JP2004222270A - Transmission control method, communication device, communication system and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide technology for appropriately discriminating whether a confirmation response received in a transmitter is corresponding to an original data segment or to a resent data segment without increasing an information quantity of data segments and without requiring a considerable design change for the transmitter and a receiver of a server device. <P>SOLUTION: A first time from a transmission time point of a data segment to a receiving time point of the confirmation response to the transmission is measured, probability distribution of the first time with respect to a client device 50 is learnt on the basis of the statistic result totalizing the measured first times, and the learnt result is used to estimate a data segment corresponding to the confirmation response. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

  The present invention relates to a technique for controlling retransmission in data communication performed via a communication network.

  In recent years, devices that transmit and receive data according to TCP (Transmission Control Protocol) have become widespread. TCP is well known as a typical transport layer communication protocol in the OSI reference model. In this TCP, a data stream passed from an upper layer is divided into segments and transmitted to a receiving device that is a communication partner. The TCP provides a function for reliably sending the divided data segments to the receiving device in order. Specifically, in TCP, a number (hereinafter, referred to as a sequence number) indicating the order of data is assigned to each data segment. When transmitting the data segment, the transmitting device sets the sequence number in the header and transmits the data segment. The transmitting device sets a timer at the same time as transmitting each data segment. If the acknowledgment is not received before the timer reaches the timeout value, the transmitting device makes a tentative determination that the data segment has been lost without reaching the receiving device, and considers that the data segment has been lost. The received data segment is retransmitted to the receiving device. Here, the timeout value is determined based on a time (a predicted round-trip time of the data segment) that is expected to receive an acknowledgment for the data segment after transmitting the data segment.

  The transmitting device determines which data segment the received acknowledgment is for the acknowledgment by referring to the acknowledgment number included in the header of the acknowledgment. Hereinafter, this determination will be described. First, when the receiving device receives a data segment and transmits an acknowledgment for the data segment, the receiving device sets a sequence number expected to be received next (hereinafter referred to as an acknowledgment number) in a header of the acknowledgment. For example, when a sequence number starting from 500 and continuing to 1000, 1500,... Is used, the acknowledgment number transmitted to the transmitting device when a data segment having the sequence number 500 is received is 1000. By receiving the acknowledgment having such an acknowledgment number, the transmitting device determines that the data segment transmitted by the transmitting device has been successfully received by the receiving device, and transmits the subsequent data segment.

  Note that the case where the transmitting device receives the acknowledgment having the acknowledgment number 1000 means that the transmitting device has not transmitted the data segment having the sequence number 1000, or the data segment is transmitted to the receiving device even if the data segment has been transmitted. It is when it has not arrived. This acknowledgment number is always 1000 until the receiving device receives the data segment having the sequence number 1000. For example, even if the data segment having the sequence number 1500, which is the subsequent segment, has arrived at the receiving device earlier because the data segment having the sequence number 1000 has been lost, etc. Until receiving the data segment having the sequence number 1000, the acknowledgment number of the acknowledgment transmitted from the receiving apparatus is set to 1000. That is, the acknowledgment number is always set to the sequence number of the youngest segment of the data segment that the receiving device has not yet received.

  In the transmission of the data segments described above, in order to surely transmit each data segment, a method of transmitting one data segment, receiving an acknowledgment response thereto, and then transmitting the next data segment is adopted. Is preferred. However, although this method is reliable, the data transmission efficiency deteriorates. In order to improve transmission efficiency, TCP adopts a method of transmitting a number of data segments determined by a window at a time. The window is the number of bytes of data allowed to be transmitted to the receiving device before receiving the acknowledgment, that is, the number of data segments. The size of this window is determined by the transmitting device within a range not exceeding the free space of the buffer in the receiving device. When the transmitting device receives an acknowledgment for the transmitted data segment, the window slides by the number of data segments for which the acknowledgment was received at that time, and subsequent data segments are transmitted by the number of sliding windows. This is the “sliding window method”. In the sliding window method, the data flow rate can be controlled by controlling the size of the window.

  By the way, after a plurality of data segments are transmitted to the receiving device according to the sliding window method, communication may be stopped due to a congestion state of a communication path between the transmitting device and the receiving device. In particular, when the communication path includes a wireless section, the radio wave environment is highly likely to deteriorate. Even if the transmitting device transmits a data segment in such a situation, the data segment does not reach the receiving device and is lost in the middle, or somewhere in the communication network until the communication state of the communication network is restored. It will be temporarily stored in the node. Thereafter, when the wireless environment is improved or communication is resumed, the data segment temporarily stored in the network may reach the receiving device after being delayed.

  Here, if the time during which the state of the communication path is deteriorated is long, the following situation occurs. First, after transmitting the data segment, if the acknowledgment is not received before the time-out occurs, the head segment of the data segment that has been transmitted and has not received the acknowledgment is retransmitted. As a result, both the data segment staying in the network (hereinafter, referred to as the original data segment) and the retransmitted data segment (hereinafter, referred to as the retransmission data segment) reach the receiving device. Then, the receiving device transmits an acknowledgment response to the original data segment when receiving the original data segment, and transmits an acknowledgment response to the retransmission data segment when receiving the retransmitted data segment.

  The acknowledgment for the original data segment and the acknowledgment for the retransmitted data segment usually have the same acknowledgment number. For this reason, when the transmitting device receives the first acknowledgment, it cannot determine whether the acknowledgment is for the original data segment or the retransmission data segment. In such a case, when the transmitting device receives an acknowledgment for the first time after transmitting the retransmitted data segment, the transmitting device regards the acknowledgment as an acknowledgment for the retransmitted data segment and determines that the original data segment did not reach the receiving device. I do. Hereinafter, the reason why such a determination is made will be described in detail with reference to examples.

  FIG. 9 is a sequence chart illustrating an example of a case where packet communication is performed between a conventional server device 10 ′ (transmitting device) and a client device 50 (receiving device). In the figure, the four-digit number shown to the right of the start point of each arrow in the server device 10 'is the sequence number of the data segment transmitted from the server device 10' by packet communication. The four-digit number shown on the left side of the starting point of each arrow in the client device 50 is an acknowledgment number included in the acknowledgment transmitted from the client device 50. Here, it is assumed that the initial window size is “3”. That is, three data segments can be transmitted before receiving the acknowledgment.

  In FIG. 9, original data segments S1, S2, and S3 having sequence numbers 0, 1000, and 2000 are transmitted from the server device 10 'to the client device 50. At this time, a timer is set in the server device 10 '.

  In the illustrated example, the original data segments S1, S2, and S3 are sequentially received with a delay by the client device 50 due to the deterioration of the communication environment in the network. Upon receiving the original data segment S1, the client device 50 transmits an acknowledgment R1 having an acknowledgment number 1000. Similarly, the client device 50 transmits an acknowledgment R2 having an acknowledgment number 2000 when receiving the original data segment S2, and transmits an acknowledgment R3 having an acknowledgment number 3000 when receiving the original data segment S3.

  In the example illustrated in FIG. 9, the server device 10 'does not receive an acknowledgment even if the elapsed time measured by the timer reaches the timeout value. For this reason, in the server device 10 ′, when the timeout occurs, it is determined that the original data segment S1 has not been received by the client device 50, and the data segment having the sequence number 0 (retransmission data segment S′1) is retransmitted. Is done. At this time, since the window size is reduced to the minimum value due to the occurrence of the retransmission timeout, only one data segment is retransmitted.

  After that, the server device 10 'receives the acknowledgment R1 having the acknowledgment number 1000. In the above-described situation, whether the acknowledgment R1 corresponds to the original data segment S1 or the resend data segment S'1 It cannot be determined whether it is a corresponding one. As a result, the data segment having the sequence number 1000 (retransmission data segment S'2) is transmitted as being regarded as corresponding to the retransmission data segment S'1. At this time, the window size is increased by one data segment due to the reception of the acknowledgment R1, so that the data segment having the sequence number 2000 (retransmission data segment S'3) is also continuously transmitted.

  Next, in the server device 10 ', the acknowledgment R2 having the acknowledgment number 2000 transmitted from the client device 50 is received. As a result, the data segment having the sequence number 3000 (the original data segment S4) and the subsequent data segments are received. Are sequentially transmitted.

  In the illustrated example, the original data segments S2 and S3 are retransmitted (retransmission data segments S'2 and S'3) even though the client device 50 has actually received them successfully. That is, since both the data segments S2 and S3 are received twice by the client device 50, it can be said that as a result, useless retransmission of the data segment has been performed. Also, as is clear from FIG. 9, since acknowledgments R'2 and R'3 for retransmission data segments S'2 and S'3 are sent, retransmission of a total of four data segments is wasteful as a result. Met.

If it is not possible to determine whether the first acknowledgment after retransmission of the data segment is for the original data segment or for the retransmitted data segment, if it is determined that it is for the original data segment, such waste No retransmission is performed.
However, if such a determination is made, there may be a case where an acknowledgment to the original data segment is regarded as an acknowledgment to the retransmitted data segment in fact. In this case, the window is slid behind the arrangement of the data segments, and the subsequent original data segments are transmitted. If this occurs repeatedly, the number of data segments transmitted from the server device 10 'and not arriving at the client device 50 may increase.

To avoid the risk of such a situation, if it cannot be determined whether the acknowledgment is for the original data segment or for the retransmitted data segment, it is determined that the acknowledgment is for the original data segment. -ing
However, when such a determination method is adopted, useless retransmission of the data segment cannot be avoided.

  Techniques for solving such a problem are described in Non-Patent Documents 1 and 2. Non-Patent Document 1 discloses a technology that makes it possible to accurately determine whether an acknowledgment is for an original data segment or a retransmission data segment by using a time stamp option (RFC 1323). Has been described.

Non-Patent Document 2 discloses a technique for estimating whether an acknowledgment is for an original data segment or a retransmission data segment using statistical information in packet communication via a wired packet communication network. .
In this document, the smallest round-trip time among round-trip times obtained by actually measuring the time from when a data segment is transmitted to when an acknowledgment arrives when a communication connection between a transmitting device and a receiving device is established is obtained. It has been proposed to perform the above estimation using 1/2 of the trip time as a threshold value. In other words, if the elapsed time from the transmission of the retransmission data segment to the reception of the first acknowledgment is equal to or greater than the threshold (or exceeds the threshold), it is regarded as an acknowledgment for the retransmission data segment. (Or below the threshold) is considered a confirmation response to the original data segment.

The use of a threshold value of を of the minimum round trip time has been proposed based on the circumstances listed below.
[Situation 1] When the statistics are obtained, the period from the transmission of the retransmission data segment to the elapse of 1/2 of the minimum round trip time, the period of 3/4, and the elapse of 1/1 It has been found that there is no great difference in the probability of receiving an acknowledgment for the original data segment within the time period before the acknowledgment.
[Situation 2] When statistics are obtained, it is found that the probability of receiving an acknowledgment for the retransmitted data segment rapidly increases after about half of the minimum round trip time has elapsed since the retransmitted data segment was transmitted. Was.

R. Ludwig, et. Al., "The Eifel Detection Algorithm for TCP" (draft-ietf-tsvwg-tcp-eifel-alg-04.txt) ), [Online], July 24, 2002, Internet <URL: http: // www. watersprings. org / pub / id / draft-ietf-tsvwg-tcp-eifel-alg-04. txt> Mark Allman, et. Al., "On Estimating End-To-End Network Path Properties", ACM SIGCOMM '99, 1999 October, Vol. 29, No. 4, p. 263-274, (Section 2.8)

  However, in the technique of Non-Patent Document 1, even in a good communication environment, the transmitting device adds time stamp information to the original data segment, and the receiving device adds time stamp information to the acknowledgment. Will be added. That is, the information amount of both data segments increases. Therefore, if this technique is applied to packet communication via a mobile packet communication network on the premise of pay-as-you-go billing, in which the communication fee changes according to the traffic, the communication fee will increase. This is an unnecessary increase in the amount of communication in a favorable communication environment in which retransmission of data is rarely required. Needless to say, the increase in the communication fee is not what the user of the receiving device or the user of the transmitting device desires.

  Instead, a method is also conceivable in which the technique is applied to set information in reserved bits in a TCP header used at the time of communication, instead of adding information to a data segment. However, when such a method is used, the existing communication system cannot cope with it, and therefore, a significant design change is required for the transmitting device and the receiving device in order to use the method.

  Further, even if the technique of Non-Patent Document 2 is used in a communication environment in which the delay of a data segment in a wireless section is extremely large, such as a mobile communication environment via a mobile communication network, an optimum determination result cannot be obtained.

  The present invention has been made in view of the circumstances described above, and appropriately determines whether an acknowledgment received by a transmitting device is for an original data segment or a retransmitted data segment. The purpose is to provide technology that can do. According to this technique, a slight change in the design of the transmission device is sufficient without increasing the information amount of the data segment, and without requiring any change in the design of the reception device.

  In order to solve the above problem, the present invention provides a first generation process of generating first probability distribution information indicating an appearance probability of a round trip time of a data block, and a plurality of data blocks from a transmission device to a reception device. And a retransmitting a data block that has not received an acknowledgment signal from the receiving device among a plurality of data blocks transmitted to the receiving device, and measuring an elapsed time. Activating the timing means, acquiring an elapsed time measured by the timing means when receiving an acknowledgment signal from the receiving device, and acquiring the value of the acquired elapsed time and the first probability. Based on the distribution information, it is estimated whether the acknowledgment signal is an acknowledgment for one of the plurality of data blocks transmitted in the transmission process. An estimating step, wherein, when the acknowledgment signal is estimated to be an acknowledgment for one of the transmitted plurality of data blocks, the plurality of data blocks transmitted in the transmitting step And a transmission control method in a communication network, characterized by transmitting a data block following the data block. An example of the data block is a data segment in TCP. Also, preferably, the receiving device is a mobile communication terminal that performs packet communication via a mobile communication network.

  In a preferred aspect, after transmitting the subsequent data block, if it is determined that the estimation performed in the estimation step is correct, the appearance time is used to determine the appearance of the elapsed time using the value of the elapsed time acquired in the acquisition step. The method further includes a second generation step of generating second probability distribution information indicating a probability, wherein the estimation step is based on a value of the elapsed time acquired in the acquisition step, and the first and second probability distribution information. Then, the estimation is performed. In this case, in the second generation step, when the acknowledgment signal requesting the subsequent data block is received at least twice from the receiving device, the estimation performed in the estimation step is determined to be correct. Is also good.

  In another preferred aspect, when transmitting the plurality of data blocks to a plurality of the receiving devices, the first and second generation processes may include a plurality of the plurality of data blocks set between the plurality of receiving devices. Generating the first and / or second probability distribution information for each of the connections, and the estimating step includes a step of acquiring the acknowledgment signal from any of the plurality of receiving apparatuses, The estimation may be performed based on a time value and the first and / or second probability distribution information generated for the connection set up with any one of the receiving devices. Further, in another preferred aspect, when transmitting the plurality of data blocks to a plurality of the receiving devices each belonging to a different sub-network, the first and / or second generation process may include the sub-network. Generating the first and / or second probability distribution information every time, and the estimating step includes the step of calculating the elapsed time acquired by the acquiring unit when an acknowledgment signal is received from any of the plurality of receiving devices. The estimation may be performed based on a value and the first and / or second probability distribution information generated for the subnetwork to which any of the receiving apparatuses belongs.

  Preferably, the first and second probability distribution information are statistical information obtained on a trial basis before transmitting the plurality of data blocks to the receiving device, In the process, after the transmission of the plurality of data blocks is started, the first and second probability distribution information may be generated by updating the statistical information obtained on a trial basis. Alternatively, the first and second probability distribution information is statistical information obtained in advance before starting communication for transmitting the plurality of data blocks to the receiving device, and the first and second probability distribution information are provided. In the generating step, after the communication for transmitting the plurality of data blocks is started, the first and second probability distribution information may be generated by updating the statistical information obtained in advance. .

  Further, the present invention provides a first generation unit for generating first probability distribution information indicating an appearance probability of a round trip time of a data block, a transmission unit for sequentially transmitting a plurality of data blocks to a reception device, A timer for measuring the elapsed time, and among the plurality of data blocks transmitted to the receiver, retransmitting a data block that has not received an acknowledgment signal from the receiver; and Activating means, acquiring means for acquiring an elapsed time measured by the clocking means when receiving an acknowledgment signal from the receiving device, and acquiring the value of the acquired elapsed time and the first probability distribution information. Estimating means for estimating whether or not the acknowledgment signal is an acknowledgment for one of the plurality of data blocks transmitted by the transmitting means based on the acknowledgment signal. And transmitting the data block subsequent to the plurality of data blocks transmitted by the transmitting means when the acknowledgment signal is estimated to be an acknowledgment response to one of the plurality of transmitted data blocks. There is provided a communication device characterized by transmitting.

  In a preferred aspect, after transmitting the subsequent data block, when the estimation made by the estimating means is determined to be correct, the appearance time of the elapsed time is obtained by using the value of the elapsed time acquired by the acquiring means. The apparatus further includes a second generation unit that generates second probability distribution information indicating a probability, wherein the estimation unit is configured to calculate the second probability distribution information based on the value of the elapsed time acquired by the acquisition unit and the first and second probability distribution information. Then, the estimation may be performed. In addition, the communication device according to the present invention has means for executing various aspects of the transmission control method.

Further, the present invention provides a communication system capable of realizing the transmission control methods in the above various aspects by providing at least two devices with the means of the communication device.
In addition, the present invention provides a program for causing a communication device to execute the transmission control methods in the various aspects described above. This program can be recorded on other recording media such as a magnetic tape, a magnetic disk, a floppy (registered trademark) disk, an optical recording medium, a magneto-optical recording medium, a DVD (Digital Video Disk), a RAM, and the like in which the program is stored.

  According to the present invention, it is possible to optimally determine whether the acknowledgment data segment received by the transmission device is for the original data segment or the retransmission data segment, and as a result, Unnecessary transmission can be prevented. According to this technique, a slight change in the design of the transmission device is sufficient without increasing the information amount of the data segment, and without requiring any change in the design of the reception device.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In all the drawings for describing the embodiments, components having the same function are denoted by the same reference numerals, and repeated description thereof will be omitted.

[First Embodiment]
(1. Configuration)
<Configuration of Communication System 1>
FIG. 1 is a block diagram illustrating a configuration of a communication system 1 according to the first embodiment of the present invention.
The communication terminal 40 is connected to the client device 50 and communicates with the client device 50.
The mobile packet communication network 30 provides a packet communication service to communication terminals 40 accommodated in the mobile packet communication network 30. The communication section between the mobile packet communication network 30 and the communication terminal 40 includes a wireless section (not shown). Communication in this wireless section is performed using radio waves.
The server device 10 performs packet communication with the client device 50 via the communication terminal 40, the mobile packet communication network 30, and the Internet 20, and transmits a data segment to the client device 50 by packet communication. In the present embodiment, packet communication is performed according to TCP (Transmission Control Protocol).

<Configuration of server device 10>
Next, the configuration of the server device 10 will be described. Since the configuration of the server device 10 is the same as that of a general computer, only the configuration according to the present invention will be described with reference to FIG.
The CPU 100 controls each unit of the server device 10 by executing a program stored in the storage unit 105. Further, the CPU 100 includes timers 100a and 100b.

The timer 100a outputs a trigger signal when the timeout value set by the CPU 100 elapses. Further, the timer 100a measures a time (hereinafter, referred to as a round trip time) from the transmission of a data segment to the reception of an acknowledgment for the data segment. As in the conventional method, the round trip time obtained by the timer 100a is used for calculating a timeout value. This timeout value is a variable value determined according to the communication environment that changes every moment. In the present embodiment, the round trip time measured by the timer 100a is also used when generating probability distribution data (see FIG. 3, which will be described in detail later) indicating the appearance probability of the round trip time.
Although the round trip time is generally measured for one data segment for each window, the round trip time may be measured for all data segments and used for updating the timeout value or the probability distribution data. .

  Timer 100b operates only when a retransmission data segment is transmitted. Specifically, the timer 100b starts measuring the time according to the instruction of the CPU 100 when transmitting the retransmission data segment, and ends the measurement according to the instruction of the CPU 100 when the first acknowledgment is received after the transmission of the retransmission data segment. I do. The time measured by the timer 100b is notified to the CPU 100. By comparing this time with the probability distribution data, CPU 100 estimates whether the received acknowledgment is for the original data segment or for the retransmitted data segment.

The storage unit 105 includes a random access memory (RAM) 102, a read only memory (ROM) 103, and a hard disk (HD) 104.
The ROM 103 stores a program for causing the CPU 100 to perform data segment transmission control processing. After transmitting a data segment (hereinafter referred to as an original data segment) to the client device 50 according to this program, the CPU 100 waits for reception of an acknowledgment response corresponding to the data segment. Further, the CPU 100 sets a timeout value in the timer 100a when transmitting the original data segment, and causes the timer 100a to measure an elapsed time thereafter.

  Here, when the radio wave environment in a wireless section between the communication terminal 40 and the mobile packet communication network 30 is deteriorated, a situation may occur in which an acknowledgment is not received even after the time corresponding to the timeout value elapses. . In such a case, if a trigger signal indicating that the timeout value has elapsed is output from the timer 100a without receiving an acknowledgment, the CPU 100 determines that the data segment has not been received by the client device 50. I do. Then, the data segment is retransmitted, and the timer 100a starts measuring the elapsed time again. Since the data segment has been retransmitted, the measurement of the elapsed time by the timer 100b is also started.

  It is assumed that the communication via the mobile packet communication network 30 is restarted due to the improvement of the radio wave environment, and the transmission and reception of the data segment and the acknowledgment are enabled again. Accordingly, upon receiving the confirmation response, the CPU 100 resets the timer 100a and causes the timer 100b to stop measuring the elapsed time. Then, using the elapsed time measured by the timer 100b and the probability distribution data described later, it is determined whether the received acknowledgment has a high probability that it corresponds to the original data segment or corresponds to the retransmitted data segment. It is determined whether the probability is high.

  If it is determined that the probability that the original data segment corresponds to the original data segment is high, the CPU 100 estimates that the original data segment has been received by the client device 50, and subsequently transmits a subsequent original data segment. If it is determined that the probability that the data segment corresponds to the retransmission data segment is high, the CPU 100 estimates that the original data segment has not been received by the client device 50, and then determines whether or not there is a subsequent retransmission data segment. I do. If the determination result is affirmative, the subsequent retransmission data segment is transmitted. If the determination result is negative, the subsequent original data segment is transmitted.

The HD 104 stores probability distribution data (FIG. 3) indicating the appearance probability of the round trip time measured by the timer 100a.
This probability distribution data is data indicating a probability distribution of a round trip time until an acknowledgment for a data segment is received when communication is normally performed. FIG. 3 shows the appearance probability of the round trip time of each data segment starting from the time at which the data segment was transmitted.

  In a preferred embodiment, when communication is normally performed, a round trip time (hereinafter referred to as O1-RTT) until an acknowledgment for the original data segment is received, and an acknowledgment for the retransmitted data segment are received. The respective probability distributions of the round trip times up to (hereinafter, referred to as S1-RTT) are treated as being the same as described below.

  FIG. 4 shows, with a solid line, the probability distribution of the arrival time until the acknowledgment for the original data segment is received, starting from the time at which the original data segment was transmitted, until the acknowledgment packet for the retransmitted data segment is received. Is indicated by a dotted line. In this figure, time t1 is the time when the retransmission data segment was transmitted. Time t2 is a time when the appearance probability of the arrival time until the acknowledgment for the original data segment is received is equal to the appearance probability of the arrival time until the acknowledgment for the retransmission data segment is received. As shown in the figure, generally, an acknowledgment for the original data segment is received until a certain time elapses (until time t2 in the illustrated example) from the transmission of the original data segment. After that, there is a high probability that an acknowledgment for the retransmitted data segment will be received. Therefore, in the present embodiment, it is estimated whether the acknowledgment received after transmission of the retransmission data segment is for the original data segment or for the retransmission data segment by using two probability distributions that draw the same curve as illustrated. It is carried out.

Further, the HD 104 stores a learning program for causing the CPU 100 to perform the above-described learning process of the probability distribution. In this learning process, the CPU 100 causes the timer 100a to measure a round trip time from the transmission of the original data segment to the client device 50 to the reception of an acknowledgment corresponding to the data segment. Then, the value of the probability distribution data stored in the HD 104 is sequentially updated by calculating the measured appearance probability of each round trip time. As described above, the CPU 100 learns the above-described probability distribution.
In a preferred embodiment, when it is determined that the acknowledgment received first after transmitting the retransmission data segment after the timeout corresponds to the retransmission data segment, the probability distribution is calculated using the time measurement value of the timer 100b. The data may be updated. When it is determined that the acknowledgment corresponds to the retransmission data segment, the value measured by the timer 100b is longer than the elapsed time until time t2 in FIG. 4 and is the same acknowledgment as the first acknowledgment received. This is the case where the second acknowledgment having the number is not received.

The CPU 100 may start learning the probability distribution after starting communication between the server device 10 and the client device 50, or may learn the probability distribution based on a statistical result obtained in advance. good. Specifically, three methods are conceivable.
In the first method, after starting communication between the server device 10 and the client device 50, a period for transmitting a test data segment is provided, and probability distribution data is generated based on the data segment transmitted and received during the period and the acknowledgment. I do. Then, after a certain period of time, the transmission of the data segment to be transmitted is started, and in the event that the data segment needs to be retransmitted, the probability distribution data obtained using the test data segment is used. The above determination is made. Even after the transmission of the data segment to be transmitted starts, the probability distribution data is updated each time the timer 100a acquires the round trip time.

In the second method, the transmission of the data segment to be transmitted is started immediately after the communication between the server device 10 and the client device 50 is started, and since the probability distribution data does not exist for a while, the above-described determination is not performed. This is a method of learning only the distribution. In this method, the transmission control method of the present invention is used after a certain period of time has elapsed since the start of communication. Therefore, if the method is used for communication for a relatively long time, an effect can be expected.
The third method is to start transmission of a data segment to be transmitted immediately after starting communication between the server device 10 and the client device 50, and to use the above-described probability distribution data measured in advance using a similar communication environment. This is a method of making a judgment. In this case, learning is performed by sequentially updating past probability distribution data using the round trip time acquired after the start of communication.
When there are a plurality of client devices 50 (not shown), the CPU 100 learns the above-described probability distribution for each IP address of the client device 50.

  As described above, in the present embodiment, the server device 10 measures the elapsed time until the acknowledgment is received after the retransmission data segment is transmitted, and uses the learned probability distribution data to measure the elapsed time O1 Determining whether the acknowledgment corresponds to the original data segment or to the retransmission data segment by determining whether the probability of being an RTT or S1-RTT is high; presume.

<Configuration of Client Device 50>
Since the client device 50 is similar to a general computer, only the functions according to the present invention will be described.
When receiving the data segment from the server device 10, the client device 50 has a function of transmitting an acknowledgment indicating that the data segment has been received to the server device 10. Specifically, the client device 50 sets the sequence number of the data segment expected to be received next in the acknowledgment number field of the acknowledgment header, and transmits the acknowledgment to the server device 10.

(2. Operation)
Next, the operation of the present embodiment will be described.
FIG. 5 is a sequence chart illustrating an example of a case where packet communication is performed between the server device 10 and the client device 50. FIG. 6 is a flowchart illustrating a data segment transmission and reception operation of the server device 10 according to the present embodiment. In the present embodiment, the data segment is transmitted using the sliding window method, and the window size at the start of communication is “3”. It is also assumed that the capacity of the data segment that can be received by the client device is sufficiently larger than the size of the window set by the transmitting device. Also, for simplicity of explanation, the window size is fixed at “3”.
Further, the CPU 100 of the server device 10 executes the learning program stored in the HD 104 and performs the above-described learning process of the probability distribution every time a data segment is transmitted and received while performing packet communication with the client device 50. It is assumed that

First, in FIG. 5, each data segment having the sequence numbers 0, 1000, and 2000 (original data segments S1, S2, and S3) is transmitted from the server device 10 to the client device 50. The reason that three data segments are transmitted is that the window size is “3”.
In the illustrated example, the radio wave environment in a wireless section included in a communication section between the communication terminal 40 and the mobile packet communication network 30 deteriorates, and the original data segments S1, S2, and S3 temporarily stay in the communication network. are doing. Thereafter, the radio wave environment is improved, and the original data segments S1, S2, S3 are transmitted to the client device 50 via the mobile packet communication network 30. That is, the original data segments S1 to S3 arrive at the client device 50 with a large delay from the time of transmission. Then, first, the client device 50 that has received the original data segment S1 transmits an acknowledgment R1 having an acknowledgment number 1000 corresponding to the original data segment S1 to the server device 10.

Next, the operation of the server device 10 up to the point described above will be described with reference to FIG.
In step S10, the CPU 100 of the server device 10 transmits the original data segments S1, S2, and S3. Next, it waits for an acknowledgment transmitted from the client device 50 to the original data segment S1. For this reason, the CPU 100 sets the timeout value in the timer 100a and causes the timer 100a to measure the elapsed time thereafter (step S11).
Then, the CPU 100 determines whether or not any confirmation response has been received (step S12). If the determination result is “NO”, the CPU 100 determines whether or not a timeout of the timer 100a has occurred (step S13). ). If the result of the determination in step S13 is "NO", the flow returns to step S12. Thereafter, the CPU 100 repeats the determinations in steps S12 and S13 until an acknowledgment is received until the timer 100a times out.

  Then, it is assumed that a timeout occurs before receiving the confirmation response R1 transmitted from the client device 50. In this case, without receiving an acknowledgment, the elapsed time measured by the timer 100a reaches the timeout value, and a trigger signal is output (step S12; NO, step S13; YES). Then, CPU 100 resets the value of timer 100a and advances the process to step S14.

  In step S14, the CPU 100 determines that the original data segment S1 has not been received by the client device 50, and retransmits the data segment having the sequence number 0 (retransmission data segment S'1 in FIG. 5) to the client device 50. A timeout value is set in the timer 100a, and the timer 100a measures the elapsed time. At this time, the timer 100b also starts measuring the elapsed time.

Then, the CPU 100 determines whether or not any confirmation response has been received (step S15). If the determination result is “NO”, the CPU 100 determines whether or not a timeout of the timer 100a has occurred (step S16). ). If the result of the determination in step S16 is "NO", the flow returns to step S15. Thereafter, the CPU 100 repeats the determination of steps S15 and S16 until an acknowledgment is received until the timer 100a times out.
Thereafter, it is assumed that the above-described acknowledgment R1 is received before the timeout value elapses, that is, before the trigger signal is output from the timer 100a. In this case, when the process proceeds to step S15, the determination result is "YES", and the process proceeds to step S17.

  In step S17, the CPU 100 causes the timer 100b to stop measuring the elapsed time. Then, using the measured elapsed time and the probability distribution data stored in the HD 104, it is determined whether the acknowledgment response R1 has a high probability of corresponding to the original data segment S1 or corresponds to the retransmission data segment S'1. A process is performed to determine whether the probability is high (step S18).

  More specifically, as described in the above <Configuration of server device 10> column, the elapsed time measured in step S17 is calculated as S1-RTT, focusing on the relationship between the two types of probability distribution data in FIG. It is determined whether a certain probability is high or a probability of being O1-RTT is high.

  In a preferred embodiment, the HD 104 stores one type of probability distribution data as shown in FIG. Then, the CPU 100 compares the value of the elapsed time measured by the timer 100b with the probability distribution data, and obtains the appearance probability of S1-RTT corresponding to the measured value. Further, a value set to 100a as a timeout value triggered by retransmission data segment transmission is added to the value measured by the timer 100b, and the appearance probability of O1-RTT corresponding to the measured value + timeout value is calculated. As a result, if the probability of occurrence of S1-RTT is equal to or greater than the probability of occurrence of O1-RTT, the acknowledgment R1 is considered to have been transmitted for the retransmitted data segment. Conversely, if the former is smaller than the latter, the acknowledgment R1 is considered to have been sent for the original data segment.

  In another aspect, the HD 104 stores two types of probability distribution data as shown in FIG. The CPU 100 adds the value set to 100a as the timeout value that triggered the retransmission data segment transmission to the value measured by the timer 100b, and the probability distribution data indicating the occurrence probability of O1-RTT (solid line in FIG. 4). Then, the appearance probability of O1-RTT corresponding to the measured value + timeout value is obtained. Similarly, the measured value + timeout value is compared with probability distribution data (dotted line in FIG. 4) indicating the appearance probability of S1-RTT, and the appearance probability of S1-RTT corresponding to the measured value + timeout value is obtained. If the occurrence probability obtained by referring to the probability distribution data of S1-RTT is equal to or greater than the occurrence probability obtained by referring to the probability distribution data of O1-RTT, the acknowledgment R1 is included in the retransmission data segment. Is assumed to have been sent. Conversely, if the former is smaller than the latter, the acknowledgment R1 is considered to have been sent for the original data segment.

When the probability distribution data of S1-RTT and the probability distribution data of O1-RTT shown in FIG. 4 are used, the intersection of the two curves (time t2 in FIG. 4) is set as the threshold, and the timeout value is added to the measurement value of the timer 100b. If the calculated value is smaller than the threshold value, the acknowledgment R1 may be regarded as corresponding to the original data segment, and if equal to or larger than the threshold value, it may be regarded as corresponding to the retransmission data segment.
For example, as shown in FIG. 4, if the value obtained by adding the timeout value to the elapsed time measured in step S17 is t'1, it can be understood that the probability of being O1-RTT is high.

  Accordingly, the CPU 100 determines that the probability that the confirmation response R1 corresponds to the original data segment S1 is high, and proceeds to step S21. In this case, since the acknowledgment response to S1 is received, the window slides by one data segment, and one subsequent original data segment can be transmitted. Therefore, in step S21, the original data segment having the subsequent sequence number 3000 (S4 in FIG. 5) is transmitted. Then, the process proceeds to step S11 and step S12.

Here, returning to FIG. 5, as a process corresponding to S21 in FIG. 6, the original data segment S4 is transmitted from the server device 10 to the client device 50. Then, the process returns to step S11, and measurement of the elapsed time by the timer 100a is started.
Thereafter, the CPU 100 repeats the determinations in steps S12 and S13 described above until an acknowledgment is received until the timer 100a times out.
On the other hand, while the above processing is being performed in the server device 10, the original data segment S2 is received by the client device 50, and the acknowledgment R2 having the acknowledgment number 2000 corresponding to the original data segment S2 is sent to the server device 10. Sent to 10.

When the acknowledgment R2 having the acknowledgment number 2000 reaches the server device 10, the determination result of the CPU 100 of the server device 10 in step S12 is “YES” in FIG. As a result, the window further slides by one data segment, and the original data segment having the subsequent sequence number 4000 (S5 in FIG. 5) is transmitted (step S10).
As a process corresponding to this step, the original data segment S5 is transmitted from the server device 10 to the client device 50 in FIG.

As described above, the data segment is transmitted from the server device 10 to the client device 50.
When the CPU 100 determines in step S19 in FIG. 6 that the probability that the data corresponds to the retransmission data segment is high, the process proceeds to step S20. In step S20, it is determined whether or not there is a subsequent retransmission data segment. When it is determined that there is, the process proceeds to step S14, and when it is determined that there is no process, the process proceeds to step S21.

[Second embodiment]
Next, another example of performing packet communication between the server device 10 and the client device 50 having the same configuration as in the first embodiment will be described with reference to FIGS.

In the second embodiment, in addition to the probability distribution used in the first embodiment (hereinafter referred to as probability distribution 1), another probability distribution (hereinafter referred to as probability distribution 2) is used in combination, and the determination in step S18 in FIG. Perform processing.
Probability distribution 2 is exemplified in FIG. 7, and starts from the transmission time point of the retransmission data segment due to the occurrence of the timeout, and receives the acknowledgment when the first acknowledgment received after the timeout is an acknowledgment for the original data segment. This indicates the appearance probability of the time required to perform the operation (hereinafter referred to as a temporary round trip time). On the other hand, the probability distribution 1 shows the appearance probability of the round trip time obtained from the transmission time of the original data segment as a starting point. Not always the same. For this reason, in the determination processing of step S18, when determining the probability that the confirmation response corresponds to the original data segment, the reliability based on the probability distribution 2 is higher than that based on the probability distribution 1. A judgment result is obtained.
Under such circumstances, in the present embodiment, when the probability that the received acknowledgment is the acknowledgment for the retransmitted data segment is obtained, the probability distribution 1 is used, and when the probability that the received acknowledgment is the acknowledgment for the original data segment is obtained. Uses the probability distribution 2.

(1. Configuration)
Since the configuration of the second embodiment is the same as that of the first embodiment in the main part, only the additional functions used in the second embodiment will be described.
First, as in the first embodiment, the CPU 100 of the server device 10 according to the second embodiment has two types of timers 100a and 100b (see FIG. 1). Both the timers 100a and 100b have the same function as the first embodiment. However, in the second embodiment, the value of the elapsed time measured by the timer 100b is not only used for the determination process in step S18, but also for the probability distribution. 2 is also used for learning.

The HD 104 stores probability distribution 2 data in addition to probability distribution 1 data. Similarly, the HD 104 has a learning program for learning the probability distribution 1 described in the first embodiment (hereinafter, referred to as a first learning program), and also has the CPU 100 perform a learning process for the probability distribution 2. A learning program (hereinafter, referred to as a second learning program) is stored.
In the process according to the second learning program, when it is determined that the acknowledgment received after resending the data segment is highly likely to correspond to the original data segment, the CPU 100 determines the value of the elapsed time measured by the timer 100b. Is used to update the value of the probability distribution 2 data stored in the HD 104. Specifically, the probability that the elapsed time measured by the timer 100b is a provisional round trip time for the original data segment is determined, and the value of the probability distribution 2 data stored in the HD 104 is sequentially determined using the determined probability. Update.
In addition, as a specific method of updating the probability distribution 2 data, various methods are conceivable, such as performing recalculation together with a value measured by the timer 100b in the past, and can be arbitrarily changed.

  Here, the above-mentioned "case where it is determined that the acknowledgment received after resending the data segment has a high probability that it corresponds to the original data segment" is as follows. When the CPU 100 receives both the acknowledgment for the original data segment (R1 in FIG. 5) and the acknowledgment for the retransmitted data segment (R'1 in FIG. 5) after transmitting the retransmission data segment, the acknowledgment R1 is received first, The acknowledgment R'1 will be received later. The determination process of step S18 in FIG. 6 is performed when the acknowledgment R1 is received. At this time, it cannot be concluded that it is possible to estimate which segment the acknowledgment is for. When the acknowledgment R'1 is received later, it can be determined that the acknowledgment received earlier is for the original data segment and the acknowledgment received later is for the retransmitted data segment. As described above, the correctness of the result obtained by the determination processing in step S18 is determined only when the second confirmation response R'1 is received, and therefore, when the second confirmation response is received, the CPU 100 The probability distribution 2 data is updated using the measurement value of 100b.

  In practice, as shown in FIG. 5, the retransmission data segment S′1 is received by the client device 50 after receiving the original data segments S1, S2, S3, and is therefore included in the acknowledgment R′1. The acknowledgment number is different from the first acknowledgment R1. Therefore, the CPU 100 of the server device 10 updates the probability distribution 2 data when two acknowledgments (R3 and R'1 in FIG. 5) having the same acknowledgment number are received.

  Note that, similarly to the learning process of the probability distribution 1 described in the first embodiment, the probability distribution 2 may be learned by using any one of the first to third methods. The first method is to provide a period for transmitting a test data segment after starting communication between the server device 10 and the client device 50, and to provide a temporary round acquired by the timer 100b when the test data segment is retransmitted. This is a method of generating probability distribution 2 data based on a trip time.

  The second method is to start learning the probability distribution 2 after starting communication. In this case, since the probability distribution 2 data does not exist for a while after the start of communication, the above-described determination is not performed, and the determination process is performed after a certain time has elapsed. Here, when the retransmission data segment is transmitted, the measurement by the timer 100b is started, and the elapsed time until the first acknowledgment is received is measured. Then, when the second acknowledgment having the same acknowledgment number is received, the probability distribution data is generated using the value measured by the timer 100b.

  The third method is a method in which after the communication starts, the above determination is started immediately using the probability distribution 2 data measured in advance using a similar communication environment. Specifically, by using the time stamp option described in the related art section, it is determined whether the acknowledgment received after transmitting the retransmission data segment is for the original data segment or for the retransmission data segment, and the probability distribution is determined in advance. Learn 2 Then, the learning process of the probability distribution 2 may be performed by updating the probability distribution 2 data after the communication starts. This time stamp option is applicable to both the transmission period of the test data segment in the first method and the period in which only learning is performed immediately after the start of communication in the second method.

  Using the probability distribution 2 learned in this way and the probability distribution 1 described in the first embodiment, the following can be understood. First, when communication is normally performed, a probability distribution of a round trip time (hereinafter, referred to as S2-RTT) from transmission of a retransmission data segment to reception of an acknowledgment for the retransmission data segment is probability distribution 1 (FIG. 3). Further, a provisional round trip time (hereinafter, referred to as O2-pRTT) from the transmission of the retransmission data segment to the reception of the acknowledgment for the original data segment is represented by probability distribution 2 (FIG. 7). Therefore, in FIG. 8, the probability distribution of O2-pRTT (probability distribution 2) is indicated by a solid line, and the probability distribution of S2-RTT (probability distribution 1) is indicated by a dotted line, starting from the time when the retransmission data segment is transmitted. In this figure, time t'2, which is the intersection of the two curves, is the time when the appearance probability of O2-pRTT is equal to the appearance probability of S2-RTT. In the illustrated example, the probability that an acknowledgment for the original data segment is received is high until the time t'2 elapses after the retransmission data segment is transmitted, and after the time t'2 elapses, the retransmission is performed. It can be seen that the probability that an acknowledgment for the data segment is received is high. In the present embodiment, utilizing this property, the determination in step S18 in FIG. 6 is performed based on both the probability distribution 1 data and the probability distribution 2 data.

  When there are a plurality of client devices 50 (not shown), the CPU 100 learns the above-described probability distribution 2 for each IP address of the client device 50.

(2. Operation)
Next, the operation of the second embodiment will be described.
The operation of the second embodiment is the same as that of the first embodiment except for the operation related to the determination processing in step S18 in FIG. 6, and therefore, only the part related to the determination processing will be described.

  The determination process in step S18 is specifically performed as follows. The CPU 100 compares the elapsed time measured by the timer 100b with each of the probability distribution 1 data and the probability distribution 2 data. Then, the occurrence probabilities corresponding to the values measured by the timer 100b are determined, and if the appearance probability determined by referring to the probability distribution 1 data is equal to or greater than the occurrence probability determined by referring to the probability distribution 2 data, the confirmation is performed. The response R1 is considered to have been transmitted for the retransmission data segment. Conversely, if the former is smaller than the latter, the acknowledgment R1 is considered to have been sent for the original data segment.

As another method, an intersection (t'2 in FIG. 8) of the two curves indicating each of the probability distribution 1 data and the probability distribution 2 data is determined as a threshold, and if the measured value of the timer 100b is smaller than the threshold, a confirmation is made. The response R1 may be considered to correspond to the original data segment, and if it is equal to or greater than the threshold, it may be considered to correspond to the retransmitted data segment.
In the example of FIG. 8, since the acknowledgment R1 has been received at the time t′1, it is determined that the acknowledgment has a high probability of corresponding to the original data segment (step S19; NO), and the next step Proceed to.

  Further, in the present embodiment, after performing the determination processing in step S18, when two acknowledgments having the same acknowledgment number are received, processing for updating the probability distribution 2 data is performed. This is because, as described above, the determination process in step S18 is determined only by receiving two identical confirmation responses.

[Modification]
As described above, the embodiments of the present invention have been described, but the present invention can be embodied in various other forms without departing from the main features. Note that, for example, the following modifications are possible.

  In the above embodiment, the packet communication is performed according to the TCP. However, the present invention is applicable to communication protocols other than the TCP as long as the protocol performs retransmission control of a data segment or a data block. When another protocol is used, the data segment may be transmitted using a window similar to the sliding window method adopted by TCP.

  In the above embodiment, the server device 10 has learned the above probability distributions 1 and 2 for each IP address. However, the probability distribution of the round trip time may be learned for each subnetwork. When there are a plurality of mobile packet communication networks 30, the probability distributions 1 and 2 may be learned for each mobile packet communication network. This is because, since the communication environment differs depending on the communication network, a more reliable determination result can be obtained by learning the probability distribution for each communication network and making a determination based on the probability distribution.

  The learning of the probability distributions 1 and 2 performed by the server device 10 and the estimation of the data segment in response to the confirmation response may be performed by another server device. In such a configuration, upon receiving the acknowledgment after transmitting the retransmission data segment, the server device 10 inquires of the other server devices whether the acknowledgment is for the original data segment or the retransmission data segment. The other server device performs the determination process performed by the server device 10 described above, and transmits the determination result to the server device 10. The server device 10 can estimate a data segment for the acknowledgment based on the transmitted determination result.

  In the embodiment described above, the client device 50 performs the packet communication with the server device 10 via the communication terminal 40. However, the client device 50 may have a wireless communication function and perform packet communication with the server device 10 via the mobile packet communication network 30 and the Internet 20 without using the communication terminal 40.

FIG. 1 is a block diagram illustrating a configuration of a communication system 1 according to first and second embodiments of the present invention. FIG. 2 is a block diagram illustrating a configuration of a server device 10 according to the first and second embodiments. It is a figure showing roughly the probability distribution concerning a 1st embodiment (probability distribution 1 in a 2nd embodiment). It is a figure for explaining a difference of appearance probability using a probability distribution concerning a 1st embodiment. 9 is a sequence chart illustrating an example of a case where packet communication is performed between the server device 10 and the client device 50 according to the first and second embodiments. 9 is a flowchart illustrating packet transmission and reception operations of the server device 10 according to the first and second embodiments. It is a figure showing roughly the probability distribution 2 concerning a 2nd embodiment. It is a figure for explaining a difference of appearance probability using probability distribution 1 and probability distribution 2 concerning a 2nd embodiment. 10 is a sequence chart showing an example of a case where packet communication is performed between a conventional server device 10 ′ and a client device 50.

Explanation of reference numerals

  DESCRIPTION OF SYMBOLS 1 ... Communication system, 10 ... Server device, 100 ... CPU, 100a ... Timer, 100b ... Timer, 101 ... Communication part, 102 ... RAM, 103 ... ROM , 104 HD, 105 storage unit, 111 bus, 20 Internet, 30 mobile packet communication network, 40 communication terminal, 50 client device.

Claims (15)

A first generation process of generating first probability distribution information indicating a probability of occurrence of a round trip time of a data block;
A transmitting step of sequentially transmitting a plurality of data blocks from the transmitting device to the receiving device,
Among the plurality of data blocks transmitted to the receiving device, retransmitting a data block that has not received an acknowledgment signal from the receiving device, and activating a timing unit for measuring elapsed time,
When receiving an acknowledgment signal from the receiving device, an obtaining step of obtaining the elapsed time measured by the timing unit,
Estimating, based on the acquired elapsed time value and the first probability distribution information, whether the acknowledgment signal is an acknowledgment for one of the plurality of data blocks transmitted in the transmission process. Having a process,
The transmitting step includes transmitting a data block subsequent to the plurality of data blocks transmitted in the transmitting step when the acknowledgment signal is estimated to be an acknowledgment for one of the transmitted data blocks. A transmission control method in a communication network. After transmitting the subsequent data block, if it is determined that the estimation performed in the estimation process is correct, the value of the elapsed time acquired in the acquisition process is used to indicate the appearance probability of the elapsed time. A second generation step of generating probability distribution information of No. 2;
The transmission control method according to claim 1, wherein in the estimation step, the estimation is performed based on a value of the elapsed time acquired in the acquisition step and the first and second probability distribution information. First generating means for generating first probability distribution information indicating a probability of occurrence of a round trip time of a data block;
Transmitting means for sequentially transmitting a plurality of data blocks to the receiving device,
A timing means for measuring elapsed time;
Among the plurality of data blocks transmitted to the receiving device, retransmitting a data block that has not received an acknowledgment signal from the receiving device, and activating the timing unit.
When receiving an acknowledgment signal from the receiving device, acquiring means for acquiring the elapsed time measured by the clock means,
Estimating whether the acknowledgment signal is an acknowledgment for one of the plurality of data blocks transmitted by the transmission unit based on the acquired elapsed time value and the first probability distribution information. Means,
The transmitting unit transmits a data block subsequent to the plurality of data blocks transmitted by the transmitting unit when the acknowledgment signal is estimated to be an acknowledgment response to one of the plurality of transmitted data blocks. A communication device characterized in that: When transmitting the plurality of data blocks to a plurality of the receiving device,
The first generation unit generates the first probability distribution information for each of a plurality of connections set between the plurality of reception devices,
The estimating unit, when receiving an acknowledgment signal from any of the plurality of receiving devices, the value of the elapsed time acquired by the acquiring unit and the connection set between any of the receiving devices The communication device according to claim 3, wherein the estimation is performed based on the generated first probability distribution information. When transmitting the plurality of data blocks to a plurality of receiving devices each belonging to a different sub-network,
The first generation means generates the first probability distribution information for each of the sub-networks,
The estimating unit, when receiving an acknowledgment signal from any of the plurality of receiving devices, the value of the elapsed time acquired by the acquiring unit and the one generated for the subnetwork to which any of the receiving devices belongs. The communication device according to claim 3, wherein the estimation is performed based on first probability distribution information. The first probability distribution information is statistical information experimentally obtained before transmitting the plurality of data blocks to the receiving device,
The first generation unit generates the first probability distribution information by starting transmission of the plurality of data blocks and then updating the statistical information obtained on a trial basis. Item 6. The communication device according to any one of Items 3 to 5. The first probability distribution information is statistical information obtained in advance before starting communication for transmitting the plurality of data blocks to the receiving device,
The first generation unit generates the first probability distribution information by starting the communication for transmitting the plurality of data blocks and then updating the statistical information obtained in advance. The communication device according to claim 3. After transmitting the subsequent data block, when it is determined that the estimation performed by the estimating unit is correct, a value indicating the appearance probability of the elapsed time using a value of the elapsed time acquired by the acquiring unit is used. A second generation means for generating the probability distribution information of the second,
The communication device according to claim 3, wherein the estimating unit performs the estimation based on the value of the elapsed time acquired by the acquiring unit and the first and second probability distribution information. The second generation unit determines that the estimation performed by the estimation unit is correct when the acknowledgment signal requesting the subsequent data block is received at least twice from the reception device. Item 9. The communication device according to item 8. When transmitting the plurality of data blocks to a plurality of the receiving device,
The first and second generating means generate the first and second probability distribution information for each of a plurality of connections set with the plurality of receiving devices,
The estimating unit is configured to determine a value of the elapsed time acquired by the acquiring unit when an acknowledgment signal is received from any of the plurality of receiving devices, and a connection set between any of the receiving devices. The communication device according to claim 8, wherein the estimation is performed based on the generated first and second probability distribution information. When transmitting the plurality of data blocks to a plurality of receiving devices each belonging to a different sub-network,
The first and second generating means generate the first and second probability distribution information for each of the sub-networks,
The estimating means includes a value of the elapsed time acquired by the acquiring means when an acknowledgment signal is received from any of the plurality of receiving devices, and a value of the elapsed time generated for a subnetwork to which any of the receiving devices belongs. The communication device according to claim 8, wherein the estimation is performed based on first and second probability distribution information. The first and second probability distribution information is statistical information experimentally obtained before transmitting the plurality of data blocks from the communication device to the receiving device,
The first and second generating units may start transmitting the plurality of data blocks and then update the statistical information obtained on a trial basis to update the first and second probability distribution information. The communication device according to claim 8, wherein the communication device generates the communication device. The first and second probability distribution information is statistical information obtained in advance before starting communication for transmitting the plurality of data blocks from the communication device to the receiving device,
The first and second generating means start the communication for transmitting the plurality of data blocks and then update the first and second probability distribution information by updating the statistical information obtained in advance. The communication device according to claim 8, wherein the communication device generates the communication device. Generating means for generating first probability distribution information indicating an appearance probability of a round trip time of a data block;
Transmitting means for sequentially transmitting a plurality of data blocks from the transmitting device to the receiving device,
A timing means for measuring elapsed time;
Among the plurality of data blocks transmitted to the receiving device, retransmitting a data block that has not received an acknowledgment signal from the receiving device, and activating the timing unit.
When receiving an acknowledgment signal from the receiving device, acquiring means for acquiring the elapsed time measured by the clock means,
Estimating whether the acknowledgment signal is an acknowledgment for one of the plurality of data blocks transmitted by the transmission unit based on the acquired elapsed time value and the first probability distribution information. Means,
The transmitting unit transmits a data block subsequent to the plurality of data blocks transmitted by the transmitting unit when the acknowledgment signal is estimated to be an acknowledgment response to one of the plurality of transmitted data blocks. A communication system characterized by: A process of generating first probability distribution information indicating a probability of occurrence of a round trip time of a data block;
A transmission process of sequentially transmitting a plurality of data blocks to the receiving device;
Among the plurality of data blocks transmitted to the receiving device, retransmitting a data block that has not received an acknowledgment signal from the receiving device, and a process of activating a timing unit for measuring elapsed time,
When receiving an acknowledgment signal from the receiving device, a process of acquiring the elapsed time measured by the timing unit,
Processing for estimating whether or not the acknowledgment signal is an acknowledgment for one of the plurality of data blocks transmitted by the transmission processing, based on the acquired value of the elapsed time and the first probability distribution information When,
And transmitting a data block subsequent to the plurality of data blocks transmitted by the transmission processing when the acknowledgment signal is estimated to be an acknowledgment response to one of the plurality of transmitted data blocks. Program to be executed.

Images