JP2011134240A - Content replacement system, content replacement method, and content replacement program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a content replacement system which quickly replacements contents in a distributed server without reducing a cache hit rate in the distributed server. <P>SOLUTION: The content replacement system includes: a central server 30 holding a plurality of contents; a distributed server 20 which transmits contents from the central server 30 to user terminals; a placement-popularity extraction part 11A which extracts a plurality of delete-candidate contents from the distributed server 20, and extracts a plurality of storage-candidate contents to be introduced from the central server 30 to the distributed server 20; a replacement influence calculation part 11B which extracts arbitrary one of the extracted delete-candidate contents and arbitrary one of storage-candidate contents to specify a pair of contents, and predicts transition of popularity of each of the contents on the basis of the change of past popularity; and an instruction content determination part 11C which determines the storage-candidate content as a content to be introduced to the distributed server, on the basis of the calculation result. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a system for distributing content on the Internet, and in particular, a content replacement system for replacing content in a distributed base server installed in the vicinity of a user in order to reduce the load on a central base server, and content replacement The present invention relates to a method and a content replacement program.

  With the increase in performance of computers and the increase in bandwidth of networks, VoD (Video on Demand) services that allow users to acquire and view high-quality video content when they want to view them have become widespread.

  Here, if distribution is performed from a central base server such as a data center in response to a distribution request from all users, a large volume of video traffic (information flow) is generated on all routes from the central base server to each user terminal. As a result, a high load is applied to a wide range of networks and adversely affects other communications. In addition, because it uses a best-effort network (such as the amount of information that exceeds the capacity is discarded) over a long distance, such as the Internet, throughput due to congestion on the communication path (per unit time) It is easily affected by a significant decrease in the data transfer amount of the communication line) and an increase in the delay time, and there is a possibility that a significant waiting time until the start of viewing and a viewing interruption occur and the user's quality of experience is significantly impaired.

  On the other hand, a distributed base server such as a proxy server or an edge server capable of storing a part of content is installed in the vicinity of the user, and the popular base content is placed on the distributed base server so that the traffic is in the vicinity of the user. Hierarchical content distribution systems have been formed, which have alleviated the above problems. In response to the popularity of each content that changes from moment to moment, when a distribution request for content that does not exist in the distributed base server is received or by replacing the content stored in the distributed base server appropriately at regular intervals, Has been trying to maintain and improve the probability (cache hit rate) that content requested for distribution can be distributed from distributed base servers.

  As a related technique of the content replacement method for improving the cache hit rate, a method disclosed in Patent Document 1 is known. The content replacement method described in Patent Document 1 is based on the fact that a frequently accessed content is stored in a VISP unit (Virtual Internet Service Provider) connected to the same access network as the user. The burden of connecting to a remote access server is reduced. Then, with respect to the change in popularity, the ranking of the popularity after the change is compared with the content in the distributed base server, and the difference from the current one is replaced so that it is stored in the distributed base server in order of popularity. In this way, a state in which highly popular content is stored in the distributed base server is maintained. As a result, the cache hit rate can be maintained and improved, the communication distance to the user can be shortened stochastically, and the influence of the network load due to video traffic and the throughput reduction due to congestion can be reduced.

  Moreover, the following patent document 2 thru | or patent document 4 are known as a related technique regarding the replacement | exchange of the said content.

Patent Document 2 discloses a technology that can greatly reduce the cost required for content arrangement management.
Patent Document 3 discloses a technique that can create a replica while suppressing the communication load of the entire system.
Japanese Patent Application Laid-Open No. 2004-151867 discloses a technique that can prevent a distribution control device that transmits a content data distribution instruction to an external device in response to a received viewing request from unnecessarily increasing the waiting time of a viewer who has requested viewing later. It is disclosed.

JP 2003-8664 A JP2005-31987 JP2007-141103A JP 2009-89019 A

  However, the above-described technology for replacing content has the following problems. The content replacement method maintains and improves the cache hit rate by replacing the difference between the popularity ranking after the change and the content in the distributed base server with respect to the change in popularity, and the network load due to video traffic. However, it does not take into account the temporary drop in the cache hit rate due to content replacement. Since the contents being distributed to the user terminal cannot be deleted, the procedure for replacing the contents of the distributed base server is as follows. First, in response to a distribution request to the content before replacement (deletion target) in the distributed base server, the distributed base server does not distribute but distributes from the central base server (hereinafter, this state is referred to as “content blockage”). ). Then, after the distribution of the content to be deleted from the distributed base server is completed, the content is replaced.

  That is, while the content is closed, both the content before replacement (deletion target) and the content after replacement (replication target) are distributed from the central base server, and the cache hit rate is temporarily greatly reduced. Therefore, if content with little difference in popularity is replaced, the cumulative increase in long-term cache hit rate due to replacement may be less than the cumulative amount of temporary decrease in cash hit rate. May be reduced.

  Further, the technique described in Patent Documents 1 to 4 does not solve the problem that the cache hit rate decreases when the contents of the distributed base server are replaced.

(Object of invention)
The present invention provides a content replacement system, a content replacement method, and a content replacement program capable of quickly replacing content in a distributed base server without reducing the cache hit rate in the distributed base server. Objective.

In order to achieve the above object, a content replacement system according to the present invention includes:
Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server Content replacement means for deleting one content in the distributed base server and introducing one content in the central base server to the distributed base server, and a plurality of content recorded in the distributed base server Arrangement for extracting a plurality of deletion candidate contents from the contents and a plurality of storage candidate contents that are candidates to be introduced into the distributed base server from a plurality of contents stored in the central base server -Having a popularity degree extraction unit, the extracted deletion candidate content and the storage candidate container; A replacement effect calculation unit that extracts one arbitrary content from each of the channels and identifies a set of content, and predicts the transition of the popularity of each combined content based on the change in the popularity to date Provided, the replacement effect calculation unit predicts the popularity of each content after a certain time when the popularity of the set of content approaches each other, and the popularity of the storage candidate content is the deletion candidate An introduction content determination unit is provided that determines the storage candidate content for introduction into the distributed base server when the popularity of the content is greater.

In order to achieve the above object, the content replacement method according to the present invention includes:
Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server In the content replacement system for deleting one content in the distributed base server and introducing one content in the central base server to the distributed base server, a plurality of stored in the distributed base server Placement-popularity includes a plurality of deletion candidate contents from among the contents and a plurality of storage candidate contents that are candidates to be introduced into the distributed base server from among the plurality of contents stored in the central base server. Each of the extraction units extracts the extracted deletion candidate content and the storage candidate content. One arbitrary content is extracted from each of them, a set of content is specified, and the transition of the popularity of each combined content is predicted based on the change in the popularity so far, When the popularity of the set of contents approaches each other, the replacement impact calculation unit predicts the popularity of each content after a predetermined time, and the popularity of the storage candidate content is determined by the deletion candidate content. If the storage candidate content is larger than the popularity, the introduction content determination unit determines the storage candidate content for introduction to the distributed base server, and the content replacement unit deletes one content in the distributed base server based on the determination. The content replacement means introduces one content in the central base server to the distributed base server.

In order to achieve the above object, a content replacement program according to the present invention includes:
Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server In the content replacement system for deleting one content in the distributed base server and introducing one content in the central base server to the distributed base server, a plurality of content recorded in the distributed base server Contents that respectively extract a plurality of deletion candidate contents from the contents, and a plurality of storage candidate contents that are candidates to be introduced into the distributed base server from a plurality of contents stored in the central base server Extraction function, the extracted deletion candidate content and the storage candidate content A popularity trend prediction function that extracts one arbitrary content from each of them, identifies a set of content, and predicts the trend of popularity of each combined content based on the change in popularity to date, When the popularity of the set of contents approaches each other, the popularity of each content after a predetermined time is predicted, and the popularity of the storage candidate content is determined to be greater than the popularity of the deletion candidate content The content replacement means deletes one content in the distributed base server based on the popularity of the content, and an introduction content determination function that determines the storage candidate content for introduction into the distributed base server when And a content introduction function for introducing one content in the central base server to the distributed base server. Characterized in that so as to realize the computer.

  As described above, the present invention predicts highly popular content and introduces it in advance. According to this, the possibility that the cache hit rate of the distributed base server is reduced by replacing the content is effective. Therefore, it is possible to provide a content replacement system, a content replacement method, and a content replacement program having an excellent effect that the cache hit rate can be further improved.

It is a block diagram which shows one Embodiment of the content replacement system which concerns on this invention. It is a flowchart which shows the operation | movement of the delivery to the user terminal of the content replacement system in one Embodiment disclosed in FIG. It is a flowchart which shows the operation | movement of the arrangement | positioning update of the content of the content replacement system in one Embodiment disclosed in FIG.

  The present invention relates to a content distribution system that distributes content using the Internet, in which a distributed base server is installed in the vicinity of a user for the purpose of reducing the load on a central base server. When replacing the contents in the base server, compare the long-term cache hit rate improvement amount with the temporary cache hit rate decrease amount, and only if the long-term cache hit rate improvement amount is larger The replacement is performed, and it is possible to eliminate the disadvantage that the cumulative cache hit rate is lowered due to the content replacement.

Hereinafter, an embodiment of a content replacement system according to the present invention will be described with reference to FIGS. 1 to 3.
(Overall configuration)

FIG. 1 is a block diagram showing an embodiment of the content replacement system of the present invention.
In this figure, the content replacement system possesses all content with one or a plurality of distributed base servers 20 that perform content transmission / reception, replication to a cache area, and content replacement according to a predetermined content distribution schedule, A central base server 30 that transmits content in response to a request from the distributed base server 20 or the user terminal 50 and pre-distributes content according to a predetermined content distribution schedule, and displays content and a content list to the user. And one or a plurality of user terminals 50 that request content to the distributed base server 20 or the central base server 30 according to user operations, and content to be arranged on the distributed base server 20 based on the popularity of the content. Determined and said In response to a request from the user terminal 50, the content placement management apparatus 10 that instructs the distributed base server 20 to replace the content, the content placement status is confirmed with the content placement management apparatus 10, and the placement status is confirmed by an EPG (Electronic Program). A portal site 40 that processes a program guide such as a Guide) or ECG (Electronic Content Guide) or a content list of search results (hereinafter referred to as a content list) and notifies the user terminal 50, and a network 60. ing.

  In response to the content replacement instruction from the content arrangement management device 10, the distributed site server 20 requests the corresponding content from the central site server 30. Further, the distributed site server 20 responds to a request from the content storage unit 23 for storing the content received from the central site server 30, the content replacement unit 21 for replacing the content in the content storage unit 23, and the user terminal 50. A content transmission / reception means 22 for transmitting the content is provided.

  The central site server 30 transmits content in response to requests from the content storage means 32 possessing all the contents, the distributed site server 20 and the user terminals 50, and transmits content in accordance with the determined content distribution schedule. Or it is comprised with the content transmission / reception means 31 which performs prior delivery.

  Among these, the content transmission / reception means 31 reads the requested content from the content storage means 32 and transmits it to the content replacement means 21 when receiving a content request from the content replacement means 21 of the distributed base server 20. When a content request is received from the user terminal 50, the requested content is read from the content storage unit 32 and transmitted to the user terminal 50.

  The content arrangement management device 10 manages the arrangement of contents stored in the division site 20. The content placement management apparatus 10 determines the content to be placed on the distributed base server 20 based on the popularity of the content, and instructs the distributed base server 20 to replace the content.

  In addition, the content arrangement management device 10 includes a content arrangement storage unit 12 that stores information on content currently stored in the distributed base server 20, and a content popularity degree storage unit 13 that stores information related to the popularity of each content. Blocking / staying time storage means 14 for storing the closing time and staying time of each content in advance, content information currently stored in the distributed base 2 acquired from the content arrangement storage means 12 and content popularity degree storage means The popularity of the content acquired from 13 and the blocking time of each content at the time of content replacement acquired from the blocking / holding time storage means 14 and the residence time until a certain content is stored in the distributed base server 20 and then deleted. The content to be stored in the distributed base server 20 is determined and the distributed base server 0 is composed of a replacement content determination means 11 for instructing a replacement of the content. Here, the blocking time is defined as the time taken to block each content at the time of content replacement. The dwell time is defined as the time from when a certain content is stored in the distributed base server 20 until it is deleted.

  The replacement content determination unit 11 is based on the content information currently stored in the distributed base server 20 acquired from the content arrangement storage unit 12 and the popularity of the content acquired from the content popularity storage unit 13. A list of contents that may be deleted from 20 (hereinafter referred to as deletion candidate contents) and contents that may be stored in the distributed base server 20 instead of the deletion candidate contents (hereinafter referred to as storage candidate contents) A content arrangement / popularity extracting unit 11A for creating a list of Further, the content arrangement / popularity extraction unit 11A uses the content popularity acquired from the content popularity storage unit 13 and the deletion candidate content based on the blockage time and the retention time acquired from the blockage / retention time storage unit 14. And the storage candidate content are compared, and a replacement effect calculation unit 11B that calculates the effect when the content is replaced, and an introduction content determination unit 11C that determines the content to be replaced based on the effect when the content is replaced. And.

  Among them, the arrangement-popularity extraction unit 11A stores the contents stored in the distribution base server 20 from the contents arrangement storage unit 12 when the content replacement timing is reached in the content distribution schedule set by the operator of the content distribution system. Information on the content currently stored in the means 23 is acquired, and the popularity of the content is acquired from the content popularity storage means 13. Then, the contents extracted in order from the highest popularity of the contents by the amount that can be stored in the content storage means 23 (hereinafter referred to as ideal arrangement) are compared with the currently stored contents, and the ideal arrangement is obtained. Then, contents that are not present in the currently stored contents are listed as storage candidate contents, and contents that are not in the ideal arrangement but are present in the currently stored contents are listed as deletion candidate contents. The storage candidate content list and the deletion candidate content list are notified to the replacement influence calculation unit 11B. Each list clearly indicates the deletion candidate content and the storage candidate content.

  When the replacement influence calculation unit 11B receives the storage candidate content list and the deletion candidate content list from the arrangement-popularity extraction unit 11A, the replacement influence calculation unit 11B acquires the popularity of the content from the content popularity storage unit 13, The closing time and the residence time are acquired from the residence time storage means 14. Then, based on the popularity of the content, the blocking time, and the dwell time, a combination of a set of content in the combination of the deletion candidate content and the storage candidate content is compared and these contents are replaced. Calculate the impact.

  Here, the combination of contents to be compared may be a combination of arbitrarily selecting contents one by one from the list of candidate storage contents and the list of candidate deletion contents. In addition, the combination is not limited to the combination, for example, a combination that is extracted in descending order of popularity in the storage candidate content list, and is compared with the content extracted in descending order of popularity in the deletion candidate content list. There is. In addition, there is a combination in which contents having the highest popularity in the list of candidate storage contents are extracted in order, and compared with contents extracted in the order of popularity in the list of candidate deletion contents. By preparing a selection of various methods, a method suitable for the situation can be selected.

  A method for comparing the above contents will be described below. First, for each of the storage candidate content and the deletion candidate content to be compared, a future trend of popularity is predicted based on the popularity of the content acquired from the content popularity storage unit 13.

Here, there are the following three methods for predicting popularity transition.
The first is a linear approximation that assumes that the popularity is declining at a constant rate. The popularity decay rate is obtained from a statistical value of the content popularity transition rate or set by the operator of the content distribution system. Then, it is assumed that the popularity decreases at the decay rate from the current popularity of the storage candidate content and the deletion candidate content.

  Next, it is a prediction method that assumes that the degree of popularity decays according to a power law. It is known that the popularity distribution of content tends to follow a power law. For this reason, the current popularity distribution is approximated by a power law, and the prediction is made on the assumption that the popularity ranking gradually decreases on the popularity distribution.

  The third is a prediction method that assumes that the popularity will change randomly. In this method, the current popularity degree of each of the storage candidate content and the deletion candidate content is set as a starting value, and a result obtained by changing a value by assigning a random number in a simulation is assumed as a future popularity degree.

Then, based on the popularity transition predicted by the one method of prediction, a long-term cache hit rate improvement amount and a temporary cache hit rate decrease amount due to content replacement are calculated and calculated.
A long-term cache hit rate improvement value (merit) and a temporary cache hit rate decrease value (demerit) are defined by the following equations. The popularity transition of the storage candidate is f_in (t), the popularity transition of the deletion candidate is f_out (t), the occlusion time acquired from the occlusion / retention time storage means 14 is t_rep, and the residence time is t_exp.

  When the introduction content determination unit 11C determines that the popularity of the set of contents is close to each other, the predicted popularity after a predetermined time of the storage candidate content is larger. If it is determined, the storage candidate content is determined for introduction to the distributed base server, and the content replacement means 21 of the distributed base server 20 is instructed to replace the content.

  The introductory content determination unit 11C may determine whether to replace the content based on the following criteria. The introductory content determination unit 11C performs content replacement only for the combination of the storage candidate content and the deletion candidate content in which the long-term cache hit rate improvement amount is larger than the temporary cache hit rate decrease amount. Then, the content replacement unit 21 of the distributed base server 20 is instructed to replace the content.

  In response to the content replacement instruction from the content arrangement management device 10, the content replacement unit 21 requests the relevant content from the central site server 30 and replaces the content in the content storage unit 23.

  Further, when the content replacement unit 21 receives a content replacement instruction from the replacement content determination unit 11 of the content arrangement management device 10, the content replacement unit 21 stores the deletion candidate content and the storage candidate content to be replaced. Then, each time the deletion candidate content is closed, one storage candidate content is requested from the content transmitting / receiving means 31 of the central base server 30. And when the said storage candidate content is received from the said content transmission / reception means 31, the deletion candidate content which the said occlusion complete | finished stored in the content memory | storage part 23 is deleted, and the said storage candidate content is stored in the said content storage means 23 Write.

  The content transmitting / receiving unit 22 receives content from the central site server 30 and stores it in the content storage unit 23. Further, when receiving a content request from the user terminal 50, the content transmission / reception unit 22 reads the requested content from the content storage unit 23 and transmits it to the user terminal 50.

  On the other hand, when the content transmitting / receiving unit 31 on the central site server 30 side receives a content request from the content replacing unit 21 of the distributed site server 20, the content transmitting / receiving unit 31 reads the requested content from the content storage unit 32 and the content replacing unit 21. Send to. When a content request is received from the user terminal 50, the requested content is read from the content storage unit 32 and transmitted to the user terminal 50.

  The portal site 40 includes a content list creation unit 41 that creates a content list (hereinafter referred to as a content list), and a popularity level counting unit 42 that counts the popularity of each content.

  Among these, the content list creating means 41 is configured to store the content arrangement storage whenever the arrangement information stored in the content arrangement storage unit 12 of the content arrangement management device 1 is rewritten or whenever a request is made from the user terminal 5. The arrangement information of the currently stored content is acquired from the means 12. The placement information is content placement information currently stored in the distributed bases 2. The content list is transmitted to the user terminal 50 each time a content list is requested from the user terminal 50.

  Further, when the content list creation unit 41 receives a content request from the user terminal 50, the content list creation unit 41 refers to the information on the currently stored content, and the requested content is stored in the content storage unit 23 of the distributed base server 20. If it is stored, the content transmission / reception means 22 of the distributed base server 20 is notified as a content acquisition destination. If the requested content is not stored in the content storage means 23, the content transmission / reception means 31 of the central base server 30 is notified. Notify as the content acquisition destination.

  The popularity count means 42 updates the popularity of the requested content and writes it in the content popularity storage means 13 of the content arrangement management device 10.

  When the user terminal 50 requests a content list, the user terminal 50 acquires the content list from the portal site 40 and presents it to the user. Further, when a content request is received from a user, the location of the requested content is confirmed on the portal site 40, and the content transmission / reception means 22 of the distributed site server 20 or the central site server 30 determines the location of the requested content. The content transmission / reception means 31 is requested for the requested content. Then, when the requested content is received from the content transmitting / receiving unit 22 or the content transmitting / receiving unit 31, the requested content is presented to the user.

  The network 60 is a network constituting the Internet connecting the above-described configurations.

(Overall operation)
Next, the overall operation of the above embodiment will be described with reference to FIGS. Here, FIG. 2 is a flowchart showing operations from requesting content to distribution in the present embodiment disclosed in FIG.
First, a content request from a user and distribution of the requested content will be described.

  First, in the operating state of the entire system, the portal site 40 has information on contents currently stored in the content storage means 23 of the distributed base server 20 stored in the content arrangement storage means 12 of the content arrangement management apparatus 10. Each time is rewritten, or whenever there is a request from the user terminal 50, the information of the currently stored content is acquired from the content arrangement storage means 12 (step S101 / content information acquisition step).

  Next, when a content list request is input from the user, the user terminal 50 requests the content list from the portal site 40, and the portal site 40 receives the content list request and receives the content list. Transmit to the user terminal 50. When receiving the content list, the user terminal 50 presents the content list to the user (step S102 / content list presentation step).

  Here, when a content request is received from the user, the user terminal 50 confirms the location of the requested content with respect to the portal site 40. In response to this, the portal site 40 refers to the information of the currently stored content acquired in step S11. If the requested content is stored in the content storage unit 23 of the distributed site server 20, the distributed site The content transmission / reception means 22 of the server 20 is notified as a content acquisition destination, and when the requested content is not stored in the content storage means 23, the content transmission / reception means 31 of the central base server 30 is notified as the content acquisition destination. Then, the popularity of the requested content is updated and written in the content popularity storage means 13 of the content arrangement management device 10 (step S103 / content location confirmation step).

In this case, the user terminal 50 requests the requested content from the content transmitting / receiving unit 22 of the distributed site server 20 or the content transmitting / receiving unit 31 of the central site server 30 according to the location of the requested content. When the content transmission / reception unit 22 or the content transmission / reception unit 31 receives the content request, the content transmission / reception unit 22 or the content transmission / reception unit 31 reads the requested content from the content storage unit 23 or the content storage unit 32 and transmits it to the user terminal 50. The user terminal 50 that has received the content presents the requested content to the user when the requested content is received (step S104 / requested content presenting step).
Thereafter, the steps from S101 to S104 are repeated in response to a request from the user.

Next, an operation of updating the content arrangement in the content distribution system will be described based on the flowchart shown in FIG.
When the content distribution system is operated and the content replacement schedule is set in the content distribution schedule set by the operator or the like, the arrangement-popularity extraction unit 11A moves from the content arrangement storage unit 12 to the content storage unit 23 of the distributed base server 20. Information of the content currently stored in the content is acquired, and the popularity of the content is acquired from the content popularity storage means 13. Subsequently, the arrangement-popularity extraction unit 11A compares the ideally arranged content and the currently stored content, creates a list of storage candidate content and deletion candidate content, and transmits this to the replacement influence calculation unit 11B. (Step S201 / Storage candidate and deletion candidate content list creation step).

  When the storage candidate content list and the deletion candidate content list are received from the arrangement-popularity extraction unit 11A, the replacement influence calculation unit 11B further acquires the popularity of the content from the content popularity storage unit 13, and blocks or stays. The blockage time and residence time of each corresponding content stored in advance are acquired from the time storage means 14. Then, an arbitrary set of content combinations is extracted from the combination of the deletion candidate content and the storage candidate content. Subsequently, when the introduction content determination unit 11C determines that the popularity degrees of the set of contents are close to each other, the predicted popularity degree after a predetermined time of the predicted storage candidate content is greater. If it is determined that the content is large, the storage candidate content is determined for introduction into the distributed base server, and the content replacement means 21 of the distributed base server 20 is instructed to replace the content.

  Further, the replacement content determination unit 11 may determine whether to replace the content based on the following criteria. The long-term cache hit rate improvement amount and the temporary cache hit rate decrease amount are calculated as described above for the combination of the storage candidate content and the deletion candidate content, and these are compared to determine the long-term cache hit rate. It is determined that the content replacement is performed only for the combination of the storage candidate content and the deletion candidate content with the larger hit rate improvement amount, and the content replacement unit 21 of the distributed base server 20 is instructed to replace the content (step S202 / Replacement content determination process).

  In this case, when the content replacement unit 21 receives a content replacement instruction from the replacement content determination unit 11 of the content arrangement management apparatus 10, the content replacement unit 21 stores the deletion candidate content and the storage candidate content to be replaced. Then, the content transmission / reception means 22 is instructed to close the deletion candidate content (step S203 / replacement content blocking step).

When the deletion candidate content is blocked, the content replacement unit 21 requests one storage candidate content from the content transmission / reception unit 31 of the central base server 30. Subsequently, when the storage candidate content is received from the content transmission / reception unit 31, the content replacement unit 21 deletes the deletion candidate content in the blocked state stored in the content storage unit 23, and the storage candidate The content is written into the content storage means 23. In this way, every time the replacement of the deletion candidate content and the storage candidate content is completed, the content currently stored in the content storage unit 23 stored in the content arrangement storage unit 12 of the content arrangement management device 10 is stored. Information is replaced (step S204 / content replacement step).
Thereafter, the process returns to step S202, and the steps after step S202 are repeated for the remaining contents in the deletion candidate content list (step S205 / remaining content confirmation step).

(Effect of embodiment)
Since this embodiment is configured and functions as described above, according to this, since the content having high popularity is predicted and introduced in advance, the cache hit rate of the distributed base server is increased by the content replacement. It is possible to provide a content replacement system that has an excellent effect of effectively eliminating the possibility of lowering and further improving the cache hit rate.
Furthermore, by comparing the long-term cache hit rate improvement amount and temporary cache hit rate decrease amount due to content replacement, the content is replaced only when the long-term cache hit rate improvement amount is larger In addition, it is possible to eliminate the possibility that the cumulative cache hit rate will decline due to content replacement, and to further improve the cache hit rate.
As described above, the present embodiment has an excellent effect that the cache hit rate can be further improved by content replacement.

According to the present invention, in a hierarchical content distribution system having a distributed base server in the vicinity of a user, an arrangement management device that determines replacement of content in the distributed base server, and a distribution load applied to the distribution system when the content is replaced It can be applied to uses such as an estimation device.

  As described above, one embodiment of the content replacement system according to the present invention has been described in detail with respect to its configuration and operational effects. Here, since the important points in the configuration are summarized as follows, they are listed as appendices.

(Appendix 1)
Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server Content replacement means for deleting one content in the distributed base server and introducing one content in the central base server to the distributed base server,
A candidate for introducing a plurality of deletion candidate contents from among a plurality of contents recorded in the distributed base server and a plurality of contents stored in the central base server into the distributed base server. It has an arrangement-popularity extraction unit for extracting a plurality of storage candidate contents,
One arbitrary content is extracted from each of the extracted deletion candidate content and the storage candidate content to identify a set of contents, and the transition of the popularity of each of the combined contents is determined to the current degree of popularity. Establish a replacement effect calculation unit that predicts based on changes in
When the popularity of the set of contents approaches each other, the replacement impact calculation unit predicts the popularity of each content after a predetermined time, and the popularity of the storage candidate content is determined by the deletion candidate content. A content replacement system comprising: an introduction content determination unit that determines the storage candidate content for introduction into the distributed base server when the degree of popularity is greater.

(Appendix 2)
In the content replacement system described in Appendix 1,
The replacement impact calculation unit assumes a case in which the deletion candidate content in the set is deleted from the distributed base server and the storage candidate content is introduced into the distributed base server, and a long-term cache hit rate A cache hit rate calculation function for calculating an improvement amount and a temporary cache hit rate decrease amount based on the transition of the popularity;
An introduction content determination function in which the introduction content determination unit in the previous period determines a combination of the set of contents for introduction to the distributed base server when the cache hit rate improvement amount is larger than the cache hit rate decrease amount A content replacement system comprising:

(Appendix 3)
In the content replacement system described in Appendix 2,
When the arrangement-popularity extraction unit extracts a combination of a set of contents from the plurality of deletion candidate contents and the plurality of storage candidate contents, the storage candidate contents are listed in ascending order of popularity. A content replacement system comprising a content specifying function for specifying a set of content by combining the extracted content and the content extracted in descending order of popularity in the deletion candidate content list.

(Appendix 4)
In the content replacement system described in Appendix 2,
When the arrangement-popularity extraction unit extracts a set of content combinations from the plurality of deletion candidate contents and the plurality of storage candidate contents, the arrangement-popularity extraction unit is arranged in descending order of popularity in the storage candidate content list. A content replacement system comprising a content specifying function for specifying a set of content by combining the extracted content and the content extracted in descending order of popularity in the deletion candidate content list.

(Appendix 5)
In the content replacement system described in Appendix 2,
The replacement effect calculation unit sets f_in (t) as the popularity transition of the storage candidate, f_out (t) as the popularity transition of the deletion candidate, and t_rep indicates the time taken to block each content at the time of content replacement. , Where t_exp is the time from when the content is stored in the distributed base server until it is deleted, the long-term cache hit rate improvement amount (merit) and the temporary cache hit rate decrease amount (demerit) are as follows: A content replacement system that calculates and calculates based on a formula.

(Appendix 6)
In the content replacement system described in Appendix 5,
The content replacement system, wherein the replacement effect calculation unit predicts a transition of popularity between the storage candidate content and the deletion candidate content by linear approximation in which the popularity is attenuated at a constant rate.

(Appendix 7)
In the content replacement system described in Appendix 5,
The content replacement system, wherein the replacement effect calculation unit predicts a transition in popularity between the storage candidate content and the deletion candidate content, assuming that the popularity is attenuated according to a power law.

(Appendix 8)
In the content replacement system described in Appendix 5,
The content replacement system, wherein the replacement influence calculation unit predicts a transition of popularity between the storage candidate content and the deletion candidate content, assuming that the popularity changes randomly.

(Appendix 9)
Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server A content replacement system that deletes one content in the distributed base server and introduces one content in the central base server to the distributed base server,
A candidate for introducing a plurality of deletion candidate contents from a plurality of contents stored in the distributed base server and a plurality of contents stored in the central base server into the distributed base server. The placement-popularity extraction unit extracts a plurality of storage candidate contents,
One arbitrary content is extracted from each of the extracted deletion candidate content and the storage candidate content to identify a set of contents, and the transition of the popularity of each of the combined contents is determined to the current degree of popularity. The replacement impact calculation unit predicts based on changes in
When the popularity of the set of contents approaches each other, the replacement impact calculation unit predicts the popularity of each content after a predetermined time, and the popularity of the storage candidate content is determined by the deletion candidate content. When the storage candidate content is larger than the popularity, the introduction content determination unit determines the storage candidate content for introduction to the distributed base server,
The content replacement means deletes one content in the distributed base server based on the determination, and the content replacement means introduces the one content in the central base server to the distributed base server. Method.

(Appendix 10)
In the content replacement method described in Appendix 9,
When deleting the content in the distributed base server and introducing the content in the central base server to the distributed base server,
Assuming a case where the deletion candidate content in the set is deleted from the distributed base server and the storage candidate content is introduced to the distributed base server, a long-term cache hit rate improvement amount and a temporary cache hit are assumed. The rate decrease amount is calculated based on the popularity transition predicted by the replacement impact calculation unit,
The content, wherein when the cache hit rate improvement amount is larger than the cash hit rate decrease amount, the introduction content determination unit determines the combination of the set of contents for introduction into the distributed base server Replacement method.

(Appendix 11)
In the content replacement method described in Appendix 9,
When extracting a set of content combinations from the plurality of deletion candidate contents and the plurality of storage candidate contents, the arrangement-popularity extraction unit is in order of decreasing popularity in the list of storage candidate contents. A content replacement method, wherein a set of content is specified by combining the extracted content and the content extracted in descending order of popularity in the list of deletion candidate content.

(Appendix 12)
In the content replacement method described in Appendix 9,
When extracting a set of content combinations from the plurality of deletion candidate contents and the plurality of storage candidate contents, the arrangement-popularity extraction unit is in descending order of popularity in the list of storage candidate contents. A content replacement method, wherein a set of content is specified by combining the extracted content and the content extracted in descending order of popularity in the list of deletion candidate content.

(Appendix 13)
In the content replacement method described in appendix 10,
When the change influence calculation unit calculates the transition of popularity of the storage candidate, f_in (t) represents the transition of popularity of the storage candidate, f_out (t) represents the transition of popularity of the deletion candidate, and content When the time taken to block each content at the time of replacement is t_rep and the time from when the content is stored in the distributed base server until it is deleted is t_exp, the long-term cache hit rate improvement amount (merit) and temporary A content replacement method, wherein the replacement influence calculation unit calculates and calculates a demerit of a cache hit rate based on the following formula:

(Appendix 14)
In the content replacement method described in appendix 13,
When the replacement effect calculation unit calculates the transition of the popularity of the storage candidate, the replacement effect calculation unit causes the storage candidate content and the deletion to be deleted by linear approximation in which the popularity is attenuated at a constant speed. A content replacement method characterized by predicting the transition of popularity with candidate content.

(Appendix 15)
In the content replacement method described in appendix 13,
When the replacement influence calculation unit calculates the transition of the popularity of the storage candidates, the replacement influence calculation unit calculates whether the storage influence content and the deletion candidate content are A content replacement method characterized by predicting changes in popularity.

(Appendix 16)
In the content replacement method described in appendix 13,
When the replacement impact calculation unit calculates the transition of the popularity of the storage candidates, the replacement impact calculation unit determines that the replacement impact calculation unit performs the storage candidate content and the deletion candidate content. A content replacement method characterized by predicting changes in popularity.

(Appendix 17)
Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server A content replacement system that deletes one content in the distributed base server and introduces one content in the central base server to the distributed base server,
A candidate for introducing a plurality of deletion candidate contents from among a plurality of contents recorded in the distributed base server and a plurality of contents stored in the central base server into the distributed base server. A content extraction function for extracting a plurality of storage candidate contents,
One arbitrary content is extracted from each of the extracted deletion candidate content and the storage candidate content to identify a set of contents, and the transition of the popularity of each of the combined contents is determined to the current degree of popularity. Popularity trend prediction function to predict based on changes in
When the popularity of the set of contents approaches each other, the popularity of each content after a predetermined time is predicted, and the popularity of the storage candidate content is determined to be greater than the popularity of the deletion candidate content An introduction content determination function for determining the storage candidate content for introduction into the distributed base server when
And a content introduction function for deleting one content in the distributed base server based on the popularity of the content, and introducing one content in the central base server to the distributed base server,
A content replacement program characterized by being realized by a computer.

(Appendix 18)
In the content replacement program described in appendix 17,
When deleting the content in the distributed base server and introducing the content in the central base server to the distributed base server,
Assuming a case where the deletion candidate content in the set is deleted from the distributed base server and the storage candidate content is introduced to the distributed base server, a long-term cache hit rate improvement amount and a temporary cache hit are assumed. A popularity trend calculation function for calculating a rate decrease amount based on the predicted trend of popularity,
An introduction content determination function for determining a combination of the set of contents for introduction into the distributed base server when the cache hit rate improvement amount is larger than the cache hit rate decrease amount;
A content replacement program characterized by being realized by the computer.

(Appendix 19)
In the content replacement program described in appendix 17,
When extracting a combination of a set of contents from the plurality of deletion candidate contents and the plurality of storage candidate contents, the contents extracted from the list of storage candidate contents in order of popularity and the deletion candidate contents A content identification function that identifies a set of content by combining content extracted in descending order of popularity in the list of
A content replacement program characterized by being realized by the computer.

(Appendix 20)
In the content replacement program described in appendix 17,
When extracting a combination of a set of contents from the plurality of deletion candidate contents and the plurality of storage candidate contents, the contents extracted from the list of storage candidate contents in descending order of popularity and the deletion candidate contents A content identification function that identifies a set of content by combining content extracted in descending order of popularity in the list of
A content replacement program characterized by being realized by the computer.

(Appendix 21)
In the content replacement program described in appendix 18,
The popularity transition of the storage candidate is f_in (t), the popularity transition of the deletion candidate is f_out (t), the time taken to block each content at the time of content replacement is t_rep, and the content is stored in the distributed base server A function that sets the time from deletion to deletion to t_exp, a long-term cache hit rate improvement amount (merit), and a temporary cache hit rate decrease amount (demerit) based on the following formula A content replacement program that causes the computer to realize the function to be performed.

(Appendix 22)
In the content replacement program described in appendix 18,
A function of predicting the transition of popularity between the storage candidate content and the deletion candidate content by linear approximation that the popularity is attenuated at a constant speed,
A content replacement program characterized by being realized by the computer.

(Appendix 23)
In the content replacement program described in appendix 18,
A popularity degree transition prediction function that predicts a transition of popularity degree between the storage candidate content and the deletion candidate content, assuming that the popularity degree is attenuated according to a power law.
A content replacement program characterized by being realized by the computer.

(Appendix 24)
In the content replacement program described in appendix 18,
The popularity degree transition prediction function for predicting the transition of the popularity degree between the storage candidate content and the deletion candidate content, as the popularity degree changes randomly.
A content replacement program characterized by being realized by the computer.

DESCRIPTION OF SYMBOLS 10 Content arrangement | positioning management apparatus 11 Replacement content determination means 11A Arrangement-popularity extraction part 11B Replacement influence calculation part 11C Introduction content determination part 12 Content arrangement storage means 13 Content popularity storage means 14 Blocking / retention time storage means 20 Distributed base server 21 Content replacement means 22 Content transmission / reception means 23 Content storage means 30 Central base server 31 Content transmission / reception means 32 Content storage means 40 Portal site 41 Content list creation means 42 Popularity count means 50 User terminal 60 Network

Claims (9)

Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server Content replacement means for deleting one content in the distributed base server and introducing one content in the central base server to the distributed base server,
A candidate for introducing a plurality of deletion candidate contents from among a plurality of contents recorded in the distributed base server and a plurality of contents stored in the central base server into the distributed base server. It has an arrangement-popularity extraction unit for extracting a plurality of storage candidate contents,
One arbitrary content is extracted from each of the extracted deletion candidate content and the storage candidate content to identify a set of contents, and the transition of the popularity of each of the combined contents is determined to the current degree of popularity. Establish a replacement effect calculation unit that predicts based on changes in
When the popularity of the set of contents approaches each other, the replacement impact calculation unit predicts the popularity of each content after a predetermined time, and the popularity of the storage candidate content is determined by the deletion candidate content. A content replacement system comprising: an introduction content determination unit that determines the storage candidate content for introduction into the distributed base server when the degree of popularity is greater. The content replacement system according to claim 1,
The replacement impact calculation unit assumes a case in which the deletion candidate content in the set is deleted from the distributed base server and the storage candidate content is introduced into the distributed base server, and a long-term cache hit rate A cache hit rate calculation function for calculating an improvement amount and a temporary cache hit rate decrease amount based on the transition of the popularity;
An introduction content determination function in which the introduction content determination unit in the previous period determines a combination of the set of contents for introduction to the distributed base server when the cache hit rate improvement amount is larger than the cache hit rate decrease amount A content replacement system comprising: In the content replacement system according to claim 2,
The replacement effect calculation unit sets f_in (t) as the popularity transition of the storage candidate, f_out (t) as the popularity transition of the deletion candidate, and t_rep indicates the time taken to block each content at the time of content replacement. , Where t_exp is the time from when the content is stored in the distributed base server until it is deleted, the long-term cache hit rate improvement amount (merit) and the temporary cache hit rate decrease amount (demerit) are as follows: A content replacement system that calculates and calculates based on a formula.

Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server A content replacement system that deletes one content in the distributed base server and introduces one content in the central base server to the distributed base server,
A candidate for introducing a plurality of deletion candidate contents from a plurality of contents stored in the distributed base server and a plurality of contents stored in the central base server into the distributed base server. The placement-popularity extraction unit extracts a plurality of storage candidate contents,
One arbitrary content is extracted from each of the extracted deletion candidate content and the storage candidate content to identify a set of contents, and the transition of the popularity of each of the combined contents is determined to the current degree of popularity. The replacement impact calculation unit predicts based on changes in
When the popularity of the set of contents approaches each other, the replacement impact calculation unit predicts the popularity of each content after a predetermined time, and the popularity of the storage candidate content is determined by the deletion candidate content. When the storage candidate content is larger than the popularity, the introduction content determination unit determines the storage candidate content for introduction to the distributed base server,
The content replacement means deletes one content in the distributed base server based on the determination, and the content replacement means introduces the one content in the central base server to the distributed base server. Method. The content replacement method according to claim 4,
When deleting the content in the distributed base server and introducing the content in the central base server to the distributed base server,
Assuming a case where the deletion candidate content in the set is deleted from the distributed base server and the storage candidate content is introduced to the distributed base server, a long-term cache hit rate improvement amount and a temporary cache hit are assumed. The rate decrease amount is calculated based on the popularity transition predicted by the replacement impact calculation unit,
The content, wherein when the cache hit rate improvement amount is larger than the cash hit rate decrease amount, the introduction content determination unit determines the combination of the set of contents for introduction into the distributed base server Replacement method. The content replacement method according to claim 5,
When the change influence calculation unit calculates the transition of popularity of the storage candidate, f_in (t) represents the transition of popularity of the storage candidate, f_out (t) represents the transition of popularity of the deletion candidate, and content When the time taken to block each content at the time of replacement is t_rep and the time from when the content is stored in the distributed base server until it is deleted is t_exp, the long-term cache hit rate improvement amount (merit) and temporary A content replacement method, wherein the replacement influence calculation unit calculates and calculates a demerit of a cache hit rate based on the following formula:

Based on the popularity of the central base server possessing a plurality of contents, the distributed base server that transmits one or more contents distributed from the central base server to the user terminal, and the content of the central base server A content replacement system that deletes one content in the distributed base server and introduces one content in the central base server to the distributed base server,
A candidate for introducing a plurality of deletion candidate contents from among a plurality of contents recorded in the distributed base server and a plurality of contents stored in the central base server into the distributed base server. A content extraction function for extracting a plurality of storage candidate contents,
One arbitrary content is extracted from each of the extracted deletion candidate content and the storage candidate content to identify a set of contents, and the transition of the popularity of each of the combined contents is determined to the current degree of popularity. Popularity trend prediction function to predict based on changes in
When the popularity of the set of contents approaches each other, the popularity of each content after a predetermined time is predicted, and the popularity of the storage candidate content is determined to be greater than the popularity of the deletion candidate content An introduction content determination function for determining the storage candidate content for introduction into the distributed base server when
And a content introduction function for deleting one content in the distributed base server based on the popularity of the content, and introducing one content in the central base server to the distributed base server,
A content replacement program characterized by being realized by a computer. In the content replacement program according to claim 7,
When deleting the content in the distributed base server and introducing the content in the central base server to the distributed base server,
Assuming a case where the deletion candidate content in the set is deleted from the distributed base server and the storage candidate content is introduced to the distributed base server, a long-term cache hit rate improvement amount and a temporary cache hit are assumed. A popularity trend calculation function for calculating a rate decrease amount based on the predicted trend of popularity,
An introduction content determination function for determining a combination of the set of contents for introduction into the distributed base server when the cache hit rate improvement amount is larger than the cache hit rate decrease amount;
A content replacement program characterized by being realized by the computer. In the content replacement program according to claim 8,
The popularity transition of the storage candidate is f_in (t), the popularity transition of the deletion candidate is f_out (t), the time taken to block each content at the time of content replacement is t_rep, and the content is stored in the distributed base server A function that sets the time from deletion to deletion to t_exp, a long-term cache hit rate improvement amount (merit), and a temporary cache hit rate decrease amount (demerit) based on the following formula A content replacement program that causes the computer to realize the function to be performed.

Images