KR101381199B1 - Method and System for Content Delivery and Caching - Google Patents

Abstract

The present invention is a method and system for delivering content using a chunk distribution algorithm. In the present invention, in a content delivery system for delivering content, in a method for delivering content to a host via a network including one or more entities, delivering the content from the entity storing the content to the host, And distributing the content to another entity according to the popularity of the content, thereby effectively delivering the content.

Description

Content delivery and caching method and system for same {Method and System for Content Delivery and Caching}

The present invention is a method and system for delivering content, and more particularly, a method and system for delivering and distributing (or caching) content according to the popularity of the content.

Today, as content requests explode, it is increasingly important to serve content effectively. In order to efficiently provide such a service, a CDN (Content Delivery Network) technique or P2P (Peer to Peer) technique has been suggested as a representative solution.

However, in the case of the P2P scheme, a large amount of traffic between Internet service providers (hereinafter, referred to as 'ISP') may occur, and content validity and download performance may become unstable. In addition, there is a problem in that network resources are inefficiently used, such as a duplicate connection is generated to download the same kind of content. On the other hand, in the case of the CDN scheme, the service cost is inevitably high because several servers must be used, and there is a problem of low performance in terms of traffic engineering. Accordingly, there is a need for an improved technology that can efficiently use network resources and increase user convenience by efficiently placing content.

To meet this need, by utilizing the storage contained within a network entity (e.g., a router) constituting the network while in the network, a copy of the content can be placed on a plurality of entities, or a content request can be made. An information-centric network technique has been proposed, which locates entities close to end hosts.

Referring to FIG. 1 in relation to an information-centric network, a content server 310 that stores content is connected to a host (or network between hosts) 320 via a network 200, and the network 200 is It contains a network entity (eg, a router) that contains or attaches storage. In an information centric network technique, host 320 may obtain a copy of the content from storage included or attached to the network entity, effectively obtaining the content within a short time.

That is, the information-centric network technique may use a storage that already exists in the network 200 to place a copy of the content closer to the host, thereby speeding up content delivery and increasing content validity.

However, the information-centric network scheme proposed so far does not propose a specific method of caching, considering that it can cache content to a network entity (or storage included or attached to the network entity). In the network, there is a need for a method of effectively caching content in a network.

In this regard, information-centric network schemes vary in overall performance, such as storage hit rates and storage utilization, depending on the caching and distribution schemes. In order to cache content, existing techniques for distributing the content may assume unusual topologies such as, for example, trees or hierarchies, assume cooperation between caches, or know the content request pattern in advance. According to this assumption, there is a problem that a constraint such as having to follow a predetermined topology or grasping a content request pattern in advance is applied, and a lot of overhead occurs.

Therefore, there is a need for a method of effectively caching content and increasing storage utilization in a network, that is, a method of effectively delivering and caching / distributing content.

An object of the present invention is to effectively deliver content.

Moreover, an object of this invention is to cache / distribute content effectively.

In addition, the present invention aims to reduce the delay caused by content delivery, to reduce the amount of traffic between ISPs, and to balance the load between links.

In addition, the present invention aims to increase cache hit rate and reduce cache replacement by quickly distributing highly popular content. Furthermore, an object of the present invention is to increase the cache utilization rate as a whole and reduce the overhead associated with cache management.

In addition, the present invention aims to increase the content delivery speed and increase the content validity with a simple algorithm without having to know the content access pattern. Furthermore, an object of the present invention is to deliver content quickly with fewer resources.

It is also an object of the present invention to efficiently deliver content in accordance with any routing technique.

In addition, an object of the present invention is to allow each network entity to independently determine the object to be cached and the object to be cached.

According to a feature of the present invention for achieving the above object, in a content delivery system for delivering content, in a method for delivering content to a host via a network including one or more entities, the entity for storing the content Delivering the content from the host to the host, and distributing the content to another entity, depending on the popularity of the content.

The present invention also provides a method for delivering content to a host through a network including one or more entities in a content delivery system that delivers content consisting of one or more content chunks, wherein a portion of the content chunks is provided. Forwarding a content request to a first entity that stores a second, in response to the content request, forwarding some of the stored content chunks from the first entity to the host, and from the first entity to a second entity Distributing a portion of the stored content chunks, wherein the first entity is a content server storing the content or an entity located on a path for delivering the content request from the host to the content server. And wherein the second entity is from the host to the first entity An entity located on the line that transmits the contents request.

The present invention also provides a method for delivering content from an entity to a host via a network including the entity in an entity storing one or more content chunks included in the content, via the network, Receiving a request for at least some of the stored content chunks, and in response to the request, delivering at least some of the stored content chunks to the host.

In addition, the present invention provides a method for delivering the content from the first entity to the host through a network including the first entity in a first entity storing one or more content chunks constituting content. Receiving a content request over the network, and delivering and distributing the content based on a chunk distribution algorithm.

The present invention also provides a content delivery system including a host for requesting content, a content server for storing the content, and a network connecting the content server and the host, wherein the network is a content capable of storing the content. Comprising an entity including a storage unit, the entity further comprises a content delivery unit for delivering the stored content to the host, and a content distribution unit for distributing the content to other entities according to the popularity of the content.

According to such a method and system for delivering content according to the present invention, the following effects are obtained.

The present invention can effectively deliver content and effectively cache / distribute content.

In addition, the present invention can reduce the delay caused by content delivery, reduce the amount of traffic between ISPs, and balance the load between links.

In addition, the present invention can increase the cache hit rate and reduce the cache replacement by quickly distributing the highly popular content. As a result, the present invention can increase cache utilization overall and reduce overhead associated with cache management.

In addition, the present invention can increase content delivery speed and increase content validity with a simple algorithm without having to know content access patterns. Thus, the present invention can deliver content quickly with fewer resources.

In addition, the present invention may operate in accordance with any routing technique.

And, the present invention allows each network entity to independently determine what to cache and what to replace. Therefore, the present invention can effectively deliver content even when there is no central server.

1 is a diagram illustrating a state in which content is delivered from a content server to a host.
2A illustrates a content delivery system for delivering content to a host.
2B is a block diagram showing the configuration of a network entity included in a content delivery system.
3 is a flow chart illustrating the operation of a network entity for caching content according to popularity in accordance with one preferred embodiment of the present invention.
4 is a diagram illustrating a content caching according to popularity according to an embodiment of the present invention.
5 is a flowchart illustrating the operation of an entity that has received content chunk i, in accordance with one preferred embodiment of the present invention.
6A to 6C illustrate window states and chunk windows in accordance with one preferred embodiment of the present invention.
7A to 7D are diagrams illustrating how content is delivered and distributed to network entities, in accordance with a preferred embodiment of the present invention.
8 to 14 is a view showing the performance according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. To fully disclose the scope of the invention to those skilled in the art, and the invention is only defined by the scope of the claims.

Like reference numerals refer to like elements throughout the specification.

In a preferred embodiment of the present invention, one content file to be delivered may be divided into a plurality of chunks and delivered. That is, for example, if one content file is divided into 10 chunks, the host may receive 10 chunks for the content request. It is apparent to those skilled in the art how to divide one content file into chunks. For convenience, in FIGS. 2A to 7D, it is assumed that the content requested by the host is divided into seven chunks, and an identification number is assigned to each chunk.

In an information-centric network, content files are distributed among entities located within and constituting the network (e.g., storage or storage attached to or included in a router, or storage attached or included to a content server), and the entity distributes the content files to itself. Cache the distributed content file and deliver the cached file to the host as requested by the host.

According to the present invention, whether to distribute (or cache) content chunks is determined by the popularity of the content. That is, the more content requests, the more popular content (request and popularity are proportional), and the more popular content is more distributed (or cached) to multiple entities in the network relative to the less popular content. This can increase content delivery and storage utilization.

Hereinafter, referring to FIGS. 2A to 7D, a method of distributing (or caching) content in accordance with the popularity of the content will be described.

2A illustrates a content delivery system 300 that delivers content in response to a content request from a host. The content delivery system 300 includes a content server 310, a network entity (or a network of network entities; 200), and a host (or a network between hosts; 320), and may communicate with each other wirelessly or by wire.

In addition, each of the network entities is configured as shown in FIG. 2B. For example, any network entity 200 included in the network includes an entity operation controller 211, a distributed algorithm performer 212, and a content store 215. The entity operation control unit 211 performs all operations related to the operation of the entity. For example, when the entity is a router, the entity operation control unit 211 includes all of the components necessary to perform the operation performed in the general router and to perform the operation. Take control of the entity. Meanwhile, the content storage unit 215 may cache content (or content chunk) as the above-described storage, and in particular, may cache content (or content chunk) according to the popularity of the content. The distributed algorithm performing unit 212 performs a chunk distribution algorithm described below. The distributed algorithm performing unit 212 may include a content delivery unit 213 for delivering content stored by an entity to another entity, and a content distribution unit 214 for distributing the content to another entity, in order to perform a chunk distribution algorithm. Include.

3 is a flow chart illustrating operation of a network entity caching content in accordance with popularity of the content.

3 illustrates an operation of caching the content according to the popularity of the content in one network entity (eg, 250 of FIG. 4A). In the present invention, as described above, the more content requests, the more popular content can be viewed (request and popularity proportional). Similarly, as more content is delivered to the host (in response to a content request), the content may be viewed as a highly popular content. Based on this, it is determined that the more popular the content is, the more popular the content is. . That is, the network entity determines that the content has a high popularity in proportion to the number of times of passing through the content, and increases the number of content chunks (included in the content) that are cached as the popularity increases. After receiving the content request from the host, instead of forwarding the request to the content server, the content chunk that was cached can be delivered.

The chunks to be cached among the chunks constituting the content are sequentially determined according to the identification number (i) of the chunk. For example, referring to FIG. 4A, if chunks are cached, they are cached in the order of chunk 1, chunk 2, ... chunk 7.

At this time, the number of chunks to be cached must be determined, and the number of chunks to be cached is determined by the chunk window. In other words, a chunk window is a window that distinguishes chunks to be cached.

The size of this chunk window increases exponentially as the number of times of passage of the content increases (that is, the more popular content). When the content is requested for the first time, the chunk window size is set to be small to prevent the less popular content from being cached, and when the content is determined to be highly popular according to the continuous request, the content having the high popularity can be cached. Exponentially increases the chunk window size. The chunk window size is increased by a multiplier of the dispersion index, which may be set by the user or the system. In addition, the multiplier value is initially set to 0, and increases by one if it is determined that the content is popular. In the present invention, a multiplier value is used as the window state identification number to identify the plurality of window states.

When a network entity receives a content chunk, the content chunk caching operation according to popularity begins (since receiving the content chunk means that the network entity is not caching the content chunk). As shown in FIG. 3, the network entity caches the chunk included in the chunk window among the received content chunks (S111). Thereafter, the window state and the identification number of the last cached chunk are updated (s121), and when the content chunk is received again in the future, the content chunk that needs to be cached can be identified. After this series of processes, the network entity delivers the content chunks received to the host (s131).

4A-4C illustrate a content delivery system 301 that delivers content according to a content request from a host and caches content according to popularity, according to the flow chart of FIG. 3.

The content delivery system 301 consists of a content server 311, a network entity 240, 250, and a network 321 between hosts, and can communicate with each other wirelessly or by wire. Assume that the entities included in network entity 240 are a generic router or a router that includes storage, and another network entity 250 is, for example, an access point or access router that includes storage. Also assume that the variance index of the chunk window is two.

When a host included in a network between hosts requests content to the content server 311, the content request passes through the network entity 250, and delivery of the content according to the request also passes through the network entity 250. In addition, when another host included in the network between the hosts requests the same content, the content request and the content delivery for the request also pass through the network entity 250.

That is, when hosts included in the host network 321 request the same content several times, the network entity 250 receives the same content several times. The network entity 250 may determine that the popularity of the content is higher as it receives the same content several times, and may cache chunks included in the content.

More specifically, with reference to FIGS. 4A-4C, host 321 requests content, and content is delivered to host 321 via network entity 250 in accordance with the request (401). In this process, content is delivered to entity 321 via network entity 250, which caches content chunks 411 as it receives the content. At this time, since the window state (multiplier value) is 0, 1 (= 2 ⁰ ) chunks, that is, chunk 1 are cached. The window state is then updated to 1, and the last cached chunk is updated to chunk 1.

Then, when the host 321 requests the content again, the chunk 1 that was cached in the network entity 250 is passed to the host 321 (421), and the remaining chunks (that is, chunks 2 through 7) are the content. It is forwarded from the server 311 (431). In the process, the remaining chunks (that is, chunks 2 through 7) are again passed through the network entity 250, and the network entity 250 includes two chunks included in the window state 1, that is, chunks 2 through chunks. Cache 3 (441). In addition, the window state is updated to 2, and the last cached chunk is updated to 3.

As a result, network entity 250 is caching chunks 1 through 3, and once again the same content is requested by host 321, chunks 1 through chunk 3 cached by network entity 250 become host 321. Is passed to (451). Of course, chunks 4 through 7 are obtained from content server 311 (461). Similarly, four chunks, that is, content chunks 4 through 7, are included in window state 2 and therefore cached to network entity 250 (471). Host 321 may then obtain content (Chunks 1 through 7) from network entity 250 (481).

As such, if content delivery occurs multiple times through network entity 250, network entity 250 determines that the content is popular and caches the content, resulting in any location located within host network 321. The host can quickly obtain the content from the access point 250 without having to obtain the content from the content server 311.

As such, one network entity may determine the popularity of the content itself and cache the content according to the popularity.

On the other hand, when a plurality of network entities cooperate with each other to cache content, the plurality of network entities cooperate to determine the popularity of the content and cache the content. More specifically, using the chunk distribution algorithm, content is delivered and distributed from the content server to the host. In this regard it will be described later in conjunction with Figures 5-7d.

FIG. 7A illustrates a content delivery system 300 that delivers content according to a content request of a host. The content delivery system 300 is comprised of a content server 310, a network entity 210, 220, 230, and a host (or a network between hosts; 320), and can communicate with each other wirelessly or by wire.

7A to 7D, for convenience, it is assumed that the content requested by the host is divided into seven chunks, and an identification number is assigned to each chunk, and the network entity is a router including storage. In addition, only some network entities 210, 220, 230 are described as an example among network entities included in the network of network entities (200 of FIG. 1), wherein the network entities 210, 220, 230 are provided from a host. To a content server, these entities are located on the path to which the host's content request travels (or the path from which the content is delivered from the content server). The content stored in the content server 310 is delivered to the host and distributed to the network by a request to the host 320 according to the chunk distribution algorithm described below.

Chunks distributed algorithm (chunk diffusion algorithm)

1. Dispersion Method

The chunks to be distributed among the chunks constituting the content are sequentially determined according to the identification number (i) of the chunk. For example, referring to FIG. 7A, if the chunks are distributed, they are distributed in the order of chunk 1, chunk 2, ... chunk 7. At this time, the number of chunks to be distributed must be determined, and the number of chunks to be distributed is determined by the chunk window. That is, the chunk window is a window that distinguishes the chunks to be distributed. For example, when there are several chunks in FIG. 6A, the chunk window distinguishes the chunks to be distributed.

The size of this chunk window increases exponentially as the number of requests for content increases (ie, the more popular content). When the content is requested for the first time, a small chunk window size is set to prevent the distribution of low-popular contents, and when the content is determined to be highly popular according to continuous requests, the content having high popularity may be widely distributed. To increase the chunk window size exponentially. The chunk window size is a multiplier of the dispersion index.

Dispersion index (positive integer; hereinafter referred to as 'x') determines the rate at which the chunk is distributed in the network, which may be set by the user or the system. In the following description, the dispersion index is set to 2 for convenience.

The multiplier value (hereinafter referred to as 'n') can increase the chunk window size exponentially, where the multiplier value is initially set to 0, and whenever the content is requested and delivered (that is, the content is popular) Increase the multiplier by 1 to increase the chunk window size. Hereinafter, a multiplier value is used as the identification number of the chunk window.

On the other hand, if the content requested by the host 320 is a content server first requested, it is obvious that the content is not stored in the entities 210, 220, and 230. Thus, the host 320 obtains the content from the content server. In this case, the content server distributes a part of the content to the entity network 200 in consideration that the content was once requested content. Thereafter, when the host requests the same content, the host may obtain the distributed chunk from the network 200 without transmitting the request to the content server. This content distribution technique is described in more detail with reference to FIG. 5.

FIG. 5 illustrates a process in which an entity receiving content chunk i delivers and distributes content chunk i. For convenience, assume that the entity that received the content chunk i is the entity 220 of FIG. 7A.

(즉, 분산된 청크들을 커버하는 청크 윈도우들의 사이즈 총합) 중 작은 값 (둘 다 동일한 값이라면 상기 값) 으로 업데이트된다. 예를 들어, 도 6a 에서 도시된 바와 같이, 윈도우 상태가 1 이고, 마지막으로 분산된 청크가 1 이면, 컨텐츠 청크 2 내지 3 을 전달한 이후, 엔티티는 청크 윈도우 2 내에 포함되는 청크들, 즉, 컨텐츠 청크 2 내지 3 을 분산시킨다. 그 이후 청크 윈도우 상태 및 마지막으로 분산된 청크가 업데이트되는데, 도 6b에서 도시된 바와 같이, 청크 윈도우 상태는 2 (=1+1) 으로 업데이트되며, 마지막으로 분산된 청크의 식별번호는, 마지막으로 캐싱된 청크의 식별번호 3 과, 2ⁱ(i=0~1) 의 총합인 3 (=2⁰+2¹) 중 작은 값인 3 으로 업데이트된다.When the entity receives the request for the content chunk i, it is determined whether the requested chunk i is being cached (s110). If the entity does not cache the requested chunk i, the next entity to which the content request is delivered (see FIG. 7A). In step 210, the chunk request is forwarded. If the requested chunk i is being cached, the requested chunk i is transferred to the host (s120). If the content chunk i delivered to the host is the last cached chunk (eg, the chunk with the largest identification number among the cached chunks) among the cached chunks (s130) among the cached chunks, it is included in the chunk window. And chunks that are not yet distributed are distributed to the entity that delivered the request (s140). Then update the distribution results. That is, the identification number of the last distributed chunk is updated, and the chunk window state is updated (s150). At this time, the chunk window state n is updated by increasing by 1, and the identification number of the last distributed chunk is the identification number of the last cached chunk,

(I.e., the sum of the sizes of the chunk windows covering the distributed chunks) is updated to the smaller of the values (if both are the same value above). For example, as shown in FIG. 6A, if the window state is 1 and the last distributed chunk is 1, after delivering content chunks 2 to 3, the entity has chunks included in chunk window 2, i.e. content Chunks 2-3 are dispersed. The chunk window state and the last distributed chunk are then updated, as shown in FIG. 6B, the chunk window state is updated to 2 (= 1 + 1), and the identification number of the last distributed chunk is, finally, The cached chunk is updated to the smaller of 3, and the smaller of 3 (= 2 ⁰ +2 ¹ ), which is the sum of 2 ⁱ (i = 0-1).

If the delivered chunk i is not the last chunk of cached chunks, it is determined whether the delivered chunk i is a previously distributed chunk (s160), and if the result is not the distributed chunk, the process ends.

로 변경되며 (s170) 마지막으로 분산된 청크의 식별번호는 청크 식별번호 i 에서 1 을 감한 값으로 변경된다 (s180). 예를 들어, 도 6c 에서 도시된 바와 같이, 요청된 청크 i 가 3 이라면, 윈도우 상태 n 은 1 (=

) 로 업데이트되며, 분산된 청크의 식별번호는 2 (=3-1) 로 업데이트된다.However, if the delivered chunk i was a chunk that was previously distributed, this means that chunk i is not cached in the entity that forwarded the request (230 in FIG. 7A). That is, since chunk i determined to be distributed actually means that it is not distributed to entity 230, it is necessary to update window state n so that chunk i can be distributed again, and finally update the distributed chunk. . In this case, the window state n, using the dispersion index x,

The identification number of the last distributed chunk is changed to (s170) and is subtracted from the chunk identification number i by 1 (s180). For example, as shown in FIG. 6C, if the requested chunk i is 3, the window state n is 1 (=

) And the identifier of the distributed chunk is updated to 2 (= 3-1).

Briefly expressed as a pseudo code as described above is as follows.

As described in the pseudo code above, for each content, the entity only needs the chunk window state n, and the last distributed chunk identification number sent, ie two counters. In other words, the chunk distribution algorithm can distribute the chunks using a small number of counters, thus consuming less resources.

2. Replacement target

When a new content chunk arrives with storage, for example, with a full cache, the chunk that was originally stored in the cache must be evicted to store the new content chunk. In this case, in the chunk distribution algorithm, the cache may be replaced by using various techniques. Most suitably, for example, LRU (least recently used) or LRFU (least recently / frequently used) may be used.

Maintaining all access histories of content chunks can cause a lot of overhead and waste resources. In the present invention, the access history is stored in units of content, and the chunk to be replaced is determined.

3. Caching Location

If there is a plurality of storages, for example a plurality of caches (i.e. there are a plurality of entities with the storage), it is necessary to select a cache for which the content chunks will be cached. In this case, the chunk distribution algorithm distributes the content chunks to the entity located in the direction in which the content request is received in consideration of the spatial location so that the chunk distribution is cached. For example, network entity 220 distributes and chunks of content chunks to entity 230 located in the direction in which the content request was received.

It is also possible to cache the content chunks to the entity that delivered the content request, but among the entities located in the direction in which the request was delivered, one-hop diffusion, multi-hop diffusion, for example The content of the selected entity according to one of the following schemes, i.e., half the number of hops remaining to the host), ISP relationship aware diffusion (e.g., spreading chunks to the boundaries of the ISP). You can cache chunks. Preferably, content chunks are distributed to selected entities in a hop-by-hop fashion to fully utilize storage in the network.

According to the chunk distribution algorithm described above, it can be seen that the network entity can independently determine what to cache and what to replace itself.

7A-7D are diagrams in accordance with one embodiment of the present invention relating to a method for distributing content to network entities. For convenience, it follows the assumption described with respect to FIG. 7A and follows the chunk distribution algorithm described above.

First, when the content server 310 receives a request for the first time, the host 320 has no choice but to acquire the content from the content server 310. That is, as shown in FIG. 7A, when the content server 310 receives a request for content from the host 310 through the path of the network entity 210, 220, 230, the content server 320 receives the content. Send the entire file (ie, seven chunks) to host 310 (400).

Then, if the previously requested content is requested again, as shown in FIG. 7B, the content server 310 places the chunk 1 510 of the content on the network entity 210 located on the path where the content request came in and in proximity to the content server. (410). Subsequently, if a request for the same content is received from the host 320, chunk 1 is passed from the network entity 210 to the host 320 and the remaining chunks other than chunk 1 are received from the content server 310. It is forwarded to the host 320 (430).

Then, as shown in FIG. 7C, network entity 210 distributes chunk 1 to network entity 220 (440), and content server 310 distributes chunk 2 and chunk 3 to network entity 210. (450). As a result, network entity 210 is storing chunks 1 through 3, and network entity 220 is storing chunk 1.

Again, if a content request is detected from the host 320, the request is forwarded to the content server, with chunk 1 460 from the network entity 220 located on the delivery path, chunk 2 and chunk from the network entity 210. 3 passes to the host (461), and the remaining chunks (Chunks 4-7) are passed from the content server 310 (462).

Then, as shown in FIG. 7D, likewise, a chunk distribution operation is performed, in which network entity 220 distributes chunk 1 to network entity 230, and network entity 210 chunks. 2 and chunk 3 are distributed to network entity 220 (480), and content server 310 distributes chunks 4 through 7 to network entity 210 (490).

Because of this, if the host 320 requests the same content again, the host 320 may obtain the content from entities located on the path of the host 320 and the content server 310. That is, chunk 1 can be obtained from the network entity 230, chunks 2 and 3 from the network entity 220, and chunks 4 through 7 can be obtained from the network entity 210. Through this, the host 320 may acquire the content in a faster time than obtaining the content from the content server 310.

In the following, the performance of the present invention and other techniques are compared. Other techniques here include client-server techniques, AllCache techniques (content delivered to the host is cached on all routers located between the host and the server), UniCache technique (routers located between the host and the server). Cached in one of; 1 / hop probability cached, and ProbCache technique (content delivered to the host is cached at a certain probability on one of the routers located between the host and server; the probability is 0.1 in this simulation). Cached). The simulation was performed in the following environment.

Simulator Discrete Event Based Simulator
(Discrete Event-Driven Simulator) Topology Created using GT-ITM (Georgia Tech Internetwork Topology Models)
Network topology consisting of one transit domain and five stub domains Number of routers and hosts 105 routers, 1000 hosts Content distribution 100,000 Randomly Distributed Content
(Per gigabyte / content;
Per 100 chunks / content) Content request distribution Zipf's law (parameter 1.0) Cache size 10GB / cache Content routing En-route Cache Model

First, as shown in FIG. 8, the average hop count between the host and the entity storing the content (eg, a content server or storage storing the content) is reduced. According to the present invention, the entity storing the highly popular content can be located close to the host.

In addition, as shown in FIG. 9, comparing the link stress, which is the amount of traffic over a particular link, with other techniques, it can be seen that the present invention has link stress to a similar extent as other techniques.

Furthermore, as shown in FIG. 10, it can be seen that traffic between ISPs is also reduced. Considering the cache performance described below together, it can be seen that the overall performance can be improved according to the technique according to the present invention.

11 shows the storage usage of the cache, it can be seen that the storage usage is noticeably less in the present invention. In other words, it can be seen that the present invention uses cache storage efficiently.

In addition, FIG. 12 shows the cache hit count divided by the storage usage, that is, the cache hit count normalization. As shown, it can be seen in the present invention that the number of cache hits is noticeably high.

And, as shown in Figure 13, the present invention shows that the cache replacement rate is significantly lower than other techniques, through which, it can be seen that the present invention, the storage can be used efficiently. Furthermore, by measuring the number of chunks that are not accessed and replaced, it can be seen that only 25.8% of the chunks are replaced without being accessed in the present invention, which indicates that the present invention enables efficient cache management.

In FIG. 14, relative hop counts, that is, average hop counts until the chunks are delivered to the host are divided according to the chunk identification numbers. In the present invention, it can be seen that the chunks with smaller identification numbers are distributed before the chunks with larger identification numbers, for example, in the case of the first chunk, the present invention can be located closer to the host by about 10% compared to other techniques. .

That is, the present invention, compared to other techniques, effectively caches content and increases storage utilization in the network.

Those skilled in the art may appreciate that various example logic blocks and algorithm steps described in connection with the embodiments disclosed herein may be implemented in electronic hardware, computer software, or a combination thereof. To clearly illustrate this possibility of replacing hardware and software, various illustrative steps have been described above primarily in terms of their functionality. Whether such functionality is implemented in hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but should not interpret it as causing a decision of such implementation to fall outside the scope of the present invention.

The foregoing examples are only illustrative of the present invention in detail, and the present invention is not limited to these examples. Additions, omissions, substitutions, and other modifications can be made without departing from the spirit of the present invention. The present invention is not limited by the foregoing description, but is limited only by the scope of the appended claims.

300: content delivery system 310: content server
320: host 210, 220, 230: network entity

Claims (37)

In a content delivery system for delivering content, an information-centric network comprising at least one entity and having a copy of the content located at an entity located at the one or more entities or close to a host requesting the content. In the method of delivering content to the host,
Delivering the content from the entity storing the content to the host; And
Storing the content in an entity included in the network according to the popularity of the content,
The content consists of one or more content chunks,
Wherein the storing step comprises:
Storing some of the one or more content chunks in an entity included in the network according to the popularity of the content, and as the number of requests for the content increases, the number of stored content chunks increases exponentially. Including, content delivery method. The method according to claim 1,
The popularity of the content is determined in proportion to the number of requests for the content, content delivery method. delete delete The method according to claim 1,
The entity in which the content is stored is one of the entities included in the network, the entity located on the path that delivered the request for the content. The method according to claim 1,
The entity in which the content is stored is an entity that has delivered a request for the content among the entities included in the network. The method according to claim 1,
Delivering the stored content from the entity in which the content is stored to the host. In a content delivery system for delivering content, an information-centric network comprising at least one entity and having a copy of the content located at an entity located at the one or more entities or close to a host requesting the content. In the method of delivering content to the host,
Delivering the content from the entity storing the content to the host; And
Distributing the content to other entities according to the popularity of the content,
The content consists of one or more content chunks,
The dispersing step,
Distributing some of the one or more content chunks to the other entity according to the popularity of the content, and as the number of requests for the content increases, the number of content chunks distributed to the other entity increases exponentially. Including, content delivery method. The method of claim 8,
The popularity of the content is determined in proportion to the number of requests for the content, content delivery method. delete delete The method of claim 8,
The other entity, among the entities included in the network, the content
A method of delivering content that is one of the entities located on the path that delivered the request for. The method of claim 8,
And the other entity is an entity that has delivered a request for the content among the entities included in the network. The method of claim 8,
Delivering the distributed content from the other entity that stores the distributed content to the host. In a content delivery system for delivering content consisting of one or more content chunks, a method for delivering content to a host through a network including one or more entities,
Forwarding a content request to a first entity that stores some of the content chunks;
In response to the content request, delivering some of the stored content chunks from the first entity to the host; And
Distributing some of the stored content chunks from the first entity to a second entity,
The first entity is a content server for storing the content, or an entity located on a path for delivering the content request from the host to the content server.
And the second entity is an entity located on a path for delivering the content request from the host to the first entity. The method of claim 15,
Forwarding the content request to a third entity;
And the third entity is the content server or an entity located on a path for delivering the content request from the first entity to the content server. 17. The method according to claim 15 or 16,
And the second entity and the third entity are entities selected in a hop-by-hop manner based on the first entity. In an entity that stores one or more content chunks of content, an information-centric network comprising the entity and placing the content in an entity located in one or more entities or close to a host requesting the content. In a method for delivering content from the entity to the host via an information-centric network,
Receiving, via the network, a request for one content chunk of the stored content chunks;
Delivering the requested content chunk to the host; And
If the requested content chunk is already a distributed content chunk, updating the chunk window and finally the distributed chunk so that the distributed content chunk is redistributed again. 19. The method of claim 18,
If the identification number of the requested content chunk is equal to the maximum value of the identification numbers of the stored content chunks, distributing some of the stored content chunks through the network. The method of claim 19,
Distributing some of the stored content chunks may include:
Distributing, among the stored content chunks, content chunks other than the content chunks contained in the chunk window and already distributed. 21. The method of claim 20,
Distributing some of the stored content chunks may include:
Updating the chunk window according to the distribution; And
Updating the last distributed chunk according to the distribution. delete 19. The method of claim 18,
Receiving, via the network, content chunks distributed to the entity; And
Storing the distributed content chunks. 24. The method of claim 23,
Storing the distributed content chunks,
If the storage of the entity is full, storing the distributed content chunks using an LRU scheme. In a first entity that stores one or more content chunks of content, the first entity includes the first entity and places the content in one or more entities or in close proximity to a host requesting the content. A method for delivering the content from the first entity to a host via an information-centric network, the method comprising:
Receiving a content request over the network; And
Based on a chunk distribution algorithm, delivering and distributing the content, wherein the chunk distribution algorithm is based on x and x is a positive integer as the distribution index. The method of claim 25,
Wherein the receiving comprises:
Receiving a request for the content chunk i,
The content chunk i is a content chunk requested from the host, and i is an identification number of the content chunk. 27. The method of claim 26,
And the chunk distribution algorithm is based on i. 27. The method of claim 26,
If it is not storing the content chunk i, further comprising forwarding the received request to a second entity,
And the second entity is a content server storing the content or an entity located on a path for delivering the content request from the first entity to the content server. 27. The method of claim 26,
If the content chunk i is being stored, delivering the content chunk i to the host. delete 27. The method of claim 26,
The chunk variance algorithm is based on n,
N is a chunk window state and is zero or a positive integer. 32. The method of claim 31,
The chunk distribution algorithm is based on sent,
Wherein the sent is an identification number of a content chunk last distributed by the first entity. 33. The method of claim 32,
The chunk variance algorithm is based on t,
T is an identification number of a content chunk last stored in the first entity. 34. The method of claim 33,
The chunk distribution algorithm,
If i is equal to the t, distributing the content chunk having an identification number of sent + 1 to min (x ^{n + 1} , t) among the content chunks stored in the first entity to a third entity; Including more,
And the third entity is an entity located on a path for delivering the content request from the host to the first entity. 35. The method of claim 34,
The chunk distribution algorithm,
Updating n to n + 1; And
The sent to min (

, t) updating the content delivery method. 34. The method of claim 33,
The chunk distribution algorithm,
If i is equal to or less than the sent, then n

Updating to; And
Updating the sent to i-1. A host requesting content;
A content server storing the content; And
A content delivery system comprising an information-centric network connecting the content server to the host and placing the copy of the content on one or more entities or on an entity located close to the host requesting the content. As
The network may include an entity including a content storage unit capable of storing the content according to the popularity of the content.
The entity is
A content delivery unit delivering the stored content to the host; And
According to the popularity of the content, further comprising a content distribution unit for distributing the content to other entities,
The content consists of one or more content chunks,
The content storage unit,
According to the popularity of the content, at least one of the one or more content chunks are stored in an entity included in the network, and as the number of requests for the content increases, the number of stored content chunks increases exponentially. Delivery system.

Images