JPH08237277A - Vc connecting method - Google Patents

Abstract

PURPOSE: To realize a VC(virtual channel) connecting method in which the determination of a route and the reservation/release of a band can be carried out at a greater rate without increasing the call loss rate of a whole network. CONSTITUTION: A band reservation request is sent to a two-layer sub net manager M1 from a switching node 111 to store a call originating terminal T1 in response to a VC connecting request from the call originating terminal T1 or the band reservation request R1 based on a burst information send starting request after call connection, and the band reservation request is sent successively in parallel to a low order layer direction from that two-layer sub net manager M1, and the reservation of the band is carried out for each link positioned on the desired route by the switching nodes at both the ends of that link.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a VC connection method, and more particularly to a VC connection method for determining routes between originating and terminating terminals and bandwidth reservation / release in a network composed of a plurality of switching nodes.

[0002]

2. Description of the Related Art Using a network composed of a plurality of switching nodes, voice having a CBR (Constant Bit Rate) traffic characteristic in which data is transmitted to the network at an equal speed and a large amount of data in a short period of time are transmitted. When communicating computer-processed image information having bursty traffic characteristics sent to the network, a virtual circuit (Virtual Channel: V
In response to a connection request, the route for connecting the originating and terminating terminals is determined and the bandwidth is reserved based on the required transmission capacity, and the reserved bandwidth is released in response to the VC release request after the end of communication. Has become.

Q is a user-network interface (UNI) recommended by ITU.
In 2931 and the network node interface (NNI) under discussion, as a VC connection method, from the originating node accommodating the originating terminal to the destination node accommodating the destination terminal, about each node and the link between the nodes. , And route determination and bandwidth reservation / release are performed in order. However, according to this method, it takes time to perform the VC connection process, and a series of links reaching the destination terminal are not collectively routed, and there is a limit to effective use of bandwidth resources in the network.

On the other hand, conventionally, a method has been proposed in which routes are hierarchized in a network and route determination and bandwidth reservation are processed in parallel (for example, WADoeriger et.
all, "Fast Connection Establishment in Large-Scale
Networks. "INFOCOM'93, pp489-496, 1993). In addition, for all routes connecting the originating and terminating terminals, bandwidth reservation was made in parallel from the originating terminal to the terminating terminal, and the arriving terminal was first reached. A method of adopting a route has been proposed (for example, H. Suzuki ea all, "Fast Bandwidth Reservat
ion Schene with Multi-link & Multi-path Routing in
ATM Networks. "INFOCOM'92, pp2233-2240, 1992).

[0005]

Therefore, in the conventional VC connection method as described above, according to the former, route determination and bandwidth reservation are processed in parallel due to the hierarchization of routes, so that the time required for connection processing is shortened. However, since the route is determined in the upper layer without considering the possibility of bandwidth reservation in the lower layer, there is a problem that the call loss rate becomes large and the VC connection efficiency is low. In addition, according to the latter, since bandwidth reservations are carried out simultaneously for all routes, the reserved bandwidth must be released again for routes not adopted,
There has been a problem that call processing in the entire network increases and the call loss rate in the entire network increases due to unnecessary bandwidth reservation. The present invention is intended to solve such a problem, and an object thereof is to provide a VC connection method capable of performing route determination and bandwidth reservation / release at higher speed without increasing the call loss rate of the entire network. .

[0006]

In order to achieve such an object, in the VC connection method according to the present invention, a network is composed of a plurality of switching nodes in which first layer sub-networks are mutually connected by a predetermined link. The first layer has a hierarchical structure in which the m-th layer (m is a positive integer of 2 or more) sub-network is composed of a plurality of (m-1) -th layer sub-networks interconnected by a predetermined link.
The layer sub-network includes between the switching nodes within the own sub-network and between the predetermined first-layer sub-networks connected in the same layer, between the switching nodes and the first-layer sub-network. A first-layer subnet manager that updates and holds a plurality of routes that connect to each other in a predetermined order is provided, and a sub-network of the m-th layer is provided for each predetermined sub-network of the m-th layer connected in the same hierarchy. An m-th layer subnet manager that updates and holds a plurality of routes connecting these m-th layer sub-networks in a predetermined order is provided, and the m-th layer subnet manager refers to its own content in response to a bandwidth reservation request. By doing so, the optimum route connecting the requested sub-networks of the m-th layer is determined, and the relocation located on the optimum route is determined. And a bandwidth reservation request for a route in the m-th layer subnetwork to any m-1st layer subnet manager in each m-1th layer subnetwork located on the optimum route. Is sent, and the first layer subnet manager refers to the contents held by itself in response to the bandwidth reservation request from the upper layer subnet manager, so that a predetermined first layer subnetwork as a route in the requested upper layer subnetwork. The optimal route connecting the routes is determined, the bandwidth reservation request for the link located on the optimal route is sent, and to the first layer subnet manager of each first layer subnetwork located on the optimal route,
A bandwidth reservation request is sent to the route in each layer 1 subnetwork, and the layer 1 subnet manager located on the optimum route responds to the band reservation request from any layer 1 subnet manager to hold its own content. By referring to, the optimal route connecting the predetermined switching nodes is determined as the route in the requested layer 1 subnetwork, the bandwidth reservation request for the link located on the optimal route is sent, and the calling terminal sends it. In response to the bandwidth reservation request based on the VC connection request or the burst information transmission start request after the call connection, the switching node accommodating the calling terminal sends a bandwidth reservation request to a predetermined m-th layer subnet manager, and By sequentially sending bandwidth reservation requests in parallel to the lower layers from the m-th layer subnet manager,
Bandwidth reservation is performed for each link located on a desired route by switching nodes at both ends of the link.

Further, in response to a bandwidth reservation request from the calling terminal, bandwidth reservation of a link connecting the calling terminal and the switching node accommodating this is carried out, and upon success of the link reservation with the switching node accommodating the calling terminal. Bandwidth reservations are made for each relay link and subscriber link that form a route connecting to a terminal. Also, each terminal has a network address consisting of the sub-network of each layer and the identifier of the switching node, and the switching node accommodating the calling terminal responds to the bandwidth reservation request from the calling terminal, and The network addresses of the terminals are compared in order from the upper layer identifier, and the route from the own switching node to the destination terminal is determined for the subnet manager corresponding to the identifier on the calling terminal side in the layer where the first different identifier appears. The bandwidth reservation request for the route is transmitted.

Further, when a bandwidth reservation response indicating unsuccessful bandwidth reservation is returned in response to the transmitted bandwidth reservation request, each subnet manager refers to its own content and extracts the next optimum route. Then, a new bandwidth reservation request for that route is transmitted. Furthermore, each subnet manager issues a new bandwidth reservation request for the next most suitable route in response to the bandwidth reservation failure, which is set in common to all layers or when each bandwidth reservation is unsuccessful. The number of times is repeatedly performed, and in response to the subsequent failure of the bandwidth reservation, a bandwidth reservation response indicating the failure of the bandwidth reservation is returned to the source of the bandwidth reservation request received by itself. Furthermore, the calling terminal sets the number of repetitions when bandwidth reservation fails for each layer in the bandwidth reservation request and sends it out, and each subnet manager responds to the failure of bandwidth reservation by sending the next best route. A new bandwidth reservation request for the bandwidth reservation request for the number of times when the bandwidth reservation is unsuccessful in the own layer set in the bandwidth reservation request, and the bandwidth reservation request source received by itself in response to the subsequent failure of the bandwidth reservation. A bandwidth reservation response indicating that the bandwidth reservation has failed is sent back to.

Further, the exchange node stores the link allocatable bandwidth amount holding unit for storing the link allocatable bandwidth amount of each accommodated link and the link allocatable bandwidth amount held in the link allocatable bandwidth amount holding unit. The first-layer subnet manager holds a plurality of routes connecting the switching nodes in a predetermined order for each predetermined switching node in its own subnetwork. An inter-subnet routing table that holds a plurality of routes connecting the first layer sub-networks in a predetermined order between the sub-network routing table holding unit and the predetermined first layer sub-networks connected in the same layer The number of transit nodes, which indicates the number of switching nodes that make up each route in the table holding unit and its own sub-network, is small. The contents stored in the routing table holding unit in the subnet are updated in the order of increasing link allocatable bandwidth notified from each switching node, and the adjacent sub-networks are connected between the predetermined first-layer subnet managers connected in the same layer. The link allocatable bandwidths of the boundary links are mutually exchanged, and the number of transit subnets indicating the number of first-layer sub-networks forming each route is small, In response to a bandwidth reservation request for the route between the table management unit that updates the held contents and the first layer sub-network, the held contents of the inter-subnet routing table holding unit are referred to and the requested first
An optimal route connecting between layer sub-networks is determined, bandwidth reservation requests for all links connecting individual first layer sub-networks are sent in parallel along the optimal route, and each route located on the optimal route is For the layer 1 subnet manager of the layer 1 subnetwork,
Bandwidth reservation requests for routes in each layer 1 subnetwork are sent in parallel, and in accordance with the bandwidth reservation request for routes in layer 1 subnetwork, the contents held in the routing table holding unit in the subnet are referred to, Path setting for determining an optimum route connecting switching nodes in the required first layer sub-network and sending bandwidth reservation requests in parallel to all links connecting individual switching nodes along the optimum route / And a release request unit, and the m-th layer subnet manager, for each predetermined m-th layer sub-network connected in the same layer, defines a plurality of routes for connecting these m-th layer sub-networks. An inter-subnet routing table holding unit that holds in order,
Via subnets that indicate the number of m-th layer sub-networks that make up each route by mutually exchanging the link-allocatable bandwidth of boundary links with adjacent sub-networks between the specified m-th layer subnet managers connected in the same layer A table management unit that updates the contents held in the inter-subnet routing table holding unit in the order of increasing number of link allocable bandwidths, and an inter-subnet routing table in response to a bandwidth reservation request for a route between the m-th layer sub-networks. An optimum route connecting the requested m-th layer sub-networks is determined by referring to the content held by the holding unit, and bandwidth reservation requests for all links connecting the individual m-th layer sub-networks along the optimum route. Are sent in parallel, and each m-th layer subnet word located on the optimum route Path setup / release request for sending in parallel a bandwidth reservation request for a route between the m-1 layer sub-networks belonging to each m-th layer sub-network And a part.

Further, the subnet managers of the respective layers and the switching nodes belonging to the common upper sub-network are connected via a predetermined communication line, and between the subnet managers of the same layer, between the subnet managers of the upper and lower layers, and arbitrary Messages such as request and response for call control are directly exchanged between the switching node and the subnet manager that manages the link owned by the switching node, and the link allocatable bandwidth amount calculated by the switching node is allocated to each node. The subnet manager that manages the link is notified directly. Further, the notification unit of the exchange node, for each link to accommodate, the latest link allocatable bandwidth amount calculated from the empty band of the link and the cell or packet loss rate, and is held in the link allocatable bandwidth amount holding unit. If the difference exceeds a predetermined threshold value, the link source allocation manager is notified of the new link allocatable bandwidth amount and the new link allocatable bandwidth amount is notified. The link allocatable bandwidth amount is updated and held in the link allocatable bandwidth amount holding unit.

Further, the notifying unit of the switching node, for each link to be accommodated, sets the latest link allocatable bandwidth amount calculated from the vacant bandwidth of the link and the cell or packet loss rate to the managing subnet manager of each link. It is intended to notify the. Further, one of the switching nodes connected to both ends of each link notifies the managing subnet manager of each link of the link allocatable bandwidth amount of the link. Further, the path setup / release request unit of the m-th layer subnet manager responds to the bandwidth reservation request for the route between the m-th layer sub-networks, for each of the m-th layer sub-networks located on the optimum route, Each m-1th belonging to the m-layer subnetwork
From the layer subnet manager, the one with the smallest processing load among the m-1 layer subnet managers, the one that is closest to the calling terminal as a route, and an arbitrary m-1 layer subnet manager according to a predetermined order or random number. Selected,
For each selected m-1 layer subnet manager, the m-1 layer belonging to each m-th layer sub-network
A bandwidth reservation request for a route between layer sub-networks is transmitted in parallel.

Further, when each subnet manager makes a bandwidth reservation request to a link composed of a network based on another connection procedure, each subnet manager makes a bandwidth reservation request to each switching node connected to both ends of the link. One of the switching nodes sends a predetermined call setup request to the network in response to the bandwidth reservation request, and the other switching node of the switching nodes sends a call setup request from the network. A predetermined bandwidth reservation response is returned to the corresponding subnet manager according to the above. Therefore, in response to the bandwidth reservation request based on the VC connection request from the calling terminal or the burst information transmission start request after the call connection, the switching node accommodating the calling terminal requests the band reservation to the predetermined m-th layer subnet manager. Is transmitted, and the bandwidth reservation requests are sequentially transmitted in parallel from the m-th layer subnet manager in the lower layer direction.
Bandwidth reservation is made for each link located on the desired route by the switching nodes at both ends of the link.

Further, when each subnet manager sends a bandwidth reservation request to each link on the optimum route or to any subnet manager, the route information of the route set by the bandwidth reservation request is used for individual V
It is held for each C connection or burst connection after a call connection, and by the m-th layer subnet manager, it refers to the route information that it holds in response to a bandwidth release request and is used for the corresponding VC connection or burst connection after a call connection. The bandwidth release request for the link located on the used route is sent, and the bandwidth release request for the route in the m-th layer subnetwork is sent to each m-1 layer subnet manager located on the used route. However, by referring to the route information held by itself by the layer 1 subnet manager in response to the bandwidth release request from the upper subnet manager, the position on the route used for the corresponding VC connection or burst connection after the call connection is located. The bandwidth release request for the link to A bandwidth release request for a route in each layer 1 subnetwork is sent to the layer 1 subnet manager of the network, and a layer release request from the layer 1 subnet manager is made by the layer 1 subnet manager located on the route. By referring to the route information held by itself, a band release request is sent to the corresponding VC connection or the link located on the route used for the burst connection after the call connection, and the VC release from the calling terminal In response to a request or a bandwidth release request based on a burst information transmission termination request after call connection, the switching node accommodating the calling terminal sends a bandwidth release request to a predetermined m-th layer subnet manager, and the m-th layer Used for communication by sequentially sending bandwidth release requests in parallel to the lower layers from the subnet manager By the ends switching node of the link for each link located on the route it is obtained to perform the band released.

Further, in response to a band release request from the calling terminal, a band releasing request is made for each relay link and subscriber link forming a route connecting the exchange node accommodating the calling terminal and the called terminal, and then the calling is issued. The bandwidth of the link connecting the terminal and the switching node accommodating the terminal is released. Furthermore, each subnet manager sends a bandwidth release request to each switching node connected to both ends of the link when requesting the bandwidth release to a link configured from a network based on another connection procedure, One switching node of each switching node sends a predetermined call release request to the network in response to the bandwidth release request, and the other switching node of each switching node responds to the call release request from the network. A predetermined band release response is returned to the corresponding subnet manager.
Therefore, in response to a VC release request from the calling terminal or a bandwidth release request based on a burst information transmission termination request after call connection, the switching node accommodating the calling terminal issues a bandwidth release request to a predetermined m-th layer subnet manager. Is sent, and a bandwidth release request is sent sequentially in parallel from the m-th layer subnet manager in the lower layer direction, so that for each link located on the route used for communication, switching nodes at both ends of the link are sent. The band is released by.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a hierarchical network system according to an embodiment of the present invention. In FIG. 1, T1 is a calling terminal, T2 is a receiving terminal, NW1 and NW2.
Is a two-layer sub-network having a plurality of first-layer sub-networks NW11, NW12, ... And one-layer sub-networks NW21, NW22, ... Connected as lower sub-networks, and M1 and M2 are two-layer sub-networks NW1, respectively. It is a two-layer subnet manager that manages the NW2.

First layer sub-networks NW11 and NW12
Is a plurality of switching nodes 111, 11 connected to each other.
, And switching nodes 121, 122, ..., And although omitted, other one-layer sub-networks are similar. 1st layer sub-network NW21, NW2
2 is a plurality of switching nodes 211, 21 connected to each other
, And switching nodes 221, 223, ..., And although omitted, other one-layer sub-networks are similar.

M11, M12, and M21, 22 are 1-layer sub-networks NW11, NW12, and NW, respectively.
21 and NW22 are one-layer subnet managers, and one-layer subnet managers M11, M12, ...
, M21, M22, ... Are connected via predetermined communication lines, respectively, and are connected to the first layer sub-networks NW11, N
The management information of W12, ..., And NW21, NW22 ,. Similarly, the two-layer subnet managers M1 and M2 are also connected via predetermined lines, and the two-layer sub-network N
The management information of W1 and NW2 is shared.

The switching nodes and the subnet managers of the respective layers are also connected via predetermined communication lines, and are connected to each other between the subnet managers of the same layer, between the subnet managers of the upper and lower layers, and to any switching node. As a control message for call processing, messages such as bandwidth reservation / release request and bandwidth reservation / release response to the request are directly exchanged between the subnet managers, and the link allocatable bandwidth amount calculated by the switching node. Is directly notified to the managing subnet manager that manages each link.

As described above, various control messages are directly exchanged between all the subnet managers belonging to the common upper subnetwork, and between the subnet manager and the switching node, regardless of the difference in the hierarchy. The arrival time of the control message is shortened, the VC connection time is shortened, and the processing load is reduced as compared with the case where the control message is transferred via the subnet manager. In addition, since the switching node directly notifies the management source subnet manager that manages each link assignable bandwidth amount, the notification time lag is shortened and the inter-subnet routing table is updated quickly. The reliability is improved and the call loss rate can be reduced.

The originating terminal T1 is accommodated in the switching node 111 via the link L1, and the destination terminal T2 is accommodated in the switching node 222 via the link L9. The exchange nodes 111 and 112, 121 and 122, 211
And 212 and 221 and 222 are linked to the link L, respectively.
2, L4, L6, and L8, first-layer sub-networks NW11 and NW12, NW21 and NW22 are connected via links L3 and L7, respectively, and further two-layer networks NW1 and NW2 are
It is connected via a link L5.

FIG. 2 is a functional block diagram showing the internal structure of each subnet manager and exchange node.
Is a switching node, (b) is a one-layer subnet manager,
(C) shows two-layer subnet management. Figure 2
In (a), 42 is a link allocatable bandwidth amount holding unit that holds the usage rate of each link connected to the switching node, and 41 is a link allocatable bandwidth amount held by the link allocatable bandwidth amount holding unit 42. If the difference between the free bandwidth of the link and the new link allocatable bandwidth calculated from the cell (or packet) discard rate exceeds a predetermined threshold, a new link is sent to the 1-layer subnet manager. A notification unit that notifies the allocatable bandwidth amount.

Each switching node has a call control management function for the link to be accommodated and a function for exchanging data flowing on the link, like a general ATM switch or packet switch. In particular, in response to a bandwidth reservation request for a predetermined link accommodated by itself, in the free bandwidth on the link, while securing the requested amount of bandwidth, in response to a bandwidth release request for a predetermined link accommodated by itself, The band used for the requested VC is released.

In FIG. 2 (b), reference numeral 52 denotes a switching node in the one-layer sub-network managed by itself, between the switching nodes accommodating terminals and the switching nodes adjacent to other sub-networks. For each of these, an intra-subnet routing table holding unit that holds an intra-subnet routing table that stores the optimal route in sequence, 53 is a peer one-layer sub-network connected in the same layer, and here is a self in the same two-layer sub-network. A peer layer 1 sub-network, including a switching node accommodating terminals, and the other 2
It is an inter-subnet routing table holding unit that holds an inter-subnet routing table that stores optimal routes in sequence for each one-layer sub-network of the one-layer sub-network adjacent to the layer sub-network.

Reference numeral 51 denotes a route group connecting each switching node of switching nodes accommodating terminals and switching nodes adjacent to other sub-networks, based on the notification of the link allocatable bandwidth amount from each switching node. The optimal route is extracted according to the link allocatable bandwidth amount and the number of transit nodes and stored in the inter-subnet routing table in order, and the link allocatable bandwidth amount of the boundary link with another 1-layer sub-network is Includes a switching node accommodating terminals by mutually exchanging with a connected one-layer subnet manager
For each layer 1 subnetwork of the layer 1 subnetwork adjacent to the layer 2 subnetwork and the other layer 2 subnetwork, a route group connecting the layer subnetworks is selected according to the link assignable bandwidth amount and the number of transit nodes. It is a table management unit that extracts and sequentially stores in the inter-subnet routing table.

Reference numeral 55 denotes a set route holding unit for holding route information of a route reserved according to execution of bandwidth reservation, 54
Responds to a bandwidth reservation request for an arbitrary route from the two-layer subnet manager, refers to the intra-subnet routing table holding unit 52 and the inter-subnet routing table holding unit 53, and selects each of the route groups targeted for the bandwidth reservation request. The optimum route is selected from the selected routes, bandwidth is reserved for the switching nodes and links that compose the route, the selected route is stored in the set route holding unit 55, and any route from the switching node and the two-layer subnet manager is stored. In response to the bandwidth release request for the route,
This is a path setting / releasing request unit that reads out route information of a route for which a bandwidth release request is made from the set route holding unit 55, and releases bandwidth to the switching nodes and links that form the route based on the route information.

In FIG. 2C, reference numeral 63 denotes a two-layer sub-network which is connected in the same layer and is the same as the self-layer, such as a two-layer sub-network including switching nodes accommodating terminals, and other upper sub-networks. The inter-subnet routing table holding unit that holds the inter-subnet routing table that stores the optimal route in order between each two-layer sub-networks of the two-layer sub-networks adjacent to By exchanging the link allocatable bandwidth amount notified from each accommodating switching node with a self-coordinated two-layer subnet manager connected in the same layer, a two-layer sub-network including switching nodes accommodating terminals , And a two-layer subnetwork adjacent to other higher subnetworks Each between two layers subnetwork,
A table management unit that extracts optimum routes from the routes connecting them according to the link allocatable bandwidth amount and the number of transit nodes, and sequentially stores them in the inter-subnet routing table.

Reference numeral 65 denotes a set route holding unit for holding route information of a route reserved according to execution of bandwidth reservation, 64
Responds to a bandwidth reservation request for an arbitrary route from the switching node, the inter-subnet routing table holding unit 63
, The optimum route is selected from the route groups for which bandwidth reservation requests are made, and the bandwidth reservation request is made to the 1st-layer subnet manager and the links that compose the route, and the set route holding unit is also provided. The selected route is stored in 65, the route information of the route for which the bandwidth release request is made is read from the set route holding unit 65 in response to the bandwidth release request from the exchange node for the arbitrary route, and based on the route information It is a path setting / release request unit that releases the bandwidth to the first-layer subnet manager and the links that form the route.

The configuration of each table in the in-subnet routing table holding unit 52 and the inter-subnet routing table holding unit 53 of the first layer subnet manager will be described. FIG. 3 is a block diagram showing an example of a two-layer subnetwork. In FIG.
A switching node 314 is connected to each other by links L11 to L16, and is a first layer sub-network NW31.
Is composed. Further 1st layer sub-network NW31
Are 1-layer sub-networks NW32 to N having a similar structure
W34 and the links L21 to L26 are connected to each other to form a two-layer sub-network NW3, which constitutes a hierarchical structure including one layer and two layers as a whole.

Each one-layer sub-network NW31 to NW3
4 is a link L4 connecting to another two-layer subnetwork
1, L31 to L33, and 1-layer subnet managers M31 to M34, respectively.
The layer subnet managers M31 to M34 are connected by a predetermined communication line. 4A and 4B are explanatory diagrams showing the configurations of various holding units. FIG. 4A is a link allocatable bandwidth amount holding unit 42 in the switching node 311, FIG.
(C) shows the inter-subnet routing table in the 1st-layer subnet manager M31.

First, in the switching node 311 of the first-layer sub-network NW31, as shown in FIG. 4A, the link allocatable bandwidth amount holding unit 42 causes each of the links L41, L11 to L13 connected to itself to be connected. The link allocatable bandwidth amount indicating the remaining allocatable bandwidth amount is held, and when the difference between this link allocatable bandwidth amount and the actual link allocatable bandwidth amount exceeds a predetermined threshold value, Notification unit 41
Will notify the designated upper subnet manager,
The link allocatable bandwidth amount in the link allocatable bandwidth amount holding unit 42 is updated.

As described above, when the fluctuation of the link allocatable bandwidth amount exceeds the threshold value, it is notified to the 1st layer subnet manager. The processing load on the switching node and each subnet manager is reduced, and it is possible to promptly notify a sudden change in the amount of link allocatable bandwidth. As a method of notifying the link allocatable bandwidth amount, the switching node may notify each subnet manager of the link allocatable bandwidth amount at predetermined intervals. In this case, the threshold value is given to each switching node. The configuration and processing for comparison and comparison are not required, and the configuration of the exchange node can be simplified.

Further, since the bandwidth reserved for each link is the same as seen from any switching node connected to both ends of the link, the link allocatable bandwidth is also equal. Therefore, the notification of the link allocatable bandwidth amount to the management source subnet manager that manages each link is performed by either one of the switching nodes connected to both ends of each link, such as an instruction from the management source subnet manager or between both switching nodes. The notification may be sent only from the side determined by the negotiation of
This reduces the processing load on the other switching node that does not notify, and also reduces the processing load on the management source subnet manager for managing the link allocatable bandwidth.

In the 1st layer subnet manager M31, as shown in FIG. 4B, the intra-subnet routing table in the intra-subnet routing table holding unit 52 stores a route group connecting predetermined exchange nodes in the order of optimum route. The table management unit 51 controls the switching nodes 311 to 314 in the first layer sub-network NW31.
Update management is performed at any time based on the link allocatable bandwidth amount notified from.

Particularly, in the intra-subnet routing table, a route connecting switching nodes accommodating terminals and switching nodes having boundary links with other sub-networks, for example, all routes connecting switching node 311 and switching node 312, That is, L11, L12-L14, L
13-L15, L12-L16-L15 routes,
The routes are stored in order from a route having a small number of via links and a route having a large amount of link assignable bandwidth of links constituting each route.

Therefore, the first layer sub-network NW3
1, in the case of selecting a route connecting any two terminals of the boundary links L21 to L23 and L41 with an arbitrary terminal accommodated in the 1st layer subnetwork NW31 and another 1st layer or 2nd layer subnetwork By extracting the highest-ranked route corresponding to the route connecting between the exchange nodes 311 to 314 accommodating two desired parties from the in-subnet routing table 52,
At that time, it becomes possible to easily select the via link that constitutes the optimum route.

In addition, as shown in FIG. 4C, the 1-layer subnet manager M31 uses the inter-subnet routing table in the inter-subnet routing table holding unit 53 to establish an optimum route for a route group connecting predetermined 1-layer subnets. The table management unit 51 is a switching node accommodated in each 1st-layer subnetwork and has link links from the switching nodes 311 to 314 that have boundary links to other 1st-layer subnetworks. In addition to obtaining the allocatable bandwidth amount, each layer 1 sub-network connected in the same layer, here, layer 1 sub-network NW23 in layer 2 sub-network NW3
Link L that exchanges the amount of bandwidth that can be allocated to the link between the NW 25 and the NW 25 and connects these 1-layer sub-networks.
The inter-subnet routing table is updated and managed as needed based on the link assignable bandwidth amount of 12 to L17.

In particular, in the inter-subnet routing table, the route connecting between the first layer sub-networks having switching nodes accommodating terminals and the first layer sub-networks of the first layer sub-network adjacent to other sub-networks, for example, the first layer All routes connecting the sub-network NW31 and the layer 1 sub-network NW32, that is, L
21, L22-L24, L23-L25, L22-L2
6-L25 is set as a route group, and the routes are stored in order from a route having a small number of routed links and a large amount of link assignable bandwidth of links configuring each route.

Therefore, the first layer sub-network NW3
When selecting a route connecting any two of the terminals connected to 1 and the boundary links L41 and L31 to L33 with another two-layer sub-network, the desired two parties are selected from this inter-subnet routing table 53. By extracting the highest-ranked route corresponding to the route connecting between the first-layer sub-networks NW31 to NW34 respectively accommodated, it becomes possible to easily select the via link that constitutes the optimum route at that time.

It should be noted that another one-layer subnet manager M3
2 to M34, and the intra-subnet routing table holding unit 52 and the inter-subnet routing table holding unit 53 included in the first-layer subnet management in the other two-layer sub-networks have the same configuration as described above. Next, the configuration of each table in the inter-subnet routing table holding unit 63 of the two-layer subnet manager will be described.

FIG. 5 is a block diagram showing an example of an upper (3 layer) network composed of a plurality of 2 layer sub-networks. In FIG. 5, NW3 to NW6 are links L.
It is a two-layer sub-network interconnected by 31 to L36 and constitutes an upper (three-layer) sub-network. Each two-layer subnetwork NW3 to NW6
Links L41 to L connected to other upper sub-networks
44, and each has a plurality of one-layer subnet managers as shown in FIG.
The two-layer subnet managers M3 to M6 are connected by a predetermined communication line.

FIG. 6 is an explanatory diagram showing the structure of the inter-subnet routing table in the two-layer subnet manager M3. In the two-layer subnet manager M3, as shown in FIG.
2 according to the routing table between subnets in
A route group connecting layer subnets is stored in the order of optimum routes, and the table management unit 61 is a switching node accommodated in each two-layer sub-network and has a boundary link with another two-layer sub-network. The link allocatable bandwidth amount of the boundary link is obtained from the switching node, and each two-layer subnetwork N connected in the same layer
The link allocable bandwidth is mutually exchanged with W4 to NW6, and the inter-subnet routing table is updated at any time based on the link allocable bandwidth of the links L31 to L36 connecting these two-layer sub-networks. to manage.

In particular, in the inter-subnet routing table, a route connecting between a two-layer sub-network having a switching node accommodating a terminal and each two-layer sub-network of a two-layer sub-network adjacent to another sub-network, for example, the second layer All routes connecting the subnetwork NW3 and the two-layer subnetwork NW4, that is, L3
1, L32-L34, L33-L35, L32-L36
With L35 as a route group, the routes are stored in order from a route having a small number of via links of the route and a large amount of link allocable bandwidth of links constituting each route.

Therefore, the first layer sub-network NW3
In the case of selecting a route connecting any two of the boundary links L41 to L44 with the terminal connected to the above, and the other upper sub-network, the desired two parties are respectively accommodated from the inter-subnet routing table 63. By extracting the highest-ranked route corresponding to the route connecting the layer sub-networks NW3 to NW6, it is possible to easily select the via link that constitutes the optimum route at that time.

Incidentally, another two-layer subnet manager M4
-M6, and the inter-subnet routing table holding unit 63 included in the two-layer subnet management in the other upper sub-network has the same configuration as described above. Further, when the upper subnet is configured, each subnet manager has the same configuration as the two-layer subnet management M3 with the above-mentioned one-layer subnetwork as the lower subnetwork.

In this way, the network has a hierarchical structure in which the upper sub-network is composed of a plurality of lower sub-networks, and a subnet manager is provided for each sub-network so that the sub-networks of the same peer sub-network are subordinate to the same upper sub-network. The optimal route of each network is managed individually. Especially, in the lowest level (1st layer) subnet manager, the optimal route between each switching node in the 1st layer subnetwork managed by itself is also managed individually. All the constituent links are managed in a decentralized manner, and it is possible to avoid failures caused by the concentration of management processing for switching nodes and subnet managers accommodating the originating and terminating terminals, as well as the processing capability to perform these management processing specially. A high-end node in the network. No, it is possible to quickly select the best route.

Next, referring to FIG. 1 and FIG. 7, the route determination and bandwidth reservation processing to the destination terminal T2 in response to the bandwidth reservation request from the calling terminal T1 will be described as the operation of the present invention. FIG. 7 shows V from the calling terminal T1 to the called terminal T2.
It is a sequence diagram showing a C connection procedure, in the figure,
The same parts as those in FIG. 1 described above are designated by the same reference numerals. In FIG. 7, R1 to R17 are bandwidth reservation requests, and A1 to A1.
Reference numeral 7 is a bandwidth reservation response returned in response to the bandwidth reservation requests R1 to R17.

First, in response to the transmission request generated by the calling terminal T1, the amount of bandwidth required for communication, the called address indicating the called terminal,
And a bandwidth reservation request R including its own address indicating the calling terminal
1 is sent. [Band reservation request R1] Destination: T1 → 111 Route: T1 to T2 It is assumed that the calling terminal and the called terminal each have the following addresses. Source terminal T1 address = 1; 11; 111; T1 Destination terminal T2 address = 2; 22; 222; T2

Each address is made up of a sequence of identifiers (hereinafter referred to as IDs) for identifying each sub-network and switching node, and the two-layer sub-network I from the top
The network configuration is shown in the order of D; first layer sub-network ID; exchange node ID; terminal ID. The switching node 111 sends out a bandwidth reservation request R2 for the link L1 connecting with the calling terminal T1 in response to the bandwidth reservation request R1 from the calling terminal T1. [Band reservation request R2] Destination: 111 → 111, T1 Route: L1

Actually, the bandwidth reservation request R2 for the link L1 is sent to the nodes at both ends of the link L1, that is, the switching node 111 and the calling terminal T1, and both of them request the bandwidth reservation request from the calling terminal T1. Bandwidth allocation, that is, bandwidth reservation is performed. If there is a free band equal to or larger than the desired band and the band reservation is successful, both parties return a band reservation response A2 indicating the success of the band reservation to the switching node 111, and the switching node 111 sends the calling terminal T1. It is confirmed that the bandwidth reservation for the link L1 has been completed.

If the bandwidth reservation is unsuccessful because at least one of them does not have a free bandwidth equal to or more than the desired bandwidth, the request source of the bandwidth reservation request R2 A bandwidth reservation response A2 indicating unsuccessful bandwidth reservation is returned to a certain switching node 111, and accordingly, the switching node 111 responds to the request source of the bandwidth reservation request received by itself with the unsuccessful bandwidth reservation. The bandwidth reservation response A1 shown is returned, and the originating terminal T1 notifies the user of the failure of bandwidth reservation.

In response to the completion of the bandwidth reservation for the link L1, the switching node 111 compares the addresses of the source terminal T1 and the destination terminal T2 in order from the upper layer side to the lower layer side, and the different subnetwork ID or switching node ID is the first. The bandwidth reservation request is sent to the subnet manager or the switching node of the sub-network indicated by the source terminal T1 address of the layer detected by the above.

Therefore, when the originating and terminating terminals T1 and T2 have the above-mentioned addresses, different subnetwork IDs will be detected in the hierarchy of the first two-layer subnetwork, and the switching node 111 corresponds to that hierarchy. The bandwidth reservation request R3 is sent from the sub-network of the calling terminal T1 address to the second-layer subnet manager M1 of the second-layer sub-network NW1 and the destination is 2
Store the layer subnet manager M1. [Band reservation request R3] Destination: 111 → M1 Route: 111 to T2

Thus, in response to the bandwidth reservation request from the calling terminal T1, the link L1 between the calling terminal T1 and the switching node 111 which accommodates the calling terminal T1 is preceded by another relay link.
After the bandwidth reservation is made for each link, the bandwidth reservation is made for each link forming the route connecting the switching node 111 and the destination terminal T2. Therefore, the link L1 requests the bandwidth reservation before other links. As a result, the desired bandwidth is reserved and even if the bandwidth reservation for other links is successful, the bandwidth reservation for the link with the calling terminal is unsuccessful and call loss can be prevented. A bandwidth reservation process of good quality will be performed.

In response to the bandwidth reservation request R3, the path setup / release request unit 64 of the second-layer subnet manager M1 reserves the bandwidth for the link L9 connecting the destination terminal T2 and the exchange node 222 that accommodates the destination terminal T2. Send request R4. [Band reservation request R4] Destination: M1 → 222, T2 Route: L9 Actually, similarly to the above-described band reservation request for the link L1 on the calling terminal T1 side, the switching node 222 and the terminating terminal T2 at both ends of the link L9. The bandwidth reservation is made in step S3 and the bandwidth reservation response A4 is returned in response to the success of the bandwidth reservation.

On the other hand, in response to the bandwidth reservation request R3, the path setup / release request section 64 of the two-layer subnet manager M1.
Refers to the inter-subnet routing table of the inter-subnet routing table holding unit 63 as shown in FIG. An optimal route located at the head of the inter-subnet routing table is extracted as a route connecting to the sub-network NW2.

Subsequently, the path setup / release request section 64 sends out a bandwidth reservation request R5 for all the links of the same layer forming the extracted route, here L5. [Band reservation request R5] Destination: M1 → 122, 211 Route: L5 In reality, the band reservation is performed by the switching nodes 122 and 211 at both ends of the link L5, and the band reservation response A5 is transmitted in response to the success of the band reservation. Will be returned.

In this case, the two-layer subnet manager M1
Then, the bandwidth reservation request R4 for the link L9 and the bandwidth reservation request R for the link L5 according to the bandwidth reservation request R3.
5 are transmitted independently of each other, whereby bandwidth reservations for the links L9 and L5 are processed in parallel, and efficient bandwidth reservation processing is performed. Since another link cannot be selected for the link L9 connecting the destination terminal T2 and the switching node 222, the band reservation success is confirmed by the band reservation response A4 corresponding to the band reservation request R4, and then the band is confirmed. The reservation request R5 may be transmitted, and when the bandwidth reservation for the link L9 is unsuccessful, it is not necessary to perform the bandwidth release processing for the link for which the bandwidth reservation is made by the bandwidth reservation request R5, and the processing in each subnet manager and the switching node The burden is reduced and the utilization efficiency of the lower layer sub-network is improved.

Next, the path setting / releasing request unit 64 of the second layer subnet manager M1 is one of the subnet managers of one lower layer respectively included in the second layer sub-networks NW1 and NW2 on the extracted route, that is, For any one of the first-layer subnet managers M11 and M12 and one of the first-layer subnet managers M21 and M22, the extracted route in each second-layer sub-network, that is, the second-layer sub-network NW1
Is a route connecting the 1st layer subnetwork NW11 accommodating the calling terminal T1 and the 1st layer subnetwork NW12 accommodating the link L5, and the 2nd layer subnetwork NW2 is the NW21 accommodating the link L5 and the destination terminal T2. Bandwidth reservation requests R6 and R7 are sent to routes connecting the first layer sub-network NW22 to be accommodated.

In this case, as a method of selecting an arbitrary 1st layer subnet manager in the 2nd layer sub-networks NW1 and NW2 by the path setting / release request section 64, the following 3
There are two possible ways. First, the current own processing load is mutually exchanged between the first-layer subnet managers M11 and M12 and between M21 and M22, and the path setting /
When the 1st-layer subnet manager with the least processing load is selected according to the bandwidth reservation request from the release requesting unit 64, the VC setting / release processing is distributed without being concentrated on the same 1st-layer subnet manager. As a result, the overall processing efficiency is improved.

Further, based on the calling terminal T1 address, 1
Layer subnet managers M11, M12 and M21,
When the most outgoing terminal T1 of M22 is selected, the one-layer subnet manager selected changes according to the outgoing terminal T1 accommodated in each position, resulting in a special processing load. Processing will be distributed to some extent without the need for management means.
Further, even when each of the first-layer subnet managers M11, M12 and M21, M22 is selected in order or by a random number, the processing is distributed to some extent without requiring any special means for managing the processing load. Will be done.

By any one of such selection methods, for example, the 1st layer subnet manager M11 is selected in the 2nd layer subnetwork NW1, and the 1st layer subnet manager M21 is selected in the 2nd layer subnetwork NW2. From the path setting / release request unit 64 of the subnet manager M1, the first layer subnet manager M
11, the switching node 111 accommodating the calling terminal T1
A bandwidth reservation request R6 for the route between the 1st-layer subnetwork NW11 including R1 and the 1st-layer subnetwork NW12 accommodating the link L5 is transmitted. [Band reservation request R6] Destination: M1 → M11 Route: NW11 to NW12

Further, from the path setting / release requesting unit 64 of the second layer subnet manager M1, to the first layer subnet manager M21, the first layer sub-network NW22 including the switching node 222 accommodating the destination terminal T2 and the link L.
The bandwidth reservation request R7 for the route to the first layer sub-network NW21 accommodating 5 is sent in parallel with the bandwidth reservation request R6. [Band reservation request R7] Destination: M1 → M21 Route: NW21 to NW22

In the above description, at the same time as the bandwidth reservation request R5 for the link L5 from the path setting / release request section 64 of the second layer subnet manager M1, the bandwidth reservation request R for the first layer subnet managers M11 and M21.
6 and R7 are transmitted, the bandwidth reservation response A5 returned in response to the bandwidth reservation request R5 confirms successful bandwidth reservation in the link L5 connecting the two-layer sub-networks NW1 and NW2. After
It is also possible to send the bandwidth reservation requests R6 and R7 of the routes in the respective second layer sub-networks NW1 and NW2 from the second layer subnet manager M1 to the first layer subnet managers M11 and M21.

For example, in FIG. 7, the bandwidth reservation request R5 is provided in parallel with the bandwidth reservation request R5 from the two-layer subnet manager M1.
6 and R7 are transmitted, but in this case, the second layer is confirmed after the desired band reservation success is confirmed by the band reservation response A5 returned in response to the band reservation request R5 from the second layer subnet manager M1. Bandwidth reservation requests R6 and R7 are sent from the subnet manager M1. As a result, after it is confirmed that the bandwidth reservation is successful in the link L5 connecting the two-layer sub-networks NW1 and NW2,
A bandwidth reservation request is made within the layer sub-networks NW1 and NW2.

Therefore, bandwidth reservation requests R5, R6, R
When 7 are transmitted in parallel, speedup of bandwidth reservation is realized as described above, while on the other hand, when bandwidth reservation requests R6 and R7 are transmitted after successful bandwidth reservation for the link L5. Does not need to perform the band release process for the link that has made the band reservation by the band reservation requests R6 and R7 when the band reservation for the link L5 is unsuccessful.
The processing load on each subnet manager and switching node is reduced, and the utilization efficiency of the lower layer sub-network is improved.

Next, in response to the bandwidth reservation request R6, 1
The path setting / releasing request unit 54 of the layer subnet manager M11 refers to the inter-subnet routing table in the inter-subnet routing table holding unit 53, as in the case of the above-mentioned two-layer subnet manager M1, and the first-layer sub-networks NW11 and NW12. The optimum route connecting to and is extracted, and the bandwidth reservation request R8 is transmitted to all the links that form the extracted route, here, the link L3. [Band reservation request R8] Destination: M11 → 112, 121 Route: L3 Actually, the band reservation is made at the switching nodes 112 and 121 at both ends of the link L3, and the band reservation response A8 is sent in response to the success of the band reservation. Will be returned.

Further, the one-layer subnet manager M11
The path setting / release requesting unit 54 refers to the in-subnet routing table in the in-subnet routing table holding unit 52 and configures all the optimal routes connecting the already-extracted 1-layer sub-networks NW11 and NW12. A bandwidth reservation request for the extraction route in each of the first layer sub-networks NW11 and NW12 is sent to the first layer subnet managers M11 and M12 of the first layer sub-networks NW11 and NW12.

That is, the first layer sub-network NW11
For the first layer subnet manager M11, the switching node 1 accommodating the calling terminal T1 (link L1)
A bandwidth reservation request R9 is sent to the route connecting 11 and the switching node 112 accommodating the link L3. [Band reservation request R9] Destination: M11 → M11 Route: 111 to 112 In addition, for the first layer sub-network NW12, the links L3 and L to the first layer subnet manager M12.
The bandwidth reservation request R10 is sent to the route connecting the switching nodes 121 and 122 accommodating 5 respectively. [Band reservation request R10] Destination: M11 → M12 Route: 121 to 122

Further, in response to the bandwidth reservation request R6 from the two-layer subnetwork M1 and R7 sent in parallel, 1
The path setting / releasing request unit 54 of the layer subnet manager M21 refers to the inter-subnet routing table in the inter-subnet routing table holding unit 53, as in the case of the above-mentioned one-layer subnet manager M11, and the first layer sub-networks NW21 and NW22. The optimum route connecting to and is extracted, and the bandwidth reservation request R11 is sent to all the links forming the extracted route, here, the link L7. [Band reservation request R11] Destination: M21 → 212,221 Route: L7 Actually, the band reservation is made at the switching nodes 212 and 221 at both ends of the link L7, and the band reservation response A11 is sent in response to the success of the band reservation. Will be returned.

Further, the one-layer subnet manager M12
The path setting / releasing request unit 54 refers to the intra-subnet routing table in the intra-subnet routing table holding unit 52, similarly to the above-mentioned one-layer subnet manager M11, and refers to the already-extracted first-layer sub-network NW21. 1 of all 1-layer sub-networks NW21 and NW22 existing on the optimum route connecting to NW22
1 for each layer subnet manager M21, M22
A bandwidth reservation request for the extraction route in the layer sub-networks NW21 and NW22 is transmitted.

That is, the first layer sub-network NW21
For the first-layer subnet manager M21, the switching node 2 accommodating the links L5 and L7, respectively.
Bandwidth reservation request R1 for the route connecting 11 and 212
2 is sent out. [Band reservation request R12] Destination: M21 → M21 Route: 211 to 212 Further, for the first layer sub-network NW22, the switching node 221 accommodating the link L7 and the destination terminal T2 (link The bandwidth reservation request R13 for the route connecting the switching node 222 accommodating L9) is transmitted. [Band reservation request R13] Destination: M21 → M22 Route: 221 to 222

In response to the bandwidth reservation request R9 from the first layer subnet manager M11, the first layer subnet manager M
The path setting / release requesting unit 54 of 11 refers to the inter-subnet routing table in its own inter-subnet routing table holding unit 53, extracts the optimum route connecting the exchange node 111 and the exchange node 112, and is extracted. The bandwidth reservation request R14 is sent to all the links forming the route, here the link L2. [Band reservation request R14] Destination: M11 → 111, 112 Route: L2

Similarly, in response to the bandwidth reservation request R10 from the first layer subnet manager M11, the path setup / release request section 54 of the first layer subnet manager M12
By referring to the inter-subnet routing table in its own inter-subnet routing table holding unit 53, the optimum route connecting the exchange node 121 and the exchange node 122 is extracted, and all the links forming the extracted route, in this case, the links. A bandwidth reservation request R15 is sent to L4. [Band reservation request R15] Destination: M12 → 121, 122 Route: L4

Further, the one-layer subnet manager M21
In response to the bandwidth reservation requests R12 and R13 from, the path setting / release requesting units 54 of the first layer subnet managers M21 and M22 respectively refer to the inter-subnet routing table in the inter-subnet routing table holding unit 53, A bandwidth reservation request is made to each of all the links forming the extracted routes, here, the optimum routes connecting the switching nodes 211 and 212 and the optimum routes connecting the switching nodes 221 and 222, in this case, the links L6 and L8. Sends R16 and R17.

[Band Reservation Request R16] Destination: M21 → 211, 212 Route: L6 [Band Reservation Request R17] Destination: M22 → 221, 222 Route: L8 Therefore, the bandwidth reservation request R6 from the second layer subnet manager M1. R7 is sent in parallel and each 1
Bandwidth reservation requests R8, R9, R10 and R11, 12, 13 are sent in parallel from the layer subnet managers M11, M21, and each one-layer subnet manager M1.
1, M12, M21, M22 to request bandwidth reservation R14-
R17 will be sent in parallel.

Each time each subnet manager sends a bandwidth reservation request, it sends route information of the route requested by the bandwidth reservation request, such as a link or subnetwork identifier, or a subnet manager or switching node identifier. Setting route storage 65,5
Store in 5. This route information is used to grasp the route used for communication when a band release request described later is received.

In response to these bandwidth reservation requests R14 to R17, switching nodes 111 and 112, 121 and 122, and 211 and 21 at both ends of the links L2, L4, L6 and L8.
Band reservations are made at 2, 221, and 222, respectively, and band reservation responses A14 to A17 are returned in response to successful band reservation. Each one-layer subnet manager M11, M12, M
21 and M22 respond to the bandwidth reservation responses A14 to A17, respectively, to the first layer subnet managers M11 and M, respectively.
Bandwidth reservation responses to A21, A9, A10, A12, A1
Send back 3.

Subsequently, the first layer subnet manager M11
Responds to the bandwidth reservation responses A9 and A10 to the upper-layer two-layer subnet manager M1.
6 is returned and the 1st layer subnet manager M2
In response to these bandwidth reservation responses A12 and A13, 1 returns a bandwidth reservation response A6 to the upper two-layer subnet manager M1. In response to this, the two-layer subnet manager M1 sends the bandwidth reservation request R3 received first, that is, the switching node 111 accommodating the calling terminal T1, to the bandwidth in the route from the switching node 111 to the destination terminal T2. The bandwidth reservation response A3 notifying that the reservation is successful is returned.

As a result, the switching node 111 returns to the calling terminal T1 a band reservation response A1 notifying that the band reservation in the route from the calling terminal T1 to the called terminal T2 has succeeded, and in response thereto. From the originating terminal T1 to the terminating terminal T
The transmission of the communication data to 2 is started, and the data communication using the band reserved and reserved by each link and switching node is realized.

If the bandwidth reservation is unsuccessful at any of the links constituting the route extracted by each subnet manager, a bandwidth reservation response indicating that fact is returned. In this case, , The route information of the unsuccessful route is read from the set route holding units 65 and 55, and the bandwidth release request is sent to each link, subnet manager, or switching node related to the route, so that the route is unsuccessful. The bandwidth of the link for which bandwidth reservation has already been made on the route that has become

At the same time, referring to its own inter-subnet routing table or intra-subnet routing table, the next most suitable route is extracted, and a bandwidth reservation request is sent to all the links of the same layer making up that route. Then, it is repeatedly performed a predetermined number of times depending on the failure of the bandwidth reservation. Note that the number of repetitions due to unsuccessful bandwidth reservation may be performed based on a value set in advance for each layer, that is, communication quality information, or according to individual bandwidth reservation requests from the calling terminal T1. You may make it implement.

In particular, when carrying out according to each bandwidth reservation request, the number of repetitions of bandwidth reservation in each layer is set as communication quality information (for example, QOS parameter etc.) in the bandwidth reservation request from the calling terminal T1. When the bandwidth reservation is unsuccessful, the subnet manager of each tier repeats the bandwidth reservation based on the number of repetitions of its own tier set in the bandwidth reservation request. It will be done. This makes it possible to provide a communication service based on the communication quality requested by the user, and the user can set the communication quality for each VC connection according to the importance of the communication data. , Effective use of bandwidth is realized.

Further, in the above description, the calling terminal T
The case where 1 and T2 are accommodated in different two-layer sub-networks has been described, but when they are accommodated in the same two-layer sub-network, the switching node 111 responds to the bandwidth reservation request from the calling terminal T1. Since the originating and terminating terminal addresses are compared with each other and become the different addresses for the first layer sub-network, the bandwidth reservation request R3 sent from the switching node 111 is transmitted to the first layer subnet manager M.
11, the bandwidth reservation request R4 for the link with the destination terminal T2 is transmitted from the 1st layer subnet manager M11, and the bandwidth reservation requests for the routes in the 2nd layer subnetwork NW1 are sequentially transmitted as described above. Will be done.

When the originating and terminating terminals T1 and T2 are accommodated in the same 1st-layer subnetwork, the aforementioned bandwidth reservation request R3 sent from the switching node 111 is sent to the 1st-layer subnet manager M11. Become a thing,
A bandwidth reservation request R4 for the link with the destination terminal T2 is sent from the 1st-layer subnet manager M11, and the bandwidth reservation request for the route in the 1st-layer subnetwork NW11 is sent to each switching node as described above. . Furthermore, if the originating and terminating terminals T1 and T2 are accommodated in the same switching node 111, the switching node 11
Bandwidth reservation request R4 for link from 1 to destination terminal T2
Is transmitted, and the bandwidth reservation response A1 is returned to the calling terminal T1.

Next, with reference to FIGS. 1 and 8, a band releasing process for the route to the destination terminal T2 in response to a band releasing request from the calling terminal T1 will be described as an operation of the present invention. FIG. 8 shows a VC from the calling terminal T1 to the called terminal T2.
FIG. 3 is a sequence diagram showing a release procedure, in which the same parts as those in FIG. 1 described above are designated by the same reference numerals. FIG.
, R21 to R37 are band release requests, A21 to A
Reference numeral 37 is a band release response indicating the band release success returned in response to the band release requests R21 to R37.

First, after the end of the data communication, in response to the communication end request generated by the calling terminal T1, the switching node 111, which is the accommodating source of the calling terminal T1, indicates the amount of bandwidth required for communication and the called terminal indicating the called terminal. A bandwidth release request R21 including the address and the self-address indicating the calling terminal is transmitted. [Band release request R21] Destination: T1 → 111 Route: T1 to T2 The switching node 111 sends a band release request R from the calling terminal T1.
In response to 1, the bandwidth release request R2 of the link L1 connecting with the calling terminal T1 is transmitted. [Band release request R22] Destination: 111 → 111, T1 Route: Link L1

Actually, the bandwidth release request R2 for the link L1 is sent to the nodes at both ends of the link L1, that is, the switching node 111 and the calling terminal T1, and both respond to the bandwidth release request from the calling terminal T1. Bandwidth is released. When both parties release the bandwidth successfully, both parties return a bandwidth release response A22 indicating the successful bandwidth release to the exchange node 111, and the exchange node 111 completes the bandwidth release to the link L1 with the calling terminal T1. After confirming that the bandwidth has been released, a bandwidth release response A21 for notifying that the bandwidth has been released successfully is returned to the calling terminal T1.

In response to the completion of the band release to the link L1, the switching node 111 sends the band reservation request R3 stored at the time of sending the band reservation request R3, here, 2
A band release request R23 is sent to the second layer subnet manager M1 of the layer sub-network NW1. [Band release request R23] Destination: 111 → M1 Route: 111 to T2

In the above description (see FIG. 8), after the bandwidth release from the switching node 111 to the link L1 is completed,
Band release request R23 to the second layer subnet manager M1
However, the transmission order of these band release requests is not particularly limited. For example, the band release request R23 is sent to the two-layer subnet manager M1 in parallel or is sent. After that, the bandwidth release request R22 for the link L1 may be transmitted. Generally, the bandwidth release processing for each link up to the destination terminal T2 requires more time than the bandwidth release processing for the link L1. By sending the bandwidth release request R22 at the same time as or after sending the request R23,
Both bandwidth release processes are performed in parallel, and the time required for bandwidth release is shortened.

In response to the band release request R23, the path setting / release requesting unit 64 of the two-layer subnet manager M1
Sends a band release request R24 to the link L9 connecting the destination terminal T2 and the exchange node 222 that is the accommodation source. [Band release request R24] Destination: M1 → 222, T2 Route: L9 Actually, similarly to the above-described bandwidth release request for the link L1 on the side of the calling terminal T1, the switching node 222 and the destination terminal T2 at both ends of the link L9. The band is released in step S3 and the band release response A24 is returned in response to the success of the band release.

In this way, in response to a band release request from the calling terminal T1, each link forming a route connecting the exchange node 111, which is the accommodation source of the calling terminal T1 that requires time to release the band, and the called terminal T2. After making a bandwidth release request,
Since the bandwidth is released for the link L1 between the calling terminal T1 and the exchange node 111 which is the accommodation source,
The bandwidth can be released quickly, the degree of reuse for bandwidth reservation for other outgoing calls or bursts is improved, and efficient bandwidth release processing is performed.

On the other hand, in response to the bandwidth release request R23, the path setup / release request section 64 of the two-layer subnet manager M1.
Refers to the set route holding unit 65, reads out the route used for this data communication, and sends out a band release request R25 to all the links of the same layer constituting this route, here L5. [Band release request R25] Destination: M1 → 122, 211 Route: L5 In reality, the switch nodes 122 and 211 at both ends of the link L5 perform the band release, and the band release response A25 is sent in response to the successful band release. Will be returned.

In this case, the two-layer subnet manager M1
Then, in response to the bandwidth release request R23, the bandwidth release request R24 for the link L9 and the bandwidth release request R25 for the link L5 are independently transmitted. Normally, unlike the bandwidth reservation processing, the bandwidth release processing executed in response to each bandwidth release request does not require a new resource for executing the processing, that is, the bandwidth, so that it always succeeds. .

Further, even if the switching node instructed to release the bandwidth is out of order, the adjacent switching nodes are hardly affected by the delay in the release of the bandwidth. The bandwidth is released by the restoration process. Therefore, since it is less necessary to confirm the bandwidth release response to the bandwidth release request, by sending the bandwidth release request R25 in parallel with the bandwidth release request R24, the time required for the bandwidth release processing is shortened as a whole. It

After confirming the bandwidth release responses A24 and A25, 2
Path setup / release request unit 6 of the layer subnet manager M1
4 is a two-layer subnetwork NW on the read route
1, 2 for each one of the subnet managers in the lower layer, that is, one of the first-layer subnet managers M11 and M12, and one of the first-layer subnet managers M21 and M22.
For the route used for data communication in the layer sub-network, that is, for the two-layer sub-network NW1,
A route connecting the 1st layer subnetwork NW11 accommodating the calling terminal T1 and the 1st layer subnetwork NW12 accommodating the link L5, and for the 2nd layer subnetwork NW2, 1 accommodating the NW 21 accommodating the link L5 and the destination terminal T2 Band release requests R26 and R27 for routes connecting the layer sub-network NW22 are transmitted.

In this case, for example, in the second layer sub-network NW1, the first layer subnet manager M11 is selected by the same method as the method of selecting the first layer subnet manager at the time of bandwidth reservation described above, and the second layer sub network N is selected.
In W2, the 1st layer subnet manager M21 is selected,
For the 1st-layer subnet manager M11, the 1st-layer subnetwork NW11 including the switching node 111 that accommodates the calling terminal T1 and the 1st-layer subnet that accommodates the link L5 from the path setting / release request unit 64 of the 2nd-layer subnet manager M1. A bandwidth release request R26 for the route to the network NW12 is transmitted. [Band release request R26] Destination: M1 → M11 Route: NW11 to NW12

Further, the one-layer subnet manager M21
Between the layer 1 subnetwork NW22 including the switching node 222 accommodating the destination terminal T2 and the layer 1 subnetwork NW21 accommodating the link L5 from the path setup / release request unit 64 of the layer 2 subnet manager M1. Band release request R27 for the route of
26 in parallel. [Band release request R27] Destination: M1 → M21 Route: NW21 to NW22

In the above description, the path setting / release request unit 64 of the second layer subnet manager M1 confirms the band release success of the links L9, L5 by the band release responses A24, A25, and then the first layer subnet manager. M
The case has been described where the band release requests R26 and 27 to the M11 and M21 are transmitted. However, as in the case of the band release requests R24 and R25 described above, since it is less necessary to confirm the band release response to the band release request, before confirming the band release responses A24 and A25, for example, the band release request R25 is confirmed. The bandwidth release requests R26 and R27 may be transmitted in parallel with the above, and as a result, the time required for the bandwidth release processing is shortened as a whole.

In response to the band release request R26, the path setting / release requesting unit 54 of the first layer subnet manager M11.
Similarly to the above-mentioned two-layer subnet manager M1, refers to the set route holding unit 55 to read the route connecting the first-layer sub-networks NW11 and NW12, and all the links constituting the route, here, Link L
A band release request R28 is sent to the packet No.3. [Band release request R28] Destination: M11 → 112, 121 Route: L3 Actually, the exchange nodes 112 and 121 at both ends of the link L3 perform the bandwidth release, and the bandwidth release response A28 is returned in response to the successful bandwidth release. To be done.

Further, the one-layer subnet manager M11
The path setting / release requesting unit 54 of the
For all the 1st-layer subnet managers M11, M12 existing on the route connecting the already-read 1st-layer subnetworks NW11 and NW12, in the 1st-layer subnetworks NW11, NW12 Send a bandwidth release request for the route.

That is, the first layer sub-network NW11
For the layer 1 subnet manager M11, the switching node 111 accommodating the calling terminal T1 and the link L
A bandwidth release request R29 is sent to the route connecting with the switching node 112 accommodating No. [Band release request R29] Destination: M11 → M11 Route: 111 to 112 Further, for the first layer sub-network NW12, the links L3 and L to the first layer subnet manager M12.
A bandwidth release request R30 for the route connecting between the switching nodes 121 and 122 accommodating 5 respectively is transmitted. [Band release request R30] Destination: M11 → M12 Route: 121 to 122

Further, in response to the bandwidth release request R26 from the second layer sub-network M1 and R27 sent in parallel, the path setting / release request section 54 of the first layer subnet manager M21 causes the aforementioned first layer subnet manager M11.
Similarly, by referring to the set route holding unit 55, the route connecting the first layer sub-networks NW21 and NW22 is read out, and the bandwidth release request R31 is issued to all the links forming the route, here the link L7. Is sent. [Band release request R31] Destination: M21 → 212,221 Route: L7 In reality, the switch nodes 212 and 221 at both ends of the link L7 release the band, and in response to the successful band release, the band release response A31 is sent. Will be returned.

Further, the one-layer subnet manager M12
The path setting / releasing request unit 54 refers to the set route holding unit 55, similarly to the above-mentioned 1st layer subnet manager M11, and has already read the 1st layer subnetwork N.
A bandwidth release request for the route in each of the first-layer sub-networks NW21, NW22 is sent to all the first-layer subnet managers M21, M22 existing on the route connecting the W21 and the NW22.

That is, the first layer sub-network NW21
For the first-layer subnet manager M21, the switching node 2 accommodating the links L5 and L7, respectively.
A bandwidth release request R32 for the route connecting 11 and 212 is transmitted. [Band release request R32] Destination: M21 → M21 Route: 211 to 212 Further, for the first layer sub-network NW22, the switching node 221 accommodating the link L7 and the destination terminal T2 (link The bandwidth release request R33 is sent to the route connecting the switching node 222 accommodating L9). [Band release request R33] Destination: M21 → M22 Route: 221 to 222

In response to the bandwidth release request R29 from the 1st layer subnet manager M11, the path setting / release requesting unit 54 of the 1st layer subnet manager M11 refers to its own set route holding unit 55 and exchanges with the switching node 111. The route connecting the node 112 is read out, and the band release request R34 is sent to all the links forming the route, here, the link L2. [Band release request R34] Destination: M11 → 111, 112 Route: L2

Similarly, in response to the bandwidth release request R30 from the first layer subnet manager M11, the path setting / release request unit 54 of the first layer subnet manager M12
Switching node 1 referring to its own set route holding unit 55
The route connecting the switch 21 and the switching node 122 is read out, and the band release request R35 is sent to all the links forming the route, here the link L4. [Band release request R35] Destination: M12 → 121,122 Route: L4

Further, the one-layer subnet manager M21
In response to the bandwidth release requests R32, R33 from the path setting / release requesting unit 54 of the first layer subnet managers M21, M22, the switching nodes 211, 212 are connected by referring to their own set route holding units 55, respectively. The route that has been opened and the route that connects the switching nodes 221 and 222 are respectively extracted, and bandwidth release requests R36 and R37 are sent to all the links that compose the extracted route, here, links L6 and L8. .

[Band Release Request R36] Destination: M21 → 211,212 Route: L6 [Band Release Request R37] Destination: M22 → 221,222 Route: L8 Therefore, the two-layer subnet manager M1 requests the bandwidth release R26, R27 is sent out in parallel,
Band release requests R28, R29, R30 and R31, 32, 33 are sent in parallel from the 1st layer subnet managers M11, M21, and the band release request R is made from the 1st layer subnet managers M11, M12, M21, M22.
34 to R37 are transmitted in parallel.

Actually, these band release requests R34 to R
37, switching nodes 111 and 112, 121 and 122, and 2 at both ends of links L2, L4, L6, and L8, respectively.
Band releases are performed at 11 and 212, 221 and 222, respectively, and band release responses A34 to A3 are generated in response to successful band release.
7 is returned. Each one-layer subnet manager M11,
M12, M21, M22 are these band release responses A34
~ A37, the bandwidth release responses A29, A30, to the first-layer subnet managers M11, M21, respectively.
After returning A32 and A33, the route information of the route whose band has been released successfully is deleted from the set route holding unit 55.

Subsequently, the 1st layer subnet manager M11
Responds to these band release responses A29 and A30 by returning a band release response A26 to the upper two-layer subnet manager M1, and the first layer subnet manager M21 responds to these band release responses A32 and A33. The band release response A26 is returned to the upper-layer two-layer subnet manager M1, and thereafter, the route information of the route whose band is successfully released is deleted from the set route holding unit 55.

In response to this, the two-layer subnet manager M1 sends the first received bandwidth release request R23,
That is, a bandwidth release response A notifying the switching node 111 accommodating the calling terminal T1 that the bandwidth release in the route from the switching node 111 to the destination terminal T2 has succeeded.
After returning 23, the route information of the route whose band has been released successfully is deleted from the set route holding unit 65. As a result, the switching node 111 accommodating the source terminal T1 and the destination terminal T
The bandwidth release for all the links forming the route that has been connected to the switching node 222 accommodating 2 is performed in parallel, and the bandwidth release is performed more quickly, resulting in an efficient bandwidth. Utilization is realized.

In the above description, the sub-network is described as an example of a network composed of two layers, that is, one layer and two layers, but the same applies to a network composed of three or more layers. Yes, when the number of layers is indicated by m, each lowest layer (m = 1) sub-network is composed of a plurality of switching nodes, and each lowest layer sub-network has the above-mentioned one-layer subnet manager (FIG. 2). Is provided as the lowest layer subnet manager, and each upper layer (m> 1) subnetwork above this is configured from a plurality of subnetworks of one layer lower (m-1), and each upper layer subnetwork is described above. This is realized by providing the two-layer subnet manager (Fig. 2) as the upper layer subnet manager.

In the above description, the case where the respective switching nodes and the sub-networks are connected by a predetermined link has been described. May be. In this case, of switching nodes located at both ends of a link configured by a network based on another connection procedure, one switching node responds to this network in response to a bandwidth reservation / release request from a predetermined subnet manager. The bandwidth between these switching nodes is requested by requesting the call setup / release and immediately returning a bandwidth reservation / release response from the other switching node to the subnet manager in response to the call setup / release request from this network. The reservation / release processing time can be shortened.

For example, when a network based on another connection procedure is present in a link connecting switching nodes in a sub-network, such as the link L2 in FIG. 1, the 1-layer subnet manager M11 sends the switching nodes 111, 112 Then, the bandwidth reservation request R14 is transmitted, and in response thereto, the call setting based on a predetermined connection procedure is transmitted to the network from the switching node 111 on the calling terminal T1 side. In response to this, call setup is performed between the nodes constituting this network, and the switching node 112 is switched from this network.
The call setup is sent to.

Here, in general, switching node 112 returns a call setup response to the network, which is sent back to switching node 111 via the network. After confirmation of the call setup response by switching node 111, switching node 11
The bandwidth reservation response A14 is returned to the 1st to 1st layer subnet manager M11. In this case, the destination terminal T2
The switching node 112 on the side performs bandwidth reservation processing in response to a call setup request from the network, and a bandwidth reservation response indicating the processing result is immediately sent to the first layer subnet manager M.
Returned to 11.

On the other hand, the same applies to the case where a network based on another connection procedure intervenes in a link connecting sub-networks like the link L3 in FIG. For example, the bandwidth reservation request R8 for the link L3 is sent from the 1st layer subnet manager M11 to the switching node 112 and is also sent to the switching node 121 via the 1st layer subnet manager M12, in response to a call setting request from the network Immediately from the switching node 121 through the layer 1 subnet manager M12, the layer 1 subnet manager M1
Will be returned to 1.

[0117] Here, the case of requesting bandwidth reservation (at the time of call setup) has been described as an example, but the same is true at the time of bandwidth release request (at the time of call release). Also, after receiving the bandwidth reservation / release response from the destination node switching node, the subnet manager of the bandwidth reservation / release request source sends the bandwidth reservation / release response based on the bandwidth reservation / release response sent from the originating terminal side switching node. The bandwidth reservation / release on the link via the network may be reconfirmed or the normal operation of the switching node on the calling terminal side may be confirmed.

As described above, when a network having another connection procedure is involved, the switching node on the called terminal side directly responds to the subnet manager of the bandwidth reservation / release request source in response to a call setup / release request from the network. Since the bandwidth reservation / release response is returned, the bandwidth reservation / release can be confirmed by the requesting subnet manager in a shorter time compared to the case of returning from the switching node on the calling terminal side via the network. The bandwidth reservation / release processing time between these switching nodes can be shortened.

[0119]

As described above, according to the present invention, a network has a hierarchical structure composed of a plurality of lower sub-networks and an upper sub-network, and each sub-network is provided with a subnet manager to be connected in the same hierarchy. The optimal route between the sub-networks is managed individually, and especially the lowest (1st layer) subnet manager is also configured to individually manage the optimal route between each switching node in the 1-layer sub-network which it manages. Therefore, all the links that make up the network are managed in a decentralized manner, and it is possible to avoid failures caused by centralized management processing for the switching node and subnet manager that accommodate the originating and destination terminals, and specialize in these management processing. Providing a node in the network with high processing power to execute Ku, it is possible to quickly select the best route. In addition, bandwidth reservation requests for each link and the lower subnet manager are sent in parallel and processed, respectively, and faster bandwidth reservation can be performed for the route between the originating and terminating terminals.

Also, in response to a bandwidth reservation request from the calling terminal, bandwidth reservation for the link connecting the calling terminal and the switching node accommodating this is performed, and upon success, the call is received from the switching node accommodating the calling terminal. Bandwidth reservations are made for each relay link and subscriber link that make up the route connecting the terminal, so that the link between the source terminal and the switching node that accommodates it is requested for bandwidth reservation before other links. As a result, the desired bandwidth is reserved, and even if the bandwidth reservation for other links is successful, the bandwidth reservation for the link with the calling terminal becomes unsuccessful and call loss can be prevented. Good bandwidth reservation processing will be performed.

Also, each terminal is given a network address consisting of the sub-network of each layer and the identifier of the switching node, and the switching node accommodating the calling terminal responds to this request in response to the bandwidth reservation request from the calling terminal. Comparing the network addresses of the originating and terminating terminals included in this sequence in order from the upper layer identifier, and for the subnet manager corresponding to the originating terminal side identifier in the layer where the first different identifier appears, for the route from the own switching node to the destination terminal Since the bandwidth reservation request is transmitted, it is possible to quickly and accurately extract the minimum required uppermost subnet manager for connecting the originating and terminating terminals in the network having a hierarchical structure.

When each subnet manager returns a bandwidth reservation response indicating failure of bandwidth reservation in response to the transmitted bandwidth reservation request, the contents held by itself are referred to and the next optimum route is obtained. Since a new bandwidth reservation request for that route is sent out, even if bandwidth reservation fails on a specific link, the most appropriate route that can reserve bandwidth at that point in time. Is selected, bandwidth reservation is performed, and the possibility of call loss can be reduced.

In addition, each subnet manager issues a new bandwidth reservation request for the next optimum route in response to the failure in bandwidth reservation. Repeatedly, the bandwidth reservation response indicating that the bandwidth reservation was unsuccessful was sent back to the requester of the bandwidth reservation request received by itself in response to the subsequent failure in bandwidth reservation. Uneven route lengths that would result in an enormous number of transit subnets and extension of bandwidth reservation processing time are alleviated, and efficient routes are selected in any of the layers.

Also, the calling terminal sets the number of repetitions when the bandwidth reservation request is unsuccessful in the bandwidth reservation request corresponding to each layer, and sends it out. A new bandwidth reservation request for a different route is repeatedly executed by the number of repetitions when the bandwidth reservation in the own layer set in the bandwidth reservation request is unsuccessful. Since the bandwidth reservation response indicating that the bandwidth reservation has failed is returned to the request source, it is possible to provide the communication service based on the communication quality according to the user's request, and use it for each VC connection. The person can set the communication quality according to the importance of the communication data, and the effective use of the band is realized.

Further, the switching node is provided with a link allocatable bandwidth amount holding unit for storing the link allocatable bandwidth amount of each link and a notifying unit for notifying the upper subnet manager of the link allocatable bandwidth amount, The subnet manager is notified of the latest link allocatable bandwidth amount, and the table management unit of the first layer subnet manager sets the routing table holding unit in the sub-network in ascending order of the number of transit nodes and these link allocatable bandwidth amounts. In addition to updating and holding, the table management unit of each subnet manager routes the inter-subnet routing in the descending order of the link allocatable bandwidth amount of the boundary links exchanged between the prescribed subnet managers that are connected in the same hierarchy with a small number of transit subnets. Update the contents held in the table holder As a result, individual subnet managers can always manage the latest optimal route in a distributed manner without the need for special nodes for route management or complicated configurations, and each table can be maintained in response to bandwidth reservation requests. The optimum route can be determined only by reading the routes in the order held by the section, and the optimum route at the time of securing the band can be immediately obtained from a large number of route groups.

Furthermore, the path setup / release request section of each subnet manager sends bandwidth reservation requests for all links connecting the individual sub-networks in parallel along the optimum route, and at the same time positions on the optimum route. A bandwidth reservation request for routes in these sub-networks is sent in parallel to the subnet managers of the respective lower layer sub-networks, and the path setup / release request unit of the first-layer subnet manager performs Since bandwidth reservation requests for all links connecting individual switching nodes are sent in parallel, new bandwidth reservation requests are sent only to the minimum required number of links and sub-managers in response to bandwidth reservation requests. The possibility of bandwidth reservation availability in lower layers as in the past Compared with route determination without consideration, efficient VC connection with a low call loss rate is realized, and a bandwidth reservation request is sent to a route that is unlikely to be actually used as in the past. It is possible to solve the problem that the call processing of the entire network is reduced and the call loss rate of the entire network is increased due to useless bandwidth reservation as compared with the case.

Further, the subnet managers and the switching nodes of the respective layers belonging to the common upper sub-network are connected via a predetermined communication line, and between these subnet managers and between the subnet manager and the switching nodes, a hierarchy is provided. Since various control messages are directly exchanged regardless of whether they are the same or not, the arrival time of control messages is shortened compared to the case of transferring via multiple subnet managers, the VC connection time is shortened, and the processing load is reduced. It will be reduced. In addition, since the switching node directly notifies the management source subnet manager that manages each link assignable bandwidth amount, the notification time lag is shortened and the inter-subnet routing table is updated quickly. The reliability is improved and the call loss rate can be reduced.

Further, for each link to be accommodated, the notification unit of the switching node stores the latest link assignable band amount calculated from the empty band of the link and the cell or packet discard rate, and the link assignable band amount holding unit. Compared with the retained link allocatable bandwidth amount, when the difference exceeds a predetermined threshold value, the new link allocatable bandwidth amount is notified to the management source subnet manager of each link. Since the new link assignable bandwidth amount is updated and held in the link assignable bandwidth amount holding unit,
Compared to the case of notifying the link allocatable bandwidth amount, the processing load on the switching node and each subnet manager is reduced, and it is possible to promptly notify the sudden change of the link allocatable bandwidth amount. .

Furthermore, by the notification unit of the switching node,
For each link accommodated, the latest link allocatable bandwidth calculated from the empty bandwidth of that link and the cell or packet discard rate is updated and held in the link allocatable bandwidth holding unit, and the link allocatable bandwidth of each link is stored. Since the amount is periodically notified to the managing subnet manager of each link, each switching node does not need to be configured and processed for comparison with a threshold value, which simplifies the configuration of the switching node. It becomes possible.

Further, one of the switching nodes connected to both ends of each link notifies the managing subnet manager of each link of the link allocatable bandwidth amount of the link. The processing load on the other switching node that does not notify is reduced, and the processing load on the management source subnet manager for managing the amount of link assignable bandwidth is also reduced.

In addition, the path setting / release requesting unit of the subnet manager of the layer which accommodates the sub-network in the lower layer responds to the bandwidth reservation request, and the lower subnet manager in each sub-network of the same layer located on the optimum route. When a new bandwidth reservation request is sent to, the subnet manager with the smallest processing load among the subnet managers, the one closest to the calling terminal as a route, or a predetermined order or random number is used to select an arbitrary subnet manager. Since the bandwidth reservation requests are transmitted in parallel, the processing is distributed to each VC connection without the processing being concentrated on a specific subnet manager in each sub-network. It is possible to reduce the call loss rate due to an abnormal increase in burden.

When each subnet manager makes a bandwidth reservation request to a link composed of a network based on another connection procedure, it makes a bandwidth reservation request to each switching node connected to both ends of the link. One of the switching nodes sends a predetermined call setup request to the network in response to the bandwidth reservation request, and the other switching node of the switching nodes sends a call setup request from the network. A predetermined bandwidth reservation response is sent back to the corresponding subnet manager, so that it can be sent to the requesting subnet manager in a shorter time than when it is sent back from the switching node on the calling terminal side via the network. It is possible to confirm the bandwidth reservation by using this, and it is possible to shorten the bandwidth reservation processing time between these switching nodes.

When each subnet manager sends a bandwidth reservation request to each link on the optimum route or to any subnet manager, the route information of the route set by the bandwidth reservation request is used for individual V
It is held for each C connection or burst connection after call connection, and corresponding route information is read in response to the bandwidth release request, and the bandwidth release request is issued in parallel to the link on the route and the lower subnet manager. Since it is sent sequentially, the bandwidth release request is sent accurately only to the minimum required link and subnet manager, and the bandwidth release processing is performed in parallel on each link and subnet manager. The bandwidth can be released more quickly, and efficient bandwidth utilization is realized for the entire network.

Furthermore, in response to a bandwidth release request from the calling terminal, the bandwidth of each relay link and subscriber link forming a route connecting the exchange node accommodating the calling terminal and the called terminal is released, and then the calling terminal is released. Since it is designed to release the bandwidth of the link connecting the switch and the switching node accommodating it, the bandwidth of the subscriber link between each relay link and the destination terminal can be released quickly, and the bandwidth from other terminals can be released. Even if a new communication start request is generated, the desired band can be secured immediately, and efficient band use can be realized.

When each subnet manager issues a bandwidth release request to a link composed of a network based on another connection procedure, a bandwidth release request is issued to each switching node connected to both ends of the link. One of the switching nodes sends a predetermined call release request to the network in response to the bandwidth release request, and the other switching node of the switching nodes sends a call release request from the network. A predetermined bandwidth release response is sent back to the corresponding subnet manager in response to the request, so that the request to the subnet manager can be sent to the requesting subnet manager in a shorter time than in the case of sending it back from the switching node on the calling terminal side via the network. It is possible to confirm the bandwidth release by using this method, and it is possible to shorten the bandwidth release processing time between these switching nodes.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a hierarchical network system according to an embodiment of the present invention.

FIG. 2 is a functional block diagram showing the internal configuration of each subnet manager and exchange node.

FIG. 3 is a block diagram showing a configuration example of a two-layer subnetwork.

FIG. 4 is an explanatory diagram showing a configuration of various holding units.

FIG. 5 is a block diagram showing a configuration example of a higher level network.

FIG. 6 is an explanatory diagram showing a configuration of an inter-subnet routing table.

FIG. 7 is a sequence chart showing a VC connection procedure.

FIG. 8 is a sequence chart showing a VC release procedure.

[Explanation of symbols]

NW1, NW2 ... 2 layer sub-network, M1, M2 ...
2-layer subnet manager, NW11, NW12, NW
21, NW22 ... 1-layer sub-network, M11, M1
2, M21, M22 ... One-layer subnet manager, 11
1,112,121,122,211,212,22
1, 222 ... Switching node, T1 ... Originating terminal, T2 ... Destination terminal, L1-L9 ... Link.

Claims (16)

[Claims] 1. A calling / receiving terminal via a network composed of a plurality of switching nodes in response to a VC connection / release request or a burst information transmission start / end request after call connection from a calling terminal accommodated in a predetermined switching node. When connecting between
While determining the connection route on the network,
A VC that reserves / releases a band according to a desired communication capacity for a link connecting each switching node on the connection route.
In the connection method, the network is composed of a plurality of switching nodes in which the first layer sub-networks are mutually connected by predetermined links, and the m-th layer (m is a positive integer of 2 or more) sub-networks have predetermined links. Has a hierarchical structure composed of a plurality of (m-1) th layer sub-networks connected to each other by the sub-network of the first layer, for each predetermined switching node in its own sub-network, and the same. A first-layer subnet manager is provided for updating and holding a plurality of routes connecting the switching nodes and the first-layer sub-networks in a predetermined order for each predetermined first-layer sub-network connected in a hierarchy. , The m-th layer sub-network, these m-th layer sub-networks are connected to each other in the same m-th layer sub-network. An m-th layer subnet manager that updates and holds a plurality of routes connecting between layer sub-networks in a predetermined order is provided, and the m-th layer subnet manager refers to its own held contents in response to a bandwidth reservation request. Determines the optimum route connecting the requested sub-networks of the m-th layer, sends out a bandwidth reservation request for the link located on the optimum route, and each m-1 th layer located on the optimum route. Any m-th subnetwork
A bandwidth reservation request for a route in the m-th layer sub-network is sent to the 1st layer subnet manager, and the 1st layer subnet manager refers to its own held contents in response to the bandwidth reservation request from the upper subnet manager. By doing so, the optimum route connecting the predetermined first layer sub-networks is determined as the route in the requested upper sub-network, and the bandwidth reservation request for the link located on the optimum route is sent out.
A bandwidth reservation request for a route in each layer 1 subnetwork is sent to a layer 1 subnet manager of each layer 1 subnetwork located on the optimal route, and a layer 1 located on the optimal route By the subnet manager, in response to a bandwidth reservation request from any layer 1 subnet manager, by referring to the contents held by itself, it is optimal to connect a predetermined switching node as a route in the requested layer 1 subnetwork. A route is determined, a bandwidth reservation request for a link located on the optimum route is transmitted, and the origination terminal responds to the bandwidth reservation request based on a VC connection request from the calling terminal or a burst information transmission start request after call connection. The switching node accommodating the terminal sends a bandwidth reservation request to a predetermined m-th layer subnet manager, By sending bandwidth reservation requests in parallel in the direction of lower layers from the layer subnet manager, bandwidth reservation is performed by the switching nodes at both ends of the link for each link located on the desired route. Characteristic VC connection method. 2. The VC connection method according to claim 1, wherein, in response to a bandwidth reservation request from the calling terminal, bandwidth reservation of a link connecting the calling terminal and a switching node accommodating the calling terminal is performed, and the success is achieved. According to the VC connection method, bandwidth reservation is made for each relay link and subscriber link that form a route connecting the switching node accommodating the source terminal and the destination terminal. 3. The VC connection method according to claim 1, wherein each terminal has a network address consisting of a sub-network of each layer and an identifier of a switching node, and the switching node accommodating the calling terminal is from the calling terminal. In response to the bandwidth reservation request, the network addresses of the calling and receiving terminals included in this request are compared in order from the upper layer identifier, and the subnet manager corresponding to the calling terminal side identifier in the layer in which the different identifier first appears The VC connection method is characterized in that a route from the own switching node to the destination terminal is determined and a bandwidth reservation request for the route is transmitted. 4. The VC connection method according to claim 1, wherein each subnet manager, when a bandwidth reservation response indicating unsuccessful bandwidth reservation is returned to the transmitted bandwidth reservation request, refers to its own content. Then, the optimum route is extracted, and a new bandwidth reservation request for the route is sent out. 5. The VC connection method according to claim 4, wherein each subnet manager responds to failure of bandwidth reservation by
Next, a new bandwidth reservation request for the optimum route is repeatedly executed in common for all layers or for the number of repetitions when the bandwidth reservation is unsuccessful set for each layer, and the self The VC connection method, wherein a bandwidth reservation response indicating that the bandwidth reservation has failed is returned to the received bandwidth reservation request source. 6. The VC connection method according to claim 4, wherein the calling terminal sets a band reservation request with a number of repetitions when band reservation is unsuccessful corresponding to each layer, and sends the band reservation request. Depending on the unsuccessful bandwidth reservation,
Next, a new bandwidth reservation request for the optimum route is repeatedly performed by the number of repetitions when the bandwidth reservation is unsuccessful in the own layer set in the bandwidth reservation request. A VC connection method characterized in that a bandwidth reservation response indicating failure of bandwidth reservation is returned to the received bandwidth reservation request source. 7. The VC connection method according to claim 1, wherein the switching node includes a link allocatable bandwidth amount holding unit that stores the link allocatable bandwidth amount of each link to be accommodated, and the link allocatable bandwidth amount holding unit. The first-layer subnet manager has a notification unit that notifies the upper-layer subnet manager of the retained link allocatable bandwidth amount, and the first-layer subnet manager connects these switching nodes for each predetermined switching node in its own subnetwork. A routing table holding unit in the sub-network for holding a plurality of routes in a predetermined order, and a plurality of routes for connecting the first layer sub-networks for each predetermined first layer sub-network connected in the same hierarchy. Between sub-net routing tables that hold the routes in a predetermined order and the routes that make up each route in the local sub-network. The number of transit nodes indicating the number of nodes is small, and the contents stored in the intra-subnet routing table holding unit are updated in the order of increasing link allocatable bandwidth amount notified from each switching node, and a predetermined number of nodes connected in the same hierarchy are updated. The 1st-layer subnet managers mutually exchange the link allocatable bandwidth of boundary links with adjacent sub-networks, and the number of transit subnets that indicate the number of 1st-layer sub-networks that make up each route is small In descending order, the table management unit for updating the contents held in the inter-subnet routing table holding unit, and the contents held in the inter-subnet routing table holding unit according to the bandwidth reservation request for the route between the first layer sub-networks Then, the optimal route connecting the required layer 1 sub-networks is And sends the bandwidth reservation requests for all the links connecting the individual first layer sub-networks along the optimum route in parallel, and the first layer sub-networks of the respective first layer sub-networks located on the optimum route. Bandwidth reservation requests for routes in each of the first layer sub-networks are sent in parallel to the first layer subnet manager, and the intra-subnet routing table is sent according to the band reservation requests for routes in the first layer sub-network. The optimum route connecting the switching nodes in the requested first layer sub-network is determined by referring to the content held by the holding unit, and bandwidth reservation is made for all links connecting the individual switching nodes along the optimal route. It has a path setting / release request unit that sends requests in parallel, and the m-th layer subnet manager is connected in the same layer. For each predetermined sub-network of the m-th layer, an inter-subnet routing table holding unit that holds a plurality of routes connecting these sub-networks of the m-th layer in a predetermined order, and a predetermined sub-network connected in the same hierarchy The m-layer subnet managers mutually exchange the link allocatable bandwidth amount of boundary links with adjacent sub-networks, and the number of transit subnets indicating the number of m-th layer sub-networks forming each route is small and these link allocatable bandwidths are small. A table management unit that updates the contents stored in the inter-subnet routing table storage unit in descending order of volume, and the contents stored in the inter-subnet routing table storage unit in response to a bandwidth reservation request for a route between the m-th layer sub-networks. Refer to and determine the optimal route connecting the required m-th layer sub-networks. And sends a bandwidth reservation request in parallel to all links connecting the individual m-th layer sub-networks along the optimum route, and at the same time, in any of the m-th layer sub-networks located on the optimum route. M-1 layer subnet managers belonging to each of the mth layer sub-networks
A VC connection method comprising: a path setting / releasing request unit that sends a bandwidth reservation request for a route between layer sub-networks in parallel. 8. The VC connection method according to claim 7, wherein the subnet managers of the respective layers and the switching nodes belonging to a common upper subnetwork are connected via a predetermined communication line, and between the subnet managers of the same layer, Calculated by the switching node by directly exchanging messages such as requests and responses for call control between the subnet managers in the upper and lower layers and between any switching node and the managing subnet manager of the link owned by the switching node. A VC connection method characterized in that the allocated subnet assignable bandwidth amount is directly notified to the management source subnet manager of each link. 9. The VC connection method according to claim 7, wherein the notifying unit of the switching node, for each link to accommodate, the latest link assignable band amount calculated from the empty band of the link and the cell or packet discard rate. And the link allocatable bandwidth amount held in the link allocatable bandwidth amount holding unit, and if the difference exceeds a predetermined threshold value, a new subnet manager is created for each link management source subnet manager. A VC connection method characterized in that the link allocatable bandwidth amount is notified and the new link allocatable bandwidth amount is updated and held in the link allocatable bandwidth amount holding unit. 10. The VC connection method according to claim 7, wherein the notification unit of the switching node, for each link to be accommodated, has the latest link assignable band calculated from the empty band of the link and the cell or packet loss rate. A VC connection method characterized in that the amount is periodically notified to the management source subnet manager of each link. 11. The VC connection method according to claim 7, wherein, from among switching nodes connected to both ends of each link, one of the switching nodes allocates a link to the managing subnet manager of each link. A VC connection method characterized in that the amount of available bandwidth is notified. 12. The VC connection method according to claim 1, wherein the path setup / release request section of the m-th layer subnet manager sets the optimum route on the optimum route in response to a bandwidth reservation request for a route between the m-th layer sub-networks. For each m-th layer sub-network located, from each m-1 th layer subnet manager belonging to the m th layer sub-network, the one with the smallest processing load among the m-1 th layer subnet managers, the route The closest to the calling terminal, a predetermined order or a random number, an arbitrary m-1 layer subnet manager is selected, and for each selected m-1 layer subnet manager, an individual m layer sub V is characterized in that a bandwidth reservation request for a route between the (m-1) th layer subnetworks belonging to the network is transmitted in parallel. Connection method. 13. The VC connection method according to claim 1, wherein each subnet manager is connected to both ends of the link when making a bandwidth reservation request to a link configured from a network based on another connection procedure. A bandwidth reservation request to each switching node, one switching node of the switching nodes sends a predetermined call setup request to the network in response to the bandwidth reservation request, each switching node The other switching node is a VC connection method characterized in that a predetermined bandwidth reservation response is returned to a corresponding subnet manager in response to a call setup request from the network. 14. The VC connection method according to claim 1, wherein when each subnet manager sends a bandwidth reservation request to each link on the optimum route or an arbitrary subnet manager, the bandwidth reservation request The route information of the route to be set is held for each VC connection or for each burst connection after the call connection, and the m-th layer subnet manager refers to the route information held by itself in response to the bandwidth release request, and responds. A band release request for a link located on a used route used for a VC connection or a burst connection after a call connection is sent, and an m-th layer to each m-1 layer subnet manager located on the used route. Sending a bandwidth release request for a route in the layer subnetwork, By referring to the route information held by itself in response to the bandwidth release request from the upper subnet manager, the bandwidth release request for the link located on the route used for the corresponding VC connection or burst connection after the call connection And sends a bandwidth release request for a route in each layer 1 subnetwork to the layer 1 subnet manager of each layer 1 subnetwork located on the route, and is located on the route. In response to a bandwidth release request from the layer 1 subnet manager, the layer 1 subnet manager refers to the route information held by the layer 1 subnet manager so that the route is used for the corresponding VC connection or the route used for the burst connection after the call connection. A band release request is sent to the located link, and the VC release request from the calling terminal is required. Alternatively, in response to a bandwidth release request based on a burst information transmission termination request after call connection, the switching node accommodating the calling terminal transmits a bandwidth release request to a predetermined m-th layer subnet manager, and the m-th layer Bandwidth release requests are sent sequentially in parallel from the subnet manager to the lower layers, so that the bandwidth is released by the switching nodes at both ends of each link located on the route used for communication. VC connection method characterized by. 15. The VC connection method according to claim 14, wherein, in response to a band release request from the calling terminal, each relay link and a subscriber forming a route connecting a switching node accommodating the calling terminal and a called terminal. A VC connection method, characterized in that after a request for releasing a band of a link is made, a band of a link connecting the calling terminal and a switching node accommodating the calling terminal is released. 16. The VC connection method according to claim 14, wherein each subnet manager is connected to both ends of the link when making a bandwidth release request to a link configured from a network based on another connection procedure. A bandwidth release request to each switching node, one of the switching nodes sends a predetermined call release request to the network in response to the bandwidth release request, and each switching node The other switching node among the above is configured to return a predetermined band release response to a corresponding subnet manager in response to a call release request from the network, and the VC connection method.