JP3850379B2 - Semantic information network construction method, semantic information network system, and program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a semantic information network that is a general-purpose P2P (Peer 2 Peer) platform, and more particularly to a semantic information network construction method, a semantic information network system, and a program that can freely construct a desired network.
[0002]
[Prior art]
In recent years, a new concept of “brokerless search model” has been devised in the networking industry. This brokerless search model is equivalent to P2P (Peer-to-Peer) that has gained worldwide attention, such as Gnutella, which was announced in March 2000. JXTA, announced in April 2001, also corresponds to P2P, a brokerless search model. P2P is an innovative technology since the WWW and is called the third generation. In recent years, research and development of brokerless transmission models that apply the P2P concept to the transmission layer and brokerless policy models that apply the P2P concept to the policy layer have become active. The dimensions of the brokerless (distribution) model are shown in FIG. One of these brokerless search models is SIONet (Semantic Information Oriented Network).
[0003]
SIONet is a meta network that delivers an event composed of data and semantic information of the data based on semantic information (metadata). As shown in FIG. 24, the event receiver registers semantic information indicating the condition of the event to be received in advance in SIONet. When an event is transmitted from the event sender to the SIONet, the SIONet collates the semantic information of the event with the registered semantic information, and transmits the event to the event receiver based on the collation result. In this way, sending an event to SIONet is called “stimulation”, and the registered semantic information matches the semantic information of the event, and notifying the event recipient who registered the semantic information of the event. It is called “ignition”.
[0004]
As a result, a specific entity can be dynamically searched and discovered from an unspecified number of entities (terminal devices) super-distributed on the network. That is, SIONet is a network that delivers events based on semantic information (sending to whom) instead of the destination address (sending to whom) used in the conventional network.
[0005]
SIONet is an autonomous distributed cooperative network (autonomous distributed computer) in which all entities including the components of SIONet perform autonomous distributed cooperation to self-organize the network. SIONet network components include "Semantic Information Switch (SI-SW)", "Semantic Information Router (SI-R)", "Semantic Information Gateway (SI-GW)", "Event Place", "Session", etc. These can be self-organized as needed to build a secure and scalable P2P network with a bottom-up approach.
[0006]
The basic concept and principle of SIONet is simple and unified. By repeating the simple operation of "registering a filter by an entity and dispatching an event", that is, "entity" stimulation "and" ignition "" chain reaction "between SIONet network components, all entities The thing that controls the behavior of this chain reaction is the semantic information registered in the SIONet by the above-mentioned event receiver, and this is called the “filter”. The method of chain reaction of entities can be dynamically controlled by the filter value registered in the filter. Another basic concept in SIONet is event place. An event place is an entity group connected to each other by a shared link, thereby limiting the scope of a chain reaction.
Details of SIONet are described in Non-Patent Documents 1 and 2. The applicant of the present invention has previously filed an application relating to SIONet according to Japanese Patent Application No. 2001-394980.
[0007]
[Non-Patent Document 1]
Takanari Hoshiai, Keiichi Koyanagi, Birge Sukubatar, Jun Kubota, Hiroshi Shibata, Takamichi Sakai: "Semantic Information Network Architecture", IEICE Transactions B, Vol.J84-B, No.3, pp.4,11 -4,2,4 (2000.7 reception, 2001-3 publication).
[Non-Patent Document 2]
Takanari Hoshiai, Hiroshi Shibata, Takamichi Sakai, Keiichi Koyanagi: “Semantic Information Network Architecture: SION Architecture”, NTT R & D, Vol.50, No.3, pp.157-164 (2000.12 reception, published in 2001.3).
[0008]
[Problems to be solved by the invention]
By the way, in the construction of a conventional network system, the network system is designed in accordance with a predetermined network shape. For this reason, the system must be redesigned every time the network shape changes. was there.
The present invention has been made in view of such circumstances, and its purpose is to cope with networks of various shapes, and therefore semantic information that does not require redesigning the system even if the shape of the network changes. To provide a network construction method, a semantic information network system, and a program.
[0009]
[Means for Solving the Problems]
The present invention has been made to solve the above-described problems, and the invention according to claim 1 is a method for constructing a semantic information network in which a plurality of entities are connected via a shared link. A process in which an entity wishing to participate in applying to an entity already participating in the semantic information network, and the number of entities already accepted by the entity that has received the participation application at that time are defined in advance. A method for constructing a semantic information network, comprising: a step of establishing whether a shared link is established with respect to an entity that has applied for an application, .
[0010]
According to a second aspect of the present invention, in the semantic information network construction method according to the first aspect, the entity that wishes to participate checks whether or not to participate in a plurality of entities when applying for participation. If it is not an application for participation in an entity, an application for participation is made to an entity constituting the semantic information network.
[0011]
According to a third aspect of the present invention, in the semantic information network construction method according to the first aspect, the entity participating in the semantic information network includes the entity that applied for participation and the application for participation in itself. It holds the list of entities that received
[0012]
The invention according to claim 4 is a semantic information network system in which a plurality of entities are connected via a shared link, and the entity accepts an entity for which it has requested participation and a request for participation in itself. Storage means for storing a list of entities, and shared with respect to an external entity that has applied for participation when the number of entities that have received a participation request recorded in the list does not reach a preset maximum number A semantic information network system comprising shared link establishment means for establishing a link.
[0013]
According to a fifth aspect of the present invention, in the semantic information network system according to the fourth aspect, the maximum number of receptions is “1”.
A sixth aspect of the present invention is the semantic information network system according to the fourth aspect, wherein the maximum number of acceptance is “2”.
[0014]
According to a seventh aspect of the present invention, in the semantic information network system according to the fourth aspect, the maximum number of receptions is N (N: a positive integer) and the maximum number of receptions is “0”. It is characterized by.
[0015]
The invention according to claim 8 is the semantic information network system in which a plurality of entities are connected via a shared link, and the entity that has received the participation request recorded in the internal list in the computer constituting the entity. The procedure for checking whether the number has reached the preset maximum number of acceptance, the procedure for establishing a shared link for the external entity that has applied for participation, and the participation And a procedure for writing the number of entities that have received a request to the list.
[0016]
In this specification, an entity is a general term for all computers such as a personal computer, a workstation, an information portable terminal, a mobile phone, and a wearable computer.
Shared link is software that connects two entities, and refers to software that can mutually set data transfer requirements.
[0017]
DETAILED DESCRIPTION OF THE INVENTION
[Details of SIONet]
<Definition of SIONet entity>
Here, all computers such as personal computers, workstations, portable information terminals, mobile phones, and wearable computers are collectively referred to as hosts. Furthermore, a host that implements SIONet software is called a “SIONet entity” or simply an “entity”. As shown in FIG. 2A, an entity (entity device) is a virtualized host in which SIONet software is installed as a unit of autonomous distributed cooperation in SIONet. SIONet software provides a mechanism for individual entities to perform autonomous distributed cooperation. This SIONet software allows each host to become an autonomous distributed computer.
[0018]
Entities are mainly classified into the following three types as shown in FIG.
・ Applications that act as services (service entity: SE, service entity part) are included inside the entity.
· Modules that act as network components (network entity: NE) are included inside the entity
・ Including both SE and NE in the entity
[0019]
<Entity structure and state transition>
FIG. 3 is a diagram illustrating state transition of an entity. As shown in this figure, there are three states of an entity, “Non-Existent”, “Suspend”, and “Active”.
The “Non-Existent” state represents the entity state when the SIONet software is installed on the host.
By executing (launching) the SIONet software, a transition is made from the “Non-Existent” state to the “Suspend” state. At this time, internal modules such as an entity control unit, a control panel, a network entity factory (NE factory), and a network entity (NE) are generated in the entity as shown in FIG. Below, each role is demonstrated easily.
[0020]
The entity control unit is an entity entity, and accepts various requests (session establishment request, shared link establishment request, etc., which will be described later) from other entities and the control panel. Note that a request for establishing a session or a request for establishing a shared link cannot be made from another entity to an entity in the “suspend” state.
[0021]
The control panel provides a GUI (Graphical User Interface) to owners of entities such as people. For example, the control panel prompts the owner of the entity to input the entity name, and the entity control unit assigns the entity name to the entity based on the input entity name. In the example of FIG. 4, the entity name is “entity 2”. This entity name may be referred to as a “local entity name”, particularly distinguished from a “global entity name” described later. The local entity name can be arbitrarily set by the owner of the entity.
[0022]
The NE (network entity) factory has a function of dynamically generating an NE.
NE (Network Entity) means SI-R (Semantic Information Router), SI-GW (Semantic Information Gateway), alive entity, fault handling entity, statistics collection entity, etc. It is a general term for modules that behave in order to achieve this.
[0023]
The SE (service entity) is an application embedded as an SE in the entity using a plug-in mechanism provided by the entity, and is a virtualized P2P application operating on the SIONet. An application can be embedded in an entity as an SE using a plug-in mechanism provided by the entity. Also, the SE can establish an arbitrary number of sessions with the SI-SW. The SE can send and receive events only through a session. SE is not involved in SIONet network construction and operation. A plug-in mechanism for incorporating SE into an entity will be described later.
[0024]
For example, the entity (entity 2) in the “suspend” state in FIG. 3 generates an event place as shown in FIG. 5 or joins an existing event place as shown in FIG. When the entity (entity 2) belongs to an arbitrary event place and becomes a member of an entity group, the entity (entity 2) transitions from the “suspend” state to the “active” state. Only entities in the “active” state advertise their existence (public, details will be described later), and can accept session establishment requests and shared link establishment requests from other entities.
[0025]
Here, after the software is installed on the host and an entity including an NE, an NE factory, an entity control unit, a control panel, and the like is generated, an operation until the entity enters an “active” state will be described. Assume that an entity receives an instruction to create an event place or participate in another event place via the control panel from the owner of the entity. The entity control unit of the entity requests the NE factory of the entity to generate SI-SW (Semantic Information Switch). When the NE factory generates SI-SW, the entity control unit establishes a session between the generated SI-SW and an internal module such as NE. This makes it possible to accept advertisements, session establishment requests from other entities, shared link establishment requests, and the like.
[0026]
Here, the role of SI-SW and session will be described with reference to FIG.
A session is a connection between SI-SW and SE, and SE can send and receive events only through the session. There are three types of sessions: event transmission session, reception session, and transmission / reception session.
SI-SW provides a switching mechanism that switches events based on semantic information. The SI-SW accommodates NEs and internal entities such as SEs and NEs incorporated in the entities by a plug-in mechanism through a session in a star shape. In the following, SE will be described as an example, but the same applies to other internal entities included in an entity such as NE.
[0027]
Here, the configuration of the event will be described. As shown in an example in FIG. 8, the event includes a control information part, a semantic information part, and a data part. Transmission data is stored in the data portion. Various data and programs such as text data, binary data, reference, proxy, and agent can be stored as transmission data.
[0028]
The semantic information portion stores semantic information (vocabulary and its value) of transmission data and a vocabulary concept (event type) of semantic information. Here, the semantic information is metadata describing the characteristics of transmission data and is an instance of an event type. The event type is a semantic information template. You can define inheritance relationships between event types. FIG. 9 shows an example of the semantic information system. In this figure, the semantic information “title; yesterday”, “price; $ 30”, “artist name; beatles” belongs to the event type “popular” and also belongs to the event type “music”. Yes. In addition, the meaning information of “Title; Night of Kiyoshiko”, “Price; $ 20”, “Situation; Christmas” belongs to the event type “BGM” and also belongs to the event type “Music”. .
[0029]
As a description language of the semantic information system, there is XML (Extensible Markup Language). Note that some event type names (all names starting with SIONet) are reserved for network entities.
The event control information part shown in FIG. 8 is a control field used for SIONet execution control, and only this is not open to the SE who is the user of SIONet. The control information section includes matched filter identifiers, matched filter verification scores, synchronous statistical information collection flags, TTL (Time To Live) values within event places, between event places, within event places, and between event places. Control information such as the number of hops and hop attributes is set.
[0030]
In the entity configuration as shown in FIG. 7, the function of the SE will be described. The SE registers event acquisition conditions in the SI-SW via the reception session. This is called a “filter”. In the filter, set the vocabulary concept (event type) of the event you want to acquire, and the matching condition with the semantic information (for example, the vocabulary “Price” is in the range of “$ 20 to $ 40”) . Note that collation conditions such as “complete match” with all vocabularies defined in the event type, “partial match” with some vocabularies, and “weighted match” can be selected in units of filters.
[0031]
The SE can register a plurality of filters from one reception session, but the filters registered through the same session have an “or” relationship. That is, for one event, one receiving session is fired at most once. If a wildcard is specified for the event type of the event to be acquired, all event types are targeted for acquisition. Further, when 1 (logical value is true) is set as the matching condition with the semantic information, it means that the matching condition with the semantic information is satisfied unconditionally.
[0032]
In addition, the SE transmits an event to the SI-SW via a transmission session. At this time, the SI-SW collates the semantic information part of the event with the filter. Specifically, first, it is checked whether or not the event type is desired to be received. If the event type is satisfied, the semantic information and the matching condition are checked. As a result of the collation, when the semantic information satisfies the collation condition, the entity that registered the matched filter is activated and the event is notified.
In SIONet, event transmission is referred to as “stimulation”, that the filter matches the event as “reaction”, and that the entity that registered the matched filter is activated to notify the event is referred to as “ignition”.
[0033]
<Entity name assignment method>
As shown in FIG. 5, the case where the entity 2 generates an event place named “event place A” will be described. The owner of the entity 2 inputs an event place name to the entity 2 through a control panel or the like. Here, the input event place name is “event place A”. Then, the entity control unit requests the NE factory to generate SI-SW, and establishes a session between the generated SI-SW and an internal module such as NE. Further, the entity control unit stores the name of the generated SI-SW as “event place name + local entity name”, that is, “event place A + entity 2”. This name is called “global entity name”. Creation of this global entity name corresponds to generation of an event place.
[0034]
That is, in SIONet, some management entity is not necessarily generated by generating an event place. That is, in SIONet, member management of entities belonging to an event place is realized by a shared link between entities described later, and therefore there is no centralized management unit itself that performs member management. For this reason, even if the entity that generated the event place leaves the event place, the entity remaining in the event place autonomously self-organizes, whereby the operation of the event place is continued. In other words, the removal of the last entity from the event place corresponds to the disappearance of the event place.
[0035]
This global entity name is used when a shared link or session is established. An operation in which the entity 2 as shown in FIG. 5 further generates an event place B and establishes a shared link from the entity 1 will be described with reference to FIG.
[0036]
First, an operation in which the entity 2 already participating in the event place A generates the event place B will be described. The entity control unit of entity 2 causes the NE factory to generate a new SI-SW, as described above. The entity control unit assigns “event place B + entity 2” as the global entity name to the generated new SI-SW. Thereby, the event place B is generated. At this time, the entity 2 has two global entity names (SI-SW names) “event place A + entity 2” and “event place B + entity 2”.
[0037]
Next, an operation in which the entity 1 joins (joins) the event place A will be described. As described above, the entity 2 participating in the event place A and the event place B is in the “active” state. Entities in the “active” state can advertise their existence to other entities, with the global entity name being published.
[0038]
When entity 1 discovers entity 2 and makes a join request to entity 2, entity 1 indicates that the global entity name of entity 2 that is the join destination entity is “event place A + entity 2”. Entity 2 is notified.
[0039]
In this way, the global entity name can uniquely identify the SI-SW that establishes the shared link, and thus can participate in the event place A. In this way, the SI-SW, which is the shared link or session establishment destination, is virtualized as a global entity name, so that the entity owner or the like is not directly aware of the SI-SW.
[0040]
<SE plug-in method>
The operation of incorporating SE into the entity shown in FIG. 11 will be described below.
(1) The entity owner instructs an application plug-in to the control panel. At this time, the executable file name of the application to be plugged in is given as a parameter.
(2) The control panel notifies the entity control unit that the plug-in instruction has been received and the execution file name of the plug-in application.
[0041]
(3) The entity control unit stores an execution file name to be plugged in, starts an application using the given execution file name, and establishes a session between the SI-SW and the started application.
That is, the plug-in in SIONet is nothing but to establish a session between the application and SI-SW. In SIONet, all operation entities such as SE and NE cooperate through a session. Therefore, plug-in / plug-out (detachment) can be easily realized, and the plug-in / plug-out does not affect other operation entities (super loose coupling). By extending this concept, SE sharing between entities as described later can be realized.
[0042]
<PREFERENCE architecture>
SIONet adopts the concept of a reference model as shown in FIG. The PREFERENCE architecture shown in Fig. 12 is an architecture for building a listening society (a listening-oriented cyber society) that registers semantic information in a filter and sets distribution conditions (matching conditions) in the semantic information section of an event. Provides total solutions for less-less distribution model, broker-less search model, and broker-less policy model. Specifically, it consists of a stream interface, SIONet, COMNet, and the like.
SIONet is an event transmission layer and provides a network interface to higher layers. The middleware layer is a community (community network: COMNet), and corresponds to the intelligence layer of SIONet. In this layer, information localization, authentication, security, information right guarantee, policy control, and the like are performed. The highest layer is the application layer. SE is implemented in this application layer. Examples of the SE include a personal TV station and a smart messenger for a message exchange service.
[0043]
In this way, SIONet provides a common P2P network infrastructure and a plug-in mechanism for all P2P services (SEs provide services) such as distributed computing, information exchange, collaboration, and message delivery. To support efficient application development.
[0044]
<Role of entity in SIONet>
SIONet is a form of constructing a volunteer network by helping each other's entities to help each other, and each network is self-organizing through autonomous distributed cooperation. For example, in the configuration of FIG. 6, an operation in which the entity 2 participates in the event place generated by the entity 1 and then the entity 1 leaves the event place will be described.
[0045]
In advance, the entity 1 generates an event place and participates in the event place. Here, when entity 2 makes a join request to event place for entity 1, as shown in FIG. 13, SI-SW is generated for entity 2, and SI-SW of entity 1 and entity 2 is further generated. A shared link is established between them and an entity group is formed.
In this way, after the shared link is established between the SI-SWs of the entity 1 and the entity 2 and the entity 2 participates in the event place, the entity 1 that is the creator of the event place leaves the event place. Then, the shared link between SI and SW is released, and thereafter, the event place construction / operation is continued only by entity 2. Thus, since each entity self-organizes, even if an entity leaves, disappears, etc., another entity performs the function of the entity.
[0046]
In SIONet, various forms of P2P networks can be constructed by arranging and combining entities according to the purpose. It is also possible to seamlessly link different forms of P2P networks. This is achieved by realizing:
(1) All entities are autonomously distributed and coordinated using a simple and simple mechanism called “chain reaction based on stimulation and firing”.
(2) Reducing the difference in P2P network form to the problem of entity placement and management
In this way, one of the features of SIONet is the ability to build various forms of P2P networks with a single mechanism.
[0047]
<Shared link>
The “shared link” is a concept for performing bidirectional event sharing between a plurality of different entities. For example, as shown in FIG. 14, when entity 2 makes a request for establishment of a shared link (SL) to entity 1, as shown in FIG. 15, SI-SW2 and SI-SW1 A shared link is established between them and a new entity group is formed.
That is, the shared link is a mechanism for forming an entity group, and is also a multi-hop route.
[0048]
In FIG. 14, the entity 2 that has successfully established the shared link includes the event place name (global entity name) from the entity 1, the event transfer method (routing method) in the event place, and the description (description) of the event place. Various information such as SE information that is plugged in is transmitted. Various event routing methods can be dynamically set between SI-SWs by setting filter values using SI-R, which will be described later. In the following, the operation of establishing a shared link will be described, and the role and mechanism of SI-R will be described in detail.
[0049]
<Establishment of shared link>
A mechanism up to establishment of a shared link will be described with reference to FIGS. 14 and 15. First, in FIG. 14, consider a case where entity 2 issues a shared link establishment request to entity 1.
(1) The entity 2 issues a shared link establishment request to the entity 1.
[0050]
(2) As shown in FIG. 15, the entity control unit of entity 1 establishes SL1,2 (shared link for receiving an event from entity 2) based on semantic information in the NE factory. SI-R1,2 that performs event routing (event transfer) between SI and SW is dynamically generated inside entity 1. Further, the entity control unit stores that the entity 2 is a shared link establishment request source. Strictly speaking, the entry point of the entity 2 and the global entity name (SI-SW2) are stored as a shared link establishment request source.
[0051]
(3) SI-R1,2 establishes an event reception session for SI-SW2 by requesting entity 2 to establish an event reception session. At this time, a filter indicating a condition of an event received by the entity 1 is registered in the SI-SW 2. Furthermore, SI-R1 and 2 establish a session for event transmission with respect to SI-SW1 to which SI-R1 and 2 belong. Such a combination of transmission and reception sessions established by SI-R is called a shared link. As a result, SL1,2 is established. For example, an event sent by SE4 to SI-SW2 is also sent to SI-SW1 via SI-R1,2.
[0052]
(4) At this time, the event transfer method can be dynamically controlled by the set value of the filter registered by SI-R1,2 to SI-SW2. An example is shown below.
(A) The event sent to SI-SW2 is unconditionally transferred to SI-SW1. That is, SI-R1,2 fires for all events. This is made possible by setting “wildcard for the acquired vocabulary concept and true for the matching condition with semantic information (vocabulary)” as a filter that SI-R registers in SI-SW. This is called “unconditional routing”. This setting only needs to be performed once when the shared link is established. In this method, there is a possibility that unnecessary event transfer and event collation processing overhead at the transfer destination may occur. However, since there is no overhead for event path establishment (routing information setting) described later, the number of registered filters is small. This method is effective when it is sufficiently larger than the number of event transmissions. This routing method is mainly used for multi-hop broadcast communication.
[0053]
(B) Only events necessary for entity 1 are transferred from SI-SW 2 to SI-SW 1. Thereby, unnecessary event transfer is not performed. That is, SI-R1,2 fires only for a specific event. This is because SI-R sets only the vocabulary concept (matching condition is always true) in the filter as “event routing by vocabulary concept”, and sets the matching condition between the vocabulary concept and vocabulary in the filter as “event routing by vocabulary” It is divided roughly into. The former is mainly used for multicast communication with a multi-hop type attribute, and the latter is mainly used for multi-hop type unicast communication and multi-hop type multicast communication. This is used in failure processing notifications, statistical information notifications, alive notifications, reply notifications, advertisement notifications, and the like.
[0054]
In addition to multi-hop communication such as multi-hop broadcast communication, multi-hop attribute multicast communication, multi-hop unicast communication, and multi-hop multicast communication, the communication destination can be set according to the filter value set by SI-R. Direct communication using entity entry points is also possible. Multi-hop broadcast communication, multi-hop attributed multicast communication, multi-hop unicast communication, and multi-hop multicast communication will be described later.
[0055]
(4) The procedures (2) to (3) described above are performed in the same way in the entity 2 that is the shared link establishment request source. That is, the entity control unit of entity 2 stores entity 1 (SI-SW1) as a shared link establishment request destination and dynamically sets SI-R2,1 to the NE factory in order to establish SL2,1. Generate. This generated SI-R2,1 registers a filter in SI-SW1 in the same manner as described above, and establishes a shared link. Thereby, a bidirectional shared link is established between SI-SW1 and SI-SW2, and events can be shared among entities. Since SI-R does not register the filter, it is possible to establish a one-way shared link, that is, to prevent SI-R from being fired.
[0056]
<Establishing an event path>
Route selection information for event forwarding (event routing) (a set of filters that SI-R registers for event sharing) is called an “event path”. For example, as shown in FIG. 15, in a configuration in which entity 2 and entity 3 each establish a shared link with entity 1, consider a case where SE3 of entity 2 registers a filter for SI-SW2. At this time, the SI-R2,1 of the entity 2 registers the filter registered by the SE3 in the SI-SW1 of the entity 1 through the reception session. Similarly, the SI-R1,3 of the entity 1 registers the filter in the SI-SW 3 of the entity 3 through the reception session. In this way, with the filter registration by SE3 as a trigger, the SI-R registered filter sequentially spreads to adjacent SI-Rs based on the shared link, thereby establishing an event path. This is called “event path setting or spreading”.
[0057]
Here, the establishment of the event path for the above-described “event routing based on the vocabulary concept” will be described. In FIG. 15, in order to establish an event path for “event routing by vocabulary concept”, SI-R2,1 is the event type only in the filter registered by SE3 of entity 2 with respect to SI-SW2. Set the matching condition is always true) as a filter to entity 1's SI-SW1. That is, the matching condition registered by SE3 as a filter is not used. In other words, by using only the event type, for example, when SE1 of entity 1 sends an event, only the event type is verified in SI-SW1, and as a result, if SI-R2,1 fires, SI-R2, 1 sends the event to SI-SW2, and collates with semantic information in SI-SW2.
[0058]
In addition, establishment of an event path for the above-mentioned “event routing by vocabulary” will be described. In FIG. 15, when establishing an event path for “event routing by vocabulary”, SI-R2,1 of entity 2 sets the filter value registered by SE3 as the filter value to SI-SW1 of entity 1 as it is. To do. That is, SI-R2,1 is a form in which both event type and semantic information registered by SE3 as a filter are registered in SI-SW1 of entity 1 and complete filtering is performed in SI-SW1.
[0059]
Therefore, the event routing method based on vocabulary does not cause any unnecessary event transfer, but the event path setting overhead for routing becomes enormous, so this method is effective when the number of event transmissions is sufficiently larger than the number of registered filters. It is. On the other hand, the routing method based on the event type is positioned as a compromise between the two methods described above. In other words, it is effective when the number of filter registrations and the number of event transmissions are about the same, or when the ratio between the number of filter registrations and the number of event transmissions cannot be predicted. With these mechanisms, multi-hop communication of events based on semantic information is realized. An event routing method can be specified when generating an event place or registering a filter from an SE.
[0060]
The event path setting request is propagated to all SI-Rs in the event place, but the propagation range can be limited by the TTL value. The operation will be briefly described with reference to FIG. Here, for convenience of explanation, it is assumed that an event path is established for vocabulary concept routing, but the present invention is not limited to this. Further, each entity advertises (filter registration) the same vocabulary concept. The description will be made assuming that the TTL value is 2.
[0061]
In FIG. 16A, the SI-R of the entity 2 starts an event path setting request. Here, since the TTL value is 2, the event path setting request is propagated only to entity 1, entity 3, and entity 4, and as a result, the event path is established as shown in FIG. The
[0062]
Similarly to the above, when the SI-R of the entity 11 starts an event path setting request, an event path is established for the entity 10 and the entity 9. In the configuration shown in FIG. 16, when the SI-R of entity 2 and the SI-R of entity 11 make an event path setting request, the event path is not set in entity 5. Is not shared. That is, the event that has occurred from the entity 2 does not cause a chain reaction with the entities 5 to 11 and the event is not transferred to the entities 5 to 11.
[0063]
However, as shown in FIG. 16B, when the SI-R of the entity 5 initiates an event path setting request, the event path for the entity 4, the entity 5, the entity 6, the entity 7, and the entity 9 is determined. Is established. As a result, as shown in FIG. 16C, an event path is established for all entities in the event place except the entity 8.
[0064]
Thus, if the vocabulary concept is frequently used (reputed or popular) in the event place, the event path will eventually be established in the event place. Become. On the other hand, vocabulary concepts that are not so popular will eventually be deceived, which allows a mild chain reaction to be realized. Note that the numbers on the event path in FIG. 16 indicate the multiplicity.
[0065]
An entity can publish the multiplicity of event paths as an attribute of an entity property. On the other hand, an entity having a high multiplicity of event paths can be found by entity property discovery. At this time, by re-sharing the shared link with respect to the entity, it is possible to gradually gather like-minds (like entities) in the vicinity (localization of like-entities). Details of the entity property will be described later.
[0066]
From the viewpoint of distributed object technology, the event path can also be interpreted as follows. Entities that are super-distributed around the world have some “correlation” between them. There are various properties such as an entity name, a group name, and attributes (location, interest, reputation, fashion, service) that give a correlation. Based on the correlation, the entities have a connection between the entities and perform distributed cooperation. This entity that gives an entity correlation is called an entity property.
[0067]
In SIONet, this correlation is expressed by the vocabulary concept and vocabulary and set as an event path based on the shared link. The strength of the correlation corresponds to generalization / specialization of vocabulary concepts, multiplicity of event paths, and the like. That is, in SIONet, the correlation between entities is dynamically controlled and managed by an event path. This is the control of the chain reaction by the filter. Therefore, the entity does not have a fixed entity identifier.
[0068]
For example, an IP address is a fixed identifier of an entity based on a position. In SIONet, instead of this, an entity property described as a vocabulary concept / vocabulary is used as an entity identifier. Entities register these as filters in SI-SW, thereby declaring entity properties (entity identifiers) and advertising their properties based on shared links. This corresponds to the propagation of the event path. Thereby, an event path is established.
[0069]
As described above, SI-R is a network entity having both aspects of event transmission and event reception, and is not fundamentally different from a general service entity. In SIONet, an entity used for SIONet control such as SI-R is called a network entity (NE), particularly distinguished from a service entity (SE). In SIONet, all service entities and network entities are treated as a common entity, and autonomously operated according to a simple and consistent common logic of event transmission and event reception, that is, chain reaction of stimulus and firing. , Provide a super-distributed and super-loosely coupled architecture in which all entities can cooperate autonomously and distributedly.
[0070]
<Release and re-establish shared link>
In cases where an entity participating in the event place falls into a failure or leaves the event place, the remaining entity is not able to participate in the operation of the network for various reasons. By doing so, it is necessary to be able to continue the network service.
In such a case, SIONet releases the shared link established between SI and SW, and re-establishes the shared link. The operation will be described with reference to FIG.
[0071]
In FIG. 17, the first numbers (1) and (2) in the shared link establishment request indicate the order of those requests. That is, in order of the shared link establishment request, first, the shared link establishment request is made from the entity 2 to the entity 1. Next, a request for establishment of a shared link is made from entity 3 to entity 2. Finally, a shared link establishment request is made from entity 4 to entity 2. Here, “Each entity can make a shared link establishment request at most once in the same event place, but establishes a shared link based on a link topology that can accept an unlimited number of establishment requests. Propose a link establishment method. This guarantees that no loop will occur between entities linked by a shared link. That is, by using this method, an open link topology can be easily realized only by link re-establishment between adjacent entities.
[0072]
If the establishment request is successful in the above-described order, a shared link is established between each SI-SW by the above-described procedure. When the shared link is established, the entity control unit of each entity stores the entity that has made the establishment request and the entity that has accepted the establishment to itself. For example, the entity 2 holds a list of the entity 1 that has requested establishment and the entities 3 and 4 that have accepted the establishment of the entity 2.
In this situation, for example, when the entity 2 is moved out or removed, a shared link release request is issued to its own SI-SW 2 to release the shared link. Thereafter, each entity establishes a new shared link. Hereinafter, the operation will be described.
[0073]
(1) When the entity 2 leaves, the entity control unit of the entity 2 notifies the one-hop entity (entity 1, entity 3, entity 4) that the shared link is released. This can be done by sending an event with the TTL value of the event set to 1. At this time, for the entity (entity 3 and entity 4) that has accepted the establishment of the shared link to itself, the entity 1 that has made the establishment request is used as the establishment request destination of the new shared link on behalf of itself. teach. If there is no entity that has made an establishment request, an arbitrary entity is selected from the establishment requesting entities (one hop ahead) to itself, and this is used for the shared link on behalf of itself. The establishment request destination.
[0074]
(2) In order to release the shared link (SL2,1, SL2,3, SL2,4) established by itself, the entity control unit of entity 2 notifies SI-R2,1, SI-R2,3 , Notify SI-R2,4 (not shown). SI-R2,1, SI-R2,3, and SI-R2,4 are SI-Rs of entity 2, and, as described above, shared link SL2, to entity 1, entity 3, and entity 4 1, SL2,3, SL2,4 are established. These SI-Rs use this as a trigger to release the session of entity 2 to SI-SW2.
[0075]
(3) The entity control unit of entity 1 instructs SI-R1,2 (not shown) to release shared link SL1,2. SI-R1,2 releases shared link SL1,2. The entity 1 waits for establishment requests from the entities 3 and 4.
(4) As with entity 1, entity 3 and entity 4 release all shared links (SL3,2 and SL4,2) that they have established for SI-SW2 (SI-SW2 of entity 2). Then, the shared link establishment request is made to the new entity (entity 1) taught from the entity 2 as the shared link establishment request destination, and the shared link is reestablished. That is, as described above, each entity control unit causes the NE factory to generate SI-R3,1 and SI-R4,1 respectively, and the SI-R3,1 and SI-R4,1 share the shared link SL3,1. 1. Establish SL4,1.
[0076]
As described above, the shared link establishment process and re-establishment process are performed only between adjacent entities and do not affect other entities. That is, there is no need to reconstruct links for all entities.
[0077]
The shared link is not only requested to release the shared link by following proper procedures such as moving out or deinstalling the entity, but also due to an entity failure, power failure, session (physical communication path) failure, etc. There are cases where you must re-establish. However, in such cases, it may not be possible to teach alternative entities that are needed when reestablishing a shared link. In SIONet, in order to cope with such a situation, each entity transmits an event of n hops (n is an arbitrary natural number) to grasp an alternative establishment request destination entity. In SIONet, finer hop control is possible by using TTL values and hop attributes. Examples of hop attributes include:
[0078]
(1) Only the entity requesting shared link establishment is targeted for hopping.
(2) Only the entity requesting the establishment of a shared link is a hop target.
(3) All entities are targeted for hopping.
[0079]
An example of an event flow when the hop attribute of (1) is designated is indicated by a broken line in FIG. In FIG. 18, (i / j) is shared link establishment information stored in each entity. i indicates the entity to which the entity has made a shared link establishment request, while j indicates the entity that has accepted the shared link establishment request. For example, (2/5, 6) of the entity 3 indicates that the entity 3 makes an establishment request to the entity 2 and accepts the establishment requests from the entities 5 and 6. Each entity holds shared link establishment information, so that, for example, when the entity 3 makes a further shared link establishment request to another entity, the request can be rejected as an error. This guarantees the consistency of the open link topology described above. In this situation, entity 6 has 3 hops and sends an event having the above-mentioned hop attribute (1), so that it becomes possible to know the existence of entity 1 and entity 2, which are alternatives when entity 3 fails. Become an entity.
[0080]
The hop attribute described above is used not only for searching for an alternative entity at the time of failure, but also for searching for an entity that can accept establishment of a shared link, and for searching for a top entity (nobody establishes a shared link). It is valid. In SIONet, only the top entity can establish a federation (corresponding to a shared link between event places) for other event places.
[0081]
<Purpose of online increase / decrease>
The purpose of the entity increase / decrease in SIONet is mainly divided into the following two.
(1) From the viewpoint of improving the total processing capacity of the event place, the number of entities in the event place is increased, and the event filtering process is load-balanced. From the opposite point of view, the entity is reduced. This is mainly used in the operation of hybrid P2P and backbone P2P networks.
(2) A flexible and global P2P network is built from the bottom up by flexibly establishing a shared link for dynamically generated entities. This is mainly used in the operation of a pure P2P network.
[0082]
Note that a hybrid P2P network is a network that is configured by, for example, an operator such as a network provider generating an event place in advance on a host or the like, and a service entity such as a personal terminal connecting to the event place through a session. . The pure P2P network is a network configured by connecting entities that are individual transmission / reception terminals via shared links, and among these networks, the smallest unit for sharing an event is an event place. The backbone P2P network is a connection form between networks, for example, a network in a state where pure P2P networks in a plurality of regions are connected via a hybrid P2P network.
[0083]
<Type of increase / decrease>
SION provides several types of increase / decrease configurations, but here, typical configurations shown in FIGS. 19 and 20 will be described.
(1) Combining and separating event places
As shown in FIG. 19A, a plurality of event places can be synthesized. Here, composition refers to collecting entities belonging to a plurality of event places as members in one event place. As a typical example, service integration (information sharing) based on a business alliance between different service operators can be considered.
[0084]
This is because a request for establishing a shared link is issued to any entity belonging to the event place or an entity in the event place where the request is made, by issuing a composition request. As a result, a shared link is established between SI and SW, and the combination of both is realized. The request source and the request destination of synthesis may be either an entity or an event place. On the other hand, in the case of division, the established shared link is released and separated into event places.
[0085]
(2) Participating in and leaving the event place
As shown in FIG. 19B, an SI-SW is generated in the request source entity by making a Join request to the entity in the event place or the event place. Then, by issuing a shared link establishment request to the request destination entity, the shared link is established between the SI and SW, and can participate in the event place.
When the entity leaves the event place, the shared link between the entities is released, the shared link is reconstructed, and the entity leaves the event place as described above. At this time, the state of the entity that has moved out transitions to the suspended state.
[0086]
By requesting the entity to establish a shared link from the event place side, the entity can be taken into the event place. This is called absorption. The opposite is called splitting.
(3) Increase / decrease of entities (SI-SW)
As shown in FIG. 20 (c), when an entity expansion request is made to an entity in an event place or an event place, an SI-SW is newly generated for the specified entity, and an existing SI-SW is generated. A shared link is established with the SW. On the other hand, when an entity reduction request is made to the event place, a shared link release request is issued to the specified entity, and after the shared link between SI and SW is released, the shared link is reestablished. , The entity is deleted. At this time, the state of the entity transitions to Non-Existent.
[0087]
(4) Federation between event places
As shown in Fig. 20 (d), a SI-GW that transfers events between event places is dynamically generated by making a federation (cooperation) request to the entities in the event place or to the event place. Both event places cooperate through the session. Note that the federation request source and request destination may be either an entity or an event place.
[0088]
<Entity advertisement>
With reference to FIGS. 21 and 22, the advertisement (publication) of an entity will be described. There are the following two aspects to entity advertisement in SIONet.
Viewpoint 1: Entrance advertisement
Viewpoint 2: Entity property advertisement
Hereinafter, each viewpoint will be described.
Viewpoint 1: Entrance advertisement
The base event place is an event place serving as a base (starting point) for searching for (discovering) an optimal event place based on an entity. In other words, the base event place is the entrance to SIONet. Therefore, entities participating in the base event place can make the SIONet entrance public. Here, opening the entrance means advertising the entry point and the global entity name of the entity that is the shared link establishment request destination. This public information is found by a search using a broadcast described later.
[0089]
Viewpoint 2: Entity property advertisement
In any entity group connected by a shared link (all event places including the base event place), each entity can advertise its entity properties. This corresponds to the propagation of the event path described above. This public information is discovered by sending out a discovery event described later.
[0090]
The flow from entity disclosure and search to entity group formation will be described below. In the following, base entity, entity 1, entity 2, entity 3, entity 4, entity 5, and entity 6 belong to the base event place, and entity 3, entity 11, entity 12, and entity 13 belong to event place α. It will be described as belonging.
[0091]
(1) Install SIONet software on the host as shown in FIG. The state of the entity at this point is “Non-Existent” as described above.
For convenience of explanation, this entity is assumed to be entity Y.
(2) By executing the SIONet software, the entity Y transitions from the “Non-Existent” state to the “Suspend” state. An entity in this state has not yet been recognized by SIONet.
[0092]
{Circle around (3)} The entity Y in the “suspend” state searches for the entrance (entry point and global entity name) of other entities belonging to the base event place in order to participate in the network as a component of SIONet. Specifically, a nearby entity is searched from among entities participating in the base event place by broadcasting. The broadcast method depends on the implementation. For example, when implemented in a wireless network, all entities within the wireless reach are targeted for search. On the other hand, when implemented in an IP network, IP broadcast or IP multicast is performed. FIG. 21 shows that the entity 1 (entry point of the entity 1 and the global entity name) is found by the search by broadcast. If a nearby entity cannot be found by broadcasting, a well-known entity can be used. Well-known entities are called base entities. Note that the base entity can exist in all event places including the base event place.
[0093]
{Circle around (4)} The entity Y makes a join request to the base event place for the discovered entity 1. That is, the entity Y makes a Join request to the entry point of the entity 1 using the global entity name of the entity 1 as a parameter. The entity 1 that has received the Join request from the entity Y accepts the Join request only when its own state transition is “active” and the entrance and the entity property are in the public mode. The entity Y cannot issue a Join request exceeding the maximum number of Joins. The maximum number of Joins can be set by the owner of the entity Y or the like on the control panel of the entity Y. When the entity 1 receives the join request of the entity Y, a shared link is established between the entity Y and the SI-SW of the entity 1 by the same operation as described above. Thereby, the entity Y is self-organized as a network component in the base event place. At this time, the entity Y transitions from the “suspend” state to the “active” state.
[0094]
Entity Y joined to the base event place can open the entrance. You can also advertise entity properties. Entity properties include global entity name (and entry point), nickname, group name, alive (only assertion exists, entry point and global entity name as shared link establishment information are not disclosed, entity is not disclosed (Even if it is a mode, there are disclosure targets), descriptions (descriptions of entities), attributes, etc. The attributes include plug-in SE information, event path multiplicity, event place information, and the like. The entity property description language includes XML.
[0095]
In FIG. 21, for example, the entity control unit of the entity 3 can publish the event place α in which the entity 3 participates simultaneously as an attribute of the entity property. Furthermore, the entity control unit can disclose not only the event place that is currently joined but also the event place information of the event place that has been joined in the past. This event place information is, for example, a description of the event place (description of the event place, explanation information), or the global entity name of the entity (semantic information switch) that established the shared link when participating in the event place ( Event place name + entity name, connection establishment information), entry point (entry point information), and the like. The number of entities that can hold the event place information joined in the past is arbitrary. For example, the entity owner or the like can be set by the control panel. The entity that has obtained the public event place information presents the event place information to the entity owner by outputting the event place information through a control panel or the like. Thereby, the entity owner can recognize the existence of another event place.
[0096]
In SIONet, there is no function to centrally manage event place information such as where and what event place and what service is provided in that event place, so if you want to get event place information, Event place information needs to be obtained from entity properties of other entities. By making public the event place information that participated in the past as well as the event place that you are currently participating in, the chances of acquiring the event place information will increase, and the chances of participating in the event place that provides the service you want will increase. .
[0097]
When the entrance of the entity Y is released, these entity properties are registered as filters in the semantic information switch in the entity, whereby the entity properties are propagated to other entities based on the shared link, and an event path is established. Basically, the event path establishment request is propagated to all the entities in the event place, but the propagation range of the event path establishment request can be limited by the TTL value. The global entity name is used as an identifier during multi-hop type unicast communication, the group name is used as an identifier during multi-hop type multicast communication, and the attribute is used as an identifier during multicast communication with a multi-hop type attribute. For example, as a typical example of multi-hop unicast communication, there is a reply notification from an event reception entity to an event transmission source entity.
[0098]
In other words, the global entity name is multi-hop unicast communication, which is a communication method for transmitting an n-hop event from the SI-SW of the entity specified by the global entity name. Multi-hop multicast communication is a communication method for transmitting n-hop events to SI-SWs of entities belonging to a group specified by a group name. Multi-hop attributed multicast communication is a communication method for transmitting an n-hop event to an SI-SW of an entity having the same attribute as that of itself. Multi-hop broadcast communication is a communication method for transmitting an n-hop event to SI-SW of an arbitrary entity.
[0099]
(5) In FIG. 22, the entity Y who participated in the base event place releases “entity properties that meet his / her needs” by sending out a discovery event to search for the “event place where he / she wants to participate”. Search for the entity This discovery event is an event used for control of SIONet, and is equivalent to an event transmitted by SE. A discovery event is an event for igniting an NE, and a matching condition with an entity property is set in the semantic information portion of the event. In other words, adding a new function in SIONet means defining a new vocabulary concept and vocabulary (not adding a mechanism, but adding a chain reaction condition). As a result, various functions can be added only by a single mechanism (chain reaction).
[0100]
Here, it is assumed that the entity 3 is found as a result of the search. As described above, the entity 3 participates in the event place α simultaneously with the base event place. By discovering the entity 3, the entity Y can know the existence of the event place α.
(6) The entity Y issues a Join request to the event place α to the entity 3 and participates in the event place α. Note that, as described above, the entity 3 accepts a Join request (shared link establishment request) when its state transition is “active” and in the public mode.
[0101]
(7) Furthermore, the entity Y can send an event for the same search at the event place α. Thereby, it is possible to find a new entity that meets its own needs. For example, it is assumed here that the entity 12 has been discovered. That is, entity Y means that entity 12 that could not be found in the base event place can be found via entity 3. By repeating such an operation, it is possible to gradually reach an event place that meets its needs.
[0102]
The advertisement method shown here has the following effects.
(1) Since there is no broker (management unit) that manages public information of entities, it is possible to construct a self-organizing network that is strong in fault tolerance at low cost. In addition, it is not realistic to manage a huge number of public information (entry points and entity properties) with a broker.
(2) Since the event place suitable for the entity can be narrowed down to the formula to be determined, a desired event place can be efficiently searched. For example, even if the spillover range of the chain reaction (the number of hops of the discovery event and the spillover range of the event path setting request for the entity property) is limited by the TTL value, as described above, the entity 12 via the entity 3 As a result, another event place to which the entity 12 belongs can be found. It should be noted that network traffic (event transfer count, event path setup request transfer count) can be reduced by limiting the spread range of the chain reaction.
[0103]
Note that entities that simply receive information from other entities without providing information (for example, without disclosing entity properties) will be forced to release the shared link as a penalty, As long as you are not satisfied, you can prevent them from participating in the event place again.
[0104]
<SE sharing method (automatic distribution)>
In the following, the mechanism of the SE sharing method will be described by taking SE (game application program) sharing in the game event place as an example.
(1) The owner of the entity 2 who is an operator of the game event place gives an event place name via the control panel of the entity 2. It is assumed that the event place name assigned here is “game”.
[0105]
(2) The entity control unit of entity 2 gives the global entity name “game + entity 2” to the SI-SW generated by the NE factory, as described above. Thereby, an event place is generated. At this time, the entity 2 that is the creator of the game event place automatically participates in the event place when the event place is generated.
(3) The owner of the entity 2 instructs the control panel of the entity 2 to plug in the game application program (SE). At this time, the executable file name of the application to be plugged in and the presence or absence of SE sharing are given as parameters.
[0106]
(4) The control panel of entity 2 notifies the entity control unit of this fact.
(5) The entity control unit of entity 2 stores the name of the executable file to be plugged in, and starts the application (SE) using the given executable file name, and between SI-SW and the application as described above Establish a session on That is, the plug-in in SIONet means establishing a session between the application and SI-SW.
[0107]
(6) The entity control unit of entity 2 advertises (publishes) the entity property of entity 2. Examples of the entity property public form of entity 2 include the following.
Entity 2 participates in the base event place and publishes the entity properties of entity 2 in the base event place.
Entity 2 discovers and participates in an event place that is closely related to the game event place by performing an expression search, and publishes the entity property of entity 2 in the event place.
-Entity 2 is a base entity. That is, the entity properties (entry point and global entity name) of entity 2 are made known.
[0108]
(7) The entity 1 that knows the existence of the entity 2 requests the entity 2 to participate in the game event place.
(8) If this request is approved, entity 1 requests entity 2 to establish a shared link. Then, a shared link is established between entity 1 and entity 2 as described above.
(9) When the shared link is established, the shared link establishment request destination (entity 2) returns information necessary for transmission / reception of the event in the event place to the establishment request source (entity 1). Specifically, for example, an event place name (global entity name), whether or not the event place that participated is a base event place, an event transfer method (routing method) in the event place, a description of the event place, a plug-in SE information (when there is SE sharing) and the like.
[0109]
Specifically, the SE information returned to the shared link establishment request source when sharing the SE share is, for example, an execution file name of the application to be plugged in or a file of the application.
(10) The entity control unit of the entity 1 that has received the file activates the application using the given executable file name as described above, and establishes a session between the SI-SW in the entity 1 and the application. Establish. As a result, the game application program is plugged into the entity 1 as SE.
[0110]
In other words, participation in an event place means that an entity group connected by a shared link is formed, and a participant shares a plug-in-completed application program. As a result, an entity that is a member in the event place can automatically use the plugged-in application simply by participating in the event place. Note that when the entity 1 leaves the game event place, the plugged-in game application program is plugged out. At that time, the entity 1 can continue to hold the received file, or the file may be deleted at the time of plug-out.
[0111]
The operation described above is application sharing in a pure P2P network environment. Therefore, the distribution of the application program is performed only between the two parties, that is, the shared link establishment destination entity and the establishment source entity, and thus does not affect other entities that are already coupled by the shared link. That is, since the processing is localized between the two parties for file transfer, scalable SE sharing can be achieved.
[0112]
[Embodiment of the Invention]
A network construction method (link topology) according to an embodiment of the present invention will be described below with reference to FIG.
In FIG. 1A, E1 to E4 are entities and have semantic information switches SW1 to SW4, respectively. The link topology in SIONet defines the rules for establishing shared links between entities. SIONet supports two topologies as link topologies: open topology and ring topology (closed topology). Below, each link topology is demonstrated.
[0113]
(1) Open topology
In FIG. 1A, numerals (1) to (3) indicate the order of shared link establishment requests. That is, first, the entity E2 makes a shared link establishment request to the entity E1, then makes a shared link establishment request from the entity E3 to the entity E2, and finally, the shared link is transmitted from the entity E4 to the entity E2. This shows a case where an establishment request is made.
Here, SIONet establishes a shared link based on the following link topology rule.
“Each entity can issue a shared link establishment request at most once in the same event place, but establishes a shared link based on a link topology that can accept an unlimited number of establishment requests.”
[0114]
When this establishment request is successful, a shared link is established between the switches (see FIG. 1B) based on the establishment procedure shown in the establishment of the shared link (see FIGS. 14 and 15) described above. The shared link establishment based on the open topology is performed between the establishment request source entity and the request destination entity, and the processing does not affect other entities.
[0115]
In addition, the entity E2 holds a list of the entity E1 that has made a request for establishment, and the entities E3 and E4 that have accepted establishment of the entity E1 (E1 / E3, E4). On the other hand, the entity E1 stores that it has not made an establishment request, that is, that the entity E1 is positioned at the “top” and that it has accepted the establishment request from the entity E2 (− / E2). Similarly, the entity E3 and the entity E4 store that they have issued an establishment request to the entity E2, and that they are not accepting establishment requests from other entities, that is, “bottom”. (E2 /-).
[0116]
By performing such topology management, for example, when the entity E3 makes a request for establishing a further shared link to another entity, the request is rejected as an error, so that In addition to guaranteeing consistency, it is guaranteed that infinite cycles of events will not occur between entities linked by shared links.
For example, when the entity E2 leaves the event place, the entity E3 and E4 can be notified that the shared link re-establishment entity is E1.
[0117]
By using such a topology rule, an open link topology can be easily realized simply by establishing a shared link between adjacent entities. There is one exception to this rule. For example, the entity E3 can establish a shared link with the entity E1 positioned upstream as a countermeasure when the entity E2 falls into a failure. In this case, an infinite circulation of events generally occurs. SIONet has adopted a preventive measure combined with the hop attribute to solve this problem.
[0118]
Another feature of the link topology in SIONet is that all open topologies can be realized in a unified mechanism based on the topology rules described above, in combination with the “maximum number of operations accepted”. For example, as shown in FIG. 25, by setting the number of shared link probabilities that can be accepted for each entity to 1, a bus-type link topology can be realized. In addition, as shown in FIG. 26, a star-type link topology can be realized by combining the acceptable number n entities and the acceptable number 0 entities. Further, as shown in FIG. 27, for example, by setting the acceptable number to 2, a tree (binary tree) type topology can be constructed.
In this case, the maximum number of receptions may be determined using an operating environment attribute that is one of the property items. For example, the maximum accepted number is determined based on the processing capability (CPU power, memory amount, communication bandwidth) of the entity.
[0119]
(2) Ring topology
The ring topology is a well-known closed topology and has a cyclic topology structure. Each entity issues a shared link establishment request, whereby a shared link is established based on a ring topology as shown in FIG.
[0120]
In the ring topology, the semantic information router establishes an event path for unconditional routing when a shared link is established. This only needs to be done once when the shared link is established and does not change until the shared link is re-established. This event path is for realizing a communication method in the broadcast mode. That is, the event sent out by the entity E1 is always transferred to the entity E2, the entity E3, and the entity E4. Further, in this ring topology, an infinite cycle of events occurs. Therefore, the event transmission source entity discards the entity when the event is completed.
[0121]
The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design changes and the like within a scope not departing from the gist of the present invention.
[0122]
【The invention's effect】
As described above, according to the present invention, it is possible to cope with networks having various shapes, and as a result, even if the shape of the network changes, an effect can be obtained that can cope with the system without redesigning the system. .
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a semantic information network system according to an embodiment of the present invention.
FIG. 2 is a diagram for explaining entities in SIONet (semantic information network), which is a prerequisite technology of the present invention.
FIG. 3 is a diagram for explaining entity state transition in SIONet;
FIG. 4 is a block diagram showing an internal configuration of an entity in SIONet.
FIG. 5 is a diagram for explaining generation of an event place in SIONet.
FIG. 6 is a diagram for explaining an operation in which an entity in SIONet participates in an event place.
FIG. 7 is a diagram for explaining a semantic information switch SI-SW and a session in SIONet.
FIG. 8 is a diagram showing a configuration of events in SIONet.
FIG. 9 is a diagram for explaining a semantic information system in SIONet.
FIG. 10 is a diagram for explaining an operation in which an entity participating in a certain event place further participates in another event place in SIONet.
FIG. 11 is a diagram for explaining a service entity plug-in in SIONet;
FIG. 12 is a diagram for explaining a reference model in SIONet.
FIG. 13 is a diagram for explaining a shared link in SIONet.
FIG. 14 is a diagram for explaining a shared link in SIONet.
FIG. 15 is a diagram for explaining a shared link in SIONet;
FIG. 16 is a diagram for explaining an event path in SIONet;
FIG. 17 is a diagram for explaining re-establishment of a shared link in SIONet.
FIG. 18 is a diagram for explaining a hop attribute in SIONet;
FIG. 19 is a diagram for explaining event place and entity increase / decrease in SIONet;
FIG. 20 is a diagram for explaining event place and entity increase / decrease in SIONet;
FIG. 21 is a diagram for explaining entity advertisement in SIONet;
FIG. 22 is a diagram for explaining entity advertisement in SIONet;
FIG. 23 is a diagram showing dimensions of a P2P model.
FIG. 24 is a diagram for explaining the concept of SIONet.
FIG. 25 is a diagram showing an example of a network constructed by a semantic information network construction method according to an embodiment of the present invention.
FIG. 26 is a diagram illustrating an example of a network constructed by a semantic information network construction method according to an embodiment of the present invention.
FIG. 27 is a diagram showing an example of a network constructed by a semantic information network construction method according to an embodiment of the present invention.
FIG. 28 is a diagram showing an example of a network constructed by a semantic information network construction method according to an embodiment of the present invention.
[Explanation of symbols]
E1 to E4 ... entities
SW1 to SW7 ... switch
SL ... Shared link

Claims (5)

In a method of constructing a semantic information network formed by connecting multiple entities via a shared link,
When an entity wishing to participate in the semantic information network applies to an entity that has already participated in the semantic information network, it is checked whether or not it is participation in a plurality of entities. The process of applying to the entities that make up the semantic information network ,
The entity that has received the application for participation checks whether or not the number of entities already accepted at that time has reached a predetermined maximum number of reception, and if not, The process of establishing a shared link for
A process in which an entity participating in the semantic information network maintains a list of entities that have applied for participation and entities that have accepted applications for participation ;
Have a,
A method for constructing a semantic information network, wherein a network is constructed in a bus type, tree type or star type link topology in an open topology by a combination of the maximum number of received entities . When an entity with the maximum number of receptions “1” is combined, a bus-type link topology is formed. When an entity with a maximum number of receptions “2” is combined, a tree-type link topology is formed, and the maximum number of receptions is “n” and “ The method of constructing a semantic information network according to claim 1 , wherein when the entities of “0” are combined, a star-type link topology is formed . Multiple entities Ri name connected via a shared link, the entity that wishes to participate, upon participation application, examine whether or not it is the participation in the multiple entities, meaning if it is not subscribe to multiple entities In the semantic information network system that applies to the entities that make up the information network,
The entity is
Storage means for storing a list of entities that have requested participation and a list of entities that have accepted the participation request;
Shared link establishment means for establishing a shared link for an external entity that has applied for participation when the number of entities that have received a participation request that has been recorded in the list does not reach a preset maximum number of acceptance ,
Have a,
A semantic information network system in which the shared link establishing means constructs a network in a bus type, tree type, or star type link topology in an open topology by a combination of the maximum number of accepted entities . When the shared link establishment unit combines the entities with the maximum number of receptions “1”, the bus-type link topology is used. When the entity with the maximum number of receptions “2” is combined, the tree-type link topology has the maximum number of receptions “ 4. The semantic information network system according to claim 3, wherein when the entities of “n” and “0” are combined, a network is constructed as a star-type link topology . A semantic information network program for causing a computer to function as the semantic information network system according to claim 3.

Images