KR100243391B1 - Method for controlling multi-cast graph - Google Patents

Abstract

The present invention is to enable multicast transmission of a message to form a group of nodes on the network to receive the same message as a group and to send a message to the technology to implement the group communication based on all currently provided protocols The present invention relates to a control method for generating a multicast graph that supports message multicast communication. As described above, the present invention is based on a technique for performing message multicast, and by using the acquired technique, it overcomes the drawbacks of one-to-one and one-to-many communication techniques, and efficiently uses network resources. Can be used as In the multicast communication technology, the multicast scheme according to the present invention will greatly contribute to the development of new communication applications, particularly the use of message multicast such as a multiparty conferencing system.

Description

CONTROL METHOD FOR CREATING MULTICAST GRAGH WHICH SUPPORT MESSAGE MULTICASE COMMUNICATION

The present invention relates to a control method for generating a multicast graph that supports message multicast communication. Particularly, as a part enabling multicast transmission of a message, nodes on a network that want to receive the same message are formed in a group. The present invention relates to a control method for generating a multicast graph that supports message multicast communication to implement group communication based on all protocols currently provided as a technology for sending a message to it.

In general, a method for multicast transmission of a message has been mostly reconfigured with a new phase for only multicast transmission of a message with nodes wishing to communicate on the Internet, and multicast transmission of a message using the message. Depending on the topology used, message multicast routing and the accompanying techniques have been proposed.

At present, group communication has been proposed as the best communication model for multicast support of messages. Group communication forms a group of nodes that receive the same message, so that if a message is sent only to a node that sends a message or to one node in a group that receives the message, no matter how many nodes it needs to receive, By sending a message, you can reach the node you want to receive.

Without using the concept of group communication, if you send a message to all nodes that want to receive a message, the node that sends the message either sends it in a one-to-one manner to all nodes that you want to receive, or if it exists on the network. Sending a message to all nodes wastes a lot of network resources.

The group communication proposed in the direction to improve the above problem is that the node which wants to send a message to one group must be a member of the receiving group, even if it is not an element of the close group and the receiving group. It can be divided into open groups that can send messages.

Group communication using closed groups can selectively select nodes that participate in the transmission and reception of messages on the network, which provides the advantage of natural security, while identifying sender and receiver nodes that prevent fast, real-time message transmission. It has the disadvantage of including the process of opening, and the open group is a group form in which the advantages in the closed group are disadvantages, and the disadvantages are advantages. The group type that is widely used is also important for security on the network, but the open group rather than the closed group is preferred because fast message transmission in real time is required, and much research is being conducted.

A message multicast system using a virtual topology based on an open group includes a system using a propagation tree virtual topology. In a system using this propagation tree, destination nodes that want to receive messages can be included in multiple multicast groups. Nodes included in multiple multicast groups are formed into a multicast group, that is, a meta group without overlapping, and then formed into a single phase to select a message path using the formed meta group.

The shape of the formed phase forms a "propagation tree" structure. The structure of the propagation tree has a small amount of traffic concentrated in the root group, which is a disadvantage of the tree structure, but it goes through a dummy group that does not have to go through sending a message to the destination group. It is a structure.

There is a process of optimizing the configured virtual topology to reduce the delay of these messages, and attempts to set up route assignments through as few dummy groups as possible. Although the optimization process can achieve fast message transmission through a small number of dummy groups, this optimization process is not valid for all message transmissions, but is used only for a specific message.

Therefore, the propagation tree initially configured according to the message to be transmitted may be used as it is, and sometimes the virtual topology optimized for the initially formed propagation tree should be used, resulting in a decrease in system efficiency.

An object of the present invention is to generate a multicast graph with a virtual topology used in a message multicast system, and do not need to newly optimize the virtual phase according to a transmitted message, and to provide information on an alternative path when a defect occurs due to message transmission. If a new multicast group is created or a new multicast group is added by splitting an existing multicast group, or nodes not included in the multicast group are generated between the transmitting node and the receiving node, they already exist. The present invention provides a control method for generating a multicast graph that supports message multicast communication to generate a virtual phase that minimizes the propagation delay caused by message transmission by maintaining the virtual phase while minimizing the change of the virtual phase. .

Means for achieving the object of the present invention comprises the first step of determining the cardinality for each meta group after forming a meta group without overlap of nodes from the multicast group, and determined in the first step A second step of selecting a meta group connected by using the concentration and the tree trunk, and a meta group having a concentration smaller than the concentration among the meta groups including the concentration determined in the first step and an exclusion relationship with the concentration. A third step of selecting a meta group having a meta group selected by the concentration determined in the first step, the meta group connected to the concentration and the tree trunk, and connected to each other by a hyper trunk to move from the concentration at a time; Step 4, and all meta-groups formed by repetitive work are included in the virtual topology. It comprises a fifth step of repeating the algorithm.

1 is a system configuration for supporting multicast communication in accordance with the present invention.

2 is a block diagram of a message multicast system protocol used in the present invention

3 is a multicast graph generation algorithm used in the present invention.

4 is an exemplary diagram of a multicast graph generated using a multicast graph generation algorithm provided by the present invention.

<Explanation of symbols for main parts of drawing>

11a through 11n: system environment 12: user interface

13: Message Multicast Routing Table 14: Network Systems

15: network 201: application

202: group communication 203: TCP / IP

204: broadcast network

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

1 is a block diagram of a message multicast communication system according to the present invention. The multicast graph generator that supports the message multicast method of this structure operates in a distributed environment, and in order to efficiently transmit multicast messages by any one transmitting node under the distributed environment, each user system environment (11a to 11n) is used. ) Uses the multicast graph 13.

In a distributed environment, a multicast system that transmits the same message to multiple meta groups based on the multicast graph information 13 provides information on adjacent meta groups. Adjacent meta-group information is provided only by the representative node of each meta-group, and transmits a multicast message with information connected to the hyper trunk and tree trunk.

Therefore, there is a difference between the information that should be present in the representative node and other general nodes in each meta group. General nodes included in the meta group need only have a connection with the representative node of each meta group. But this does not mean a physical direct connection. These environments 11a to 11n provide a system capable of transmitting the same message to multiple nodes as a system supporting message multicast.

First, the user interface part 12 serves as a link between a user who uses the system and the system, and the user recognizes the order of currently received messages rather than providing information about the virtual topology used in the network. It consists of parts that can create new messages.

In a multicast system that supports message multicast, the meta group containing the current node is connected to one meta group and a hyper edge in a virtual topology generated for all nodes in the network, and to another meta group using a tree edge. It is a part that stores the route assignment table 13 which tells whether or not it is, and sends the message generated by the user to the network system 14 and shows the received message from the network system on the user interface 12. A state plays a role.

The network system 14 is responsible for sending messages to all nodes connected to the network 15. The network system 14 is a representative of a meta group including nodes, rather than a part having information about all nodes connected to the entire network. It has information about the node and connects to this representative node only. In this system 14, the reliability of the message delivery can be ensured by using a network that guarantees the reliability of the message.

FIG. 2 is a message multicast system protocol block diagram used in the present invention, in which the contents of FIG. 1 are expressed in a network protocol dimension.

The broadcast network 204 at the bottom of FIG. 2 represents a physical network connection and is used as a path for delivering a message generated at one node to a destination.

The portion 203 immediately above uses a protocol such as TCP / IP 203 as the portion for sending messages over the configured physical network. This portion 203 is responsible for the reliability portion of the message that group communication should have. This protocol also guarantees the reliability of the transmitted message as part of sending the message originating from the message sending node to the desired destination through the physical communication network.

The reliability portion of the message, which is a part of the group communication technique used in the present invention, should be implemented at a higher level 202 than this level, or include a protocol such as TCP / IP to support multicast communication.

The next protocol layer 202 is the most important part for multicast message transmission. This layer is a part that delivers a message received through the layer to the application 201 on the upper side. Through communication with all nodes in all networks, all data necessary for generating a multicast graph, which is a virtual topology used in the present invention, may be transmitted.

All nodes formed as meta groups select a representative node as described above. There are many papers on how to select representative nodes. This part also accepts the process of receiving a message at the TCP / IP layer 203 by a receive primitive, and delivers it to the application 201 by a delivery primitive.

This series of procedures ensures the reliability of the message in group communication 202. When the representative node of the meta group receives a signal that all nodes have received a message delivered to each node by the receiving primitive, the application moves to the forwarding primitive. At this time, if there are nodes in the meta group that do not receive the message, the reception of the message itself is canceled. This characteristic is called atomicity or all-or-nothing.

On the contrary, in order to transmit a message generated by the application 201 to other nodes, the message generated by the application 201 is first transmitted to the group communication 202 using an order primitive. Sending this back from the group communication layer 202 to the TCP / IP layer 203 uses a send primitive.

When transmitting to the representative node of the meta group by this primitive, a time stamp is attached and transmitted. The order of the messages is also determined by the group communication layer 202. The receiving primitive is responsible for transmitting to all nodes in the metagroup in the order of the messages arriving at the group communication layer 202 of the representative node. The application layer 201, which is the top layer of FIG. 2, is a layer that delivers a message arriving at the lower layer 202 to the user and a message generated from the user to the lower layer 202. The performance of the message multicast system may vary depending on the method implemented in this layer 201.

3 illustrates a multicast graph generation algorithm used in the present invention.

This algorithm forms nodes connected to the network into meta groups without overlapping, and generates a virtual topological multicast graph composed of the meta groups thus formed.

The first step is to form a meta group with no duplication of nodes from the multicast group and then determine the cardinality for each meta group. The concentration is the number of multicast groups involved to form a meta group. Every meta group has an integer concentration value, and the meta group having the highest concentration value is defined to be the root of the virtual phase multicast graph. The algorithm defines this as mG (301).

The next step is to select the meta group connected by using the mG and the tree trunk (302). Among meta groups including mG, meta groups 303 and 304 having a smaller concentration than mG and meta groups having an exclusion relationship with mG are selected (305 and 306). The meta group which is betrayed with mG is a meta group having the same concentration as that of mG. Next, the meta groups selected by mG are connected to mG and tree trunks (307), and to each other by hyper trunks (308). This selects meta groups that can be moved at once from mG, and then repeats the algorithm until all meta groups formed by repetitive tasks are included in the virtual topology. The method of selecting another mG for repetition is as follows (309). The meta group with the highest concentration is selected among the meta groups connected by the tree trunk.

However, this condition can make mG difficult in this algorithm, which selects metagroups in a manner similar to the breadth-first search. So, the algorithm assigns a vector value to every meta group for selecting a new mG, whose role is a vector indicating which meta group the selected meta group is selected by the iteration of the algorithm during execution. This is to select the meta group having the smallest vector value first in selecting a new mG.

If two or more meta groups having the same vector value exist, the meta group having a large concentration is selected as in the method of selecting mG for the first time, and even if the same meta group exists, the selection is performed in any order. All of these vector values are initially set to '0'. Therefore, if there is a meta group whose value of this vector is '0' during the execution of the algorithm, it means that there is still a meta group that is not included in the virtual phase.

If the vector values are not all '0', it means that the virtual phase for message multicast is completed. At the end of the virtual phase thus formed, there is a meta group having the smallest concentration of meta group. If a meta group having a concentration of '1' is formed, the meta group is a prototype of the multicast group used to form the meta group. to be.

When the message is transmitted using the formed virtual phase, the meta group at the end of the virtual phase becomes the starting point of the message transmission.

4A and 4B illustrate exemplary diagrams of a multicast graph generated using a multicast graph generation algorithm provided by the present invention.

Given the following multicast group, it is possible to form a meta group and use the formed meta group to create the following multicast graph. The given multicast group is formed of a multicast group consisting of A, B, C, and D, as shown in Fig. 4A. This multicast group consists of groups that must receive the same message.

There are multiple nodes in each multicast group, and these nodes can be elements of more than one multicast group.

As shown in (b) of FIG. 4, meta groups formed from the multicast groups shown above include <ac>, <cd>, <ab>, <ad>, <bd>, <acd>, and <abd>. , <a>, <b>, <c>, and <d>.

According to the present invention, a base technology for performing message multicast is acquired, and the use of the acquired base technology overcomes disadvantages of one-to-one and one-to-many communication technology and efficiently utilizes network resources. Can be. In the multicast communication technology, the multicast scheme according to the present invention will greatly contribute to the development of new communication applications, particularly the use of message multicast such as a multiparty conferencing system.

Claims (5)

Determining a cardinality for each meta group after forming a meta group with no duplication of nodes from the multicast group; A second step of selecting a meta group connected by using the concentration determined in the first step and the tree trunk; A third step of selecting a meta group having a concentration smaller than the concentration and a meta group having an exclusion relationship with the concentration among the meta groups including the concentration determined in the first step; A fourth step of connecting meta-groups selected by the concentration determined in the first step to the concentration and the tree trunk, and selecting meta-groups that can be moved from the concentration at once by connecting each other to the hyper trunk; And a fifth step of continuously repeating the algorithm until all metagroups formed by the repetitive tasks are included in the virtual topology. The method of claim 1, And the concentration in the first step is the number of multicast groups that participated in forming a meta group. The method of claim 1, The meta group having a betrayal relationship with the concentration in the third step is a meta group having the same concentration as that of the concentration determined in the first step. Control method. The method of claim 1, The fifth step includes a first step of assigning a vector value to all meta groups for selecting a new concentration; If there is more than one meta group having the same vector value, a meta group having a large concentration value is selected, and if there is a meta group having the same concentration value, the second step of selecting in any order is a message. Control method for generating multicast graphs that support multicast communication. The method of claim 4, wherein The vector value allocated in the first process indicates whether the selected meta group is a meta group selected by the repetition of the execution of the algorithm, and for generating a multicast graph supporting message multicast communication. Control method.

Images