JP2010109733A - Packet communication method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a packet communication method which does not increase a load of an application layer, but is fast at high reliability. <P>SOLUTION: An IP layer is located at an upper level of a MAC layer, a UDP layer is located at an upper level thereof, and an UECP (Unique Error Control Protocol) is located at an upper level thereof. A UECP header is reserved in an area of an upper 12 bytes in a data field of a UDP packet, and includes a sequence number field, a confirmation response number field, a code bit field, and a window size field. An MSS of the UECP is set to 1,398 bytes or less, and an MTU thereof is set to 1,438 bytes or less. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a packet communication method, and more particularly to a packet communication method that enables high-speed and reliable packet communication.

  TCP (Transmission Control Protocol) and UDP (User Datagram Protocol) are known as typical transport layer protocols used in IP networks. TCP (RFC1122 and RFC1123) is a connection-oriented protocol, and in order to avoid data communication errors due to packet loss and to guarantee reliability, the sender sends an acknowledgment signal (ACK ), The same data is transmitted again (retransmission control) when a timeout occurs. This timeout period is called a retransmission timeout period (RTO). If the response is not continued, retransmission is repeated. In this case, the retransmission interval is exponentially increased according to the number of retransmissions. When transmission packet loss repeatedly occurs, transmission is suppressed by increasing the retransmission timeout period, and control (congestion control) is performed to reduce the probability of packet loss. On the receiving side, the packet arrival order is guaranteed by setting the sequence number and the acknowledgment number of the successfully received packet to ACK and returning them (order control).

On the other hand, UDP is a connectionless protocol, and the above retransmission control, congestion control, order control, etc. are not performed. For this reason, TCP is used for WWW, e-mail, FTP, etc. that require data quality over speed, and UDP is used for voice calls and video distribution where speed (real time) is prioritized over data quality. Conventional TCP and UDP are disclosed in Patent Document 1, for example.
JP 2001-136202 A

  In RFC2988, it is recommended that the TCP RTO be incremented to 1 second when the retransmission timeout is less than 1 second. Therefore, if a TCP packet is lost, the re-transmission takes at least 1 second, and the real-time property of communication is impaired.

  TCP also has a keepalive function. Regardless of whether there is actual data transmission or not, the connection is sent at regular intervals to monitor the response, and if the TCP connection is not valid, it is disconnected. However, the default keepalive time is 2 hours, so if you want to perform non-communication monitoring at a desired interval for each communication partner, you must perform non-communication monitoring in an application layer higher than TCP, and the monitoring interval is short. There is a technical problem that the monitoring load in the application layer increases.

  An object of the present invention is to solve the above-described problems of the prior art and provide a high-speed and highly reliable packet communication method without increasing the load on the application layer.

  In order to achieve the above object, the present invention is characterized in that a packet communication method for controlling packet switching in an upper layer of UDP includes the following procedure.

  (1) A protocol header and each field of data are secured in the data field of the UDP packet, and the protocol header includes at least a sequence number field, an acknowledgment number field, and a code bit field for packet identification. Retransmission control and order control are performed based on the number registered in the response number field.

  (2) The protocol header includes a window size field.

  (3) A time-out value that prescribes a time-out of the keep-alive response to the keep-alive request is stored, and if the keep-alive response cannot be received even after the fourth time-out value has elapsed, the session is closed, and this fourth time-out value T4 Including a procedure for setting.

According to the present invention, the following effects are achieved.
(1) Packet communication with the same reliability as TCP and a transfer rate equivalent to UDP in actual use is possible.
(2) Since the protocol is implemented in the UDP data field, it is possible to utilize a conventional layer located below.
(3) Since the window size can be specified, efficient data transfer becomes possible.
(4) Since the keep-alive transmission interval can be set freely, if this is set to an appropriate short time, it is possible to quickly detect the disconnection of the LAN cable or the power-off of the communication partner in the transport layer. Frequent monitoring in the application layer is unnecessary, and the load on the application can be reduced.

  Hereinafter, the best embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the main part of an IP button telephone apparatus 100 to which the present invention is applied. Here, the configuration unnecessary for the description of the present invention is omitted.

  The IP button telephone apparatus 100 can be used not only for legacy lines (conventional / old models) such as analog lines, number display contract lines, network number dial-in lines, analog dedicated lines, digital dedicated lines and ISDN lines as external lines, NGN lines can be used. In addition to legacy extension terminals such as general analog terminals, ISDN terminals, hotel phones, digital button telephones and fax machines as extension terminals, it accommodates IP extension terminals such as softphones and IP telephones. Control the communication between.

  In the IP button telephone apparatus 100, a main control unit CCU (Central Control Unit) 1 controls the entire system according to a main control program. The TDM control unit TCCU2 controls time division multiplexed communication (TDM) between each legacy extension terminal and the IP network, and between each legacy extension line and the IP extension terminal.

  The customer edge unit CEU (Customer Edge Unit) 3 has a function as a high-speed broadband router that multiplexes voice data and control data to perform IP communication, and has high sound quality (bandwidth of about 50 to 7000 Hz) as a holding sound. Music data and standard sound quality (bandwidth of about 300 to 4000 Hz) are stored. When a call between the extension terminal and the external terminal is put on hold by an operation on the extension terminal side, a high-quality or standard sound quality hold sound is sent to the external terminal on the held side according to the media capability at the time of establishment of the call .

  The voice conversion unit VCU (Voice Conversion Unit) 4 converts voice packets of a call from high sound quality to standard sound quality and vice versa, and holds high sound quality music data and standard sound quality music data as hold sound To do. When a call between extension terminals is put on hold by operating one of the extension terminals, the other hold-side extension terminal that does not have a hold tone has a high sound quality or standard sound quality depending on the media capability at the time of call establishment. Send music on hold.

  The TCCU 2 includes various types of legacy external lines such as a fax terminal 81, a hotel phone terminal 82, a digital key telephone 83, and a single telephone 84, as well as general external lines, ISDN net lines, digital dedicated lines, analog dedicated lines, and the like. And a TDM bus 5 for TDM communication. The TDM bus 5 includes a D channel for transmitting / receiving control signals and a B channel for transmitting / receiving PCM converted audio signals.

  The CCU 1 and TCCU 2 are connected by two LANs, a voice LAN 8 and a control LAN 9. CCU1, CEU3 and VCU4 are connected by a control LAN. Further, a maintenance computer 95 is connected to the CCU 1 via the maintenance LAN 10 and a hub (hub) 7 is connected via the IP extension LAN 6. The HUB 7 is connected to an IP key telephone 91 and an audio conference device 92 via a dedicated interface, and a wireless access point (AP) is accommodated with an IP cordless phone 93, and a computer on which a soft phone 94 application is further mounted. Is connected.

  The CCU 1 includes a CPU 11, a DDR memory (RAM) 12, a flash memory 13, a dedicated chip 14 specialized for packet transfer control, a removable Compact Flash (CF) memory 15 and a layer 2 (L2) switch. 16 is the main configuration. When the CCU 1 is powered on or reset, first, the boot loader (software) in the flash memory 13 operates, and the OS stored in the file system of the CF memory 15 is loaded into the DDR memory 12 and activated. Next, the OS loads and starts the application (main body software) of the CF memory 15 in the DDR memory 12, and the main control program starts operating.

  The TCCU 2 mainly includes a CPU 21, a DDR memory 22, a flash memory 23, a phi (PHY) 24, a highway switch 25 for controlling the B channel, a control signal communication chip 26 for the D channel, and a DSP 27. The PHY 24 is an LSI that controls the physical layer of Ethernet (registered trademark) (not shown).

  The CEU 3 mainly includes a CPU 31, a DDR memory 32, a flash memory 33, a dedicated chip 34 specialized for packet transfer control, and an L2 switch 35. The dedicated chip 34 is the same as the dedicated chip 14 of the CCU 1.

  The VCU 4 has a CPU 41, a DDR memory 42, a flash memory 43, an L2 switch 44, and a DSP 45 as main components. For example, when a high-quality voice call is transferred to a legacy extension terminal that supports only the standard sound quality, the DSP 45 converts the high-quality voice packet into a standard sound quality and sends it to the legacy extension terminal, while sending from the legacy extension terminal. The standard voice quality voice is converted into a voice packet for high quality voice and sent to the other party.

  In this embodiment, load distribution is realized by distributing the CPUs 11, 21, 31, and 41 to the CCU1, TCCU2, CEU3, and VCU4, respectively, and the CCU1 and CEU3 further have a dedicated chip 14 specialized for packet transfer. , 34 is mounted, the processing of packet transfer in the CPUs 11 and 31 is reduced, and further load distribution is realized.

  In such a configuration, when a legacy extension terminal (for example, a single telephone 84) makes / receives a call via NGN, a PCM converted voice signal is transmitted / received between the legacy extension terminal and the TCCU 2 via the TDM bus 5 by TDM communication. Is done. A communication session is established between TCCU2 and CCU1, and data packets of voice signals and control signals are exchanged in this communication session. The CEU 3 is connected to the NGN via the L2 switch 35, and the CCU 1 and the CEU 3 exchange packets using the dedicated chips 14 and 34, respectively.

  In addition, when the IP extension terminal makes / receives a call via a legacy external line (for example, ISDN network), control information, voice information, or a voice packet is transmitted / received via the TDM bus 5, TCCU2, CCU1, and HUB7. Further, when an IP extension terminal makes / receives a call via NGN, a SIP message, control information or voice packet is transmitted / received via CEU 3, CCU 1 and HUB 7.

  Next, the unique packet communication protocol of the present invention employed for communication between CCU 1 and each IP unit, such as between CCU 1 and TCCU 2, between CCU 1 and CEU 3, and between CCU 1 and VCU 4, will be described. In addition, between CCU1 and TCCU2, it applies only to communication of the control signal by LAN9 for control, and is not applied to communication of the audio signal by LAN8 for audio | voices.

  FIG. 2 is a diagram showing a protocol hierarchy to which the packet communication method according to the present invention is applied. The IP layer is located above the MAC layer, the UDP layer is located above it, and the present invention is located above the layer. An error control protocol (hereinafter referred to as “UECP: Unique Error Control Protocol”) is located.

  FIG. 3 is a diagram illustrating a structure of the UECP header. The UECP header is secured in the upper 12 bytes of the data field of the UDP packet, and includes a sequence number field, an acknowledgment number field, a code bit field, and a window size field. In this embodiment, UECP MSS (Maximum Segment Size) is set to 1398 bytes or less, and MTU (Maximum Transmission Unit) is set to 1438 bytes or less. These are based on the fact that the MTU of B FLET'S and FLET'S ADSL (registered trademark), which are typical ISP services, is 1454 bytes or less, and the MTU of FLET'S Hikari Premium (registered trademark) is 1438 bytes or less.

In the sequence number field of the UECP header, the sequence number of the first data octet of this UECP transmission / reception unit (segment) is registered. The initial sequence number (ISN) is set to “0” when the session is opened. Arithmetic sequence number is performed in two ³² modulo.

In the acknowledgment number field, when an acknowledgment (ACK) packet is received, a value of a sequence number that is assumed to be received next by the transmitting side of the segment is registered. Arithmetic acknowledgment number is also performed in two ³² modulo. In the window field, the number of bytes of data that can be accepted by the receiving side of this segment without an acknowledgment (ACK) is registered starting from the number of bytes indicated by the acknowledgment number. As shown in FIG. 4, various flags for identifying a packet are set in the code bit field.

  A NAK (Negative Acknowledgment) flag is set in a NAK packet that requests retransmission to the transmission source when the sequence number of the received packet is different from the receivable sequence number. The KPA (Keep Alive) flag is used for a keep alive request / response. The ACK (Acknowledgement) flag is set in an ACK packet that is returned when the acknowledgment number is valid. The PSH (Push) flag indicates whether or not the received data is immediately passed to the upper application. The received data in which this bit is set must be immediately passed to the upper application. The RST (Reset) flag is set when the session is forcibly disconnected. The SYN (Synchronize) flag is set when a session is established (opened). The FIN (Fin) flag is set when the session is disconnected (closed).

  FIG. 5 is a diagram showing an implementation example of a standby port number in UECP (a port number in which CCU 1 waits for a session open request and a port number in which an IP unit issues a session open request), and a UDP port number is used. The destination port number is an arbitrary fixed port number, and the source port number is an ephemeral (short-lived) port number.

  FIG. 6 is a diagram showing an implementation example of a communication port number (port number for transmitting and receiving data after establishing a session) in UECP. A UDP port number is used, and a destination port number and transmission are performed regardless of the communication direction. An ephemeral (short-lived) port number is used for any of the original port numbers. As for the port number on the CCU 1 side, as shown in FIG. 7, a standby port number may be used similarly to TCP.

  FIG. 8 is a diagram showing a sequence of session opening by UECP. Here, a case where a session opening is requested from the IP unit to CCU 1 will be described as an example.

  If the IP unit cannot receive the session open response packet (SYN + ACK) from the CCU within the first timeout value T1 after sending the session open request packet (SYN), it sends an RST packet as a session open failure and opens the session. Repeat the procedure from sending the request packet. If the CCU does not receive an acknowledgment (ACK) packet from the IP unit within the timeout value T1 after sending the session open response packet, the CCU sends an RST packet as a session open failure and waits for the session open request packet to be received. Try again. The IP unit enters the session establishment state when the acknowledgment packet is transmitted, and the CCU enters the session establishment state when the confirmation response packet is received.

  9 and 10 are diagrams illustrating a session closing sequence. In this embodiment, the session can be closed from either the CCU or the IP unit. FIG. 9 shows a procedure for requesting session close from the IP unit, and FIG. 10 shows a procedure for requesting session close from the CCU.

  If the transmission side of the session close request packet does not receive the session close response packet within the second timeout value T2 after the transmission, the session close request is not completed but the session close request packet is not retransmitted. Treat as session closed. As will be described in detail later, in this embodiment, if the transmitting side and the receiving side do not transmit a keep-alive request / response for a certain period of time, the session is closed. There is no problem.

  11, 12, and 13 are diagrams showing a basic sequence of data transmission. FIG. 11 is a transmission sequence from the CCU to the IP unit, FIG. 12 is a transmission sequence from the IP unit to the CCU, and FIG. The transmission sequence is shown.

  When the transmission side transmits m bytes of data with the sequence number n, the reception side that has received it normally returns an acknowledgment (ACK) with an acknowledgment number (n + m).

  FIG. 14 is a diagram showing a sequence of confirmation responses. In UECP, when data transmitted from the transmission side is received at the reception side, the reception side notifies the reception of transmission data by transmitting an acknowledgment (ACK) to the transmission side. When the transmission data or confirmation response is lost, the transmission side retransmits the data. The acknowledgment packet and the data transmission packet may be separated, or may be transmitted as a single packet.

  FIG. 15 is a diagram showing a window control sequence. In data transmission in a session established state in UECP, it is not necessary to wait for ACK reception for each data transmission, and data can be transmitted continuously within a window size (fixed value). The receiving side can transmit ACK for each data reception, but may transmit ACK for a plurality of data receptions within a window size.

  FIG. 16 is a diagram showing a sequence of sequence control. In conventional UDP, the receiving side may not be able to receive data in the order of data transmission on the transmitting side. However, in the UECP of the present invention, the sequence number and the acknowledgment number are used, and only the data according to the sequence number is received, not according to the sequence number. By discarding the data, it is possible to receive the data in the order even when crossed data is received.

  17 and 18 are diagrams illustrating a retransmission control sequence. In the UECP of the present invention, (1) the data transmission packet is lost (ACK cannot be received within the reception timeout value T3), (2) ACK is lost (ACK cannot be received within T3), (3) invalid Resending is performed in the four cases of receiving an acknowledgment number ACK and (4) receiving a NAK. FIG. 17 shows the sequence in the case (2), and FIG. 18 shows the sequence in the case (4).

  In the case of ACK reception timeout, the first retransmission is after T3 has elapsed, the second retransmission is after (T3 × 2) has elapsed, and the third retransmission is after (T3 × 3) has elapsed. Although the number of retransmissions is not limited, there is no limit to the number of times because there is keep-alive monitoring. The reception timeout value T3 can be arbitrarily adjusted by an operation from the maintenance computer 95 or an application setting, but is preferably set to about 500 ms or less.

  19, 20, and 21 are diagrams illustrating a keepalive sequence. FIG. 19 shows a sequence when there is no transmission data between the CCU and the IP unit, and the timer for the keep alive interval starts at the same time as the session is established. FIG. 20 is a sequence when there is transmission data from the CCU to the IP unit, and FIG. 21 is a sequence when there is transmission data from the IP unit to the CCU. When there is transmission data, the timer for the keep alive interval starts from the last transmission timing or reception timing.

  The UECP of the present invention performs keep alive for each session at the protocol level, and confirms the life and death of communication partners and the connection state of LAN cables (including IP devices such as HUBs and routers on the way). The keep-alive request transmission interval T5 can be set for each port number. In this embodiment, it can be arbitrarily set in increments of 50 ms within the range of 50 to 65500 ms, and is set to an initial value of 3 seconds here.

  In addition, the timeout value T4 of the response to the keep-alive request can be set for each port number. In this embodiment, the keep-alive number can be arbitrarily set within the range of 1 to 65535, and here, the initial value is set to 4 times. . Therefore, the timeout value T4 is 12 seconds (= T5 × 4) seconds on the IP unit side, and 14 seconds (= T5 × 4 + 2) seconds on the CCU side in consideration of the transmission time of the keep-alive request. The transmission interval T5 and the number of keep-alives can be arbitrarily adjusted by an operation from the maintenance computer 95 or application settings, but it is desirable that the timeout value T4 be about 12 to 14 seconds or less.

  In the data packet transmitted from the CCU and the IP unit, PSH, KPA and ACK are simultaneously set to “1”, and only when there is no data to be transmitted, KPA and ACK are set to “1”. However, PSH is set to “1” in the keep-alive request packet transmitted from the CCU to the IP unit. The “single keepalive request packet sent by CCU” is a packet with a data size of “0”, and the “keepalive response packet sent by the IP unit” is an acknowledgment (ACK) packet with a KPA of “1”. A packet having a data size of “0” is not retransmitted at the UECP level.

It is a block diagram of an IP button telephone apparatus to which the present invention is applied. It is the figure which showed the structure of the packet communication method (UECP) based on this invention. It is the figure which showed the structure of UECP header. It is the figure which showed the content of the code bit field of a UECP header. It is the figure which showed the example of mounting of the standby port number in UEPC. It is the figure which showed the example of mounting of the communication port number in UEPC (the 1). It is the figure which showed the example of mounting of the communication port number in UEPC (the 2). It is the figure which showed the session open sequence of UECP. It is the figure which showed the session close sequence (the 1) of UECP. It is the figure which showed the session close sequence (the 2) of UECP. It is the figure which showed the basic sequence (the 1) of the data transmission of UECP. It is the figure which showed the basic sequence (the 2) of the data transmission of UECP. FIG. 11 is a diagram illustrating a basic sequence (part 3) of UECP data transmission. It is the figure which showed the sequence of the confirmation response of UECP. It is the figure which showed the sequence of the window control of UECP. It is the figure which showed the sequence of the order control of UECP. FIG. 5 is a diagram showing a sequence (part 1) of UECP retransmission control. FIG. 6 is a diagram illustrating a UECP retransmission control sequence (part 2). It is the figure which showed the sequence (the 1) of UECP keep alive. It is the figure which showed the sequence (the 2) of UECP keep alive. FIG. 6 is a diagram showing a UECP keep-alive sequence (part 3);

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Main control unit (CCU), 2 ... TDM control unit (TCCU), 3 ... Customer edge unit (CEU), 4 ... Voice conversion unit (VCU), 100 ... IP button telephone apparatus

Claims (8)

In the packet communication method for controlling packet switching in the upper layer of UDP,
Secure the protocol header and data fields in the UDP packet data field.
The protocol header includes at least a sequence number field, an acknowledgment number field, and a code bit field for packet identification;
A packet communication method, wherein retransmission control and order control are performed based on numbers registered in the sequence number field and the acknowledgment number field.   The packet communication method according to claim 1, wherein the protocol header includes a window size field.   A session open response to a session open request and a first timeout value T1 that defines a timeout of an acknowledgment response to the session open response are stored, and the session open response or the acknowledgment response is stored even after the first timeout value T1 has elapsed. 3. The packet communication method according to claim 1 or 2, wherein if the packet cannot be received, the session opening procedure is redone.   If a second timeout value T2 that defines the timeout of the session close response to the session close request is stored and the session close response cannot be received even after the second timeout value T2 has elapsed, the session close request is not retransmitted. The packet communication method according to claim 1 or 2, characterized in that the session is handled as a closed session.   3. A third time-out value T3 that defines a time-out of an acknowledgment for data transmission is stored, and data is retransmitted if an acknowledgment cannot be received after the elapse of the third time-out value T3. 3. The packet communication method according to 2.   6. The packet communication method according to claim 5, wherein the third timeout value T3 is approximately 500 milliseconds or less. A fourth timeout value T4 that defines a timeout for a keepalive response to a keepalive request is stored, and if a keepalive response is not received after the fourth timeout value T4 has elapsed, the session is closed,
The packet communication method according to claim 1 or 2, further comprising a step of setting the fourth timeout value T4.   The packet communication method according to claim 7, wherein the fourth timeout value T4 is 12 to 14 seconds or less.

Images