JP2006352613A - Voice communication method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that a communication technique preventing an increase in delay even if HTTP or TCP is used is sought because companies etc. want to perform voice communication via the Internet irrespective of the presence of a firewall. <P>SOLUTION: A voice communication method in a voice communication terminal 10 using a communication protocol for performing retransmission control of data has a judging step in which a retransmission judging section 14 predicts a time when a packet transmission/reception section 13 receives a reception acknowledgment, and transmits retransmission instruction information to the section 13 when the reception acknowledgement is not received from an opposite party voice communication terminal 20 within a predetermined time from the predicted time; and a retransmission step in which the transmission/reception section 13 retransmits the packet to the voice communication terminal 20 on the basis of the retransmission instruction information. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a voice communication technique for retransmitting data in anticipation of a reception confirmation notification time.

  In a conventional technique, a protocol that performs retransmission control represented by the TCP protocol detects packet discard and performs packet retransmission. At this time, in order to avoid congestion deterioration due to packet retransmission, for example, the following method is used as a packet discard detection method.

  As shown in FIG. 4, when there is a missing packet in the receiving TCP, a reception confirmation notification (ack) indicating that the sequence number given to the packet before the missing packet has been received is transmitted. To the side. The reception confirmation notification is generally transmitted every two received packets. However, when packet discard is detected, it is common to perform switching so that a reception confirmation notification is transmitted for each reception. For example, if a packet of 4, 5 is received after receiving a packet with a sequence number of 1, 2, 2, 3 in the middle is missing. Therefore, every time 2, 4, and 5 are received, a reception confirmation notification indicating that up to 2 has been received is transmitted to the transmission side. In this case, when the reception confirmation notification indicating that up to 2 has been received is received twice, it can be predicted that the packet of sequence number 3 has been discarded. However, assuming that the order of arrival of the packets is changed, the packet with the sequence number 2 is often retransmitted when the reception confirmation notification that up to 2 has been received is received three times. Here, the sequence numbers are expressed as 1, 2, and 3 for simplicity, but in actuality, the sequence numbers are given based on the number of bytes of data (for example, see Non-Patent Documents 1 and 2).

  The above technique is an operation in the case where there is a missing reception packet and the transmission side receives a duplicate reception confirmation notification. In some cases, packet discard does not occur in some cases, but all packets after a certain packet are discarded. In this case, since the reception confirmation notification is not returned at all, retransmission is performed by the following method.

  As shown in FIG. 5, the TCP on the transmission side measures the round-trip time of the packet from the transmission packet and the reception confirmation notification, and calculates a timeout value from the past round-trip time history until retransmission. The timeout value is generally a value obtained by adding four times the average deviation to the evaluation value (predicted value) of the round-trip time. The reason for adding four times the average deviation is to avoid unnecessary retransmissions.

  TCP is designed so that data can be transmitted at the maximum speed of the communication width as long as the processing on the receiving side permits, by transmission amount control using window control. When packet discarding occurs while transmitting at the maximum speed, there is a high possibility that the network is congested. Therefore, if retransmission is performed wastefully, the congestion may be further deteriorated. Therefore, a mechanism that does not perform useless retransmissions described so far is implemented.

  On the other hand, when the operation of reading data higher than TCP is applied to the description of the packet discard of the sequence number 3 described above, the operation is as follows.

  Data higher than TCP cannot read data after sequence number 3 until the TCP layer receives 3 packets retransmitted in the TCP layer. That is, as shown in FIG. 4, even if the sequence numbers 4 and 5 are received by the TCP layer, if the sequence number 3 is not received by the TCP layer, the data of the sequence numbers 4 and 5 is read by the upper layer from the TCP. I can't. As described above, TCP does not perform retransmission at an early stage in order to avoid congestion, and is not suitable for communication with a small delay.

  On the other hand, in the Internet using the TCP / IP protocol group, a company or the like often provides a firewall to connect to the Internet. When there is a firewall, it is generally limited to only HTTP using TCP that can connect to the Internet from within a company. When there is a firewall, HTTP is connected to the Internet via an HTTP proxy server.

Even if there is a firewall, there is a strong demand for voice communication using the Internet in companies and the like. However, HTTP relayed by an HTTP proxy server uses TCP and is not suitable for communication with a small delay.
V. Jacobson, “Modified TCP Consistency Avidance Algorithm”, Apr. 30, 1990 end2 end-interest mailing list. W. Richard Stevens (Author), "Detailed TCP / IP <Vol.1>Protocol"

  Even if there is a firewall in a company or the like, there is a demand for voice communication through the Internet. Therefore, there is a demand for a communication technology that does not increase delay even if HTTP and TCP are used.

  Accordingly, in order to solve the above-described problem, the invention according to claim 1 is a voice communication method using a communication protocol for performing packet retransmission control, and an expected time at which a reception confirmation notification is received from a counterpart performing voice communication. And a packet corresponding to the reception confirmation notification is transmitted to the other party when the reception confirmation notification is not received within a predetermined time from the predicted time.

  The invention according to claim 2 is a voice communication method in a voice communication system using a communication protocol for performing packet retransmission control, and is expected to receive a reception confirmation notification from the voice communication system of the other party that performs voice communication. A time is calculated, and when a reception confirmation notification is not received within a predetermined time from the expected time, a packet corresponding to the reception confirmation notification is transmitted to the voice communication system of the other party.

  According to a third aspect of the present invention, there is provided a voice communication method in a voice communication system using a communication protocol for performing data retransmission control, wherein the retransmission judgment unit predicts the time when the packet transmission / reception unit receives a reception confirmation notification. A determination step of transmitting retransmission instruction information to the packet transmission / reception unit when a reception confirmation notification is not received from the other party's voice communication system within a predetermined time from the expected time, and the packet transmission / reception unit includes the retransmission instruction information And a retransmission step for retransmitting the packet to the voice communication system of the other party based on the above.

  The invention according to claim 4 is characterized in that the expected time of the reception confirmation notification of the nth packet is predicted by transmission interval × (n−1) + fluctuation. The voice communication method described.

  The voice communication system uses a server including a packet transmission unit and a retransmission determination unit, or a computer including a packet transmission / reception unit and a retransmission determination unit, and a voice communication terminal including a speaker and a microphone. It is possible to use a system such as

  According to the first to fourth aspects of the present invention, since packet retransmission is detected at an early stage and data retransmission is performed, an increase in delay due to data retransmission can be suppressed.

  Moreover, the said effect can be acquired by using invention of Claims 1-4 only about the transmission side.

  Hereinafter, embodiments will be described with reference to the drawings.

(Embodiment 1)
FIG. 1 shows a configuration diagram of the transmission-side voice communication terminal 10 and the reception-side voice communication terminal 20. As shown in FIG. 1, the voice communication terminal 10 includes a communication unit 11, a control unit 12, a packet transmission / reception unit 13, and a retransmission determination unit 14.

  The voice communication terminal 20 includes a communication unit 21, a control unit 22, a packet transmission / reception unit 23, a voice data processing unit 25, a D / A conversion unit 27, and a speaker 27.

  The communication units 11 and 21 perform data transmission / reception between terminals via a communication line. For the communication unit 11, for example, a network card or the like can be used.

  The packet transmission / reception unit 13 receives the reception confirmation notification and notifies the retransmission determination unit 14 to that effect. The packet transmitting / receiving units 13 and 23 rearrange the packets that are the received data and packetize the data to be transmitted, and transmit the packets to the counterpart terminal. For the packet transmission / reception units 13 and 23, for example, an OS provided with a TCP stored in an HDD, a nonvolatile memory, a ROM, or the like can be used.

  The retransmission determination unit 14 stores the time at which voice communication is started in a memory, and a reception confirmation notification corresponding to the nth voice data transmitted by the packet transmitting / receiving unit 13 is constant from the time at which the packet transmitting / receiving unit 13 starts voice communication. If not received within the time, an n-th packet retransmission instruction is transmitted to the packet transmitting / receiving unit 13. When calculation of fluctuation described later is necessary, the reception time of the reception confirmation notification is stored in the memory, and fluctuation is determined from the average of dispersion of the reception time. For the retransmission determination unit 14, for example, a retransmission determination program stored in an HDD, a nonvolatile memory, a ROM, or the like can be used.

  For example, when the data received from the packet transmitting / receiving unit 23 is digitally compressed, the audio data processing unit 25 performs processing such as decompressing and transmitting the data to the D / A conversion unit 26. For the audio data processing unit 25, for example, a program stored in an HDD, a nonvolatile memory, a ROM, or the like can be used.

  The control units 12 and 22 control the packet transmission / reception units 13 and 23, the retransmission determination unit 14, and the audio data processing unit 25. For the control units 12 and 22, for example, a CPU can be used.

  The packet transmitting / receiving unit 13 of the voice communication terminal 10 on the transmission side transmits data of a certain size at regular intervals. Here, the transmission interval is 60 ms. In addition, a sequence number is assigned to the header portion of the audio data packet, and the data length is also assigned.

  The reception confirmation notification is transmitted from the packet receiving side to the packet transmitting side every time a packet is received. When the packet transmitting / receiving unit 13 of the voice communication terminal 10 on the packet transmitting side transmits a packet every 60 ms, if the communication delay is constant, a reception confirmation notification should be received every 60 ms. However, there is actually a fluctuation in communication delay, and the reception confirmation notification is not received accurately every 60 ms. The reception time of the n-th reception confirmation notification can be expressed by the following equation. The time starts from zero.

60 × (n−1) + Δ (n) [ms], Δ (0) = 0
Note that n is an integer of 1 or more and represents the order of reception confirmation notification. The first received packet notification is received at time zero. Δ indicates fluctuations.

  The predicted value of fluctuation Δ is calculated as follows. For example, the fluctuation Δ may use an average value of past reception time variances, or may be weighted to a closer reception time variance instead of a simple average.

The predicted reception time of the (n + 1) th reception confirmation notification is
60 × n + Δ (n + 1) [ms]
It can be expressed.

  The retransmission of the voice data by the packet transmitting / receiving unit 13 is performed when the retransmission determining unit 14 determines that it will not receive the (n + 1) th reception confirmation notification at the following time, which is the predicted reception time of the (n + 1) th reception confirmation notification. Do.

60 × n + Δ (n + 1) + t
Note that t is a set value.

  Here, consider the case where the packet 3 is discarded as shown in FIG.

  If the packet transmitting / receiving unit 13 does not transmit ack3 from the packet transmitting / receiving unit 23 beyond 60 × 3 + Δ (3 + 1) + t from the start of audio data transmission, the packet 3 is retransmitted even if ack3 is not transmitted. It is possible to adjust the time required for retransmission by adjusting the value of t.

  Note that it is also possible to perform the processing performed by the retransmission determining unit 14 in the packet transmitting / receiving unit 13 without providing the retransmission determining unit 14.

(Embodiment 2)
Next, a case where a reception confirmation notification is transmitted every time two packets are received will be described.

When the reception confirmation notification is transmitted every time two packets are received, the following three types of patterns can be considered for occurrence of packet discard.
[1] When only the first packet for which the acknowledgment notification is not transmitted is discarded.
[2] When only the second part to which the acknowledgment notification is sent is discarded.
[3] When both are discarded.

  If only the first one is discarded, if a second packet is received, a reception confirmation notification is sent immediately, and it can be seen that the first packet has been discarded. There is no need to use this embodiment as soon as it is received.

  When only the second one is discarded, the reception confirmation notification is not transmitted as shown in FIG. In this case, only the packet 4 is discarded. However, since it is impossible to distinguish whether both 3 and 4 are discarded or only 4 is discarded on the transmitting side, both 3 and 4 are immediately retransmitted. Even when both are discarded, the reception confirmation notification is not transmitted, so both 3 and 4 are immediately retransmitted.

(Other)
In the above embodiment, audio data has been described as an example, but the above embodiment can be applied to any information communication that transmits and receives packets. As information to be transmitted, for example, information such as video and audio recorded in a database accessible by the communication terminal 10 can be used.

  In addition, it is conceivable that a retransmission determination unit, a speaker, a microphone, an A / D converter, and a D / A converter are installed in both of the audio playback terminals 10 and 20 and realized as an IP phone. In this case, it can be configured as a system including a voice communication terminal including a microphone and a speaker, and a computer including a communication unit, a control unit, a packet transmission / reception unit, and a retransmission determination unit. When implemented as an IP phone, voice input from a microphone is converted into digital data by an A / D converter, and this data is packetized by the packet transmitter / receiver and transmitted to the other party.

The block diagram of a voice communication terminal. The figure which shows the audio | voice data resending of Embodiment 1. FIG. The figure which shows the audio | voice data resending of Embodiment 2. FIG. The figure which shows audio | voice data resending using TCP. The figure which shows audio | voice data resending using TCP.

Explanation of symbols

10, 20 Voice communication terminal 11, 21 Communication unit 12, 22 Control unit 13, 23 Packet transmission / reception unit 14 Retransmission determination unit 25 Audio data processing unit 26 A / D conversion unit 27 Speaker

Claims (4)

A voice communication method using a communication protocol for performing packet retransmission control,
Calculate an expected time for receiving a reception confirmation notification from the other party that performs voice communication, and send a packet corresponding to the reception confirmation notification to the other party when the reception confirmation notification is not received within a predetermined time from the predicted time. A voice communication method characterized by the above. A voice communication method in a voice communication system using a communication protocol for performing packet retransmission control,
An expected time for receiving a reception confirmation notification from the voice communication system of the other party that performs voice communication is calculated, and when the reception confirmation notification is not received within a predetermined time from the predicted time, a packet corresponding to the reception confirmation notification is calculated. A voice communication method characterized by transmitting to the voice communication system. A voice communication method in a voice communication system using a communication protocol for performing retransmission control of data,
When the retransmission judgment unit predicts the time when the packet transmission / reception unit receives the reception confirmation notification and does not receive the reception confirmation notification from the other party's voice communication system within a certain time from the predicted time, the retransmission instruction information is sent to the packet transmission / reception unit. A decision step to send to
A voice communication method comprising: a retransmission step in which the packet transmitting / receiving unit retransmits a packet to the voice communication system of the other party based on the retransmission instruction information. The expected time of the reception confirmation notification of the nth packet is
The voice communication method according to claim 1, wherein the prediction is performed based on transmission interval × (n−1) + fluctuation.

Images