JP4527627B2 - Content distribution apparatus, content distribution network, and content distribution method - Google Patents

Description

  The present invention relates to a content distribution technique, and more particularly to a content distribution apparatus that distributes a large-capacity multimedia content file used in, for example, a digital cinema to a plurality of distribution destinations in time for the distribution date.

  Multicast technology is a method for efficiently distributing the same content to a plurality of distribution destinations.

  In IP multicast, which is one of the multicast technologies, IP packets are duplicated by a router, and multiple receiving terminals receive packets addressed to a predetermined IP address at the same time. Can be transmitted. [Non-Patent Document 1] and [Non-Patent Document 2] outline IP multicast technology and multicast setting protocol.

  On the other hand, in IP multicast, a reliable transmission protocol such as TCP that has been used in conventional unicast communication cannot be used as it is. New transmission protocols including corrections need to be implemented. [Non-Patent Document 3] outlines a protocol for realizing reliable data distribution on a multicast network, and [Non-Patent Document 4] describes a technique for extending TCP for multicast. is doing.

  In IP multicast, receiving terminals other than the delivery destination can also receive packets freely. To ensure that only a specific delivery destination can receive data, it is necessary to encrypt the data and send it to users who have the right to receive the data. Security measures such as encryption key distribution, encryption key update, and user and data authentication processing are essential. [Non-Patent Document 5] outlines a technique relating to security in multicast, particularly, a data encryption key management method.

  As mentioned above, IP multicast is one of the most efficient methods for realizing multicast, but there are problems in terms of reliability and security. Furthermore, implementation of multicast function in router and multicast setting Due to the difficulty of management, it has not yet been widely used for content distribution.

  As another method for realizing multicast, there is a method for realizing multicast at the application level. In application-level multicast, normally, each distribution destination has a function for caching, copying, and redistributing data, and the distribution destination is made to function as a relay node to form a distribution tree. In order to manage the distribution tree and the distribution destination, it is necessary to manage the distribution route and the distribution destination information. This method includes a method for centrally managing information in one or a plurality of directory servers prepared in advance. Various methods have been proposed, such as a method using hierarchical processing or distributed processing. Further, in order to exchange management information, it is common to construct a control network in addition to the data distribution network. [Non-patent document 6] and “Non-patent document 7” outline application-level multicast technology.

  In application-level multicast, compared to unicast communication, the data transmission bandwidth from the distribution center can be greatly reduced, while the network load on the subscriber system connected to the distribution destination increases, and the distribution route becomes redundant. There is a problem that the delivery delay increases. In particular, when the file size of content is very large as in a digital cinema, it is difficult to redistribute data files from a distribution destination to a plurality of distribution destinations through a subscriber network.

  One method of realizing application-level multicast is to configure a multicast distribution network between mirror servers that perform data caching. [Non-Patent Document 8] shows an example. In addition, this document proposes an on-the-fly transfer method that transfers data to the next leaf node without waiting for the entire file to be received as a technique for solving the increase in delivery delay, which is a problem in application-level multicast. ing. The method of this document is advantageous in terms of stability and optimization of the distribution schedule because the distribution network is configured by a dedicated server, but it is necessary to maintain and manage a plurality of mirror servers in the network. In addition, the on-the-fly transfer method has a drawback that the recovery process becomes complicated if the server goes down or the line is disconnected during the transfer because the data being transferred is further transferred to another mirror server.

  Many of the multicast technologies described above are targeted at distributing data to an unspecified number of receiving terminals, such as thousands to tens of thousands, and a relatively small number of fixed distribution destinations such as data distribution in digital cinema. It is not necessarily optimal when distributing data according to a predetermined schedule. In addition, it is necessary to implement and set up multicast functions in routers and distribution destination application programs, and to maintain and manage mirror servers and the like provided in the distribution network.

  Furthermore, in general multicast technology, it is assumed that the transmission bandwidth of the network to which the distribution server is connected is limited, but as shown in [Non-patent Document 9], as the optical multiplexing technology advances, the backbone network Therefore, when the distribution center is connected to the backbone network, the transmission band of the transmission path on the distribution center side does not always become a bottleneck.

  On the other hand, in content distribution such as digital cinema, it is necessary to finish distributing the content in time for the screening date. For this reason, the contents need to be transmitted in an optimal order in consideration of the delivery date. [Non-Patent Document 10] proposes Deadline Credit Algorithm (DC) as a data transfer scheduling method in real-time video transmission. DC distributes a plurality of video streams, and assigns a priority based on the distribution delay and discard rate of each stream, thereby controlling the delay amount and discard rate of each data stream to be the same value. is there.

[Non-Patent Document 11] describes a scheduling technique for transmitting content requested by a user at an optimal timing when content is distributed to a mirror server arranged in a network using satellite communication. . However, in a content distribution device that transfers a plurality of contents via a distribution network constituted by a plurality of transmission paths having different transmission bands, the distribution order is set as efficiently as possible and in time without causing network congestion. Regarding the scheduling method, the optimal technique is not known.
“Internetworking Technology Handbook Chapter43: Internet Protocol Multicast”, CISCO SYSTEMS Internetworking Handbook. (Http://www.cisco.com/univercd/cc/td/doc/cisintwk/ito_doc/ipmulti.htm#xtocidl) MARIA RAMALHO, “INTRA- AND INTER-DOMAIN MULTICAST ROUTING PROTOCOLS: A SURVEY AND TAXONOMY”, IEEE Communications Surveys & Tutorials First Quarter 2000, vol.3 no.1. (Http://www.comsoc.org/livepubs/surveys /public/1q00issue/ramalho.html) Brian Neil Levine, JJ Garcia-Luna-Aceves, “A comparison of reliable multicast protocols”, Multimedia Systems 6: pp.334-348 (1998). Sam Liang, David Cheriton, “TCP-SMO: Extending TCP to Support Medium-Scale Multicast Applications”, IEEE Infocom, June 2002. (http://www-dsg.stanford.edu/papers/tcp-rtm/tcp-smo -liang-infocom02.pdf) Paul Judge and Mostafa Ammar, “Security Issues and Solutions in Multicast Content Distribution: A Survey”, IEEE Network January / February 2003. (http://www.cs.virginia.edu/~jaz2n/papers/mcastsurvey.pdf) Fereydoun Hormozdiari, “A Survey on Application Layer Video Multicasting”, CMPT 880 Selected Topics on Multimedia and Networking, Spring 2005. (http://www.cs.sfu.ca/~jcliu/cmpt880/Report/ALM_Revised.pdf) Zhi Li and Yongjoo Shin, “Survey of Overlay Multicast Technology”, 2nd June 2002. (http://networks.cs.ucdavis.edu/~lizhi/papers/ECS289.pdf) Junya Akiba, Hirofumi Abe, “Scalable Content Distribution Technology”, NTT Technical Journal pp.16-20, 2003.8. Masayuki Murata, Kenichi Kitayama, Hideo Miyahara, “Possibility of eliminating bottlenecks in the Internet with 1000-wavelength WDM,” IEICE Technical Report (PNI2000-6), May2000. Adamantia Alexandraki, Michael Paterakis, “Performance Evaluation of the Deadline Credit Scheduling Algorithm for Soft-Real-Time Applications in Distributed Video-on-Demand Systems”. Cluster Computing Journal Volume 8 Number1 p.61-75, January 2005. Reuven Cohen, Liran Katzir and Danny Raz, “Scheduling Algorithms for a Cache Pre-Filling Content Distribution Network”, Infocom'2002, New-York, June 2002.

  As mentioned above, conventional multicast technology is not suitable for distributing multiple different contents with multiple schedules with high reliability, such as digital cinema distribution. There was a problem.

  The present invention solves the above-mentioned problems of the prior art, and in a content distribution network that distributes a large-capacity content file to a plurality of distribution destinations by unicast communication, the content file is high-speed while maintaining high reliability by the delivery date. It is intended to optimally control the distribution order of content files so as not to cause network congestion at the time of distribution.

  In order to solve the above-mentioned problems, the present invention is characterized in that a plurality of unicast communications are performed simultaneously and scheduling of each unicast communication is performed. In addition to the data delivery date, the data delivery order is determined in consideration of the transmission bandwidth and the use bandwidth in the network.

  Specifically, the present invention provides a content distribution network including a distribution device that transmits content, a plurality of distribution destination devices that receive the content, and a network that connects them. A storage server configured to store data loaded from the data server and transmit the data to a plurality of distribution destination devices, and a distribution scheduler that performs distribution scheduling according to a distribution date, Based on at least the size of data to be transmitted to each distribution destination device, the throughput actually measured in the transmission device, or both of the information, and the distribution route information, distribution to the distribution destination device using the transmission device Means for selecting data are provided.

  Alternatively, the distribution scheduler loads the transmission device based on at least the size of data to be transmitted to each distribution destination device, the throughput actually measured in the transmission device, or both of the information, and the distribution route information. Means for controlling a data loading order and a transmission order of data transmitted from the transmitting apparatus.

In particular, according to the present invention, the distribution scheduler includes a two-stage scheduling unit including a first scheduling unit and a second scheduling unit . Before SL first scheduling means, by evaluating the difference between the reduced amount of network load due to increased volume and unloaded data of network load due to the loaded data to the transmitting apparatus, a distribution object data to said transmission device Evaluating the effect of reducing the network load that occurs when loading, selecting data to be loaded into the transmission device according to the evaluation result, the second scheduling means, among the untransmitted data loaded into the transmission device, Data having a sufficient transmission path usage rate with respect to the delivery destination device and having the greatest effect of reducing the network load evaluated for the loaded untransmitted data is selected as transmission target data.

  As described above, the distribution scheduler is configured, and in selecting the data to be loaded into the transmission device, scheduling is performed on the evaluation scale having a longer time scale using the data size to be transmitted to each distribution destination device, and the data is transmitted from the transmission device. In selecting the data to be transmitted, it is possible to further optimize the scheduling of the data to be distributed by performing the scheduling that emphasizes the load fluctuation of the network with a shorter time scale using the throughput actually measured from the transmitting device. .

  In addition, when the content file size exceeds the memory capacity of the transmitting device, the data server divides the content data file to be equal to or smaller than the memory size of the transmitting device, The distribution destination apparatus includes means for adding information for recombining them. Thus, a content file having a size exceeding the memory capacity of the transmission device is divided by the data server. For this reason, it is possible to transmit a file having a size exceeding the memory capacity of the transmission apparatus.

  In addition, when the same data file has a plurality of delivery dates, the delivery scheduler transmits a single data file with a plurality of delivery schedules. Therefore, a plurality of data IDs determined according to the delivery dates are assigned to the same data file. A means for adding is provided. Thereby, it is possible to simplify the control of managing the schedule for each data ID and distributing the same content to a plurality of distribution destinations with different schedules.

  In addition, the distribution scheduler detects congestion occurring in the network by monitoring the discard rate or delay amount of the data packet during transmission of the data file in the transmission device, and further uses the distribution route information. A means is provided for scheduling data transmission so as to avoid congestion by specifying a transmission path in which congestion occurs and setting a transmission band used for content distribution on the transmission path to be low. This makes it possible to perform highly reliable data distribution control that avoids high-use transmission lines.

  The distribution scheduler detects a disconnected transmission path by using distribution path information when a network disconnection is detected during transmission of a data file by the transmission device, and the transmission path is used as a distribution path. Means for temporarily deleting the data transfer to the delivery destination device included in the data management information of the data to be delivered. As a result, for a transmission line that is considered to be disconnected, the distribution destination that includes the transmission line in the distribution path is temporarily removed from the distribution target until it can be detected that the transmission line becomes usable again. The delivery schedule can be avoided.

  The distribution scheduler includes means for selecting a route with the lowest transmission path usage rate from the plurality of distribution routes when there are a plurality of distribution routes to the distribution destination device in the network. As a result, it is possible to perform distribution using an optimum distribution route among a plurality of distribution routes.

  According to the present invention, it is possible to efficiently distribute data via a network so as to meet the delivery date, and it is possible to deliver content having an important delivery date with high reliability.

  In particular, the present invention uses a two-stage delivery scheduling based on an evaluation measure according to the length of the time scale, that is, scheduling of the load order of data to the transmitter and scheduling of the order of transmission of data from the transmitter. It is possible to optimize the scheduling of data to be distributed.

  That is, the present invention can transfer the data loaded from the data server to the memory of the transmission device to a plurality of distribution destination devices in the distribution device installed in the distribution center. Since the input / output load can be reduced and data is transferred between memories, the input / output speed of a storage device such as a magnetic disk does not limit the transmission speed. Also, the distribution scheduler selects the data to be distributed to the distribution destination device based on the size of the data that flows through the transfer of content on each transmission path in the network, and controls the distribution destination of the data so as to avoid the high usage rate transmission path Therefore, the possibility of causing congestion in the delivery route can be reduced. In addition, the distribution scheduler selects the data to be distributed to the destination device based on the measured throughput in each transmission path in the network, so that the error rate increases in the transmission path when there is a failure in the network. Even in such a case, it is possible to realize highly reliable data distribution.

  A content distribution network according to an embodiment of the present invention includes a distribution device that transmits content, a plurality of distribution destination devices that receive the content (hereinafter, also simply referred to as “distribution destination”), and a network that connects them. The distribution device includes a data server for loading the data file from the storage device to the transmission device, a plurality of transmission devices for transmitting the loaded data file to the network in parallel, and the loading and transmission order of the data file. A distribution scheduler is provided for optimization. The distribution destination is provided with a receiving device for receiving data from the network and a receiving side data server for saving data from the receiving device to the storage device.

  In order to optimize the distribution order of data files, the distribution scheduler is provided with means for holding distribution route information, transmission band information, and data management information. Distribution route information consists of a list of all distribution destinations and information about distribution routes to each distribution destination, and transmission bandwidth information consists of a list of all transmission paths in the network and information about the transmission bandwidth of each transmission path. The data management information is used to determine the list of combinations of data files and their delivery destinations, the file size for each item, the delivery date, and whether the file is loaded on the sending device or not sent. Consists of distribution status information.

  In order to parallelize the distribution of the content file and realize reliable data transfer, the data server of the distribution device includes means for dividing the content file into a plurality of data files having a predetermined size or less in the reception data server. , Means for recombining the divided data files and restoring the original content file, and means for transmitting and receiving the divided data files by unicast communication are provided in the transmitting device and the receiving device. Further, in order to speed up data transfer, the transmission device and the reception device are provided with a memory capable of holding a plurality of data files, and the data file between the storage device, the transmission device, and the reception device. Means for executing the transfer process in parallel with the transmission / reception process is provided. In order to further increase the parallelism of data transmission, the transmission apparatus is provided with means for executing a plurality of data file transfers in parallel in one transmission apparatus.

  Further, the transmission device is provided with means for actually measuring the transmission throughput of the data file being transmitted, and the distribution scheduler is in the distribution path of the transmission destination of each transmission device from the transmission throughput reported from each transmission device. Calculate the data bandwidth flowing through all transmission paths, and further divide that value by the transmission bandwidth of each transmission path to calculate the transmission path usage rate for each transmission path, and use that value as the transmission band information. Calculate the channel usage rate for each delivery destination by registering the value in the delivery route information and the maximum value of the usage rate of the transmission path included in the delivery route for each delivery destination. Means are provided. By calculating the transmission line usage rate, it is possible to determine the transmission path that is under load due to data transfer, and the distribution destination that includes the transmission path in the distribution path, and avoid using transmission lines with high usage rates. The delivery schedule can be optimized.

  In order to calculate the amount of network load generated by sending a data file, the distribution scheduler provides the distribution route for each registered item to the distribution destination for a plurality of registered items selected from the data management information. A means for calculating the total amount of data that flows through each transmission line until distribution of all selected registered items is completed by accumulating the size of the data file for all transmission lines, and the value for each transmission. A means for calculating the load on each transmission path by dividing by the transmission band of the path, and calculating the load on the entire network by evaluating the maximum value of the load on all transmission paths in the network, and the distribution of each distribution destination Means are provided for calculating the network load for each distribution destination by evaluating the maximum value of the transmission path load included in the path. Since the calculated load corresponds to the time required to complete the transmission of all data files, it is necessary to complete the transmission of the selected data file by using the network load to each transmission path and distribution destination. The distribution order can be scheduled so as to minimize the distribution time.

  In order to manage the data management information registration items and the data files loaded in the transmission device, the distribution scheduler displays the registration items that were being transmitted by the transmission device when the data file transfer was completed in the transmission device. A means for deleting the data file from the memory of the transmitting device is provided when the data file is deleted from the data management information and the transmission of the data file to all the delivery destinations is completed.

  In addition, in order to select a data file to be newly transmitted and a distribution destination, the distribution scheduler selects all registered items whose distribution status is already loaded from the data management information, and selects a network for the selected item. The load calculation function calculates the network load for each delivery destination. Among all the registered items loaded in the transmission device, the delivery date is early, the transmission line usage rate to the delivery destination is low, and the network for the delivery destination Means having a function of selecting a registration item having a large load as a new data transfer target is provided. By this means, a delivery destination having a low transmission path usage rate is selected with priority, so that the network is less likely to be congested. Furthermore, priority is given to data transmission to a delivery destination with a large network load, that is, a delivery destination that takes the longest time for transmission. As a result, the time required to finish delivering the data file can be shortened.

  In addition, when there is a free space to load a new data file in the memory (data file buffer) of the transmission device, the distribution scheduler first calculates the network load for all loaded data files by the network load calculation function. The network load for all unloaded data files is calculated, and for each unloaded data file, it is assumed that the same data file has been loaded on the sending device. A means for calculating the amount of increase in network load for data files and the amount of decrease in network load for unloaded data files, and among all unloaded data files, the delivery date is early and the loaded data file Network load for The amount is small, it means having the function of reducing the amount of network load for the data file unloaded to select a large data file as a load object are provided. By this means, it is possible to load a data file having a delivery destination that has an earlier delivery date and does not impose a network load on the transmission device.

  The network in the content distribution network according to the present invention can be realized by using an IP network and a TCP protocol.

  In order to set the data distribution parallelism and scheduling time unit to an appropriate size, means for setting the size of data to be divided from 100 Mbytes to 1 Gbytes is provided.

  Further, in order to distribute the same content to a plurality of distribution destinations with different schedules, the data server is provided with means for adding a plurality of data IDs with different distribution dates to the divided data files.

  Furthermore, in order to detect congestion in the network, a means for detecting congestion in the network by monitoring the discard rate and the delay amount of data packets at the time of data transmission at the transmitting device further provides a plurality of distribution destinations. If congestion is detected, means to identify the transmission path where congestion has occurred by using the transmission band information to the delivery destination, and reduce the transmission path bandwidth of the transmission path where congestion has occurred. By setting, means for scheduling the delivery order so as to avoid transmission of the data file using the transmission path in which congestion occurs is provided, and means for notifying the delivery device manager of the occurrence information of the congestion Is provided. By this means, the delivery order is scheduled so as to avoid the transmission path in which congestion has occurred, and the administrator can take measures against the congestion.

  In addition, in order to verify the connectivity to the delivery destination, a means for sending a status packet for requesting the return of a response packet to each delivery destination, and if the response packet does not return from the delivery destination, the data management information Means for canceling the delivery status of all items destined for the delivery destination registered in, and temporarily removing the delivery destination from the delivery target, and means for issuing a warning to the administrator of the delivery apparatus.

  In addition, since the distribution scheduler verifies the possibility of distribution by the distribution date of each content, it is assumed that no new items are registered in the data management information and that no congestion or the like occurs during data transmission. In addition, there is a means to perform numerical simulation of data file transfer, and if the end date and time of each data file exceeds the delivery date set in the file, delivery by the due date is impossible. Is provided for issuing a warning to notify the administrator of the distribution apparatus.

  In order to handle cases where distribution devices are connected to multiple networks, the distribution scheduler divides distribution devices into groups for each connected network, and stores distribution route information, transmission bandwidth information, and data management information for each network. And a means for optimizing the loading of data to the distribution apparatus and the selection of transmission data for a plurality of networks.

  To cope with the case where there are a plurality of delivery routes to a delivery destination in the network, when selecting a means for registering a plurality of items having different delivery routes in delivery route information for a delivery destination and a new transmission data file There are provided means for selecting the route with the lowest transmission path usage rate from among a plurality of delivery routes, and means for selecting a delivery route that reduces the size of the network load when a new file to be loaded into the transmission device is selected. It is done.

  Furthermore, in order to perform data transfer with higher reliability, when a content file is divided in the data server, means for adding an error detection and correction code to the divided file and error detection and correction processing in the receiving device are performed. Means for notifying the transmission device of the result, and means for re-sending the data file by not updating the distribution status of the data management information when an error is notified. Provided.

  FIG. 1 shows the overall configuration of a content distribution network according to an embodiment of the present invention. The content distribution network includes a distribution device 1, a network 2, and a plurality of distribution destination devices 3 installed in a distribution center. The distribution device 1 includes a distribution scheduler 14, a data server 15, a storage device 16, and a plurality of transmission devices 17. Consists of.

  The content data file stored in the storage device 16 such as a magnetic disk is loaded from the storage device 16 to the transmission device 17 by the data server 15, and the transmission device 17 distributes the loaded data file via the network 2. Transmit to the destination device 3. Here, transmission of the data file from the transmission device 17 to the distribution destination device 3 is performed by one-to-one unicast communication. In addition, the distribution scheduler 14 uses the distribution path information 11, the transmission band information 12, and the data management information 13 stored in the storage device to load the data file into the transmission device 17, and the transmission from the transmission device 17 to the network 2. Control the order.

  FIG. 2 shows a configuration example of the transmission apparatus. In the configuration example of FIG. 2, the transmission device 17 has two network interfaces 171 and 172 for receiving a data file from the data server 15 and for transmitting the data file to the network, and stores the data file. The main memory 174 includes a CPU 173 that performs data transmission / reception processing, data file management, transmission throughput monitoring, and the like. The main memory 174 is divided into a plurality of data file buffer areas, and a plurality of data files can be stored in the transmission device 17.

  In the transmission device 17, since the data file is loaded on the main memory 174, data transmission can be performed at a speed higher than that in the case of transmission while reading from the disk. The data file read in 174 can be transmitted to a plurality of delivery destination apparatuses 3 without being loaded again. For this reason, it is possible to reduce the load of the data server. Since the content file having a size exceeding the memory area of the transmission device 17 is divided by the data server 15, a file having a size exceeding the size of the recording area of the main storage memory 174 of the transmission device 17 is also transmitted. It is possible.

  FIG. 3 shows a configuration example of distribution route information and transmission band information managed by the distribution scheduler. FIG. 3A shows the topology of the content distribution network in this configuration example. As shown in FIG. 3A, distribution destination IDs such as A, B, C,... Are assigned to the distribution destination device 3, and transmissions such as P1, P2, P3,. It is assumed that a road ID is given.

  As shown in FIG. 3B, the distribution path information 11 is composed of a distribution destination ID indicating the distribution destination apparatus 3 and a list of transmission path IDs indicating transmission paths in the distribution path. As shown in FIG. 3C, the transmission band information 12 is composed of a transmission path ID in the network 2 and information on the transmission band of each transmission path. The transmission bandwidth of the transmission path is the bandwidth of the physical transmission path, the bandwidth of the transmission path allocated to the network by a network device such as a router, or the usable bandwidth of the transmission path that is dynamically predicted based on network congestion information. Determined by. Information on the transmission path usage rate calculated for each transmission path can be added to the transmission band information 12.

  FIG. 4 shows a configuration example of the data management information. The data management information 13 of this configuration example is configured by a list of a combination of a data file ID (referred to as a data ID) and a delivery destination ID, and holds a delivery date, a data file size, and delivery status information. The distribution status information indicates whether the data file is loaded in the transmission device 17, which transmission device 17 is loaded, and whether the file is already being transmitted.

  FIG. 5 shows a procedure for loading the data file to the transmission apparatus. First, by referring to the data management information 13, it is searched whether there is a data file that is not loaded in the transmission device 17 (step S1). If there is an unloaded data file, it is next determined whether or not the memory of the transmission device 17 is free (step S2). If the memory of the transmission device 17 is available, the unloaded data file with the earliest delivery date is selected as the load target (step S3). Next, the network load reduction effect obtained when the data file is loaded into the transmission device 17 is evaluated for each of the selected data files (step S4). Next, the data file having the greatest effect of reducing the network load is selected as a load target to the transmission device 17 (step S5), and the load process is started (step S6). At the same time, the data management information is changed from unloaded to loaded (step S7).

  FIG. 6 shows a procedure for evaluating the effect of reducing the network load. There are two network loads: network load for loaded data and network load for unloaded data.

  First, the network load in the current state is calculated (step S10). Next, it is assumed that the data to be loaded has been loaded into the transmission device 17 (step S11), and the network load on the loaded data and the network load on the unloaded data are re-evaluated (step S12). Subsequently, the increase amount of the network load with respect to the loaded data and the decrease amount of the network load with respect to the unloaded data are evaluated (step S13), and the reduction effect of the network load is evaluated by taking the difference, for example (step S14). ). The reduction effect is evaluated for all data files to be loaded, and the maximum data is selected to determine the data to be loaded (step S15).

  FIG. 7 shows a network load evaluation procedure. First, the data size and the delivery destination ID are referred to from the data management information 13 for the evaluation target data file (step S20), and the total data transmission amount to each delivery destination is calculated (step S21). Next, the total amount of data flowing in each transmission path is evaluated by using the distribution path information 11 (step S22), and the total bandwidth is divided by the transmission band of the transmission path by using the transmission band information 12. Is calculated (step S23). The transmission line load is considered to be an estimated value of the transmission time required to finish transmitting the data file to be evaluated to all the distribution destination devices 3.

  Next, the network load is calculated from the maximum value or weighted average of all the transmission path loads in the network (steps S23 and S24). The weighted average is evaluated in such a way that the weighting for a large transmission line load is increased. This is because a transmission line with a high load becomes a bottleneck when transmitting a data file.

  FIG. 8 shows an example of network load evaluation. First, the earliest delivery date data is selected as an evaluation target. In this example, three data files with data IDs 1, 2, and 3 are selected. Each data file has two distribution destinations. Next, the transmission data amount of the transmission path is calculated by using the distribution path information 11. For example, since the transmission path P2 is a distribution path for the distribution destination devices 3A and 3B, the data amount can be calculated from the cumulative value of the data file sizes for the distribution destinations whose distribution destination IDs are A and B.

  Next, the transmission line load is calculated by dividing the transmission data amount of each transmission line by the transmission band. In this example, it can be seen that the P7 transmission line has the highest load. When the network load is evaluated by the maximum value, it is 15. For example, when the network load is calculated by the average of the square values, 8.04, and when it is calculated by the average of the cube values, 27. 05. The evaluation based on the maximum value focuses only on the transmission line with the maximum load. When the square value or the cube value is used, the weight of the transmission line with a large load is weighted. The load is being calculated.

  FIG. 9 shows a transmission procedure of the data file from the transmission apparatus. First, since there is generally a restriction on the number of distribution destination devices 3 that can be simultaneously transmitted from one transmission device 17, it is determined whether or not the number of data currently being transmitted is equal to or less than a specified value (step S30). . Next, it is determined whether there is a data file that can be transmitted in the memory of the transmission device 17 (step S31).

  If there is transmission data, the transmission path usage rate of the distribution path to each distribution destination is calculated based on the measured transmission throughput of the data file being transmitted (step S32). The evaluation of the transmission line usage rate is repeated when data transfer is newly started, when data transfer is completed, or when it is estimated that there has been a change in the network status, such as congestion. S33). If there is no data file that can be sent that includes an empty delivery destination in the delivery path, data transmission is not performed until the transmission path usage rate decreases.

  If there are multiple data files and distribution destinations with sufficient transmission line usage, calculate the network load value based on the loaded unsent data using the network load calculation method described above, and send the data file and distribution target. The amount of network load reduction that occurs when it is assumed that data transmission has started is evaluated (step S35), and the data file and distribution destination with the largest amount of reduction are selected as transmission targets (step S36). Simultaneously with starting the data file transmission (step S37), the data management information 13 is changed from being loaded to being transmitted (step S38).

  FIG. 10 shows a processing procedure at the end of data file transmission. When the transmission of the data file is completed (step S40), first, the data file whose transmission has been completed and the registration item of the distribution destination ID are deleted from the data management information 13 (step S41). Further, the data management information 13 is searched for the data ID of the data file that has been transmitted. If there is no registration, that is, if the transmission of the data file to all distribution destinations has been completed (step S42). The data file is deleted from the memory of the transmission device 17 (step S43).

  FIG. 11 shows a procedure for evaluating the transmission line usage rate. First, in each transmission device 17, the transmission throughput is evaluated for the data file being transmitted. The distribution scheduler 14 refers to the distribution path information 11, obtains a distribution path from the distribution destination ID of the data file, and accumulates transmission throughput values along the distribution path (step S50), thereby flowing through each transmission path. The data transfer band is calculated (step S51), and the value is divided by the transmission band obtained from the transmission band information 12 to determine the transmission line usage rate (step S52). Next, the transmission path usage rate for each distribution destination is calculated by evaluating the maximum value of the bandwidth used for the transmission path in the distribution path for each distribution destination (step S53).

  FIG. 12 shows an example of calculating the transmission line usage rate. In this example, distribution destination apparatuses 3 having distribution destination IDs A to E as distribution paths are connected by transmission paths with transmission path IDs P1 to P8. Among these, the data file is currently transmitted to the distribution destination devices 3 having distribution destination IDs A, C, and D, and the transmission throughputs are 180 Mbps, 90 Mbps, and 80 Mbps, respectively. The bandwidth used for the transmission path is obtained by accumulating the transmission throughput value for the transmission path along the distribution path. The transmission line usage rate is obtained by dividing the transmission line usage band by the transmission line transmission band.

  Finally, the transmission path usage rate to each distribution destination apparatus 3 is obtained by calculating the maximum value of the usage rate of the transmission path on the distribution path. In the example, most of the transmission band is used for the A, C, and D distribution destination apparatuses 3 that are transmitting, and the B and E distribution destination apparatuses 3 have a free transmission band. I know that there is. Even in the case of a distribution destination device 3 that has not transmitted data, if the distribution route is shared with the distribution destination device 3 that is transmitting data, there is a possibility that the transmission line usage rate will increase.

  FIG. 13 is a diagram for explaining a method of transmitting a large content file in the embodiment of the present invention. Since the content file 4 having a size exceeding the memory area (data file buffer) of the transmission device 17 cannot be transmitted as it is, the content file 4 is divided in the data server 15 so as to be equal to or smaller than the memory size of the transmission device 17. Each data file 41 obtained by dividing the content file 4 is added with the ID and sequence number of the content file 4 as an ID, and the original content file 4 ′ is referenced by referring to the sequence number in the distribution destination device 3 that is a receiving device. Restore.

  FIG. 14 is a diagram for explaining a management method for data files having different delivery dates in the embodiment of the present invention. In the data management information 13 of FIG. 14A, the distribution destination ID and the distribution date of the data file with the data ID = 1 are displayed. Here, the delivery date for delivery destinations with delivery destination IDs A, B, and C and the delivery due date for delivery destinations with delivery destination IDs D, E, and G are largely different, so the delivery schedule is managed separately. There is a need to. In the data management information 13 in FIG. 14B, the data IDs for the distribution destinations D, E, and G are changed to 2, and different distribution schedules can be applied to the same data file. It becomes possible. Of course, the correspondence information between the data ID and the actual data file in the storage device 16 is managed by the data server 15, and the correspondence information is updated as the data ID is changed.

  FIG. 15 is a diagram for explaining an example of a management method for network congestion information and network reachability information in the embodiment of the present invention. If the network bandwidth guarantee is not sufficient, there is a possibility that congestion will occur in the delivery route. In this content distribution network, when a congestion or transmission impossible state is detected, a transmission path in which congestion or transmission path disconnection is found from the distribution path information 11, and the distribution schedule is changed based on the information.

  In this example, congestion is detected for the distribution destination devices 3 with the distribution destination IDs A and B, and transmission to the distribution destination device 3 with the distribution destination ID E becomes impossible. The distribution route information 11 makes it possible to determine that congestion has occurred in the transmission path P2 sharing the route among the distribution destination devices 3A, 3B, and 3F, and data cannot be transmitted to the distribution destination device 3E. From the above, it can be estimated that the line is disconnected on the transmission line P8.

  As a means for avoiding congestion, it is possible to reduce the setting of the transmission band of the transmission line in which the congestion has occurred, thereby increasing the evaluation of the used band of the transmission line. For transmission lines that are considered to have been disconnected, take the method of temporarily removing the distribution destination that includes the transmission line from the distribution target until it can be detected that the transmission line is available again. Can do.

  FIG. 16 is a diagram for explaining a distribution route management and a transmission path usage rate evaluation method when there are a plurality of distribution routes for one distribution destination. In this example, there are a distribution path that passes through the transmission path P2 and a distribution path that passes through the transmission path P9 for the distribution destination apparatuses 3 having distribution destination IDs A and B. The respective transmission path usage rates are 40% and 50%. In this case, the transmission path usage band for the distribution destination IDs A and B is evaluated not for the transmission path P9 but for the transmission path P2. Therefore, the transmission path usage rate with the distribution destination ID A is 40% of P2, which has the highest usage rate among the transmission paths P1, P2, and P3, and the transmission path usage rate with the distribution destination ID B is the transmission path P1. , P2 and P4, the usage rate is 60% of the largest P4.

  As described above, the content distribution network according to the present embodiment includes the distribution center distribution device 1 that transmits the content, the plurality of distribution destination devices 3 that receive the content, and the network 2 that connects them. The distribution device 1 includes a data server 15 that stores data files in the storage device 16, a plurality of transmission devices 17 that transmit data files to the network 2, and a distribution scheduler 14. The loaded data file is stored in an internal memory and transmitted to a plurality of distribution destination devices 3.

  The distribution scheduler 14 includes distribution route information 11 to the distribution destination device 3, transmission band information 12 of each transmission line in the network, the size of the data file to be distributed, distribution destination, distribution date, and the loading status to the transmission device 17. Data management information 13 is held and managed.

  Further, the distribution scheduler 14 determines the amount of data that flows through each transmission path from the size of the data file and the distribution path information 11 until the transmission of all the selected data files is completed, and the transmission divided by the transmission band. The transmission device 17 transmits the load from the distribution route information 11 and the transmission band information 12 by calculating the load and calculating the weighted average or maximum value of the transmission load of all the transmission paths in the network. It has a function of accumulating the measured transmission throughput of the data file along the distribution path and calculating the usage rate of the transmission band of the distribution path to each distribution destination device 3. There is room, the delivery date is early, and the network load reduction effect that is evaluated for unsent data files that have already been loaded is the most effective. It has a function of determining the transmission order of data files in order from the large distribution destination device 3, and the increase amount of the network load evaluated for all the data files loaded in the transmission device 17 and the unloaded all data files By evaluating the difference in the amount of decrease in network load that is evaluated, the effect of reducing the network load that occurs when a data file is loaded is evaluated. Has a function of determining the order of loading to the transmission device 17 in order.

  The present invention is used when efficiently transmitting a large-capacity multimedia content file used in a digital cinema or the like to a plurality of distribution destinations in an efficient and highly reliable manner in time for the distribution date.

It is a figure which shows the whole structure of the content delivery network based on the Example of this invention. It is a figure which shows the structural example of a transmitter. It is a figure which shows the structural example of the delivery path | route information and transmission band information managed by a delivery scheduler. It is a figure which shows the structural example of data management information. It is a figure which shows the procedure of loading to the transmitter of a data file. It is a figure which shows the evaluation procedure of the reduction effect of a network load. It is a figure which shows the evaluation procedure of a network load. It is a figure which shows the example of evaluation of network load. It is a figure which shows the transmission procedure from the transmission apparatus of a data file. It is a figure which shows the process sequence at the time of completion | finish of transmission of a data file. It is a figure which shows the evaluation procedure of a transmission-line usage rate. It is a figure which shows the example of calculation of a transmission-line usage rate. It is a figure which shows the example of a division | segmentation of a content file. It is a figure which shows the structural example of the data management information from which a delivery date differs. It is a figure explaining the example of the management method of network congestion information and network reachability information. It is a figure explaining the management method of a delivery route when there are a plurality of delivery routes, and the evaluation method of a transmission way utilization factor.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Distribution apparatus 2 Network 3 Distribution destination apparatus 11 Distribution route information 12 Transmission band information 13 Data management information 14 Distribution scheduler 15 Data server 16 Storage apparatus 17 Transmission apparatus

Claims (8)

In a distribution device that transmits content to multiple distribution destination devices via a network,
A data server for storing content;
A transmission device that accumulates data loaded from the data server and transmits the data to a plurality of distribution destination devices;
A distribution scheduler that schedules distribution according to the distribution date,
The distribution scheduler,
Reduction of network load caused when data to be distributed is loaded on the transmission device by evaluating the difference between the increase amount of the network load due to the loaded data to the transmission device and the decrease amount of the network load due to unloaded data First scheduling means for evaluating an effect and selecting data to be loaded into the transmission device according to the evaluation result;
Among the unsent data already loaded to the transmission device, there is a margin in the transmission path usage rate for the delivery destination device, and the data with the greatest effect of reducing the network load evaluated for the loaded unsent data, A content distribution apparatus comprising: second scheduling means for selecting data to be transmitted . When the content file size exceeds the memory capacity of the transmitting device, the data server divides the data file of the content to be equal to or smaller than the memory size of the transmitting device, and distributes the divided data file to each divided data file. The content distribution apparatus according to claim 1, further comprising means for adding information for recombining them in the destination apparatus. When the same data file has a plurality of delivery dates, the delivery scheduler adds a plurality of data IDs determined according to the delivery dates to the same data file in order to send one data file with a plurality of delivery schedules. The content distribution apparatus according to claim 1, further comprising: means. The distribution scheduler detects congestion occurring in the network by monitoring a discard rate or a delay amount of data packets during transmission of a data file in the transmission device, and further uses the distribution route information to detect congestion. 2. A means for scheduling data transmission so as to avoid congestion by specifying a transmission path that occurs and setting a transmission band to be used for content distribution on the transmission path to a low level. The content distribution apparatus according to claim 3 . The distribution scheduler detects a disconnected transmission path by using distribution path information when a network disconnection is detected during transmission of a data file by the transmission device, and includes the transmission path in the distribution path. The content distribution device according to any one of claims 1 to 4, further comprising means for temporarily deleting data transfer to a distribution destination device from data management information of data to be distributed. . The delivery scheduler comprises means for selecting a route with the lowest transmission path usage rate from a plurality of delivery routes when there are a plurality of delivery routes to a delivery destination device in the network. Item 6. The content distribution device according to any one of Item 5 to Item 5 . In a content distribution network composed of a distribution device that transmits content, a plurality of distribution destination devices that receive content, and a network that connects them,
The distribution device includes a data server that stores content, a transmission device that stores data loaded from the data server and transmits the data to a plurality of distribution destination devices, and a distribution scheduler that schedules distribution according to distribution dates And
The distribution scheduler,
Reduction of network load caused when data to be distributed is loaded on the transmission device by evaluating the difference between the increase amount of the network load due to the loaded data to the transmission device and the decrease amount of the network load due to unloaded data First scheduling means for evaluating an effect and selecting data to be loaded into the transmission device according to the evaluation result;
Of the unsent data loaded to the transmitting device, there is a margin in the transmission line usage rate for the delivery destination device, and the data with the greatest effect of reducing the network load evaluated for the loaded unsent data is A content distribution network comprising: second scheduling means for selecting data to be transmitted . In a content distribution method in a content distribution network including a distribution device that transmits content, a plurality of distribution destination devices that receive content, and a network that connects them,
The distribution device includes a data server that stores content, a transmission device that stores data loaded from the data server and transmits the data to a plurality of distribution destination devices, and a distribution scheduler that schedules distribution according to distribution dates And
The distribution scheduler evaluates the difference between the increase amount of the network load due to the loaded data to the transmission device and the decrease amount of the network load due to the unloaded data by the first scheduling means, so that the distribution target data is Evaluating the effect of reducing the network load that occurs when loading to the transmitter, selecting data to be loaded to the transmitter according to the evaluation results,
The data server loads the selected data into the appropriate transmitting device;
The delivery scheduler has a margin in the transmission line usage rate for the delivery destination device among the unsent data already loaded to the transmission device by the second scheduling means, and is evaluated for the loaded unsent data. Select the data that has the greatest effect of reducing the network load as the transmission target data,
The content distribution method, wherein the transmission device transmits the selected data to the corresponding distribution destination device.

Images