the same content that are redundantly held in a plurality of nodes to a delivery destination node via a network. According to the present invention, when multiple link failures can occur, the number of failure links in which all of the k paths fail and content is not transferred to the delivery destination node is maximized, and the number of such combination patterns of failure links It is possible to calculate a highly reliable content distribution route that minimizes the number of

DESCRIPTION OF SYMBOLS 1 ... Distribution route determination apparatus, 2 ... Root node selection means, 3 ... Route search means, 4 ... Route establishment means, 5 ... Link distance correction means

Claims (10)

In a route determination device for determining a plurality of content distribution routes for distributing and distributing content from a plurality of distribution source nodes to a distribution destination node on a network,
Root node selection means for sequentially selecting root nodes as starting points of route search without duplication;
For each adjacent node of the selected root node, route search means for searching an upstream route as the shortest route from the adjacent node to any of the distribution source nodes;
Path establishing means for connecting the upstream path searched for each adjacent node and the downstream path connecting the adjacent node and the distribution destination node to establish the content distribution path;
The root node selection means includes:
A first step of selecting the delivery destination node as a first root node;
After the first step, a second step of sequentially selecting a root node from a set of adjacent nodes searched for upstream paths with the distribution destination node as a root node;
After the second step, execute a third step of repeatedly selecting the root node sequentially from the set of adjacent nodes searched for the upstream route until the number of established content distribution routes reaches the target number. A delivery route determination device characterized by:   The root node selection means selects the next root node after the content distribution path is established for all the adjacent nodes of the selected root node, and in the third step, the root node is hopped from the distribution destination node. The delivery path determination device according to claim 1, wherein the selection is performed in ascending order of numbers.   The route search means, for each node that is an adjacent node of the selected root node and is not a node constituting any of the downstream routes before the selection of the root node, from the adjacent node The delivery route determination device according to claim 1, wherein an upstream route is searched as the shortest route to reach any one of the delivery source nodes.   The route search means, for each node that is an adjacent node of the selected root node and does not exist on the content distribution route already established by the route establishment means, any one of the distributions from the adjacent node. The delivery route determination device according to claim 1, wherein an upstream route is searched as the shortest route to reach the original node.   In the second step and the third step, the root node selection unit sequentially selects a root node from the set of searched adjacent nodes in order of increasing distance of the upstream route in the second step and the third step. The delivery route determination device according to any one of claims 1 to 4.   Each time the content distribution route is established by the route establishment unit, the link search unit searches for the upstream route as the shortest route that is the link distance of each link constituting the already established content distribution route group. 6. The delivery route determination device according to claim 1, further comprising link distance correction means for correcting a link distance referred to.   The link distance correcting unit corrects the link distance of each link to a value proportional to the number of the content distribution routes already established by the route establishing unit passing through the link. 6. The delivery route determination device according to 6.   Each time the content distribution route is established, the link distance correction unit configures the link distance of each link in common with all the content distribution routes already established by the route establishment unit passing through the link. A difference set obtained by subtracting a second link set, which is a set of links constituting the downstream path of the content distribution path to be established next to the established content distribution path, from the first link set, which is a set of links. The delivery path determination device according to claim 6, wherein the distribution path determination device corrects the value to be inversely proportional to the number of links constituting the third link set.   9. The delivery route determination device according to claim 1, wherein the route search means searches for the upstream route by a Dijkstra method. In a route determination method for determining a plurality of content distribution routes for distributing and distributing content from a plurality of distribution source nodes to a distribution destination node on a network,
A root node selection step for sequentially selecting root nodes as starting points for route search without duplication;
For each adjacent node of the selected root node, a route search step for searching for an upstream route as the shortest route from the adjacent node to any one of the distribution source nodes;
A path establishment step of connecting the upstream path searched for each adjacent node and a downstream path connecting the adjacent node and the distribution destination node to establish the content distribution path;
The root node selection step includes:
A first step of selecting the delivery destination node as a first root node;
After the first step, a second step of sequentially selecting a root node from a set of adjacent nodes searched for upstream paths with the distribution destination node as a root node;
After the second step, a third step of repeatedly selecting the root node sequentially from the set of adjacent nodes searched for the upstream route until the number of established content distribution routes reaches the target number. A delivery route determination method characterized by the above.

Images