JP2004072534A - Method for calculating multicast transfer path - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To set a transfer path for reducing inter-user delay variation in a network consisting of two hierarchies that are an access network and a core network. <P>SOLUTION: An edge node holds intensive information about delay values in the access network and distributes a delay value to a node that performs a multicast path calculation. A node in the core network notifies a node that calculates a multicast transfer path of information obtained by measuring a delay which passing information suffers. The relevant node holds two pieces of delay information, and in the calculated transfer path, a multicast transfer path for reducing the difference between receivers each other of a delay which a sender and a receiver suffer is calculated with the held delay information. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a multicast transfer path calculation method, and more particularly, to a multicast transfer path calculation method in a core network connecting an access network accommodating terminals and an access network in multicast communication in a case where networks are hierarchically configured.
[0002]
[Prior art]
2. Description of the Related Art Multicast communication for delivering moving images and audio to a specified number of users on a computer network has attracted attention. In this communication method, information is copied at a portion where a route is divided among routes connecting a sender and each recipient, and information is distributed to each recipient. When information is delivered using unicast communication in which one-to-one communication with a sender is performed to a specified number of recipients, the sender needs to prepare information for the number of recipients. Therefore, by using the multicast communication, the amount of information in the network is reduced. In multicast communication, a specified number of recipients are managed in a management unit called a multicast group, and one transfer path is set for the multicast group. This transfer path is set so as to connect all the receivers belonging to the multicast group from the sender. Further, a user who wants to acquire information transferred to a certain multicast group acquires information by joining the multicast group. Therefore, the transfer route changes according to the user's participation status.
[0003]
Applications using multicast communication include video conferencing, online games, and moving picture distribution of movies, televisions, and the like. In a video conference or an online game, a plurality of users are data receivers, and are also senders to return responses to data received at the same time. In such an application, attention has been paid to a technique for making each user's utterance opportunity equal by making each user's response time to data transmitted by a certain user approximately the same.
[0004]
As a means for realizing the equalization of the response time, there is a means for designing a transfer route of information from a sender to each receiver so that all the receivers suffer the same delay. Here, the difference in the transfer delay of the information incurred on the transfer path when the information reaches each recipient is called inter-user delay dispersion. Currently, papers dealing with this problem include: 1) "G. Rouskas, et al.," Multicast Routing with End-to-End Delay and Delay Variation Constraints, "IEEE Journal, Selected Journals, Essential Information, Sealed Relations, and more. 1997), 2) "Pi-Rong Sheu, et al.," A Fast and Efficient Heuristic Algorithm for the Delay-and Efficient Heuristics Algorithmic Dimensional Day-to-Date Delivery Date- 001 "have been published.
The above-mentioned paper 1) is the first to solve this problem from the viewpoint of selecting a transfer path, and the paper 2) proposes a method that requires less computation than the method proposed in the paper 1). , Reducing the calculation time of the route.
[0005]
In these papers, it is an object of the multicast transfer path, which is a calculation target under the situation where the inter-user delay dispersion is minimized, to minimize the total sum of the costs assigned to the links constituting the path. The algorithm proposed in the above-mentioned paper operates under a situation where the entire network configuration up to the node receiver which calculates the route for transferring the multicast information is understood. 2. Description of the Related Art Currently provided computer networks are roughly classified into two layers, an access network accommodating users and a core network connecting the access networks. The core network and the access network are connected by a node called an edge node. When information is transferred from one access network A to another access network B, the information passes through an edge node connecting the access network A and the core network. Then, the data is transferred via the core network via an edge node existing between the access network B and the core network.
[0006]
The edge node also has a function of relaying information flowing on the above-mentioned transfer path when it becomes a part of a transfer path formed between other edge nodes.
[0007]
The core network includes an edge node and a relay node that only relays data.
[0008]
[Problems to be solved by the invention]
However, the above conventional technique has the following problems.
[0009]
Hereinafter, the nodes in the core network include an edge node and a relay node. A node that performs transfer route calculation and exists in each access network holds only configuration information of the network to which the node belongs. The nodes existing in the core network hold information obtained by aggregating the configuration information of each network, and do not hold detailed network configuration information. This is because it is difficult for each node to hold information on all nodes and links because the number of nodes and links constituting the entire computer network is large. For this reason, it is practically difficult for all nodes that calculate the transfer route to know the network configuration. The present invention has been made in view of the above points, and is provided from an access network in a network configured by two layers, an access network and a core network, under a situation where aggregated information regarding transfer delay in each access network can be used. It is an object of the present invention to provide a multicast transfer path calculation method capable of setting a transfer path for reducing delay dispersion between users in a core network based on the aggregated information.
[0010]
It is another object of the present invention to provide a multicast transfer path calculation method capable of realizing a minimum delay required for a calculated path to be transmitted from a sender to all receivers.
[0011]
A further object of the present invention is to provide a multicast transfer path capable of calculating a transfer path such that the sum of costs assigned to nodes and links constituting a multicast transfer path as a calculation target is as small as possible. The purpose is to provide a calculation method.
[0012]
[Means for Solving the Problems]
The present invention resides in a hierarchical computer network in a multicast transfer path calculation method in a core network connecting between an access network accommodating terminals and an access network in multicast communication when the network is hierarchically configured. When a node calculates a transfer path used for delivery by multicast communication for delivering information, an edge node connecting an access network and a core network, a terminal of a receiver existing in the access network connected to itself. Measure the delay incurred by the route through which information passes when delivering to each receiver terminal,
From the measured delay value, calculate the delay value incurred in the access network,
Hold as aggregated information about the calculated delay value,
Deliver the calculated delay value to the node that performs the multicast route calculation,
At the nodes in the core network, measure the delay experienced by the information passing through it,
Notify the measured information to a node having a function of calculating a multicast transfer path,
In a node having a function of calculating a multicast transfer path, holding two pieces of delay information distributed from an edge node and a node in a core network,
The calculated transfer path calculates a multicast transfer path that reduces the difference between the receiver terminals and the delay incurred between the sender terminal and the receiver terminal based on the held delay information.
[0013]
Further, the present invention provides an edge node for connecting an access network and a core network,
When delivering to a terminal of a receiver existing in an access network connected to the edge node itself, a delay incurred in a route through which information is passed is measured for each terminal of the receiver,
This is stored as aggregated information regarding the maximum value of the measured delay values.
[0014]
Further, the present invention provides an edge node for connecting an access network and a core network,
When delivering to a receiver terminal existing in an access network connected to the edge node itself, a delay incurred in a route through which information passes is measured for each receiver terminal,
The average value of the measured delay values is held as aggregated information regarding delay.
[0015]
Also, the present invention provides a node for calculating a multicast transfer route,
The delay incurred from the egress edge node, which is the egress to the access network in which the terminal of the receiver of the multicast group exists, to the ingress edge node, which is the ingress from the access network in which the terminal of the sender transmitting the multicast group exists. Calculating the minimum value for each of the exit edge nodes;
Each exit edge node distributes, sums up the aggregate delay information in the access network connected to the own node and the minimum value of the measured delay, searches for the minimum value,
It is determined that the multicast transfer path for setting the found minimum value is a delay to be realized.
[0016]
Also, the present invention calculates a transfer path for a multicast transfer path connecting a terminal of a sender and an egress edge node for each egress edge node,
The calculated delay value to be realized by the multicast transfer path minus the aggregated information on the delay of the multicast group in the access network connected to the egress edge node, plus the delay difference between the terminals of the allowed receivers Is set as the upper limit, the subtracted value is set as the lower limit, and the route that realizes the delay value between them is held as a transfer route candidate,
From the held candidates, a candidate having the largest number of links overlapping with the route selected as the transfer route at the time of selection is selected as the transfer route.
[0017]
As described above, the multicast transfer path setting method of the present invention is for realizing reduction of delay dispersion between users in a network having a hierarchical structure, and the path calculated by the present invention is used in an access network. From the calculated maximum delay between the edge node of the network and the receiver of the multicast group to be calculated existing in the network and the delay recorded by the node in the core network and suffering information by itself. The first feature is that it is calculated to reduce transfer delay between users.
[0018]
In addition, the transfer delay from the edge node connected to the access network in which the sender exists to the receiver farthest from the edge node among the receivers participating in the multicast group is a target to be achieved by the information sent to the corresponding multicast group. The second feature is to use a delay value.
[0019]
Further, a third feature is to adopt a route having the least number of links through the routes found.
[0020]
The present invention is significantly different from the above-described conventional technology in that the target network is changed from a single network to a hierarchical network. For this reason, a transfer path that reduces the delay dispersion between users without maintaining the positions of the nodes of the entire computer network and the connection relationships of the links, which need to be known in the related art in order to reduce the dispersion delay between users. Is very different.
[0021]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0022]
FIG. 1 is a diagram of a network configuration in a mesh form.
[0023]
The present invention is directed to interactive applications, such as video conferences, online games, stock information, etc., that interact with each other as in normal dialogue. When implementing such an application, a transfer path is set for one sender's terminal toward a specific number of receivers' terminals.
[0024]
The shape of this transfer path is called point-to-multipoint. By setting such a transfer path for each application user, these applications are realized. Here, this form is called a mesh form.
[0025]
FIG. 2 is a network configuration diagram of a route sharing mode.
[0026]
The network shown in FIG. 1 is realized by collecting information from each user, setting a transfer path as shown in FIG. 2 from a server that manages the information, and delivering it to each user. In this case, the multicast transfer path is set from the information management server to the terminal of each receiver. Here, this mode is called a route sharing mode.
[0027]
Such an application is realized on a computer network. The computer network includes an access network that accommodates users and a core network that connects the access networks. FIG. 3 shows such a network configuration.
[0028]
The information delivered to each receiver terminal includes, from the sender terminal, an edge node 30A and a core network 20 connecting the access network 10A and the core network 20 to which the sender terminal belongs, and furthermore, the core network 20 and the receiver Is delivered to the receiver's terminal via the edge nodes 30B, 30C, 30D connecting the access networks 10B, 10C, 10D to which this terminal belongs. Of the edge nodes 30 shown here, the former (edge node 30A) is called an entrance edge node, and the latter (edge nodes 30B, 30C, 30D) is called an exit edge node.
[0029]
In multicasting, since there are a plurality of receiver terminals, the transfer route has a tree structure. At such a branch point in the tree structure, the delivery information is copied and delivered to each recipient terminal.
[0030]
As a method of reducing the transfer delay dispersion between users for such a transfer mode, there is a method of making the delay incurred on the transfer path of the multicast group to be calculated to be almost the same for each receiver terminal. These transfer routes are calculated by the nodes constituting the core network, particularly, the ingress edge node for the multicast group, or a node dedicated to route calculation prepared in the core network or the access network. The calculation result is notified to the node that sets the transfer route. Examples of the node to which the transfer route is set include an ingress edge node and an egress edge node, or a dedicated node for instructing the setting of a node existing in the core network or the access network. The node for performing the route calculation and the node for setting the route are determined depending on the management policy of the network administrator.
[0031]
In order to reduce the delay dispersion between users, it is necessary for the node that performs the route calculation to grasp the transfer delay from the terminal of the sender to the terminal of each receiver. The forwarding delay is the delay from the sender's terminal to the ingress edge node, the delay from the ingress edge node to the egress edge node to the access network to which each recipient's terminal belongs, and the delay from the egress edge node to each recipient's terminal. It consists of a delay and a delay from the egress edge node to the terminal of each receiver. Among them, the delay related to the delay difference occurring between the terminals of each receiver is the delay from the ingress edge node to the egress edge node to the access network to which the terminal of each receiver belongs, and the delay from the egress edge node to the terminal of each receiver. Until the delay. In the present invention, a route is calculated in consideration of these two delays.
[0032]
First, consider the delay to the egress edge node to the access network to which each receiver's terminal belongs.
[0033]
The geographical scale of the access networks, as well as the user density, are different for each access network. In addition, the locations of the receiver terminals belonging to the same access network are different for each receiver terminal. For this reason, the delay required to transmit information from the egress edge router to each recipient terminal differs for each recipient terminal.
In the present invention, these different delay information are aggregated into one piece of information, and the information is notified to a node that calculates a route in the core network.
[0034]
FIG. 4 is a diagram for describing aggregated information according to an embodiment of the present invention.
As this aggregated information,
Case 1) Delay from the egress edge node to the terminal of the receiver that suffers the greatest delay;
Aggregation delay d (k) = Max {d (Ri)}
Case 2) Average value of delay values incurred by information transmitted from the egress edge node to each receiver terminal;
Is raised.
[0035]
Here, Ri is a calculation target (participant of the multicast group) belonging to the access network a, and d (Ri) indicates a delay from the egress edge node to the receiver terminal Ri.
[0036]
FIG. 5 is a diagram for explaining an information collection mechanism according to one embodiment of the present invention. In the figure, Ri is a calculation target (participant of the multicast group) belonging to the access network, D (Ri) is a delay from the egress edge node to the receiver terminal Ri, and mi is the receiver's terminal Ri. Indicates the time when the information arrived at the terminal Ri.
[0037]
The egress edge node manages this information about the access network to which it is connected. As a mechanism to collect information,
1) Existing route calculation protocol (routing protocol);
2) Information collection mechanism prepared by the network administrator;
And the like.
[0038]
As an example of the above 2), a time stamp is provided in the information to be delivered, the time when the information passed through the egress edge node is recorded in the time stamp, and the transfer delay from the egress edge node is measured at each receiver terminal. There is a method of transmitting measurement information directly to the exit edge node.
[0039]
The collected aggregate information notifies the exit edge node that is present in the core network and performs the route calculation. As a method of notifying such information,
1) Existing route calculation protocol (running protocol);
2) Information mechanism prepared by the network administrator;
And the like.
[0040]
Next, consider the delay from the ingress edge node to the egress edge node.
[0041]
In order to equalize the transfer delay from the sender's terminal to the receiver's terminal joining the multicast group to be transmitted, the ingress edge node to the egress edge node depend on the aggregated information about the delay collected in the access network. The set target delay up to must be set for each egress edge node. An algorithm for calculating the set delay will be described below.
[0042]
First, the node that performs route calculation checks the minimum value I (k) of the delay required to deliver information from the ingress edge node to each egress edge node k for the multicast group to be calculated. Information necessary for this investigation includes information on delays that information suffers at each node. Currently existing such nodes have the ability to measure this delay. Therefore, as a method of investigating the minimum value of the delay to each exit edge node,
1) A node that performs route calculation collects information on delays caused by each node from nodes existing in the core network independently;
2) The information collection node collects this information and notifies the node that performs the calculation;
Is raised.
[0043]
FIG. 6 is a diagram illustrating a method of calculating a delay value according to an embodiment of the present invention. A calculated value is obtained by adding the calculated minimum delay I (k) for each egress edge node k and aggregated information of transfer delays held by the edge node and related to the multicast group in the access network under its control. Of these calculated values, the smallest one is the minimum value of the delay value required for transmission to the terminals of all recipients. Therefore, this value is set as the delay value Δ to be realized by the multicast transfer path. That is,
Δ = min {d (k) + I (k)}
It becomes.
[0044]
By setting the delay value Δ that is the target to be achieved in the multicast transfer path, the target delay value δ (k) that is incurred in transferring information from the ingress edge node to each egress edge node is calculated. That is,
δ (k) = Δ−d (k)
It becomes. The node performing the route calculation searches for a route having a transfer delay of δ (k) from the ingress edge node to the egress edge node. However, since it is difficult to find a path having a transfer delay equal to δ (k), the delay value to be satisfied by the transfer path is δ (k) −α ≦ ΣD (i) ≦ δ (k) + α
A route that satisfies is searched. Here, α is the value of the user dispersion delay allowed in the application. ΣD (i) is a delay incurred when the delivery information passes through a route to be searched.
[0045]
As an example of the route search method, there is a method in which all possible routes are searched and a route that satisfies the above-described delay condition is secured as a candidate. When a plurality of routes are held as candidates, a route having many links in common with the route already adopted as the transfer route is adopted as the transfer route. This example is shown in FIG. In this example, a route X passing through the nodes A and B and a route Y passing through the node C are searched as candidates. In this case, ΣD (i) for path X is
ΣD (i) = D (A) + D (B)
ΣD (i) for the path Y is D (C). When counting the number of links common to the path through which the two pass and the search path that has already been adopted as the transfer path, the path X is two and the path Y is zero. Therefore, the adopted path is X. This operation is all performed for the egress edge node. A transfer path created as a result of performing these operations is a transfer path set in a core network that reduces delay between users in a hierarchical network.
[0046]
Finally, a series of operations in the above system will be described.
[0047]
FIG. 8 is a diagram for explaining a series of operations according to the embodiment of the present invention. It is assumed that the symbols and operation numbers in FIG.
[0048]
First, each edge node collects information on a transfer delay to a terminal of a receiver of a multicast communication to be calculated for a route existing in an access network to which the edge node is connected (operation 1). In the example shown in the figure, it is assumed that, among the access networks 10A, 10B, 10C, and 10D, a terminal of a sender exists in 10A and a terminal of a receiver exists in 10B, 10C, and 10D. Therefore, the entry edge node is 30A, and the exit edge nodes are 30B, 30C, and 30D. The egress edge nodes 30B, 30C, and 30D collect delay information of the terminals of the receivers existing in the access networks 10B, 10C, and 10D, respectively. This operation is performed independently at each edge node.
[0049]
From the collected information, each edge node calculates aggregated information of delays incurred in the access network (operation 2). The method of collecting such information and the method of calculation are based on the aforementioned method. The update of the aggregated information is appropriately performed when the information provided from the delay information collection mechanism changes due to a change in the delay state or the like in the access network.
[0050]
Each edge node transmits its own calculated aggregated information of delays in the access network to nodes that calculate routes in the core network, or collects aggregated information by an information collection mechanism that operates in the core network (Operation 3). This operation is appropriately performed when each edge node updates the delay aggregation information. As a result, the node performing the route calculation holds the new delay information. The node that performs the route calculation calculates a route that achieves the object of the present invention to reduce the delay dispersion between the terminals of the receivers, based on the aggregated delay information from each edge node that is held (operation 4).
[0051]
If the node performing the route calculation is the same as the ingress edge node of the multicast communication to be calculated, the ingress edge node sets the information transfer route using the calculation result. If not, the node performing the route calculation notifies the ingress edge node in the corresponding multicast group of the calculation result (operation 5).
[0052]
The ingress edge node receiving the result sets a route based on the result (operation 6).
[0053]
By performing the above operation when the delay information is changed, it is possible to provide a route that always satisfies the required conditions during information delivery.
[0054]
It should be noted that the present invention is not limited to the above embodiment, and various changes and applications are possible within the scope of the claims.
[0055]
【The invention's effect】
As described above, according to the present invention, reduction of delay between users can be achieved in a hierarchical network. For this reason, by setting the response time of each user to certain data, such as a video conference or an online game, at the same time, it is possible to achieve the requirements for an application that requires fairness of the user's speech opportunity. Become.
[0056]
Further, according to the present invention, the above requirements can be achieved without holding information on the entire computer network. Therefore, it is possible to reduce the amount of information held by the route calculation node, which is required in the conventional method. For this reason, the performance of the computer required when implementing the present invention does not depend on the size of the network, and the application can be provided even on a large-scale network.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating a network configuration in a mesh form.
FIG. 2 is a network configuration diagram of a route sharing mode.
FIG. 3 is a diagram illustrating a network configuration according to an embodiment of the present invention.
FIG. 4 is a diagram for explaining aggregated information according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating an information collection function according to an embodiment of the present invention.
FIG. 6 is a diagram illustrating a method for calculating a delay value according to an embodiment of the present invention.
FIG. 7 is a diagram for explaining a method of selecting a transfer path according to an embodiment of the present invention.
FIG. 8 is a diagram for explaining a series of operations in one embodiment of the present invention.
[Explanation of symbols]
10 access network 20 core network 30 edge node

Claims (5)

In a multicast communication path calculation method in a core network connecting between an access network accommodating terminals and an access network in multicast communication in a case where networks are hierarchically configured,
When calculating a transfer path to be used for delivery by multicast communication for delivering information at a node existing in a hierarchical computer network,
In an edge node connecting the access network and the core network, a delay incurred in a route through which information passes when delivering to a terminal of a receiver present in the access network connected to the access network is added to each terminal of the receiver. Measure,
Calculating a delay value incurred in the access network from the measured delay value;
Hold as aggregated information about the calculated delay value,
Delivering the calculated delay value to the node performing the multicast route calculation;
At the nodes in the core network, measure the delay incurred by the information passing through it,
Notify the measured information to a node having a function of calculating a multicast transfer path,
In a node having a function of calculating the multicast transfer path, holding two delay information distributed from the edge node and a node in the core network,
The calculated transfer path calculates a multicast transfer path for reducing a difference between delays of a sender terminal and the receiver terminal between receiver terminals based on the held delay information. A multicast transfer path calculation method, characterized in that: In the edge node connecting the access network and the core network,
When delivering to a terminal of a receiver existing in the access network connected to the edge node itself, a delay incurred in a route through which information is passed is measured for each terminal of the receiver,
2. The multicast transfer route calculation method according to claim 1, wherein the multicast transfer route calculation method stores the aggregated information regarding the maximum value of the measured delay values. In the edge node connecting the access network and the core network,
Measure the delay incurred in the route through which information passes when delivering to the terminal of the receiver present in the access network connected to the edge node itself, for each terminal of the receiver,
2. The multicast transfer path calculation method according to claim 1, wherein an average value of the measured delay values is held as aggregated information regarding delay. In a node that calculates the multicast transfer path,
The delay incurred from the egress edge node, which is the egress to the access network in which the terminal of the receiver of the multicast group exists, to the ingress edge node, which is the ingress from the access network in which the terminal of the sender transmitting the multicast group exists. Calculating the minimum value for each of the exit edge nodes;
Each exit edge node is distributed, the aggregated information of the delay in the access network connected to the own node and the measured minimum value of the delay are added, and the minimum value is searched for,
4. The multicast transfer path calculation method according to claim 1, wherein the multicast transfer path for setting the searched minimum value is determined as a delay to be realized. The multicast transfer path calculates a transfer path connecting the terminal of the sender and the egress edge node for each egress edge node,
The calculated delay value to be realized by the multicast transfer path minus the aggregated information about the delay of the multicast group in the access network connected to the egress edge node is subtracted from the calculated delay value, and the allowable difference between the receiver terminals is added. The upper limit is set as the upper limit, the lower limit is set as the subtracted value, and a route that realizes a delay value between them is held as a candidate of the transfer route,
5. The multicast transfer route calculation method according to claim 4, wherein a candidate having the largest number of links overlapping the route currently selected as the transfer route from the held candidates and being selected is selected as the transfer route.

Images