JP2003069615A - Communication controller and communication control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system which reduces an influence of delay in a line without changing existing protocols to realize high-speed communication. SOLUTION: One or more routers 70a and 70b are arranged between a transmission source host 50 and a transmission destination host 60. The router 70a generates and transmits an acknowledge packet in response to a packet sent from the transmission source host 50 and terminates and divides a connection between the transmission source host 5 and the transmission destination host 60. The address of the transmission destination host is used as the start point address of the acknowledge packet. The packet having the address of the transmission source host as the start point address is delivered to the transmission destination host 60. Thus both hosts are made to recognize one virtual connection.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
Control Protocol (Transport ControlProtoco
l, hereinafter referred to as TCP protocol), a communication control device, a system, and a communication control method for realizing higher-speed communication.

[0002]

2. Description of the Related Art The TCP protocol has various functions such as error correction and flow control and is widely used as a protocol for highly reliable data communication. In the TCP protocol, a data reception confirmation response is performed between the transmission side host and the reception side host.

When performing TCP communication via a line with a large delay, it takes a very long time until data is sent from the sending host to the receiving host and the sending host receives the confirmation response returned by the receiving host. It costs.

Regarding the TCP protocol, RFC (Re
quest for Comments) 793.

[0005]

Due to the characteristics of the TCP communication described above, in the TCP communication via the line with a large delay, the TCP window of the sending host is immediately filled up and sent from the receiving host. You will not be able to send more data until you receive an acknowledgment. For this reason,
Even if the transfer rate of the delay line is high, it is difficult to fully utilize the bandwidth due to the structure of the TCP protocol.

As a solution to such a problem, T
It is conceivable to improve the CP protocol itself and change it so as to reduce the influence of line delay. In this case, although it is technically possible to improve the TCP protocol, it cannot be said to be a practical solution because the tens of millions or more communication devices related thereto must be improved at the same time. .

It is an object of the present invention to provide a system capable of reducing the influence of delay on the line and performing higher speed communication without changing the existing protocol.

[0008]

According to the present invention, a relay for transferring information between a host device which is a transmission source of information distributed via a network and a host device which is a transmission destination of the information. The device is placed. The relay device transmits the information sent from the source host device to the destination host device. Further, the relay device has a function of terminating the TCP connection between the hosts, and T
It has means for terminating the CP connection on the way and dividing it, and for the host device, it has means for making the divided connection virtually appear as one TCP connection.

More specifically, the relay device generates an acknowledgment for the information sent from the source host device and sends it to the source host device. As the source address information of the confirmation response, the address information of the destination host device of the information is used.

In one aspect of the present invention, the relay device includes a first relay device provided on the source host device side of the network and a second relay device provided on the destination host device side of the network. . The confirmation response to the source host device is generated and transmitted by the first relay device. The first relay device has means for converting the destination address of the received packet into the address assigned to itself in order to terminate the packet from the source host device. Further, the second relay device has means for converting the source address of the packet including information into the address of the source host device in order to make the destination host device recognize one TCP connection. Have.

According to another aspect of the present invention, there is provided a method of processing a communication packet in a communication device arranged between a first computer as a transmission source of information and a second computer as a transmission destination. It Specifically, the communication device receives a communication packet from the first host device whose start point address is the address of the first computer and whose end point address is the address of the second computer. The communication device generates a confirmation response packet in which the start point address is the second computer and the end point address is the first computer, and transmits the confirmation response packet to the first computer. Further, the communication device transfers a packet containing the data of the communication packet sent from the first computer to the second computer. This establishes communication between the first computer and the second computer.

According to the present invention, data is sent through a plurality of TCP connections, which are viewed by the user as one connection. Since the TCP connection is terminated by the router, the confirmation response can be returned immediately, and high-speed TC is possible even when the delay line part is included in the communication path without changing the existing protocol.
P communication can be realized.

The communication system according to the present invention is particularly suitable for T
Suitable when CP communication is performed via a network system and the time until the window size held by the source host reaches the upper limit is longer than the time until the TCP acknowledgment is received. Is.

[0014]

1 is a block diagram of a TCP communication system according to an embodiment of the present invention.

The TCP communication system of this embodiment includes a satellite receiving station and a satellite transmitting station. The satellite receiving station is a receiving gateway 10 (RGW), a satellite receiver 1
1 (IRD), router 70b, and destination host or personal computer (PC) 60. The satellite transmission station has a transmission side gateway 20 (SGW), an uplink station 21, a router 70a, and a transmission source host or server 50.

One-way communication can be performed between the satellite transmitting station and the satellite receiving station via the communication satellite 30 from the satellite transmitting station side to the satellite receiving station side. Also, Internet network 4
It is possible to perform bidirectional communication via 0. In the present embodiment, the RGW 10 and the SGW 20 realize asymmetric routing of transmission packets in a network including a one-way route.

In this embodiment, the RGW 10 and the SGW 20 are used.
Causes the packet transmitted from the PC 60 to the server 50 to be “PC 60 → local network → router b (70b) → R
GW10 → Internet network 40 → SGW20 → router a (70a) → local network → server 50 ”. In addition, the packet transmitted from the server 50 to the PC 60 is “server 50 → local network → router a”.
(70a) → SGW20 → uplink station 2
1 → communication satellite 30 → IRD11 → RGW10 → router b
(70b) → local network → PC60 ”. Details of the asymmetric routing using the RGW 10 and SGW 20 are described in, for example, Japanese Patent Laid-Open No. 11-
No. 313109 is disclosed in the public information.

The router 70 (routers 70a, 70b) is
It has a function of terminating a TCP connection between hosts and creating a single virtual TCP connection. Since the router 70 is the terminal of the TCP connection, an acknowledgment can be immediately returned between networks other than the delay line. As a result, high-speed communication can be performed without filling the TCP window.
It takes a lot of time for the acknowledgment to be returned during the delay line, but the influence of the delay can be reduced by increasing the window size of the router 70. Also,
Since the user sees only one end-to-end virtual TCP connection, the user can perform TCP communication as in the conventional case without being aware of the division of the TCP connection. This virtual TCP connection is realized by the packet address conversion performed by the router 70.

It should be noted that the SGW 20 and the router 70a or the RG constituting the TCP high speed system according to the present embodiment.
Although the W10 and the router 70b are described as devices independent of each other, they may be configured as one communication device physically housed in the same housing. Further, one or more PCs 60 may be connected to the local network, and communication between the server 50 and the PCs 60 is not limited to one-to-one, but information can be simultaneously transmitted from one server 50 to a plurality of PCs 60. Further, as a communication path, the Internet network 40,
It is possible to use a communication path other than the communication satellite 30, and a plurality of different communication paths may exist.

Next, details of the operation of the router 70 will be described. The function of the router 70 is the same as that of the transmitting side router 70a and the receiving side router 70b. In the following, description will be made without distinguishing between the transmission side router 70a and the reception side router 70b.

In the router 70, a processing program including three modules of a dazzler (call virtualization module), a coupler (call connection module) and a mediator (call establishment module) is executed.

Dazzler virtualizes the TCP connection by performing address conversion on the packet. Dazzler has an address conversion information table and performs address conversion according to this table. When a packet not registered in the address translation information table arrives, a new registration request (active registration request) is sent to the mediator, and new address translation information is received from the mediator and processed. When a request for deleting address translation information is received from the mediator, the information is deleted from the address translation information table according to the instruction. Here, "active" refers to the host on the side that sent the connection establishment request. The active-side host and the passive-side host dynamically change according to the situation, but in the present embodiment, any host can be the active-side or the passive-side.

The mediator is a module which instructs other modules to establish and release a TCP connection. The mediator has a virtual connection information table. When the active registration request is issued from the dazzler, the passive registration request is sent to the opposite mediator, the active connection request is sent to the coupler, and the address translation information update request is sent to the dazzler. When receiving a passive registration request from the mediator of the opposite router 70, the mediator sends a passive connection request to the coupler and an address translation information registration request to dazzler. When the connection release request is sent from the coupler,
A connection release request is sent to the opposing mediator, a connection release request is sent to the coupler, and an address translation information deletion request is sent to Dazzler.

The coupler has a function of establishing a connection with the opposite coupler, copying the packet data sent from the host, and sending it to the opposite coupler. The coupler operates as a separate process for each individual connection. When an active connection request or a passive connection request is received from the mediator, the parent process connects to the opposing coupler and creates a child process. The child process has a function of copying data between TCP connections disconnected on the router 70. When copying is completed, a connection release request is sent to the mediator. The parent process terminates the child process when a connection release request is received from the mediator.

FIG. 2 is a simplified block diagram showing the hardware configuration of the router 70. The router 70 has a CPU 1001, a memory 1002, a disk controller 1003, a hard disk 1100, a console controller 1004, a console 1200, and a network controller 1005, as shown in the figure.

The CPU 1001 controls the operation of the entire router 70. The memory 1002 stores programs and data executed by the CPU 1001. The disk controller 1003 controls the hard disk 1100. The hard disk 1100 stores programs and data. Console controller 1004
Controls a console 1200 that performs input / output with a user. Network controller (a) 1005a
Communicates with the SGW 20 or the RGW 10. The network controller (b) 1005b communicates with the local network.

The operation of the router 70 is realized by the program processing to be described later by the CPU 1001. As described above, this program includes three modules, which are a dazzler (call virtualization module), a coupler (call connection module), and a mediator (call establishment module).

FIG. 3 is a data configuration diagram of the address translation information table 1500 used by the dazzler program in the process.

As shown in FIG. 3, the address translation information table 1500 has an outer IP address field 1501.
An outer port number field 1502, an inner IP address field 1503, an inner port number field 1504, and a boundary IP address field 1505.
And a boundary port number field 1506.

In the outer IP address field 1501 and outer port number field 1502, the router 70
The IP address and the port number of the host located outside are stored. In the inside IP address field 1503 and inside port number field 1504, the router 7
The IP address and the port number of the host inside when viewed from 0 are stored. Border IP address field 1505
And in the boundary port number field 1506, the network controller (b) installed in the router 70.
The IP address and port number of 1005b are stored.

Here, the outside means a network connected to the local network side with the router 70 as a boundary, and the inside means a network connected to the opposite side from the local network with the router 70 as a boundary. Say that.

In the present embodiment, the dazzler performs two kinds of address conversion depending on the packet transmission direction. First,
For a packet sent from the outside to the inside, if the end point address of the packet matches the inside address and the start point address matches the outside address, the end point address is converted into a boundary address. For a packet sent from the inside to the outside, if the end point address of the packet matches the outside address and the start point address matches the boundary address, the start point address is converted to the inside address.

FIG. 4 is a conceptual diagram showing the principle of packet transfer in this embodiment. In FIG. 5, the source host 8
The content of the TCP packet when the TCP packet is transmitted from 0 to the destination host 90 is shown. 4 and 5, the first router 70d and the second router 70e are installed between the source host 80 and the destination host 90. The first dazzler 101 and the first coupler 201 are the first
The second dazzler 102 and the second coupler 202 are in the second router 70e.

Hereinafter, how the TCP packet sent from the source host 80 to the destination host 90 is processed and delivered to the destination host 90 will be described.

First, the transmission source host 80 sets the start point address to the transmission source host 80 and the end point address to the transmission destination host 9
The TCP packet set to 0 is transmitted to the first router 70d (step 2901). In FIG. 5, IP-PC1
And P-PC1 are the IP address and port number of the source host 80, and IP-PC2 and P-PC2 are the destination host 9
An IP address of 0 and a port number are shown.

When the TCP packet is delivered to the first router 70d, the packet is sent to the first dazzler 101. The first dazzler 101 performs the address translation of the TCP packet according to the translation rule registered in advance in the address translation information table 1500. Specifically, the end point address of the packet is converted into the boundary address of the first coupler 201. After that, the first Dazzler 101
Sends the address-converted packet to the first coupler 201 (step 2902). In FIG. 5, IP-C1
E and P-C1E are the boundary IP address and port number of the first coupler 201.

The first coupler 201 terminates the TCP connection from the source host 80. Because the end point address of the TCP packet is the first coupler 2
This is because the boundary address is 01. Therefore,
Here, an acknowledgment is created for the TCP packet, and the first coupler 201 sends the acknowledgment packet to the source host 80. That is, the first coupler 201 passes the confirmation response packet whose start address is the boundary address of the first coupler 201 and whose end address is the source host 80 to the first dazzler 101 (step 2903). In this confirmation response packet, an ACK flag indicating that it is a confirmation response is added in the TCP header.

The first dazzler 101 again performs address translation of the confirmation response packet according to the address translation information table 1500. Specifically, set the start point address to the first
From the boundary address of the coupler to the address of the destination host 90. After that, the address-translated confirmation response packet is transmitted from the first dazzler 101 to the transmission source host 80.
(Step 2904). The transmission source host 80, which has received this confirmation response packet, recognizes that it has received a confirmation response from the transmission destination host 90.

On the other hand, the TCP packet that has reached the first coupler 201 must be delivered to the destination host 90. Therefore, data is copied between the first coupler 201 and the second coupler 202.
Communication between these two couplers is not limited to TCP and can be realized by any protocol. Here, it is assumed that the TCP protocol is used.

The first coupler 201 transmits a TCP packet whose start address is the internal address of the first coupler and whose end address is the internal address of the second coupler 202 to the second coupler 202 (step 2905). In FIG. 5, IP-C1I and P-C1I are the first coupler 20.
1 internal IP address and port number, IP-C2I and P
-C2I is the internal IP address and port number of the second coupler 202.

The second coupler 202 that has received this TCP packet transmits an acknowledgment packet for the TCP packet to the first coupler 201 (step 29).
06). In this way, packets are exchanged between the two couplers.

Then, the second coupler 202 passes the TCP packet whose start address is the boundary address of the second coupler 202 and whose end address is the destination host 90 to the second dazzler 102 (step 2907). In FIG. 5, IP-C2E and P-C2E are the second coupler 202.
Border IP address and port number.

The second dazzler 102 performs address translation of the TCP packet according to its own address translation information table 1500. Specifically, the start address is
The boundary address of the second coupler is converted to the address of the source host 80. After this, the address-converted TC
The P packet is transmitted from the second dazzler 102 to the destination host 9
0 (step 2908). The destination host 90 receiving this TCP packet recognizes it as if it received the TCP packet from the source host 80.

Destination host 9 that received the TCP packet
0 returns an acknowledgment packet for it. Specifically, a confirmation response packet whose start point address is the destination host 90 and end point address is the source host 80 is transmitted to the second router 70e (step 2909).

When the acknowledgment packet is delivered from the destination host 90 to the second router 70e, the packet is sent to the second dazzler 102. The second dazzler 102 performs the address conversion of the TCP packet according to the conversion rule registered in advance in the address conversion information table 1500. Specifically, the end point address of the packet is converted into the boundary address of the second coupler 202. afterwards,
The second dazzler 102 transmits the address-converted packet to the second packet.
To the coupler 202 (step 2910).

In the second coupler 202, the destination host 90
From the TCP connection is terminated. Because T
This is because the end point address of the CP packet is the boundary address of the second coupler 202.

As described above, the first dazzler 101 and the first
Coupler 201, second dazzler 102, second coupler 2
02 performs processing, the original TCP connection "source host 80-destination host 90" is "source host 80-first coupler 201", "first coupler 201".
-Second coupler 202 "and" second coupler 202-destination host 90 ". Moreover, the source host 8
Since the packet transmitted by 0 and the packet received by the destination host 90 are exactly the same, the source host 80 and the destination host 90 are as if the “source host 80
~ It seems that the TCP connection of the destination host 90 "is established. Each host can perform conventional TCP communication without being aware of the existence of the router 70. In addition, since the router returns the confirmation response, there is no need to wait for the confirmation response via the delay line, so that faster TCP communication can be realized.

In the system configuration of this embodiment, the delay line is "first coupler 201-second coupler 202".
Or “second coupler 202 to destination host 90”
When existing between them, the effect of improving the TCP communication speed is great. When a delay line exists between the "source host 80 and the first coupler 201", the TCP communication speed cannot be expected to improve.

As described above, in the present embodiment, the original TCP connection "source host 80 to destination host 90" causes the "source host 80" to be processed by the processing of the first router 70d and the second router 70e. ~ First router 70
d "," first router 70d to second router 70e ",
It is divided into "second router 70e-destination host 90". However, the source host 80 and the destination host 90
From the source host 80 to the destination host 9
It seems that the TCP connection of "0" is established. In this specification, such an apparent TCP connection is called a virtual TCP connection. Each host can perform conventional TCP communication without being aware of the existence of the router 70.

When a data packet is transmitted from the destination host 90 to the source host 80, the packet is transferred in the same way. Therefore, the description thereof is omitted here.

FIG. 6 is a flowchart of the dazzling program.

The dazzler program first checks whether a command has been received from the mediator (step 1601). When the command from the mediator is received, it is checked whether or not the command is address translation information registration (step 1602), and if so, the address translation information passed to the command is converted into the address translation information table 1500. (Step 1603). If the command is not address translation information registration, it is checked whether or not the command is address translation information deletion (step 1604). If the command is the address translation information deletion, the address translation information specified by the command is deleted from the address translation information table 1500 (step 1
605). If the command is not the address translation information deletion, it is further checked whether or not the command is the address translation information update (step 1606). in that case,
Address translation information table 150 specified by command
The contents of 0 are updated with new information (step 160).
7).

Next, the network controller (a) 1
005a or network controller (b) 100
It is checked whether a TCP packet is received from any of 5b (step 1608). If so,
It is checked whether the set of the IP address and the port number of the packet is registered in the address translation information table 1500 (step 1609). If so, the address of the packet is translated according to the information in the address translation information table 1500, and the packet is output (step 16).
10). If the combination of the IP address and the port number of the packet is not registered in the address translation information table 1500, it is checked whether the packet arrives from the outside and is a connection establishment request packet (step 1611). Whether the TCP packet is a connection establishment request is determined by whether the SYN bit of the TCP header is set and the ACK bit is cleared. Details are described in RFC793.
If the condition of step 1611 is met, an active registration request is sent to the mediator (step 1612).

FIG. 7 is a data structure diagram of the virtual connection information table used by the mediator program in its processing.

Virtual connection information table 2500
Is an active side host IP address field 2501, an active side host port number field 2502, an active side boundary IP address field 2503, an active side boundary port number field 2504, an active side internal IP address field 2505, and an active side internal port number field 250.
6, passive side host IP address field 2507, passive side host port number field 2508, passive side boundary IP address field 2509, passive side boundary IP address field 2510, passive side internal IP address field 2511, and passive side internal port number 2512
Has a field.

Active Host IP Address Field 25
01 and the active side host port number field 2502 store the IP address and port number of the host that is actively connected. Active side boundary IP address field 250
3 and the active side boundary port number field 2504, the IP of the network controller (b) 1005b installed in the router 70 on the actively connected host side.
Address and port number are stored. The active-side internal IP address field 2505 and the active-side internal port number field 2506 store the IP address and port number of the network controller (a) 1005a installed in the router 70 on the host side that is actively connected. The passive-side host IP address field 2507 and the passive-side host port number field 2508 store the IP address and port number of the passively connected host. The passive side boundary IP address field 2509 and the passive side boundary port number field 2510 store the IP address and port number of the network controller (b) 1005b installed in the router 70 on the passively connected host side. Passive side internal IP address field 2511 and passive side internal port number field 25
12 is a network controller (a) 100 installed in the router 70 on the host side that is passively connected.
The IP address and port number of 5a are stored.

FIG. 8 is a flowchart of the mediator program.

In the processing of the mediator program, when a command is received (step 2601), it is checked whether or not the command is an active registration request (step 2602).
If it is an active registration request, it sends a passive registration request to the mediator of the opposite router 70, an active connection request to the coupler, and an address translation information update request to dazzler. Further, the connection information is registered in the virtual connection information table 2500 (step 2603). If the command is not an active registration request, it is checked whether the command is a passive registration request (step 2604). If there is a passive registration request,
A passive connection request is sent to the coupler, an address translation information registration request is sent to Dazzler, and a virtual connection information table 2500 is sent.
The connection information is registered in (step 2605).
If the command is not a passive registration request, it is checked whether the command is a connection release request (step 260).
6). If it is a connection release request, it sends a connection release request to the opposite mediator, a connection release request to the coupler, and an address translation information deletion request to Dazzler. Further, the connection information specified by the command is deleted from the virtual connection information table 2500 (step 2607).

As described above, there are two types of couplers, a parent process (parent coupler) and a child process (child coupler).
The operation of the coupler will be described below.

FIG. 9 is a flow chart of the parent coupler.

In the processing of the parent coupler, when a command is received from the mediator (step 3601), it is checked whether the command is an active connection request (step 36).
02). If the command is an active connection request, a socket is created on the border side of the router and a passive connection is made to the active host. After that, a socket is created inside the router and active connection is made to the inside address of the opposite router 70. Then, a child coupler is generated (step 360).
3). If the command is not an active connection request, it is checked whether the command is a passive connection request (step 360).
4). If it is a passive connection request, a socket is created inside the router and passive connection is made to the internal address of the opposite router 70. After that, a socket is created on the boundary side of the router and is actively connected to the passive side host address to create a child coupler (step 3605). If the command is not a passive connection request, it is checked whether the command is a connection release request (step 3606). If it is a connection release request, the corresponding child process is terminated (step 3607). The socket refers to a set of an IP address and a port number that identify an end point (end point) of the connection. For details, see, for example, UNIX Network Programming Second Edition (W. Richard Stevens, 1999, Pearson Education Publishing Co., ISBN 4-89471-205-9).
There is a description in.

FIG. 10 is a flow chart of the child coupler.

When the parent coupler creates a child coupler by an active connection request or a passive connection request (step 4601),
The child coupler checks whether or not there is received data in the boundary side socket (step 4602). If there is received data, that data is received and sent to the inner socket (step 4).
603). If there is no received data in the boundary side socket, it is checked whether or not the inside socket has received data (step 4604). If there is data in the inner socket, that data is received and sent to the boundary side socket (step 46).
05). If there is no data in the inner socket, it is checked whether the boundary socket is closed (step 4606),
If the border side socket is closed, close the inner socket,
Send connection release request to mediator (step 460)
7). Otherwise, it is checked whether the inner socket is closed (step 4608). If the inner socket is closed, the boundary socket is closed and a connection release request is sent to the mediator (step 4609).

FIG. 11 is a system configuration diagram of a communication system in the second embodiment of the present invention.

The communication system according to the present embodiment includes a receiving side gateway 10 (RGW) and a satellite receiver 11 (IR).
D), transmitting side gateway 20 (SGW), uplink station 21, communication satellite 30, Internet network 40, server 50, personal computer (PC) 6
0, a router 70c is included. In the communication system according to the first embodiment, the router 70 is provided on each of the satellite transmitting station side and the satellite receiving station side, but in the present embodiment, the router 70c is arranged only on the satellite transmitting station side. Further, instead of installing the router 70c on the satellite transmitting station side, the router 70c can be installed on the satellite receiving station side, or the router 70 can be installed in the communication satellite 30.

The router 70c in this embodiment is the first
The first router 70d and the second router 70e in the above embodiment are combined into one communication device. The router 70c includes a first router 70d and a second router 70d.
The role of the router 70e is performed only in one communication device, and in principle, the same processing as that of the router of the first embodiment shown in FIG. 4 is performed. Also in this embodiment, a virtual TCP connection can be realized by converting the TCP packet address in the router 70c, dividing the TCP connection, and making an acknowledgment response.

FIG. 12 is a conceptual diagram showing the principle of packet transfer in this embodiment, and FIG. 13 is a T transmitted at various places.
It is a conceptual diagram which shows the content of CP packet.

In the figure, the router 70 is the source host 8
It is installed between 0 and the destination host 90. A first dazzler 101, a coupler 200, and a second dazzler 102 are inside the router 70.

First, the transmission source host 80 sets the start point address to the transmission source host 80 and the end point address to the transmission destination host 9
The TCP packet set to 0 is transmitted to the router 70 (step 2801). In FIG. 13, IP-PC1 and P-
PC1 is the IP address and port number of the source host 80, and IP-PC2 and P-PC2 are the I of the destination host 90.
The P address and port number are shown.

When the TCP packet is delivered to the router 70, the packet is sent to the first dazzler 101. The first dazzler 101 performs address conversion according to the conversion rule registered in advance in the address conversion information table 1500. Specifically, the end point address of the packet is converted into the source side boundary address of the coupler 200.
After that, the first dazzler 101 sends the address-converted packet to the coupler 200 (step 2802). Figure 1
3, IP-CA and P-CA indicate the boundary IP address and port number of the coupler 200.

The coupler 200 terminates the TCP connection from the source host 80. This is because the end point address of the TCP packet is the source side boundary address of the coupler. Here, an acknowledgment for the TCP packet is created, and the coupler 200 sends the acknowledgment packet to the source host 80. That is, the coupler 2
00 sends an acknowledgment packet whose start address is the source side boundary address of the coupler and end point address is the source host 80 to the first dazzler 101 (step 280).
3). In this confirmation response packet, an ACK flag indicating that it is a confirmation response is added in the TCP header.

The first dazzler 101 again performs address conversion of the confirmation response packet according to the address conversion information table 1500. Specifically, the start point address is changed from the source side boundary address of the coupler 200 to the destination host 9
Convert to an address of 0. After that, the address-converted confirmation response packet is sent from the first dazzler 101 to the transmission source host 80 (step 2804).

The source host 80, which has received the confirmation response packet, recognizes as if it has received the confirmation response from the destination host 90.

On the other hand, the TCP packet that has reached the coupler 200 must be properly delivered to the destination host 90. Therefore, the coupler 200 passes a TCP packet having the start address as the destination boundary address of the coupler and the end address as the destination host 90 to the second dazzler 102 (step 2805). In FIG. 13, IP-CB and P-CB represent the destination side boundary I of the coupler.
Indicates the P address and port number.

The second dazzler 102 performs address conversion of the TCP packet according to its own address conversion information table 1500. Specifically, the start address is
From the destination boundary address of the coupler, the source host 80
Address. After that, the address converted T
The CP packet is sent from the second dazzler 102 to the destination host 90 (step 2806).

Destination host 9 that received the TCP packet
0 recognizes as if a TCP packet was received from the source host 80.

Destination host 9 that received the TCP packet
0 returns an acknowledgment packet for it. Specifically, the router 7 sends an acknowledgment packet whose source address is the destination host 90 and whose destination address is the source host 80.
0 (step 2807).

When the acknowledgment packet is delivered to the router 70, the packet is sent to the second dazzler 102. Second
Dazzler 102 has an address translation information table 1500.
According to the conversion rules registered in advance in TC
Address conversion of P packet is performed. Specifically, the end point address of the packet is converted into the destination side boundary address of the coupler 200. After that, the second dazzler 102 sends the address-converted packet to the coupler 200 (step 2808).

The coupler 200 terminates the TCP connection from the destination host 90. This is because the end point address of the TCP packet is the destination boundary address of the coupler.

As described above, since the first dazzler 101, the coupler 200, and the second dazzler 102 perform processing, the original TCP connection "source host 80-destination host 90" changes to "source host 80". ~ Coupler 200 ",
It is divided into "coupler 200 to destination host 90". Packets sent by the source host 80 and destination host 90
Since the packet received by is the same, the source host 8
From 0 and the destination host 90, it looks as if the TCP connection of “source host 80 to destination host 90” is established. Is a virtual TCP connection. Each host can perform conventional TCP communication without being aware of the existence of the router 70. Since the router returns the confirmation response, it is not necessary to wait for the confirmation response from the reception source via the delay line, and thus faster TCP communication can be realized.

In this embodiment, the delay line is "coupler 20".
When it exists between 0 and the destination host 90, the effect of improving the TCP communication speed is great. If there is a delay line between the "source host 80 and the coupler 200", it is almost T.
The CP communication speed cannot be expected to improve.

FIG. 14 is a conceptual diagram showing a TCP connection between the source host 80 and the destination host 90.
The figure shows how a connection is established between a source host 80 and a destination host 90 via a conventional router. The broken line in the figure represents the TCP connection.

FIG. 15 is a conceptual diagram showing the communication architecture of communication equipment.

In FIG. 15, the communication architecture is expressed in a hierarchical form, and from the bottom, "data link / physical layer",
It is divided into "network layer", "transport layer", and "application layer". Each layer uses the functions of the layers to provide unique functions. The TCP protocol belongs to the transport layer, and the IP protocol belongs to the network layer.

When the TCP connection between the source host 80 and the destination host 90 passes through a conventional router,
As shown in FIG. 14, the destination address (that is, IP address) is determined at the network layer, and the packet is transferred to the appropriate node. In the router, the TCP connection is never disconnected.

FIG. 16 is a conceptual diagram showing the system configuration in the first embodiment in communication architecture.

The first router 70d uses the source host 80
The TCP packet sent from the device is subjected to address conversion processing at the data link / physical layer (dazzling). The address-converted TCP packet reaches the transport layer, where the TCP connection terminates. The coupler of the first router 70d establishes a connection with the coupler of the second router 70e. As described above, the TCP connection is used for this connection in the first embodiment. The coupler of the second router 70e is connected to the destination host 9
Connect TCP connection with 0. The dazzler of the second router 70e performs the address conversion process, and the TCP packet is delivered to the source host 90.

As described above, in the first embodiment, "source host 80 to first router 70d", "first router 7"
0d-second router 70e "," second router 70e-
Three TCP connections for the destination host 90 "are created.

FIG. 17 shows the system configuration of the second embodiment in terms of communication architecture.

The router 70 performs address conversion processing on the TCP packet sent from the transmission source host 80 at the data link / physical layer (dazzling). The address-converted TCP packet reaches the transport layer, where the TCP connection terminates. The coupler of the router 70 establishes a TCP connection with the destination host 90. The dazzler of the router 70 performs the address conversion process again, and the TCP packet is delivered to the source host 90.

As described above, in the second embodiment, two TCP connections of "source host 80-router 70" and "router 70-destination host 90" are created.

In the above-described embodiment, the destination host returns a confirmation response in response to data reception from the source host including TCP, and the amount of data that can be transmitted at one time like the TCP window size is limited. When communication is performed using existing protocols and a broadband line with a large delay is used, the effect of delay is reduced and high-speed communication is realized without changing existing protocols.

Further, the router 70 uses the source host 80
It can be installed anywhere (including inside the communication satellite 30) in the communication path connecting the destination host 90 and the destination host 90.

The above description and drawings are intended to represent examples and are not meant to limit the invention. The present invention can be subjected to various changes and modifications without departing from the spirit of the invention.

[0095]

According to the present invention, it is possible to construct a system capable of reducing the influence of delay in the line and performing higher speed communication without changing the existing protocol.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a TCP communication system according to an embodiment of the present invention.

FIG. 2 is a simplified block diagram showing a hardware configuration of a router 70.

FIG. 3 is a data configuration diagram of an address conversion information table.

FIG. 4 is a conceptual diagram showing the principle of packet transfer in the first embodiment.

FIG. 5 is a conceptual diagram showing contents of a TCP packet to be transmitted.

FIG. 6 is a flowchart of a dazzler program.

FIG. 7 is a data configuration diagram of a virtual connection information table.

FIG. 8 is a flowchart of a mediator program.

FIG. 9 is a flowchart of a parent coupler program.

FIG. 10 is a flowchart of a child coupler program.

FIG. 11 is a system configuration diagram of a communication system according to a second embodiment of the present invention.

FIG. 12 is a conceptual diagram showing the principle of packet transfer in the second embodiment.

FIG. 13 is a conceptual diagram showing the contents of a TCP packet to be transmitted.

FIG. 14 shows T between a source host 80 and a destination host 90.
It is a conceptual diagram showing CP connection.

FIG. 15 is a conceptual diagram showing a communication architecture of a communication device.

FIG. 16 is a conceptual diagram showing a communication architecture of the system configuration according to the first embodiment.

FIG. 17 illustrates a system architecture in the second embodiment by a communication architecture.

[Explanation of symbols]

10 ... Receiving side gateway (RGW) 11 ... Satellite receiver (IRD) 20 ... Sender gateway (SGW) 21 ... Uplink station 30 ... Communication satellite 40 ... Internet network 50 ... Source host 60 ... Destination host 70a, 70b ... Router

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Minoru Koizumi             1099 Ozenji, Aso-ku, Kawasaki City, Kanagawa Prefecture             Ceremony company Hitachi Systems Development Laboratory F-term (reference) 5K030 GA03 HA08 HD03 HD08 HD09                       LA02 LA08                 5K034 AA03 FF02 FF11 KK21 LL01

Claims (16)

[Claims] 1. A communication control device which is arranged between a first device and a second device and transfers a communication packet between the first device and the second device according to a predetermined protocol. Transmitted by a device, the first device being the source,
A communication packet including address information designating the second device as a destination is received, and a response packet corresponding to the communication packet is generated including response packet source address information corresponding to the destination address of the data packet. The communication control device further comprises a processor operable to transmit the response packet to the first device. 2. The communication control device according to claim 1, wherein the predetermined protocol is a transport control protocol (TCP). 3. A communication system comprising at least one communication control device according to claim 1. 4. A communication control method in a communication control device which is arranged between a first device and a second device and transfers a communication packet between the first device and the second device. Receiving a communication packet including address information specifying the first device as a transmission source and the second device as a transmission destination, the second device being the transmission source, and the first device. By generating a response packet corresponding to the communication packet, which includes address information designating as a destination, and transmitting the response packet to the first device,
In order to establish a communication path between the first device and the communication control device, and to establish a communication path between the second device and the communication device, the data included in the received communication packet A communication control method comprising: transmitting a second communication packet including at least a part to the second device. 5. In the step of transmitting the second communication packet, the second communication packet is transmitted to a second communication control device and included in the first packet by the second communication control device. The communication control method according to claim 4, wherein a third communication packet including at least a part of the data to be transmitted is transmitted to the second device. 6. A communication system in which communication according to TCP is performed via a network, the first relay device being connected between a transmission source host device as a source of information and the network, A transmission destination host device as a transmission destination of the information, and a second relay device connected between the network, and in the first and second relay devices, the transmission source host device and the transmission destination. T with host device
A CP connection is terminated, and a TCP connection between the source host device and the destination host device is
A communication system characterized in that it is divided into a plurality of TCP connections including two sections of the source host device-the first relay device, and the second relay device-the destination host device. 7. The first relay device converts the end point address of a TCP packet transmitted from the source host device into an address of the first relay device,
Means for transmitting a second packet having the contents of the packet to the second relay device, and means for issuing an acknowledgment to the source host device in response to the TCP packet having the translated end point address. The second relay device includes means for converting a start address of a packet sent from the network into an address of the source host device and transmitting the converted address to the destination host device. The communication system according to claim 6. 8. The first relay device is arranged at a boundary between a local network system to which the source host device belongs and the network, and the second relay device belongs to a local network to which the destination host device belongs. Communication system characterized in that it is arranged at the boundary between a simple network system and the network. 9. A communication system in which communication according to TCP is performed between a transmission source host device as a transmission source of information and a transmission destination host device as a transmission destination of the information via a network, A relay device that relays the information is provided between the source host device and the destination host device, and the relay device establishes a TCP connection between the source host device and the destination host device. A communication system comprising: a plurality of connections, and means for making the source host device and the destination host device look like the plurality of connections as one virtual connection. 10. The faking means relays a TCP packet transferred from the source host device to the destination host device, and makes an acknowledgment response to the source host device in response to the TCP packet, 10. The communication system according to claim 9, further comprising means for terminating the TCP connection from the transmission source host device. 11. The network includes a network portion having a relatively large delay, and the relay device is arranged in a portion closer to the source host device than the network portion having a large delay. Item 10. The communication system according to Item 10. 12. The transmission source host device is provided on a local system separated from the network,
12. The relay device is provided at a boundary between the local system and the network.
The communication system described. 13. The relay device includes a first relay device connected to a local network to which the source host device belongs, and a second relay device connected to a local network to which the destination host device belongs. The communication system according to claim 10, wherein: 14. A router device, which is arranged between a source host device which is a source of a packet including information to be transferred and a destination host device which is a destination of the information, and which transfers the packet. , The connection between the source host device and the destination host device is terminated on the router device and divided, and the divided connection is virtualized to the source host device and the destination host device. And a means for recognizing it as a single connection. 15. The recognizing means, in response to a packet from the transmission source host device, generates a confirmation response packet, and, as starting point address information of the confirmation response packet, an address of the transmission destination host device. 15. The router device according to claim 14, further comprising means for acquiring and setting 16. The router device further includes address translation means for translating an end point address set in the packet from the source host device into an address assigned to the router device, and the response packet 16. The router device according to claim 15, wherein the means for generating the confirmation response packet generates the confirmation response packet in response to the packet whose end point address is translated by the address translation means.