KR101570933B1 - System and method of node discovery using dynamic bloom filters for dds - Google Patents

Abstract

The present invention relates to a node searching system using a dynamic Bloom filter (DBF) in a data distribution service (DDS) and a method thereof. More specifically, a node stores end point presence information to the DBF having the size of a table dynamically adjusted by using the topic of end points, and broadcasts the DBF information and participant information notifying the presence of the node to a network. Only the nodes matched to the topic of the end points can transmit and receive the information of the corresponding end points. Accordingly, the system and the method according to the present invention enable the node and end point information to be accurately managed and enhance the accuracy of node searches in a large-scale network system such as a battle management system which includes a large number of nodes and frequently adds and deletes the nodes and end points.

Description

TECHNICAL FIELD The present invention relates to a node search system and method using a dynamic bloom filter in a data distribution service,

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a node search system and method using Dynamic Bloom Filters (DBF) in a data distribution service (DDS), and more particularly, The present invention relates to a node search system and method using a DBF in a DDS for searching a node based on a DBF.

A Combat Management System (CMS) is a system that controls and manipulates weapons and various devices mounted on traps by a computer system. The battle management system must manage multiple messages, including monitoring information for multiple devices, and various processes that manipulate them. In addition, since the information processing in the battle management system is triggered by the data collected from the sensor or by the operator's instruction, the communication characteristic of the data center is strong and it is often propagated to a plurality of nodes at the same time. In the battle management system, communication middleware based on data distribution service (DDS) is used for consistency of monitoring large-scale real-time messages and distributed environments.

DDS is a middleware for data-centric publish / subscribe communication between distributed applications. Thus, in DDS, applications interested in a particular topic can send and receive data through participating publications and subscriptions to the topic.

1 is a block diagram showing the flow of data communication in a general DDS.

The DDS is composed of a domain 100, which is a data sharing domain, and a domain participant. The domain participant is divided into a publisher 110 and a subscriber 130 for each piece of data. In one domain 100, a plurality of publishers 110 and subscribers 130 exist and exchange samples of a topic, which is a promised data type. Each of the publishers 110 may have a plurality of data writers 120 for transmitting data and each subscriber 130 may include a plurality of data readers 140 for receiving data . Accordingly, the same topic is used between the issuer 110 and the subscriber 120 communicating. More specifically, data (messages) are transmitted and received between the data transmitter 120 of the publisher 110 and the data receiver 140 of the subscriber 130, which are connected to each other through a topic.

In such a DDS, a Simple Discovery Protocol (SDP) is used in which participants participating in communication (i.e., issuers and subscribers) automatically search each other in order to enable data communication through a topic . The basic search method (SDP), which is a conventional node search method, is roughly divided into two processes: a participant search process and an end point search process (data sender and data receiver). After the participant search process is performed, the end point search process is performed.

The basic search method, which is a conventional node search method, will be described in more detail. Each participant searches the participant search process and the endpoint (data transmitter, data receiver) of the other participant who wants to publish or subscribe to the same topic as himself / End point search process. That is, the new participant informs the network of the new participant's own information, and all participants participating in the network receive the information of the new participant and store it in their local database. The local database stores a list of other participants, their publication information, and a list of subscription information. The new participant then sends all endpoint information of the new participant to all participating participants for endpoint matching with all participating participants. All participating participants receive all the endpoint information transmitted from the new participant and confirm whether there is an endpoint that matches.

In this conventional node search method, each participant has to independently store all the information of all the participants and their end points, and each time a participant is newly created or deleted, and an existing participant's end points are updated The updated information is transmitted to all the other participants. Therefore, the amount of data transmitted and received in the network increases, so that the complexity of the network traffic increases greatly, and the memory usage of each participant increases.

The conventional node search method may be suitable for a small-sized network having a small number of nodes. However, in a large-scale network system having a large number of nodes, such as a battle management system, or a large- There is a problem of increasing the memory usage of each participant, and the node search time is long, so that the reliability of the system is significantly reduced.

Korean Patent Laid-Open No. 10-2012-0048742 (2012.05.16)

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems of the prior art, and an object of the present invention is to provide a method and apparatus for managing existence information of an endpoint using a dynamic bloom filter (DBF) And to provide a node search system and method using DBF in DDS in which the node search is performed quickly and the accuracy of node search is further improved by searching the node based on the dynamic bloom filter.

It is another object of the present invention to provide a dynamic bloom filter in which the size of a table is dynamically changed by storing the existence information of the end point using the topic of the end point and information of the participant information and the dynamic bloom filter, The node searching system using the DBF in the DDS, which reduces the complexity of the network traffic by reducing the number of messages transmitted and received in the node search by transmitting and receiving the information of the corresponding endpoint only by the nodes matching the topics of the endpoints Method.

Another object of the present invention is to increase the number of filters to increase the length of the dynamic bloom filter when the capacity threshold of the dynamic bloom filter is exceeded at the endpoint update at each node, By reducing the length, it is possible to accurately manage the information of nodes and end points in a large-scale network system such as a combat management system including a large number of nodes and frequently adding and deleting nodes and end points, And to provide a node search system and method using DBF in DDS.

According to another aspect of the present invention, there is provided a method of searching a node using a DBF in a DDS according to an embodiment of the present invention includes: generating a dynamic bloom filter by a node and calculating a length of a table of a dynamic bloom filter according to the number of end points of the node itself A dynamic bloom filter forming step of storing the endpoint existence information in the dynamic bloom filter by changing the dynamic bloom filter information and the dynamic bloom filter information including the end point name and the topic name stored in the dynamic bloom filter, Receiving a remote participant information broadcast from a remote node and dynamic bloom filter information of a remote node, the method comprising: a topic search step of searching for a matched topic; and a subscription request message requesting endpoint information of the matched topic To the remote node.

According to another aspect of the present invention, there is provided a node search system using a DBF in a DDS according to another embodiment of the present invention, when a first node joins a new node or updates an endpoint, A first node for updating the first dynamic bloom filter by varying the length and broadcasting the updated first dynamic bloom filter information and the participant information, and upon receiving the participant information and the first dynamic bloom filter information of the first node, And a second node for searching for a matching topic using a topic name included in one dynamic bloom filter information, and for requesting end point information of the matched topic.

The present invention increases the length of the dynamic bloom filter by increasing the number of filters when the capacity threshold of the dynamic bloom filter is exceeded at the time of updating the end point at each node, By using the counter value when storing or deleting the endpoint information in the dynamic Bloom filter, it is possible to accurately manage the update status of nodes and endpoints in real time and to improve the probability of false positives of Bloom filter Thus, the accuracy of the node search is further improved in a large scale network system such as a combat management system including a large number of nodes and frequent addition and deletion of nodes and end points, thereby improving system stability and reliability. .

The present invention manages the existence information of an end point by using a dynamic bloom filter in which the size of a table is dynamically changed and searches nodes and end points based on the dynamic bloom filter to quickly and accurately check whether topic matching is performed on the end point In the case of a network with a large endpoint-to-participant ratio, such as a combat management system, it is possible to save considerable computation time, reduce network latency, Only the matched nodes transmit and receive the information of the corresponding endpoints, thereby reducing the amount of information to be transmitted on the network, thereby reducing network traffic.

1 is a block diagram showing the flow of data communication in a general DDS
2 is a block diagram of a node implemented in a trap of a combat management system to which the present invention is applied
3 is a diagram showing a data structure of a dynamic bloom filter (DBF) according to the present invention;
4 is a flowchart illustrating a method of searching a DDS node using DBFs on a specific node side according to an embodiment of the present invention.
5 is a flowchart showing a detailed method of the DBF forming step S10 of FIG. 4
6 is a flowchart illustrating a node search method of a node search system using a DBF in a DDS according to another embodiment of the present invention.
FIG. 7 is a diagram showing an algorithm implementing the node search method of FIG. 4; FIG.
FIG. 8 is a diagram showing an algorithm implementing a node operation method in the case of deletion of an end point among endpoint existence information management methods using the DBF of FIG. 5; FIG.

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

FIG. 2 is a block diagram of a node implemented in a trap of a combat management system to which the present invention is applied. In the battle management system, all information is transmitted through a Combat System DataBus (CSDB), and an Intelligent Processing Network (IPN) 210, which performs server functions of the system, Function. The ICU (Interface Control Units) 220 controls the sensors and the weapons. A multi-function display control (MFC) controller 230 performs a monitoring function.

The battle management system is responsible for commanding and controlling by a command node, such as IPN 210, receiving and processing the enemy information from the sensor and then delivering the battle command via the ICU 220 to the weapon.

The present invention manages existence information of a node and an end point using a dynamic bloom filter (DBF) that dynamically changes the size of a table by adjusting the number of bloom filters, and based on the dynamic bloom filter, Search.

3 is a diagram showing a data structure of a dynamic bloom filter (DBF) according to the present invention.

A Bloom filter is a probabilistic data structure that is typically used to check whether an element belongs to a set. As shown in FIG. 3 (a), a typical Bloom filter has a bit array structure of m bits in size. The Bloom filter uses k different hash functions to indicate whether there is an input element as a hash result value. Since the size of the table is fixed, the maximum number of keys that the filter can accept must be known in advance. When a key begins to be added beyond the acceptable range of the table, the probability of a false positive, which is determined to be nonexistent, increases rapidly.

As shown in FIG. 3 (b), the dynamic bloom filter according to the present invention increases the size of the table by generating and adding a bloom filter when the acceptable capacity threshold is exceeded, thereby preventing the generation of a positive error will be.

4 is a flowchart illustrating a method of searching a DDS node using a DBF on a particular node side according to an embodiment of the present invention. A method for searching a DDS node on a specific node side according to an embodiment of the present invention includes a DBF forming step (S10) of generating a dynamic bloom filter (DBF) and storing end point information of the node itself in a DBF, A topic search step (S14, S16) of searching for a matching topic when receiving the DBF information of the remote node and the remote participant information from the remote node, a broadcasting step (S12) of broadcasting the DBF information together to the network, (S18, S20) for transmitting to the remote node a join request message for requesting end point information of a topic to be transmitted to the remote node.

The DBF information includes an endpoint name stored in the DBF and a topic name of the endpoint.

5 is a flowchart showing a detailed method of the DBF forming step S10 of FIG.

(Step S31) of storing the endpoint existence information in the active bloom filter and increasing the corresponding counter value, and in the case of adding the end point, if the capacity threshold of the active bloom filter is exceeded, the DBF Storing the endpoint presence information and increasing the corresponding counter value, and if not, storing the endpoint presence information in the active bloom filter and increasing the corresponding counter value (S32) In the case of deletion, endpoint existence information is deleted from the active bloom filter and the counter value is decreased (S34). If the active bloom filter is empty, the empty bloom filter is deleted to reduce the table length of the dynamic bloom filter Step S35.

The DDS node search method using the DBF on the specific node side according to the embodiment of the present invention shown in FIGS. 4 and 5 is applicable to the node search of the combat management system of FIG.

FIG. 7 is a diagram illustrating an algorithm implementing the node search method of FIG. 4, and FIG. 8 is a diagram illustrating an algorithm implementing a node operation method in the case of end point deletion among the end point existence information management method using the DBF of FIG. FIG.

The operation flow of the DDS node search method using a DBF on a particular node side according to an embodiment of the present invention will be described later. First, referring to FIG. 5, when there is an update such as addition or deletion of an end point in a specific node A method of managing presence information of an endpoint using a dynamic bloom filter (DBF) will be described in detail.

In a system to which DDS is applied, for example, when a new node participates in the battle management system's network, the node generates and initializes the DBF. The node generates a bloom filter BF having a capacitance threshold c as shown in FIG. 3 (b) and sets it as an active bloom filter S ₁ (S30).

In the present invention, a node stores its own endpoint information in a database, manages endpoint existence information using a DBF, and can quickly and accurately check whether a specific endpoint exists using the DBF.

The node stores the endpoint existence information in the bit array of the active BF based on the hash function result value using the topic name of the endpoint as a key, and increments the counter value corresponding to the stored bit by one (S31). The topic name is determined to be unique on the network. The endpoint name is configured to include the topic name.

Thereafter, if there is an endpoint addition, the node checks if the capacity threshold of the active BF has been exceeded and, if exceeded, generates a Bloom filter (BF). The node sets the generated BF to the new active BF (S ₂ ) and connects it to the previous active BF (S ₁ ) to increase the length of the table of the DBF (S ₃₂ ). The node stores the presence information for the added end point in the new active BF (S ₂ ) and increments the counter value of the corresponding bit by 1 (S33).

If the key exceeds the acceptable capacity threshold of the DBF table and the key starts to be added, the false positive detection probability to be determined to exist does not substantially exist, which significantly increases the reliability. However, the present invention increases the length of the DBF table by confirming whether the range of the DBF table can be accommodated or not, and then adds the existence information of the endpoint so that node management is performed in a large- And node search can be performed very accurately.

If the endpoint is deleted, the node deletes the endpoint existence information from the DBF using the hash result value of the deleted endpoint, and decrements the counter value of the corresponding bit to be deleted by 1 (S34).

The node deletes the existence information of the end point, checks whether there is an empty bloom filter among the bloom filters constituting the table of the DBF, and if there is an empty bloom filter, deletes the bloom filter to reduce the length of the table of the DBF (S35).

As described above, according to the present invention, since the DBF table length is managed by varying the length of the DBF table according to the number of endpoints to be stored and the number of hash functions, even in a system including a large number of nodes and endpoints, It is possible to accurately manage the information of the node and the end point.

Now, the operation flow of the DDS node search method on the specific node side according to an embodiment of the present invention will be described with reference to FIG.

When the node participates in the network of the battle management system using the DDS, the node generates a dynamic bloom filter (DBF) and stores the endpoint existence information of the node itself in the DBF (S10). If the number of endpoints is large and the current capacity threshold of the DBF is exceeded, the length of the DBF table is increased and the endpoint existence information is stored and the counter value of the corresponding bit is increased.

When the DBF is formed, the node broadcasts the DBF information together with the participant information to the network, and notifies the presence of the DBF to the network (S12). The DBF information includes an endpoint name stored in the DBF and a topic name of the endpoint.

Thereafter, upon receiving remote participant information and DBF information of the remote node from the remote node, the node searches for a matching topic (S14, S16).

If there is a matching topic, the node sends a join request message to the remote node requesting endpoint information of the matching topic (S18, S20). If there is no matching topic, the node does not receive the message sent from the remote node.

As such, the present invention can reduce the size of a message broadcast on a network by transmitting DBF information including an endpoint name and a topic name along with participant information, rather than the entire information of the endpoint.

The present invention also provides a method and apparatus for receiving a message requesting information of an endpoint using a matching topic only when a node receives DBF information together with participant information to easily search for a matching topic, By transmitting to the remote node, only the nodes that match the topic of the endpoint transmit and receive the information of the corresponding endpoint, thereby reducing the number of messages transmitted and received during the node search, thereby reducing the complexity of the network traffic.

6 is a flowchart illustrating a node search method of a node search system using a DBF in a DDS according to another embodiment of the present invention. The node search method of the DDS node search system using the DBF according to another embodiment of the present invention shown in FIG. 6 is also applicable to the battle management system.

As shown in FIG. 6, the node search method of the DDS node search system using the DBF according to another embodiment of the present invention reflects the information of the updated end point when the end point is updated in the first node, The method includes a first step (S50, S52) of updating 1 dynamic bloom filter (DBF) by updating the first DBF information and participant information, updating the first DBF information and the participant information, receiving the participant information of the first node and the first DBF information A second step (S54-S62) of searching for a topic to which the second node matches, and transmitting a join request message requesting endpoint information of the matched topic, and a second step And a third step (S64, S66) of establishing a connection with two nodes and transmitting / receiving endpoint information of a matched topic.

If the capacity threshold of the active bloom filter is exceeded in the case of adding an end point, the first step (S50, S52) increases the length of the table of the DBF, and the existence information of the end point Storing the existence information of the additional end point in the active bloom filter and increasing the counter value (S32) if the end point is not exceeded, and deleting the end point from the active bloom filter (S34) of deleting existence information of the endpoint and decrementing the corresponding counter value; and when the active bloom filter is empty after the existence information of the end point is deleted, the empty bloom filter is deleted to obtain the table length of the dynamic bloom filter (S35).

The first step (S50, S52) further comprises a step of generating a DBF and setting it as an active bloom filter, storing endpoint existence information in an active bloom filter and increasing a corresponding counter value (S31).

According to another aspect of the present invention, there is provided a method of searching a node in a node search system using a DBF in a DDS, the method comprising: a first node storing a name and a topic name of an endpoint stored in a first dynamic bloom filter (DBF) A fourth step (S40, S42) of broadcasting the participant information together with the first DBF information to the network, and a fourth step (S40, S42) of storing the name and the topic name of the end point stored in the second dynamic bloom filter And a fifth step (S44-S48) of broadcasting the participant information to the network together with the second DBF information.

The node search method using the DBF of the node search system in the DDS according to another embodiment of the present invention shown in FIG. 6 is applicable to the node search of the battle management system using the DDS.

Now, a node search method of a DDS node search system according to another embodiment of the present invention will be described in detail with reference to FIG. 5 and FIG.

A combat management system using DDS is configured to include a plurality of nodes to perform a node search by searching for a participant and its end point. Each of the plurality of nodes broadcasts its participant information and DBF information at a predetermined period to inform its existence. Also, a newly participating node broadcasts its participant information and DBF information to inform its existence.

For example, the first node broadcasts the participant information to the network together with the first DBF information including the name of the endpoint and the topic name stored in the first dynamic bloom filter (DBF) (S40).

The second node receiving the participant information of the first node and the first DBF information retrieves whether there is a topic matching the topic of the first node. If there is no matching topic, the end node search procedure of the first node is not performed (S42, S44).

The second node broadcasts the participant information to the network together with the second DBF information including the name of the endpoint and the topic name stored in the second dynamic bloom filter (DBF) (S46).

The first node receiving the participant information of the second node and the second DBF information also searches for a matching topic (S48).

Thereafter, when the endpoint is updated at the first node, the length of the first dynamic bloom filter DBF is updated by reflecting information of the updated endpoint.

For example, if endpoints Q, R, S, T, U, V, and W are generated at the first node, then the first node generates the endpoints Q, R, S, T, U, V, (For example, the endpoint existence information, which is a hash result value having the key Q of the end point Q as a key, is stored in the first DBF). If it is determined that the capacity threshold of the first DBF has been exceeded before storing in the first DBF, a new BF is generated to increase the table length of the first DBF. Then, . If existence information of all end points thus generated is stored in the first DBF, the first node broadcasts the updated first DBF information together with the participant information to the network. The first DBF information includes names and topic names of endpoints stored in the first DBF (S52).

The second node receiving the participant information of the first node and the first DBF information retrieves whether there is a topic matching the topic of the first node. When the matched topics Q, R, and S are retrieved, the second node sends a subscription request message (Subscribe Q) requesting information (e.g., endpoint address, endpoint port, etc.) of the endpoint Q using the topic Q, (Subscribe R) for requesting information of the end point R using R and a subscription request message (Subscribe S) for requesting information of the end point S using the topic S to the first node (S54 -S62).

The first node receiving the join request messages establishes a connection with the second node and transmits information of the matching Q, R and S endpoints Q, R and S (S64, S66).

100: data domain 110: issuer
120: Data transmitter 130: Subscriber
140: Data receiver 210: IPN
220: ICU 230: MFC

Claims (8)

A node generates a dynamic bloom filter in which one or more bloom filters are sequentially connected to one another, and the length of the table of the dynamic bloom filter is varied by additionally connecting or removing the bloom filter as the number of end points of the node is changed A dynamic bloom filter forming step of storing end point existence information indicating end points existing in the node in the dynamic bloom filter;
Broadcasting the dynamic bloom filter information including the participant information and the names of the end points and the topic name to the network to inform the existence of the node itself;
When the node receives the remote participant information broadcast from the remote node and the dynamic bloom filter information of the remote node, a topic search that searches for a matching topic using the topic name included in the dynamic bloom filter information of the remote node step; And
And if there is a topic matching the node, sending a join request message to the remote node, the node requesting information about the endpoint of the remote node related to the matching topic,
Wherein the Bloom filter is a counting bloom filter.
2. The method of claim 1, wherein the dynamic bloom filter forming step comprises:
A bloom filter is set as an active bloom filter of the dynamic bloom filter, endpoint existence information based on the hash result value using the topic of the end point of the node as a key is stored in the active bloom filter, and the corresponding counter value is increased step; And
In the case of adding an end point, if the capacity threshold of the active bloom filter is exceeded, the length of the table of the dynamic bloom filter is increased by creating and connecting a bloom filter, storing the existence information of the added blob, increasing the corresponding counter value, Storing the presence information of the endpoint in the active bloom filter and increasing the corresponding counter value if the capacity threshold is not exceeded.
3. The method of claim 2, wherein the dynamic bloom filter forming step comprises:
Removing endpoint presence information based on a hash result value using a topic name of an endpoint to be deleted as a key from the active bloom filter and decrementing a counter value of the deleted bit when the endpoint is deleted; And
And deleting the empty bloom filter to reduce the table length of the dynamic bloom filter when the active bloom filter becomes empty after the deletion, wherein the method further comprises the step of reducing the table length of the dynamic bloom filter.
The first dynamic bloom filter is formed by sequentially connecting one or more bloom filters and the first dynamic bloom filter is reflected by reflecting the information of the updated end points, A first node that broadcasts first dynamic bloom filter information and participant information that is variable in length and updates and includes an endpoint name and a topic name to inform the existence of the node itself; And
Upon receipt of the participant information of the first node and the first dynamic bloom filter information, it is checked whether there is a matching topic using the topic name included in the first dynamic bloom filter information, and if there is a matched topic, And a second node requesting the first node for endpoint information of the first node related to the first node,
Wherein the Bloom filter is a counting bloom filter.
5. The method of claim 4,
And upon receiving the endpoint information request, establishes a connection with the second node and transmits the endpoint information of the matched topic, in the data distribution service using the dynamic bloom filter.
5. The method of claim 4,
In the case of adding the end point, it is checked whether the capacity threshold value of the active bloom filter constituting the dynamic bloom filter is exceeded,
If the capacity threshold value is exceeded, the newly generated Bloom filter is connected to increase the length of the table of the dynamic Bloom filter to store the existence information of the added end point and increase the corresponding counter value,
A node search system using a dynamic Bloom filter in a data distribution service that stores presence information of an endpoint added to an active bloom filter and increases the counter value if the capacity threshold is not exceeded.
5. The method of claim 4,
In case of deletion of the end point, the existence information of the corresponding end point is deleted from the active bloom filter constituting the dynamic bloom filter, the corresponding counter value of the bit to be deleted is decreased,
A node search system using a dynamic Bloom filter in a data distribution service that removes an empty Bloom filter and reduces the table length of the Dynamic Bloom filter when the Active Bloom filter becomes empty after the deletion.
5. The method of claim 4,
When the second node newly joins or when the end point is updated, the second dynamic bloom filter is updated by changing the length of the second dynamic bloom filter by reflecting information of the updated end point,
And the second dynamic bloom filter information including the name of the endpoint and the topic name stored in the updated second dynamic bloom filter together with the participant information.

Images