JP2019012866A - Control system, available band width estimation system, device, method and program - Google Patents

Abstract

To estimate an available band width on a network, where a protocol with no receipt acknowledgement cannot be used, with high accuracy by a simple configuration of application level.SOLUTION: In the inventive control system, a transmission side device 700 includes warming-up packet stream transmission means 701 for transmitting a warming-up packet stream to a reception side device 800, before transmission of a first packet stream such as a packet train, by using a protocol performing congestion control by receipt acknowledgement and a window, window size estimation means 703 for estimating the lowest value of the congestion window size in a protocol stack after transmission of the warming-up packet stream, based on the transmission time of each warming-up packet contained in the warming-up packet stream, and the reception time in the reception side device 800, acquired as a result of transmission of the warming-up packet stream, and number of warming-up packets determination means 702 for determining the number of warming-up packets contained in a warming-up packet stream to be retransmitted, based on the estimation result of the lowest value of the congestion window size.SELECTED DRAWING: Figure 8

Description

  The present invention relates to an available bandwidth estimation system that estimates an available bandwidth, a control system, a communication device, a control method, and a control program for controlling a congestion window size.

  An available bandwidth (also referred to as an available bandwidth) of a communication line is a free bandwidth obtained by subtracting other traffic (hereinafter referred to as cross traffic) flowing through the network from the physical bandwidth of the bottleneck link of the communication line. For example, when the physical bandwidth of the bottleneck link of the communication line is 100 Mbps and the cross traffic is 30 Mbps, the available bandwidth is 100-30 = 70 Mbps.

  Estimating the current value of the available bandwidth that fluctuates from time to time is important in video chat, videophone, videoconferencing, and the like in which video data is transmitted bidirectionally between terminals. The reason is that the total packet transmission rate and cross-traffic exceeds the physical bandwidth of the bottleneck link of the communication line by limiting the packet transmission rate when transmitting video data to less than the estimated available bandwidth. This is because degradation of video quality due to packet transmission delay can be prevented.

  Regarding the available bandwidth estimation technology, for example, in Patent Document 1, a plurality of estimated packets that are available bandwidth estimation packets are collectively referred to as one packet train, and each estimated packet included in the packet train is assigned at a predetermined time interval. A method is described in which the available bandwidth is estimated on the receiving device side by transmitting. In the method described in Patent Document 1, the transmission side device prepares a packet train including a plurality of estimated packets whose packets gradually increase, and transmits the estimated packets included in the packet train in order at equal intervals. Then, the receiving side apparatus estimates the available bandwidth by detecting a change in the reception interval of each estimated packet.

  According to this method, the transmission rate of each estimated packet increases linearly within the packet train. When the packet train passes through the network, if the estimated packet transmission rate exceeds the available bandwidth of the network, the packet is temporarily queued in a device such as a router or switch on the network (hereinafter referred to as the queue). The delay due to ing is called queuing delay). When the queuing delay occurs, the reception interval of the estimated packet is increased in the reception side device than the transmission interval in the transmission side device. Using this property, the receiving side device specifies an estimated packet whose reception interval starts to increase compared to the transmission interval at the transmitting side device, and divides the packet size of the estimated packet by the transmission interval, Calculate available bandwidth.

  Further, for example, in Patent Document 2, the microprocessor of the transmission side apparatus measures the round trip time (RTT) and the packet loss rate while changing the window size of the congestion window, and these measured values are measured. And a method for estimating the available bandwidth from the data size per packet. Further, Patent Document 2 further describes a method for estimating an available bandwidth using a calculation formula that varies depending on whether or not a retransmission timeout (RTO) has occurred.

  Patent Document 3 describes a technique in which a preparatory packet is sent in advance and other traffic is pushed away in a preset time zone such as when traffic congestion is expected, and then an application is transmitted. Yes.

International Publication No. 2014/017140 Pamphlet JP 2014-165809 A Special table 2009-531912

  The method described in Patent Document 1 is preferable because it can easily be estimated in the application layer without monitoring the packet flow in the communication path, and can be easily introduced into various environments. For example, the method described in Patent Document 2 requires operations lower than the application layer, such as changing the congestion window size, measuring the RTT and packet loss rate, and determining whether or not RTO has occurred, and is expensive to introduce. There's a problem. The calculation formula considers only the transmission delay due to congestion control due to retransmission timeout based on the temporal transition of the general number of packets in congestion control. Ingress delay or the like is not taken into account, and it cannot be said that the estimation accuracy of the usable bandwidth is good.

  Note that the method described in Patent Document 1 also has a problem that the estimation accuracy decreases when an error occurs in the transmission timing of the estimated packet from the transmission side device. In order to estimate the available bandwidth with high accuracy, the estimated packets must be continuously transmitted accurately at the timing specified by the application software for calculating the estimated value. For this purpose, a UDP (User Datagram Protocol) packet whose transmission timing can be easily controlled from an application is used as the estimation packet. In order to grasp the temporal transition of the available bandwidth and the average value of the available bandwidth within a certain time or more, the packet train is often transmitted periodically.

  However, the method described in Patent Document 1 cannot be used in a network environment where communication by UDP is not possible due to a firewall or the like. Therefore, in order to cope with a network environment where UDP communication is not possible, consider simply changing the protocol from UDP to TCP (Transmission Control Protocol). In this case, although the communication itself can be performed, the TCP congestion control becomes “disturbance”, and a phenomenon occurs in which continuous transmission of packets is temporarily interrupted in the middle.

  An outline of TCP communication will be described as basic knowledge for explaining this phenomenon. In TCP communication, congestion control is performed by a TCP protocol stack driven by the OS. Hereinafter, the OS including the protocol stack for TCP is referred to as an OS for the purpose of comparison with the application layer. As congestion control, an OS on the TCP transmission side manages a congestion window size (hereinafter referred to as cwnd). In addition, since the variable holding the value of cwnd is an internal variable managed by the OS, it is often impossible to observe directly from the application.

  The OS on the TCP transmission side can continuously transmit packets by the same number as cwnd without waiting for an ACK (acknowledgment) from the protocol stack on the TCP reception side. For example, when the value of cwnd is 10, the OS on the TCP transmission side can continuously transmit up to 10 packets, and when ACK is not returned, continuous transmission is temporarily performed until ACK is returned. Interrupted. When ACK is returned, cwnd is increased, and the continuous transmission which has been temporarily suspended is resumed. In TCP communication, the continuous transmission and the temporary interruption are generally repeated.

  Further, the OS on the TCP transmission side measures the RTT of each packet from the transmission time of each packet and the reception time of the ACK corresponding to the transmission time. Each time the RTT of each packet is measured, the retransmission timeout value (hereinafter referred to as RTO generation time) is calculated according to the following formula described in the standard IETF RFC 6298 (hereinafter referred to as non-patent document 1). -Val)).

  First, when the RTT of the first packet is measured, the RTT exponential weighted moving average value SRTT, the RTT average deviation RTTVAR, and the RTO-val are initialized as follows. The left-pointing arrow represents assignment of the value on the left side to the variable on the right side. Max (A, B) is a function for extracting the larger value of A and B.

・ SRTT ← RTT
・ RTTVAR ← RTT / 2
・ RTO-val ← SRTT + max (G, K * RTTVER)

  Next, every time the RTT of the second and subsequent packets is measured, the SRTT, RTTVAR, and RTO-val are updated as follows.

・ RTTVAR ← (1-beta) * RTTVAR + β * | SRTT-RTT |
・ SRTT ← (1-alpha) * SRTT + α * RTT
・ RTO-val ← SRTT + max (G, K * RTTVAR)

  Note that G, K, α, and β are predetermined constants. The OS on the TCP transmission side activates a retransmission timer that uses this RTO-val as a timeout value when each packet is transmitted.

  Further, in TCP communication, when the OS on the TCP transmission side determines that a packet loss has occurred, the value of cwnd is lowered in order to avoid network congestion. There are two types of determination of occurrence of packet loss: packet loss due to light congestion and packet loss due to heavy congestion. The behavior of the decrease in the value of cwnd differs between them.

  FIG. 10 is an explanatory diagram illustrating an example of occurrence of packet loss due to slight congestion. FIG. 11 is an explanatory diagram showing an example of packet loss caused by severe congestion. In both examples, it is assumed that transmission of 100 packets is requested from the application side of the transmission side device. Further, it is assumed that the value of cwnd before transmission in the transmission side apparatus is 100. In FIGS. 10 and 11, in order to avoid the complexity of the drawings, the RTT corresponding to the round-trip delay from when the packet is intentionally transmitted until the ACK is returned is shortened so that the straight line of each packet exchange I try not to overlap. However, originally, the RTT is often longer than the packet transmission interval. For example, the ACK of the 90th packet is returned before the transmission of the 91st packet in FIG. 11, but originally it is returned after continuously transmitting not only the 90th packet but also a plurality of subsequent packets. Often comes.

  In FIG. 10 showing an example of occurrence of packet loss due to slight congestion, when the 90th packet reaches the receiving side device, an ACK indicating that the 90th packet has reached the transmitting side device is returned. In this example, it is assumed that a packet loss has occurred in the 91st packet. In such a case, when the 92nd packet arrives at the receiving side device, the receiving side device has not received the 91st packet, so a second ACK indicating that the 90th packet has reached the transmitting side device is received. Send back. Next, when the 93rd packet arrives at the receiving side device, the receiving side device returns an ACK indicating that it has reached the 90th time again because the 91st packet has not arrived. The transmission side apparatus determines that a packet loss has occurred in the 91st packet for the first time when ACK indicating that it has reached the 90th is received three times. When the occurrence of a packet loss is detected by receiving a predetermined number of duplicate ACKs, the transmission side apparatus retransmits the 91st packet for which ACK has not returned so far, and changes the value of cwnd. In this case, the transmission side apparatus increases the value of cwnd by, for example, 0.7 times. Note that the number of times depends on the TCP implementation. For example, in the case of CUBIC TCP used as a default in Linux (registered trademark) OS, it is 0.7 times. If the value of cwnd at the time when the occurrence of a packet loss is detected is 190, the transmission side apparatus decreases the value of cwnd from 190 to 133. Hereinafter, such packet loss detected by a predetermined number of duplicate ACKs may be referred to as packet loss due to slight congestion.

  In FIG. 11 showing an example of occurrence of packet loss due to severe congestion, when the 90th packet reaches the receiving side device, an ACK indicating that the 90th packet has reached the transmitting side device is returned. The transmission side device updates the RTO value (RTO-val) every time an ACK is received. In this example, it is assumed that packet loss occurs in the 91st, 92nd, and 93rd packets. Then, before the transmission of the 94th packet, the transmission side device times out the retransmission timer that was started when the 91st packet was transmitted (RTO occurrence). The transmission side apparatus determines that packet loss has occurred in the 91st, 92nd, and 93rd packets for which no ACK has been returned so far due to the occurrence of RTO. When the occurrence of packet loss is detected due to the occurrence of RTO, the transmission side apparatus retransmits the 91st, 92nd, and 93rd packets for which ACK has not returned so far, and changes the value of cwnd. In this case, the transmission side apparatus sets the value of cwnd to 2, for example. The number of values depends on the TCP implementation. If the value of cwnd at the time when the occurrence of a packet loss is detected is 190, the transmission side apparatus greatly reduces the value of cwnd from 190 to 2. Hereinafter, such a packet loss detected by the occurrence of RTO may be referred to as a packet loss due to severe congestion.

  As with cwnd, TCP level RTT, RTO-val, and the like are internal variables managed by the OS, and are often not directly observable from an application.

  Next, a description will be given of a phenomenon in which TCP congestion control becomes “disturbance” and continuous packet transmission is temporarily interrupted.

  Here, a case is considered in which 100 estimated packets are defined as one packet train, and a plurality of packet trains are periodically transmitted. First, it is assumed that a packet loss due to slight congestion occurs during transmission of one packet train. If the packet loss due to slight congestion is about once, it is possible that the value of cwnd at the end of packet train transmission is sufficiently large due to the occurrence of packet loss or ACK reception after the occurrence of packet loss. However, when packet loss due to slight congestion occurs a plurality of times, the value of cwnd at the end of packet train transmission may be less than 100. For example, assume that the value of cwnd at the next packet train transmission is 70. In this case, although continuous transmission is possible up to the 70th estimated packet, continuous transmission of the estimated packet is temporarily interrupted immediately after that. Then, an error occurs in the transmission timing of the estimated packet from the transmission side device, and the estimation accuracy decreases.

  Further, it is assumed that a packet loss due to severe congestion occurs during transmission of one packet train. In this case, the value of cwnd is greatly reduced at the next transmission of the packet train. For example, suppose that the value of cwnd at the time of the next packet train transmission is 2. In this case, although continuous transmission can be performed up to the second estimated packet, immediately after that, the continuous transmission of the estimated packet is temporarily interrupted, and the estimation accuracy also decreases.

  Therefore, when sending a packet train using a protocol such as TCP that has ACK transmission (acknowledgement) and window congestion control, it is important to make the value of cwnd sufficiently large by some method each time. . Otherwise, continuous transmission of the estimated packet is temporarily interrupted during transmission, and the estimation accuracy is reduced.

  Note that the method described in Patent Document 3 is a method of intentionally causing congestion by transmitting a preparation packet, thereby preventing a packet having a lower priority than an application packet from being transmitted. It is not considered at all that cwnd is made sufficiently large to a value that allows continuous transmission of packets of an application. In the first place, since Patent Document 3 does not require high-accuracy continuous transmission of application packets, it is not necessary.

  Accordingly, in view of the above-described problems, the present invention provides an available bandwidth estimation system capable of accurately estimating the available bandwidth on a network that cannot use a protocol without reception confirmation with a simple configuration at the application level, and An object of the present invention is to provide a control system, a communication device, a control method, and a control program for controlling the congestion window size.

  In addition, the present invention is not limited to the estimation of the available bandwidth, and when a predetermined packet sequence is transmitted using a protocol having reception confirmation and congestion control by a window, a congestion window is previously obtained with a simple configuration at an application level. An object of the present invention is to provide a control system, a control method, and a control program that can increase the size of the packet sequence to a size that allows continuous transmission of the packet sequence.

  A control system according to the present invention includes a transmission side device that is a communication device on the transmission side and a reception side device that is a communication device on the reception side. The transmission side device has a protocol for performing reception confirmation and congestion control by a window. A plurality of warm-up packets, which are preparation packets transmitted using the protocol, are collected before transmission of the first packet sequence composed of a plurality of packets that are used and requested to be transmitted at a predetermined transmission interval. A warm-up packet sequence transmitting means for transmitting the warm-up packet sequence to the receiving side device; a warm-up packet number determining means for determining the number of warm-up packets to be included in the warm-up packet sequence; Based on the transmission time of each warm-up packet included in the warm-up packet sequence acquired as a result and the reception time at the receiving device, the protocol stack after transmission of the warm-up packet sequence Window size estimation means for estimating the minimum value of the congestion window size, and the warm-up packet number determination means is included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. The number of warm-up packets is determined.

  The available bandwidth estimation system according to the present invention includes a transmission-side apparatus that is a transmission-side communication apparatus and a reception-side apparatus that is a reception-side communication apparatus, and the transmission-side apparatus performs reception confirmation and congestion control by a window. A plurality of warm-up packets, which are preparation packets transmitted using the protocol, are transmitted before the transmission of the first packet sequence including a plurality of packets requested to be transmitted at a predetermined transmission interval using the protocol. A warm-up packet sequence transmitting means for transmitting the combined warm-up packet sequence to the receiving side device; a warm-up packet number determining means for determining the number of warm-up packets included in the warm-up packet sequence; Based on the transmission time of each warm-up packet included in the warm-up packet sequence acquired as a result of transmission and the reception time at the receiving device, the protocol protocol after transmission of the warm-up packet sequence The number of warm-up packets to be included in the warm-up packet sequence to be retransmitted is determined based on the window size estimation means for estimating the minimum value of the congestion window size and the estimation result of the minimum value of the congestion window size. A warm-up packet number determining means; and a first packet string transmitting means for transmitting the first packet string to the receiving side apparatus, wherein the receiving side apparatus is configured to transmit the first packet string based on the reception result of the first packet string. A warm-up packet that includes retransmissions based on the estimation result of the minimum value of the congestion window size, including a usable bandwidth estimation unit that estimates a usable bandwidth in a communication path between the device and the receiving-side device. The number of warm-up packets included in the column is determined.

  The communication apparatus according to the present invention uses a protocol in which reception confirmation and congestion control by a window are performed, and before the transmission of the first packet sequence including a plurality of packets requested to be transmitted at a predetermined transmission interval. And a warm-up packet train transmitting means for sending a warm-up packet sequence in which a plurality of warm-up packets, which are preparatory packets transmitted using a packet, to a predetermined receiving side device, and a warm-up to be included in the warm-up packet train Based on the warm-up packet number determining means for determining the number of packets, the transmission time of each warm-up packet included in the warm-up packet sequence, and the reception time at the receiving-side device, acquired as a result of transmission of the warm-up packet sequence Window size estimation means for estimating the minimum value of the congestion window size in the protocol stack after transmission of the warm-up packet sequence, Determining means characterized by determining based on the estimation result of the minimum value of the congestion window size, the number of warm-up packets to be included in the warm-up packet sequence to retransmit.

  The control method according to the present invention comprises a plurality of packets in which a transmission side apparatus, which is a transmission side communication apparatus, uses a protocol in which reception confirmation and congestion control by a window are performed, and transmission is requested at a predetermined transmission interval. Before transmission of the first packet sequence, a warm-up packet sequence in which a plurality of warm-up packets, which are preparation packets transmitted using the protocol, are gathered is transmitted to a predetermined receiving device, Based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device acquired as a result of transmission of the packet sequence, the congestion window size of the protocol stack after transmission of the warm-up packet sequence Estimate the minimum value and determine the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. When the determined number of warm-up packets is 1 or more, the warm-up packet sequence including the determined number of warm-up packets is retransmitted before transmitting the first packet sequence. .

  The control program according to the present invention uses a protocol in which reception confirmation and congestion control by a window are performed on a computer, and before transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. , A process of transmitting a warm-up packet sequence in which a plurality of warm-up packets, which are preparation packets transmitted using the protocol, are gathered to a predetermined receiving device, and acquired as a result of transmission of the warm-up packet sequence, Processing for estimating the minimum value of the congestion window size in the protocol stack after transmission of the warm-up packet sequence based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device, and the congestion window The number of warm-up packets to be included in the warm-up packet sequence to be retransmitted is determined based on the estimation result of the minimum size. Characterized in that to perform the management.

  According to the present invention, it is possible to accurately estimate the available bandwidth on a network in which a protocol without reception confirmation cannot be used with a simple application level configuration. In addition, according to the present invention, when a predetermined packet sequence is transmitted using a protocol having reception confirmation and congestion control by a window, the congestion window is continuously transmitted in advance with a simple configuration at an application level. Can be increased to a possible size.

It is a network block diagram which shows an example of the network structure of the available band estimation system of 1st Embodiment. It is a block diagram which shows the more detailed structural example of the transmission side apparatus 1 and the reception side apparatus 2 which are included in an available band estimation system. It is explanatory drawing which shows the example of the data structure of the data which the data storage means 17 memorize | stores. It is explanatory drawing which shows the example of the data structure of the data which the data storage means 22 memorize | stores. It is a flowchart which shows an example of operation | movement of the available band estimation system of 1st Embodiment. It is explanatory drawing of the determination method of the packet loss by the mild congestion in the application layer of the transmission side apparatus. It is explanatory drawing of the determination method of the packet loss by heavy congestion in the application layer of the transmission side apparatus. It is a block diagram which shows the outline | summary of the control system by this invention. It is a block diagram which shows the other structural example of the control system by this invention. It is explanatory drawing which shows the example of packet loss generation | occurrence | production by slight congestion. It is explanatory drawing which shows the example of packet loss generation | occurrence | production by heavy congestion.

  Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a network configuration diagram illustrating an example of a network configuration of an available bandwidth estimation system according to the first embodiment of this invention. In the usable bandwidth estimation system shown in FIG. 1, a transmission side device 1 and a reception side device 2 are connected via a network 3.

  The transmission side device 1 and the reception side device 2 are a personal computer (PC), a portable computer (PDA), a mobile phone, a smartphone, a tablet terminal, a landline phone, a street multimedia terminal, an in-vehicle terminal, a TV with a network connection function, respectively. It may be a set-top box with a network connection function, a game machine, a printer with a network connection function, a scanner with a network connection function, or another similar device having a function of exchanging information with the outside.

  In addition, a device (not shown) other than the transmitting device 1 and the receiving device 2 may be connected to the network 3, and cross traffic may flow between these devices (not shown).

  Next, an outline of the present embodiment will be described. In the usable bandwidth estimation system according to the present embodiment, first, the transmission side apparatus 1 collects a plurality of TCP packets and preparation packets (hereinafter referred to as warm-up packets) as one group before transmitting the packet train. Send warm-up packet train. Then, the transmission-side apparatus 1 calculates the reception interval of each warm-up packet using the reception time stored in the response packet for each warm-up packet in the warm-up packet sequence, and measures the application level RTT. . Then, the transmission side apparatus 1 estimates the TCP level RTO-val from the measured application level RTT, and based on the reception interval, the RTT, and the RTO-val described above, severe congestion or minor congestion at the TCP level. It is determined whether or not packet loss has occurred. Here, when it is determined that the above packet loss has occurred, the transmitting-side apparatus 1 adjusts the number of warm-up packets to the minimum necessary number according to the degree of congestion, and then reopens the warm-up packet sequence. Send. As a result, cwnd can be increased to a sufficiently large value before transmission of the packet train.

  In addition, the transmission-side apparatus 1 of the present embodiment also calculates the reception interval of each estimated packet, measures the application level RTT, and estimates the TCP level RTO-val when transmitting the packet train. The number of warm-up packets in the next warm-up packet sequence may be adjusted based on the occurrence or non-occurrence of packet loss estimated based on the above.

  It is determined whether a packet loss has occurred at the TCP level both when sending a warm-up packet train and when sending a packet train, and the warm-up packet to be sent next depends on whether or not packet loss has occurred and the degree of congestion By adjusting the number, the number of warm-up packets transmitted can be suppressed as a whole. In the present embodiment, a case where the protocol used for packet train transmission is TCP will be described as an example. However, the protocol is not limited to TCP, and any protocol that has reception confirmation and congestion control by a window may be used.

  FIG. 2 is a block diagram illustrating a more detailed configuration example of the transmission side device 1 and the reception side device 2 included in the available bandwidth estimation system.

  The transmission side device 1 includes a packet transmission unit 10, a packet sequence generation unit 11, a warm-up packet number determination unit 12, a reception interval calculation unit 13, a timer value estimation unit 14, a packet loss determination unit 15, a congestion The window size estimation means 16 and the data storage means 17 may be included.

  The packet transmission means 10 transmits a warm-up packet train and a packet train. Hereinafter, when it is not necessary to particularly distinguish each warm-up packet in the warm-up packet sequence from each estimated packet in the packet train, it may be referred to as “each packet in the packet sequence” or simply “each packet”.

  The packet sequence generation unit 11 generates a warm-up packet sequence and a packet train.

  The warm-up packet number determining means 12 determines the number of warm-up packets included in the warm-up packet sequence.

  The reception interval calculation means 13 calculates the reception interval of each packet in the packet sequence based on the reception time of the packet recorded by the reception side device 2.

  The timer value estimation means 14 calculates the application level RTT from the application level transmission time of each packet and the reception time of the response packet for each packet, and then RTO-val which is the timer value of the TCP level RTO. Is estimated.

  The packet loss determination unit 15 uses the reception interval of each packet in the packet sequence, the calculated application level RTT, and the estimated RTO-val to packet loss due to heavy congestion or light congestion at the TCP level. It is determined whether or not has occurred.

  The congestion window size estimation means 16 estimates the value of cwnd based on the determination result of the occurrence of packet loss.

  The data storage unit 17 stores various data. FIG. 3 is an explanatory diagram showing an example of the data structure of data stored in the data storage unit 17. As shown in FIG. 3, the data storage means 17 includes the number of warm-up packets 30, one or more warm-up packet transmission intervals 31, one or more warm-up packet reception intervals 32, and one or more RTT measurements. A value 40, one or more RTO estimation values 41, a congestion window size estimation value 42, an estimated number of packets 50, one or more estimated packet transmission intervals 51, and one or more estimated packet reception intervals 52. You may remember.

  Here, the number of warm-up packets 30 is the number of warm-up packets included in the warm-up packet sequence. The warm-up packet transmission interval 31 is a transmission interval of each warm-up packet. The warm-up packet reception interval 32 is a reception interval of each warm-up packet. The RTT measurement value 40 is a calculated value of the above-described application level RTT. The RTO estimated value 41 is an estimated value of the above-described TCP level RTO-val. The congestion window size estimated value 42 is an estimated value of cwnd at the TCP level in the transmission side apparatus 1.

  The estimated number of packets 50 is the number of estimated packets included in the packet train. This number is manually set by the system user or administrator. The estimated packet transmission interval 51 is a transmission interval of each estimated packet. The estimated packet reception interval 52 is a reception interval of each estimated packet.

  Further, as shown in FIG. 2, the receiving-side device 2 may include a packet receiving unit 20, an available bandwidth estimation unit 21, and a data storage unit 22.

  When receiving a warm-up packet or an estimated packet, the packet receiving means 20 returns a corresponding response packet together with a TCP level ACK packet.

  The available bandwidth estimation means 21 estimates the available bandwidth. The available bandwidth estimation means 21 can estimate the available bandwidth from, for example, whether there is a change in the reception interval of each packet when receiving a plurality of estimated packets continuously transmitted at a predetermined time interval collected as a packet train. .

  The data storage means 22 stores various data. FIG. 4 is an explanatory diagram showing an example of the data structure of data stored in the data storage unit 22. As shown in FIG. 4, the data storage unit 22 may store an available band estimated value 59.

  The available bandwidth estimated value 59 is an estimated value of the available bandwidth.

  In the present embodiment, the packet transmission unit 10, the packet sequence generation unit 11, the warm-up packet number determination unit 12, the reception interval calculation unit 13, the timer value estimation unit 14, the packet loss determination unit 15, and the congestion window size estimation unit 16 For example, it is realized by an information processing apparatus that operates according to a program such as a CPU provided in the transmission side apparatus 1. Moreover, the data storage means 17 is implement | achieved by the memory | storage device, for example. In addition, the usable bandwidth estimation unit 21 is realized by an information processing device that operates according to a program such as a CPU provided in the reception-side device 2, for example. Moreover, the data storage means 22 is implement | achieved by the memory | storage device, for example.

  Next, the operation of this embodiment will be described with reference to FIG. FIG. 5 is a flowchart showing an example of the operation of the available bandwidth estimation system of the present embodiment. In the example illustrated in FIG. 5, it is assumed that 100 is manually set to the estimated number of packets 50 by the user or administrator of the system. In other words, in this example, it is assumed that 100 is designated as the estimated number of packets included in one packet train. In addition, desired values are set manually for the warm-up packet transmission interval 31 and the estimated packet transmission interval 51 by a user or administrator of this system. Here, the transmission interval of each warm-up packet included in one warm-up packet sequence may not be the same. Similarly, the transmission intervals of the estimated packets included in one packet train may not be the same.

  In the example shown in FIG. 5, first, in step S60, the number of warm-up packets included in the warm-up packet sequence is stored in the data storage unit 17 by the warm-up packet number determining unit 12 of the transmission side apparatus 1. Set to the same value as 30 warm-up packets. When the first packet train is transmitted, the number of packets included in each of the warm-up packet sequence and the packet train may be made the same by overwriting the number of warm-up packets 30 with the estimated number of packets 50.

  In addition, when it is assumed that a TCP mechanism called delayed ACK works in the receiving-side apparatus 2, one TCP level ACK packet is returned for two warm-up packets. The warm-up packet count 30 may be overwritten with a value twice the estimated packet count 50 so that ACKs for the estimated packet count are received by transmission. This is because the value of cwnd after transmission of the warm-up packet sequence is set to 50 or more estimated packets. In this case, the number of warm-up packets included in the warm-up packet sequence is twice the estimated number of packets included in the packet train.

  Next, the packet sequence generation unit 11 generates a warm-up packet sequence including the number of warm-up packets indicated by the number of warm-up packets 30. Then, the packet transmission unit 10 transmits the generated warm-up packet sequence. For example, the packet transmission unit 10 may transmit the warm-up packet sequence by sequentially requesting the OS layer to transmit each warm-up packet included in the warm-up packet sequence. At this time, the packet transmission means 10 holds the transmission time of each warm-up packet, calculates the transmission interval of each warm-up packet from the transmission time of each warm-up packet, and stores the data as the warm-up packet transmission interval 31 It is held in the storage means 17. The transmission time of each warm-up packet may be, for example, a timing when a transmission request is made to the OS layer or a timing when a transmission completion is received from the OS layer.

  In subsequent step S61, first, the packet receiving means 20 of the receiving side apparatus 2 receives the warm-up packet and returns a corresponding response packet. It should be noted that an ACK reply is made at the TCP level before the response packet is sent back, or the response packet itself also serves as a TCP level ACK reply. Here, in each response packet, the reception time of the corresponding warm-up packet in the reception-side device 2 is embedded.

  Next, when the reception interval calculation means 13 of the transmission side apparatus 1 receives the response packet transmitted from the reception side apparatus 2, the reception time of the corresponding warm-up packet and the previous warm-up packet included in the response packet are received. The reception interval of the corresponding warm-up packet is calculated from the reception time of. Then, the reception interval calculation unit 13 holds the calculated reception interval in the data storage unit 17 as the warm-up packet reception interval 32 of the corresponding warm-up packet.

  In the subsequent step S62, the timer value estimating means 14 of the transmission side device 1 determines each warm-up packet from the transmission time of each warm-up packet acquired in step S60 and the reception time of the response packet acquired in step S61. Then, the RTT at the application level is calculated and stored in the data storage means 17 as the RTT measurement value 40. Further, the timer value estimating means 14 estimates the TCP level RTO-val for each warm-up packet from the calculated application-level RTT for each warm-up packet, and stores it in the data storage means 17 as the RTO estimated value 41. The method described in Non-Patent Document 1 may be used for estimating the TCP level RTO-val. At this time, the RTO-val may be obtained using the application level RTT as the TCP level RTT. Also, for example, when the time lag from the application level RTT to the OS layer to the application layer is known, the value obtained by subtracting the value may be used as the TCP level RTT to obtain the RTO-val. .

  In subsequent step S63, the congestion window size estimation means 16 of the transmission side apparatus 1 determines whether or not a packet loss due to slight congestion has occurred. Specifically, the congestion window size estimation unit 16 sequentially checks the warm-up packet reception interval 32 for each warm-up packet from the first packet to the last packet in the warm-up packet sequence, and It is determined whether or not the value of the warm-up packet reception interval 32 of the packet is equal to or greater than a predetermined value (hereinafter referred to as a first threshold value) that is determined to generate packet loss due to slight congestion at the TCP level. The first threshold value is, for example, when occurrence of packet loss due to slight congestion is recognized at the time of reception of 3 duplicate ACKs, and the packet is lost from the transmission time of the packet of interest and 3 duplicate ACKs are received. It may be the time until receiving. More specifically, the RTT measurement value of the warm-up packet currently focused on is a value that is three times the warm-up packet transmission interval 31 of the warm-up packet currently focused or the average warm-up packet transmission interval 31. 40 or an average RTT measurement value 40 may be added.

  Note that the congestion window size estimation unit 16 determines that the warm-up packet reception interval 32 of the subsequent warm-up packet is equal to or less than a predetermined second threshold as a determination condition for occurrence of packet loss due to mild congestion. It may be added to the conditions. Here, the second threshold value may be determined based on an interval of reception time stamps attached to each packet when a plurality of packets are received at a time in the OS layer. For example, it may be an average value, a maximum value, a minimum value, or a predetermined value (for example, 100 μs) according to the processing speed of the OS.

  Hereinafter, conditions for determining packet loss due to slight congestion in the application layer of the transmission-side apparatus 1 will be described more specifically with reference to FIG. In the example shown in FIG. 6, it is assumed that the value of cwnd immediately before the transmission of the warm-up packet sequence is 10 or more, which is the initial value. As shown in FIG. 6, when the 90th warm-up packet arrives without loss, the OS layer of the receiving side device 2 immediately notifies the application side that the 90th warm-up packet has arrived. The application for estimating the available bandwidth of the side device 2 (for example, the packet receiving means 20) immediately recognizes that the 90th warm-up packet has been received. Then, the OS layer of the receiving side device 2 returns a TCP level ACK packet for the 90th warm-up packet.

  When the OS layer of the transmission side apparatus 1 receives the ACK packet for the 90th warm-up packet, it recognizes that the 90th warm-up packet has arrived without loss. Then, the cwnd value is incremented by 1 upon receipt of the ACK. If such a process continues without any problem from the first warm-up packet to the 90th warm-up packet, the cwnd value of the transmitting apparatus 1 at the time when the ACK for the 90th warm-up packet arrives is 10 + 90 = It can be estimated that it is 100 or more.

  However, it is assumed that the 91st warm-up packet transmitted thereafter is lost in the middle of the network. Further, it is assumed that the 92nd warm-up packet arrives without being lost in the network on the way. Since the OS layer of the receiving side apparatus 2 has not received the expected 91st warm-up packet, at this time, the application side is not notified that the warm-up packet has arrived, and the TCP level for the 90th warm-up packet The ACK packet is returned again. If the 93rd warm-up packet transmitted after that arrives without any loss in the network on the way, the OS layer of the receiving-side apparatus 2 has not received the expected 91st warm-up packet. Performs the same processing as when receiving a machine packet. That is, the OS layer of the receiving-side device 2 returns a TCP level ACK packet for the 90th warm-up packet again without notifying that the warm-up packet has arrived at the application side.

  The OS layer of the transmission-side apparatus 1 determines that a packet loss has occurred in the 91st warm-up packet when the TCP level duplicate ACK packet for the 90th warm-up packet has been received three times in total. Resend the th warm-up packet.

  When the retransmitted 91st warm-up packet arrives without loss in the network on the way, the OS layer of the receiving side device 2 immediately receives the 91st, 92nd, and 93rd warm-up packets received so far by the application side. Is transmitted, and the corresponding warm-up packet ACK packet is returned to the transmitting-side apparatus 1. The application (packet receiving means 20) for estimating the available bandwidth of the receiving side apparatus 2 recognizes that it has received the 91st, 92nd, and 93rd warm-up packets only after receiving notification from the OS layer.

  The reception time of each warm-up packet in the transmission side apparatus 1 is obtained from the response packet from the application in the reception side apparatus 2. For this reason, the reception interval between the 89th and 90th warm-up packets that were not lost (the warm-up packet reception interval 32 of the 90th warm-up packet) is between the 91st warm-up packets lost from the 90th. Compared to it (warm-up packet reception interval 32 of the 91st warm-up packet), it is greatly different. More specifically, the reception interval between the 89th and 90th warm-up packets is close to the transmission interval between the 89th and 90th warm-up packets, but the 90th and 91st warm-up packets. As shown in FIG. 6, the reception interval between the RTT measurement value 40 obtained for the 91st warm-up packet is set to a value that is three times the transmission interval between the 90th and 91st warm-up packets. It extends to a value equal to or greater than the value obtained by adding the average RTT measurement value 40.

  From the application side, the 91st, 92nd, and 93rd warm-up packets, which are a predetermined number of warm-up packets after the lost packet that was lost, appear to have been received almost simultaneously, so the 91st and 92nd The reception interval between warm-up packets and the reception interval between the 92nd and 93rd warm-up packets are extremely short values. Therefore, these extremely short reception intervals may be added to the determination conditions for occurrence of packet loss due to slight congestion.

  Returning to FIG. 5, if the congestion window size estimation means 16 determines that a packet loss due to slight congestion has occurred (Yes in step S63), the process proceeds to step S64, and if not determined ( The process proceeds to step S65 (No in step S63).

  In step S64, the congestion window size estimation means 16 estimates the TCP level congestion window size, that is, the value of cwnd in response to the determination result that the packet loss has occurred due to slight congestion. Here, the congestion window size estimation means 16 does not need to estimate a strict value, and may estimate the smallest possible value (hereinafter may be referred to as the minimum value). For example, the minimum value of cwnd is, for example, the initial value of TCP level cwnd, the transmission result of the previous packet sequence, the occurrence of the specified packet loss and the degree of congestion, the transmission result of the current packet sequence, the identification Based on the occurrence of packet loss and the degree of congestion, it follows the control of the congestion window of the protocol stack, that is, by performing the same numerical calculation as that performed for cwnd in the protocol stack. Good.

  For example, consider a case where the congestion window control in the TCP protocol stack is +1 every time an ACK is received, and is multiplied by 0.7 when a packet loss occurs due to slight congestion. The congestion window size estimation means 16 obtains the immediately preceding cwnd from, for example, the initial value of cwnd at the TCP level and the number of ACKs received by the warm-up packet sequence until the occurrence of packet loss, and is 0.7 times larger than that. By doing so, it is possible to predict cwnd after occurrence of packet loss. After that, if the remaining packet sequence is normally received, the value of cwnd at the time when transmission of the warm-up packet sequence is completed can be estimated by adding the amount of increase due to ACK reception of the remaining number of warm-up packets. Note that the value of cwnd before warm-up packet transmission may be set to 0 without exceeding the initial value. If the number of occurrences of packet loss due to slight congestion is 1, the value of cwnd after completion of warm-up packet sequence transmission is approximately warm considering the increase due to ACK reception before and after packet loss. Since it can be assumed that the number of machine packets is 0.7 times or more, it may be obtained by such simple calculation.

  Now, it is assumed that the value of cwnd is estimated to be at least 70 or more when one warm-up packet sequence has been transmitted. In this case, the value of cwnd is set to 100 or more, which is the estimated number of packets before packet train transmission.

  The warm-up packet number determination unit 12 determines the number of packets in the next warm-up packet sequence based on the congestion window size estimation result by the congestion window size estimation unit 16. Here, it may be set to the minimum required value. In this example, since 100−70 = 30 warm-up packets may be transmitted, the warm-up packet number determination unit 12 sets the warm-up packet number 30 to 30. Then, the process returns to step S60, and the transmission of the warm-up packet sequence with the updated number of packets is performed again.

  On the other hand, if it is determined that no packet loss has occurred due to mild congestion, in step S65, the congestion window size estimation unit 16 determines whether or not a packet loss due to more severe congestion has occurred. Specifically, the congestion window size estimation unit 16 sequentially checks the warm-up packet reception interval 32 for each warm-up packet from the first packet to the last packet in the warm-up packet sequence, and It is determined whether or not the value of the warm-up packet reception interval 32 of the packet is equal to or greater than a predetermined value (hereinafter referred to as a third threshold) that is determined to be packet loss due to severe congestion at the TCP level. The third threshold value may be a time obtained by adding the RTO timer value to the transmission interval of the packet of interest when, for example, occurrence of packet loss due to severe congestion is recognized due to occurrence of RTO. More specifically, the value of the warm-up packet transmission interval 31 of the warm-up packet currently focused on is added to the value of the RTO estimated value 41 or the average RTO estimated value 41 of the warm-up packet currently focused on. It may be a value.

  Note that the congestion window size estimation means 16 uses a predetermined threshold (similar to the above-described second threshold) as a determination condition for occurrence of packet loss due to severe congestion, in which the warm-up packet reception interval 32 of subsequent warm-up packets is extremely short. (Good) The following may be added to the above conditions.

  Hereinafter, conditions for determining the packet loss due to severe congestion in the application layer of the transmission side apparatus 1 will be described more specifically with reference to FIG. Also in the example illustrated in FIG. 7, it is assumed that the value of cwnd immediately before transmission of the warm-up packet sequence is 10 or more which is an initial value. As shown in FIG. 7, when the 90th warm-up packet arrives without loss, as in the example shown in FIG. 6, the OS layer of the receiving device 2 immediately receives the 90th warm-up packet on the application side. And a TCP level ACK packet for the 90th warm-up packet is returned.

  When the OS layer of the transmission side apparatus 1 receives the ACK packet for the 90th warm-up packet, it recognizes that the 90th warm-up packet has arrived without loss. Also in this example, when the first warm-up packet continues to the 90th warm-up packet without any problem, the cwnd value of the transmission side device 1 at the time when the ACK for the 90th warm-up packet arrives is 10 + 90 = 100 or more.

  However, the 91st, 92nd, and 93rd warm-up packets are lost in the middle of the network, and the TCP level ACK packet for the 91st warm-up packet is returned even if it exceeds the time length of the RTO value in TCP. Suppose there wasn't. In that case, the OS layer of the transmitting side apparatus 1 determines that a packet loss due to heavy congestion has occurred in response to the expiration of the retransmission timer (RTO has occurred) that was started when the 91st warm-up packet was transmitted. The 91st, 92nd, and 93rd warm-up packets for which no ACK has been returned are retransmitted.

  It is assumed that the 91st, 92nd, and 93rd warm-up packets retransmitted at this time have arrived without loss in the middle network. The OS layer of the receiving apparatus 2 immediately notifies the application side that the 91st, 92nd, and 93rd warm-up packets received so far have arrived, and sends the corresponding ACK packet of the warm-up packet to the transmitting apparatus 1. Reply to The application (packet receiving means 20) for estimating the available bandwidth of the receiving side apparatus 2 recognizes that it has received the 91st, 92nd, and 93rd warm-up packets only after receiving notification from the OS layer.

  Also in this example, the reception interval between the 89th and 90th warm-up packets that were not lost (the warm-up packet reception interval 32 of the 90th warm-up packet) and the 91st warm-up lost from the 90th The reception interval between packets (the warm-up packet reception interval 32 of the 91st warm-up packet) is greatly different. More specifically, the reception interval between the 89th and 90th warm-up packets is close to the transmission interval between the 89th and 90th warm-up packets, but the 90th and 91st warm-up packets. As shown in FIG. 7, the reception interval between the RTO estimated value 41 obtained for the 91st warm-up packet or the average RTO estimate is the transmission interval between the 90th and 91st warm-up packets. The value is equal to or greater than the value obtained by adding the value 41.

  Further, when viewed from the application side, since the 91st, 92nd, and 93rd warm-up packets, which are a predetermined number of warm-up packets after the lost packet, appear to be received almost simultaneously, packet loss due to slight congestion As in the case of, the extremely short reception interval of the subsequent warm-up packet may be added to the determination condition for occurrence of packet loss due to severe congestion.

  Returning to FIG. 5, if the congestion window size estimation means 16 determines that a packet loss due to severe congestion has occurred (Yes in step S65), it proceeds to step S66, and if it does not determine ( The process proceeds to step S67 (No in step S65).

  In step S66, the congestion window size estimation means 16 estimates the value of cwnd at the TCP level in response to the determination result that packet loss has occurred due to severe congestion. Here again, the congestion window size estimating means 16 does not need to estimate a strict value, and it is only necessary to predict the smallest possible value.

  For example, consider a case in which the congestion window control in the TCP protocol stack sets the value of cwnd to 2 when packet loss occurs due to severe congestion. The congestion window size estimation means 16 can estimate cwnd after occurrence of packet loss as 2, for example, when packet loss occurs due to severe congestion. Depending on the timing at which packet loss due to severe congestion occurs, the value of cwnd after completion of warm-up packet sequence transmission can take values from 99 to 2. However, the minimum value of cwnd may be set to 2 which is the worst pattern without considering these timings. The congestion window size estimation means 16 follows the control of the congestion window of the protocol stack based on various measured values and estimated values, for example, as in the packet loss determination method due to slight congestion, as in the method of determining packet loss due to slight congestion. And may be calculated.

  Now, it is assumed that the value of cwnd is estimated to be at least 2 or more when one warm-up packet sequence has been transmitted. In this case, since the value of cwnd is desired to be 100 or more, which is the estimated number of packets before packet train transmission, it can be seen that it is only necessary to increase 98 more.

  The warm-up packet number determination unit 12 determines the number of packets in the next warm-up packet sequence based on the congestion window size estimation result by the congestion window size estimation unit 16. Here, it may be set to the minimum required value. In the case of this example, 100−2 = 98 warm-up packets may be transmitted, so the warm-up packet number determining means 12 sets the warm-up packet number 30 to 98. Then, the process returns to step S60, and the transmission of the warm-up packet sequence with the updated number of packets is performed again.

  On the other hand, if it is determined that packet loss due to severe congestion has not occurred, the process proceeds to step S67, but the operation of step S67 to step S73 is merely to change the transmission target from the warm-up packet sequence to the packet train. Since the contents of the operation are essentially the same as those in steps S60 to S66, the description thereof is omitted. That is, when transmission of the packet train train is completed, the packet loss due to light or severe congestion during transmission is also determined for the packet train train. Note that the adjustment of the number of warm-up packets in step S71 and step S73 is for determining the number of warm-up packets at the next transmission of the warm-up packet sequence, and after completion of the adjustment, the process proceeds to step S75.

  Step S74 is a process that proceeds when it is determined that no packet loss has occurred during the transmission of the warm-up packet train and the transmission of the packet train. In step S74, the warm-up packet number determining means 12 may set the value of the warm-up packet number 30 to zero. Note that the number of warm-up packets to be included in the warm-up packet at the next warm-up packet sequence transmission is the estimated number of packets, regardless of the packet loss determination result at the time of warm-up packet train transmission or packet train transmission. Also good.

  In step S <b> 75, the available bandwidth estimation unit 21 of the reception side device 2 estimates the available bandwidth based on the reception result of the packet train transmitted from the transmission side device 1. Then, the available bandwidth estimation means 21 stores the estimated available bandwidth value in the data storage means 22 as the available bandwidth estimation value 59. For example, the method described in Patent Document 1 may be used for estimating the available bandwidth.

  For simplification of the system, in step S64, step S66, step S71, step S73, and step S74, the value of the number of warm-up packets 30 is set to the same value as the estimated number of packets 50 or 2 of the estimated number of packets 50. A double value may be set.

  As described above, according to the present embodiment, the number of warm-up packets is adjusted according to the presence / absence of packet loss and the degree of congestion, so that it can be accurately performed at a specified time over all estimated packets of the packet train. It is possible to minimize the number of warm-up packets required for transmission to the network. For this reason, not only can the packet train be transmitted accurately at the specified time, but also the time and transmission load overhead associated with the transmission of the warm-up packet can be reduced.

  In addition, during both warm packet transmission and packet train transmission, the number of warm packets is adjusted according to the occurrence of packet loss and the degree of congestion, thereby allowing multiple packet train transmissions. Even when it is performed, the number of warm-up packets can be reduced, and the time and communication load overhead associated with the transmission of warm-up packets can be reduced.

  In the above embodiment, the example in which the method for controlling the congestion window size according to the present invention is applied to the available bandwidth estimation system for estimating the available bandwidth is shown. However, the method for controlling the congestion window size according to the present invention is applicable. It is not limited to the band estimation application. For example, the present invention can be applied to a control system that provides a function for setting a congestion window size of a protocol stack to a predetermined value or more in the application layer. In that case, the function regarding the transmission of the packet train in the transmission side apparatus 1 and the function regarding the estimation of the available bandwidth in the reception side apparatus 2 are omitted.

  Next, an outline of the control system and the available bandwidth estimation system according to the present invention will be described. FIG. 8 is a block diagram showing an outline of a control system according to the present invention. As shown in FIG. 8, the control system according to the present invention includes a transmitting apparatus 700 that is a transmitting communication apparatus and a receiving apparatus 800 that is a receiving communication apparatus.

  Transmission-side apparatus 700 includes warm-up packet train transmission means 701, warm-up packet number determination means 702, and window size estimation means 703.

  The warm-up packet sequence transmission unit 701 (for example, the packet transmission unit 10, the packet sequence generation unit 11 and the like) uses a protocol that performs reception confirmation and congestion control by a window, and is requested to transmit at a predetermined transmission interval. Before transmitting the first packet sequence including a plurality of packets, a warm-up packet sequence in which a plurality of warm-up packets, which are preparation packets transmitted using a protocol, are collected is transmitted to the receiving-side device 800.

  The warm-up packet number determining unit 702 (for example, the warm-up packet number determining unit 12) determines the number of warm-up packets included in the warm-up packet sequence. The warm-up packet number determining means 702 determines the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size by the window size estimating means 703 described later.

  The window size estimation unit 703 (for example, the congestion window size estimation unit 16) acquires the transmission time of each warm-up packet included in the warm-up packet sequence and the reception at the reception-side device 800, acquired as a result of the transmission of the warm-up packet sequence. Based on the time, the minimum value of the congestion window size in the protocol stack after transmission of the warm-up packet sequence is estimated.

  According to such a configuration, when a predetermined packet sequence is transmitted using a protocol with reception confirmation and congestion control by a window, the congestion window is continuously transmitted in advance with a simple configuration at the application level. Can be increased to a possible size. In addition, when the first packet sequence is a packet train for estimating available bandwidth, it is possible to estimate the available bandwidth on a network with a simple application-level configuration that cannot use a protocol without reception confirmation with high accuracy. Can do.

  FIG. 9 is a block diagram showing another configuration example of the control system according to the present invention. As shown in FIG. 9, in the control system according to the present invention, in addition to the above configuration, the transmission-side apparatus 700 includes a transmission / reception time acquisition unit 704, a timer value estimation unit 705, a packet loss determination unit 706, and a first packet sequence. A transmission unit 707 may be included. Further, the receiving-side apparatus 800 may include an available band estimation unit 801.

  The transmission / reception time acquisition unit 704 (for example, the reception interval calculation unit 13 and the packet transmission unit 10) acquires the transmission time and the reception time at the reception side device of each warm-up packet included in the warm-up packet sequence.

  The timer value estimating means 705 (for example, the timer value estimating means 14) is based on the application level during transmission of the warm-up packet sequence from the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device. RTT, which is the round trip time of R, and RTO-val, which is the value of the retransmission timer at the protocol stack level.

  The packet loss determination unit 706 (for example, the packet loss determination unit 15) includes a transmission interval of warm-up packets specified from the transmission time of each warm-up packet included in the warm-up packet sequence, and each of the warm-up packet sequences. Based on the reception interval of the warm-up packet specified from the reception time of the warm-up packet, RTT, and RTO-val, whether or not a packet loss has occurred and whether the degree of congestion at that time is mild or severe judge.

  The first packet sequence transmission unit 707 (for example, the packet transmission unit 10, the packet sequence generation unit 11, etc.) transmits the first packet sequence to the reception side device 800.

  The available bandwidth estimation unit 801 (for example, the available bandwidth estimation unit 21) estimates the available bandwidth in the communication path between the transmission side device and the reception side device based on the reception result of the first packet sequence.

  When the transmission side apparatus 1 includes the first packet sequence transmission unit 707 and the reception side apparatus 2 includes the usable bandwidth estimation unit 801, the control system may be referred to as an usable bandwidth estimation system.

  Moreover, although a part or all of said embodiment can be described also as the following additional remarks, it is not restricted to the following.

  (Additional remark 1) It is provided with the transmission side apparatus which is a communication apparatus of a transmission side, and the reception side apparatus which is a communication apparatus of a reception side, and the transmission side apparatus uses the protocol by which reception confirmation and congestion control by a window are performed, and A warm-up packet in which a plurality of warm-up packets, which are preparation packets transmitted using the protocol, are collected before transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. As a result of transmission of the warm-up packet sequence, warm-up packet sequence transmission means for transmitting the sequence to the receiving side device, warm-up packet number determination means for determining the number of warm-up packets included in the warm-up packet sequence, and Based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device, the congestion window in the protocol stack after transmission of the warm-up packet sequence Window size estimating means for estimating the minimum value of the size, and the warm-up packet number determining means is a warm-up packet included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size Control system, characterized by determining the number of.

  (Supplementary Note 2) The warm-up packet number determination means includes a value obtained by subtracting the estimated minimum congestion window size from the number of packets included in the first packet sequence in the warm-up packet sequence to be retransmitted. The control system according to appendix 1, wherein the number of warm-up packets is used.

  (Supplementary Note 3) The window size estimation means obtains the transmission time of each warm-up packet included in the first packet sequence and the reception time at the receiving-side device acquired as a result of transmission of the warm-up packet sequence of the first packet sequence The warm-up packet number determining means estimates the minimum value of the congestion window size based on the estimation result of the minimum value of the congestion window size, and estimates the minimum value of the congestion window size in the protocol stack after transmission of the first packet sequence. The control system according to Supplementary Note 1 or Supplementary Note 2, wherein the number of warm-up packets to be included in a warm-up packet sequence that is transmitted before transmission of one packet is determined.

  (Supplementary Note 4) The warm-up packet number determining means determines whether the first packet of the next time when no packet loss occurs during transmission of the first packet sequence and transmission of the previous warm-up packet sequence. 4. The control system according to supplementary note 3, wherein the number of warm-up packets included in the warm-up packet sequence to be transmitted is determined to be 0 before transmission.

  (Supplementary Note 5) The transmission side device includes transmission / reception time acquisition means for acquiring the transmission time and the reception time at the reception side device of each warm-up packet included in the warm-up packet sequence, and each warm-up included in the warm-up packet sequence A timer that calculates RTT, which is a round trip time at the application level during transmission of a warm-up packet sequence, and RTO-val, which is a value of a retransmission timer at the protocol stack level, from the transmission time of the packet and the reception time at the receiving side device The warm-up packet transmission interval specified from the value estimation means, the transmission time of each warm-up packet included in the warm-up packet sequence, and the warm-up time specified from the reception time of each warm-up packet included in the warm-up packet sequence. Whether or not a packet loss has occurred based on the reception interval of the machine packet, the RTT, and the RTO-val. A packet loss determination unit that determines whether the degree of congestion is mild or severe, and the window size estimation unit estimates a minimum value of the congestion window size based on a determination result by the packet loss determination unit. The control system according to any one of the above.

  (Appendix 6) The window size estimation means estimates the minimum value of the congestion window size based at least on the number of reception confirmations received by transmission of the warm-up packet sequence and the number of occurrences of packet loss due to mild and severe congestion. The control system according to appendix 5.

  (Supplementary Note 7) The window size estimation means, when it is determined that a packet loss due to a slight congestion has occurred, where j is a decrease rate of the congestion window size due to a slight congestion in the congestion window control in the protocol stack, The control system according to appendix 5 or appendix 6, wherein the minimum value of the congestion window size in the current protocol stack is estimated to be j times the number of warm-up packets included in the warm-up packet sequence used for transmission.

  (Supplementary Note 8) The window size estimation means, when it is determined that a packet loss due to severe congestion has occurred, assuming that the value of the congestion window size after severe congestion in the congestion window control in the protocol stack is k, The control system according to any one of appendix 5 to appendix 7, wherein the minimum value of the congestion window size in the protocol stack is estimated as k.

  (Supplementary note 9) The packet loss determination means is configured such that the reception interval of a certain warm-up packet is the transmission interval of the warm-up packet, where n is the number of duplicate reception confirmations received before the protocol stack detects a slight congestion Is equal to or greater than the value obtained by adding RTT to the value obtained by multiplying n by a value as a determination condition for occurrence of packet loss due to light congestion in the warm-up packet. The control system according to any one of appendix 5 to appendix 8 for determining occurrence.

  (Supplementary Note 10) The packet loss determination means determines that a packet due to severe congestion in the warm-up packet when the reception interval of a warm-up packet is equal to or greater than a value obtained by adding RTO-val to the transmission interval of the warm-up packet. The control system according to any one of appendix 5 to appendix 9, which determines occurrence of packet loss due to severe congestion for each warm-up packet using the determination condition for loss occurrence.

  (Supplementary Note 11) The packet loss determination means further includes a predetermined number of warm-ups following the warm-up packet of interest as a determination condition for occurrence of packet loss due to light congestion or as a determination condition for occurrence of packet loss due to heavy congestion. The control system according to appendix 9 or appendix 10, which is added on condition that a reception interval of a subsequent warm-up packet which is a packet is equal to or less than a predetermined threshold value.

  (Supplementary Note 12) A transmission-side device that is a communication device on the transmission side and a reception-side device that is a communication device on the reception side, the transmission-side device uses a protocol that performs reception confirmation and congestion control by a window, and A warm-up packet in which a plurality of warm-up packets, which are preparation packets transmitted using the protocol, are collected before transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. As a result of transmission of the warm-up packet sequence, warm-up packet sequence transmission means for transmitting the sequence to the receiving side device, warm-up packet number determination means for determining the number of warm-up packets included in the warm-up packet sequence, and Based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device, the congestion window in the protocol stack after transmission of the warm-up packet sequence The window size estimation means for estimating the minimum value of the size of the window and the number of warm-up packets for determining the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size Means and a first packet sequence transmitting means for transmitting the first packet sequence to the receiving side device, the receiving side device based on the reception result of the first packet sequence, the transmitting side device and the receiving side device The warm-up packet number determining means includes a warm-up packet number to be included in a warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. An available bandwidth estimation system for determining the number of machine packets.

  (Supplementary Note 13) Using a protocol in which reception confirmation and congestion control by a window are performed, and using the protocol before transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval A warm-up packet sequence transmitting means for transmitting a warm-up packet sequence in which a plurality of warm-up packets, which are preparation packets to be transmitted, are collected to a predetermined receiving side device, and the number of warm-up packets included in the warm-up packet sequence Based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device, which is obtained as a result of transmission of the warm-up packet sequence, Window size estimation means for estimating the minimum value of the congestion window size in the protocol stack after transmission of the packet sequence, Based on the estimation result of the minimum value of the congestion window size, retransmission communication device, wherein the step of determining the number of warm-up packets to be included in the warm-up packet sequence to perform.

  (Additional remark 14) The transmission side apparatus which is a transmission side communication apparatus uses the protocol in which the reception confirmation and the congestion control by the window are performed, and includes a first packet composed of a plurality of packets requested to be transmitted at a predetermined transmission interval. Before transmitting the packet sequence, a warm-up packet sequence in which a plurality of warm-up packets, which are preparation packets transmitted using the protocol, are gathered is transmitted to a predetermined receiving-side device. Based on the transmission time of each warm-up packet included in the warm-up packet sequence acquired as a result of transmission and the reception time at the receiving side device, the minimum congestion window size in the protocol stack after transmission of the warm-up packet sequence is determined. Estimate and determine the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. If the number of machine packets is 1 or more, prior to transmission of the first packet sequence control method and performing retransmission of warm-up packet train including a determined number of warm-up packets.

  (Supplementary note 15) Before transmitting the first packet sequence composed of a plurality of packets that are requested to be transmitted at a predetermined transmission interval by using a protocol in which reception confirmation and congestion control by a window are performed in a computer, the protocol A process for transmitting a warm-up packet string, which is a group of warm-up packets, which are preparation packets transmitted using a packet, to a predetermined receiving device, and a warm-up packet acquired as a result of transmission of the warm-up packet stream Processing for estimating the minimum value of the congestion window size in the protocol stack after transmission of the warm-up packet sequence based on the transmission time of each warm-up packet included in the sequence and the reception time at the receiving side device, and the minimum of the congestion window size Based on the value estimation result, a process for determining the number of warm-up packets included in the warm-up packet sequence to be retransmitted is executed. Control program.

  While the present invention has been described with reference to the present embodiment and examples, the present invention is not limited to the above embodiment and examples. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

  INDUSTRIAL APPLICABILITY The present invention can be suitably applied to the use of estimating the available bandwidth in order to ensure video quality when performing video chat, videophone, video conference, etc., in which video is transmitted bidirectionally between terminals. It is. Further, the present invention is not limited to the usage estimation of the available bandwidth, and can be suitably applied to a case where a plurality of packets are continuously transmitted with high accuracy at a predetermined time interval. Further, the present invention is not limited to the continuous transmission application, and can be suitably applied to the case where it is desired to increase and maintain the congestion window size of the protocol stack from the application layer.

DESCRIPTION OF SYMBOLS 1,700 Transmission side apparatus 2,800 Reception side apparatus 3 Network 10 Packet transmission means 11 Packet sequence generation means 12 Warm-up packet number determination means 13 Reception interval calculation means 14 Timer value estimation means 15 Packet loss determination means 16 Congestion window size estimation Means 17 Data storage means 20 Packet reception means 21 Available bandwidth estimation means 22 Data storage means 30 Number of warm-up packets 31 Warm-up packet transmission interval 32 Warm-up packet reception interval 40 RTT measurement value 41 RTO estimation value 42 Congestion window size estimation value 50 Estimated number of packets 51 Estimated packet transmission interval 52 Estimated packet reception interval 59 Available bandwidth estimated value 701 Warm-up packet sequence transmission means 702 Warm-up packet number determination means 703 Window size estimation means 704 Transmission / reception time acquisition means 705 Timer Estimating means 706 packet loss determining unit 707 first packet train transmission means 801 usable bandwidth estimator

Claims (10)

A transmission side device that is a communication device on the transmission side, and a reception side device that is a communication device on the reception side,
The transmitting device is:
Preparation using a protocol that performs reception confirmation and congestion control by a window, and before the transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. A warm-up packet sequence transmitting means for transmitting a warm-up packet sequence in which a plurality of warm-up packets, which are packets for use, to the receiving side device;
Warm-up packet number determining means for determining the number of warm-up packets to be included in the warm-up packet sequence;
A protocol after transmission of the warm-up packet sequence, obtained as a result of transmission of the warm-up packet sequence, based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device Window size estimating means for estimating a minimum value of the congestion window size in the stack,
The warming-up packet number determining means determines the number of warming-up packets to be included in the warming-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. The warm-up packet number determining means includes a warm-up packet to be included in a warm-up packet sequence to be retransmitted by subtracting a minimum value of the estimated congestion window size from the number of packets included in the first packet sequence. The control system according to claim 1. The window size estimation means is obtained based on the transmission time of each warm-up packet included in the first packet sequence and the reception time in the reception-side device, acquired as a result of transmission of the warm-up packet sequence of the first packet sequence. A minimum congestion window size in the protocol stack after transmission of the first packet sequence,
The warm-up packet number determining means determines the number of warm-up packets to be included in the warm-up packet sequence to be transmitted before the next transmission of the first packet based on the estimation result of the minimum value of the congestion window size. The control system according to claim 1 or 2. The warm-up packet number determining means, when no packet loss occurs at the time of transmission of the first packet sequence and at the time of transmission of the previous warm-up packet sequence, before transmission of the next first packet The control system according to claim 3, wherein the number of warm-up packets included in the warm-up packet sequence to be transmitted is determined to be zero. The sending device is
Transmission / reception time acquisition means for acquiring the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device;
From the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device, the RTT that is the round trip time at the application level during the transmission of the warm-up packet sequence and the retransmission at the protocol stack level Timer value estimating means for calculating RTO-val which is a timer value;
The warm-up specified by the transmission interval of the warm-up packet specified from the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time of each warm-up packet included in the warm-up packet sequence Packet loss determination means for determining whether or not a packet loss has occurred and whether the degree of congestion at that time is mild or severe based on the packet reception interval, the RTT, and the RTO-val;
The control system according to any one of claims 1 to 4, wherein the window size estimation unit estimates a minimum value of a congestion window size based on a determination result by the packet loss determination unit. The window size estimation means estimates a minimum value of the congestion window size based at least on the number of reception confirmations received by transmission of the warm-up packet sequence and the number of occurrences of packet loss due to mild and severe congestion. The described control system. A transmission side device that is a communication device on the transmission side, and a reception side device that is a communication device on the reception side,
The transmitting device is:
Preparation using a protocol that performs reception confirmation and congestion control by a window, and before the transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. A warm-up packet sequence transmitting means for transmitting a warm-up packet sequence in which a plurality of warm-up packets, which are packets for use, to the receiving side device;
Warm-up packet number determining means for determining the number of warm-up packets to be included in the warm-up packet sequence;
A protocol after transmission of the warm-up packet sequence, obtained as a result of transmission of the warm-up packet sequence, based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device Window size estimation means for estimating the minimum value of the congestion window size in the stack;
Based on the estimation result of the minimum value of the congestion window size, the number of warm-up packets determining means for determining the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted,
First packet sequence transmitting means for transmitting the first packet sequence to the receiving side device;
The receiving side device
Based on the reception result of the first packet sequence, including an available bandwidth estimation means for estimating an available bandwidth in a communication path between the transmission side device and the reception side device;
The warm-up packet number determining means determines the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. system. Preparation using a protocol that performs reception confirmation and congestion control by a window, and before the transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. A warm-up packet sequence transmitting means for transmitting a warm-up packet sequence in which a plurality of warm-up packets, which are packets for use, to a predetermined receiving device;
Warm-up packet number determining means for determining the number of warm-up packets to be included in the warm-up packet sequence;
A protocol after transmission of the warm-up packet sequence, obtained as a result of transmission of the warm-up packet sequence, based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device Window size estimation means for estimating the minimum value of the congestion window size in the stack,
The warm-up packet number determining means determines the number of warm-up packets included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size. A transmitting device that is a communication device on the transmitting side
Preparation using a protocol that performs reception confirmation and congestion control by a window, and before the transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. A warm-up packet sequence in which a plurality of warm-up packets, which are packets for use, are sent to a predetermined receiving device,
A protocol after transmission of the warm-up packet sequence, obtained as a result of transmission of the warm-up packet sequence, based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device Estimate the minimum congestion window size in the stack,
Based on the estimation result of the minimum value of the congestion window size, determine the number of warm-up packets to be included in the warm-up packet sequence to be retransmitted,
When the determined number of warm-up packets is 1 or more, the warm-up packet sequence including the determined number of warm-up packets is retransmitted before transmitting the first packet sequence. Control method to do. On the computer,
Preparation using a protocol that performs reception confirmation and congestion control by a window, and before the transmission of the first packet sequence including a plurality of packets that are requested to be transmitted at a predetermined transmission interval. A process of transmitting a warm-up packet sequence in which a plurality of warm-up packets, which are packets for use, to a predetermined receiving device,
A protocol after transmission of the warm-up packet sequence, obtained as a result of transmission of the warm-up packet sequence, based on the transmission time of each warm-up packet included in the warm-up packet sequence and the reception time at the receiving side device Processing to estimate the minimum value of the congestion window size in the stack and processing to determine the number of warm-up packets included in the warm-up packet sequence to be retransmitted based on the estimation result of the minimum value of the congestion window size Control program to let you.

Images