JP5308364B2 - Transmission device, transmission method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To transmit data according to priority desired by a user by a simple method without changing an existing network. <P>SOLUTION: A transmission rate determination part 15 determines that congestion occurs when packet loss occurs or when the packet loss does not occur but RTT increases, and calculates a new CWND according to the priority so as to be smaller than the previously calculated CWND. Also, the transmission rate determination part determines that congestion does not occur when packet does not occur or when the packet loss does not occur and there is no change in RTT or the RTT decreases, and calculates a new CWND according to the priority so as to be larger than the previously calculated CWND. Thus, an existing network route 3 is used as it is. Also, since just a congestion control system is changed, data transmission according to the priority is achieved by a simple method. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a technique for controlling a transmission rate of each transmission in order to realize data transmission in accordance with a user's intention in a network in which a plurality of transmitting apparatuses and a plurality of receiving apparatuses are connected by sharing a band. .

  Conventionally, in an IP network typified by the Internet, transmission / reception devices transmit data by sharing the same network path. However, when the total number of packets transmitted from the transmission device exceeds the bandwidth of the network path, the packets are discarded by the router, and data cannot be transmitted normally. In addition, when packets are transmitted from the transmission device without limitation, congestion occurs in the IP network, and data is not transmitted normally. For this reason, it is necessary to avoid congestion in the transmission / reception apparatus. Therefore, the transmission / reception apparatus performs congestion control using TCP (Transmission Control Protocol) which is a typical transmission protocol used in the IP network.

  TCP has a built-in congestion control function, and various methods have been proposed as its control method. In any of the proposed methods, the transmitting device receives an acknowledgment (ACK) packet from the receiving device and detects the occurrence of a packet loss, or measures RTT (Round Trip Time). By determining the change, the packet transmission amount is determined and the transmission rate is controlled.

(New Reno method)
Among the congestion control functions incorporated in TCP, the most popular method is the TCP New Reno method (see Non-Patent Document 1). In this New Reno method, when the transmission device receives an ACK including an incrementing acknowledgment number, each time the ACK is received, the transmission amount of packets transmitted at a time is increased by one packet, and an ACK including the same acknowledgment number is received. When receiving continuously, it determines with having detected generation | occurrence | production of a packet loss, and reduces the packet transmission amount transmitted at once to 1/2. The New Reno method that adjusts the packet transmission amount by occurrence of such packet loss is called a loss-based congestion control method.

  Here, the confirmation response number included in the ACK corresponds to the packet number of the data packet transmitted by the transmission device, and is a value incremented for each ACK when normal. When the receiving device receives the data packet but determines that it is not a correct packet number (at the time of abnormality), an ACK is generated with the number to be originally received as an acknowledgment number and is transmitted to the transmitting device. That is, when the receiving device receives a data packet including a packet number that is not a correct number, the receiving device transmits an ACK with the number that should be originally received as an acknowledgment number to the transmitting device. As described above, when the receiving device continuously receives data packets including packet numbers that are not correct numbers, the receiving device continuously transmits ACKs including the same acknowledgment number to the transmitting device.

  FIG. 10 is a graph for explaining a change in the congestion window size (CWND) indicating the number of transmitted packets in the conventional New Reno method. This graph shows the results obtained by simulation. In general, congestion control incorporated in TCP is performed by adjusting a congestion window size (CWND) which is the number of packets transmitted by a transmission apparatus at a time.

In the New Reno scheme, when the transmitting device continuously receives ACKs including incrementing acknowledgment numbers from the receiving device, the transmitting device increases CWND as shown in Expression (1) for each reception.
CWND = CWND + (1 / CWND) (1)
When this is expressed on the time axis, CWND increases as in the ab section shown in FIG.

When the transmitting device receives ACKs including the same acknowledgment number three times in succession, the transmitting device determines that a packet loss has occurred and reduces CWND as shown in Equation (2). That is, the packet transmission amount is reduced to ½.
CWND = (1/2) × CWND (2)
If this is expressed on the time axis, CWND decreases as in the bc section when occurrence of packet loss is determined at time b in FIG. Then, when the transmitting device continuously receives ACK including the incrementing acknowledgment number, the transmitting device increases CWND again as shown in Expression (1) for each reception. When this is expressed on the time axis, CWND increases like a cd section shown in FIG.

  The change in CWND shown in FIG. 10 represents a state of a control phase called a congestion avoidance phase in the New Reno method, and is called a slow start phase for searching the upper limit of the network bandwidth prior to this phase. Initial control is performed. That is, the change in CWND shown in FIG. 10 represents a congestion avoidance phase that is a steady operation after the initial control is performed.

  In addition to the New Reno method, many congestion control methods described later have been proposed. In any of the methods, the transmitting device receives the acknowledgment normally and increases the CWND, and when the transmitting device does not normally receive the acknowledgment, the transmission device determines the occurrence of packet loss, or the RTT increases. , CWND is reduced.

(Vegas method)
As another typical congestion control method, there is a TCP Vegas method (see Non-Patent Document 2). This Vegas method is a method in which the transmission apparatus adjusts the packet transmission amount according to the increase or decrease of the RTT, and is called a delay-based congestion control method.

(Other methods)
Based on these two congestion control methods, many improved methods have been proposed from the viewpoint of effective use of bandwidth and fairness of transmission rate (see, for example, Patent Document 1). In addition, if there is no urgency in the content of the data being transmitted, the amount of packet transmission of that data is suppressed. If there is urgency, priority is given to data transmission with high urgency by increasing the amount of packet transmission of that data. There is also a proposed method (see Patent Document 2). Thus, the conventional congestion control method is based on the premise that the transmission device and the reception device voluntarily control the transmission rate.

  On the other hand, a method has been proposed in which the IP network is provided with a QoS (Quality of Service) function, and the packet transfer order of the router is controlled based on the priority according to the contents of each transmission. As typical methods, there are an IntServ method and a DiffServ method (see Non-Patent Documents 3 and 4). In these methods, the priority is described in the packet, and the router that configures the IP network transfers the packet with the higher priority before the packet with the lower priority.

JP 2006-217234 A JP 2006-14329 A

IETF RFC3782 L. Brakmo and L. Peterson, TCP Vegas: End to End Congestion Avoidance on a Global Internet, IEEE Journal on Selected Areas in Communication, Vol 13 No. 8, Oct. 1995 IETF RFC1633 IETF RFC2474

  The methods according to Non-Patent Documents 1 and 2 and Patent Document 1 described above are intended to control the transmission rate so that the transmission / reception apparatus uses the bandwidth of the IP network fairly. For this reason, there is a problem that data cannot be transmitted according to the priority according to the transmission content. For example, when transmitting video data and audio data that require real-time performance, if transmission of data that can be transmitted later is started and the network is congested, video or audio that requires real-time performance is transmitted. May be interrupted. Further, even when it is desired to transmit data that is determined to be urgent by the user, it is not possible to transmit urgent data at a higher speed than other data.

  In addition, the above-described method according to Patent Document 2 suppresses a transmission rate from a certain transmission device to a reception device, and gives priority to data transmission from another transmission device to the reception device. For this reason, the transmission rate cannot be controlled according to the stepwise priority desired by the user.

  Further, in the conventional New Reno method, when the transmission device receives ACK from the reception device, the transmission device increases or decreases the packet transmission amount based on the acknowledgment number included in the ACK. For this reason, when RTT is large, it takes time to increase the packet transmission amount. That is, when the RTT is large, it is relatively slow to increase the transmission rate compared to when the RTT is small. In general, the RTT increases in proportion to the distance between the transmitting device and the receiving device. For this reason, when the receiving device is used as a base point, the transmission rate with a distant transmitting device is lower than the transmitting rate with a distant transmitting device. In this case, data transmission according to the priority desired by the user cannot be realized, and data transmission of the packet transmission amount intended by the user cannot be realized.

  FIG. 11 is a graph for explaining a change in CWND due to a difference in RTT in the conventional New Reno method. This graph shows the results obtained by simulation. As shown in FIG. 11, when the RTT is large, it takes time to increase the CWND indicating the packet transmission amount and the transmission rate is lower than when the RTT is small.

  Further, the methods of Non-Patent Documents 3 and 4 described above are those in which a router configuring an IP network performs transfer according to priority. In order to realize this method, it is necessary to replace a router constituting the IP network with one corresponding to this method, and there is a problem that the existing IP network cannot be used as it is.

  In this way, according to the priority desired by the user, a high-priority data can be transmitted at high speed, and low-priority data can be transmitted at low speed by a simple method without changing the existing IP network. It has been desired to control the transmission rate of each transmission. Further, it has been desired to realize data transmission according to the user's intention without being affected by the size of the RTT.

  Therefore, the present invention has been made to solve such problems, and the object of the present invention is to realize data transmission according to the user's intention by a simple method without changing the existing network. To provide a transmission device, a transmission method, and a program. Specifically, the object of the present invention is to simplify data transmission capable of transmitting high priority data at high speed and low priority data at low speed according to the priority desired by the user. It is to be realized by a simple method. Another object is to realize data transmission according to the user's intention by a simple method regardless of the size of the RTT.

In order to achieve the above object, a transmitting apparatus according to the present invention transmits a data packet to a receiving apparatus while controlling a transmission rate according to a congestion state of a network, and receives a response to the data packet from the receiving apparatus. In the apparatus, a reception unit that receives a response to the transmitted data packet, and RTT (Round Trip Time) based on the transmission time of the data packet and the reception time of the response received by the reception unit and RTT measuring unit for measuring, on the basis of the RTT measured by the response and the RTT measuring unit received by the receiving unit, determines the congestion state of the network, the congestion state, priority, and the RTT measuring unit The transmission rate for determining the transmission rate for avoiding the congestion based on the RTT measured by A rate determining unit, and when the transmission rate determining unit decreases the transmission rate, the amount of decrease in the transmission rate is smaller when the priority is high than when the priority is low. In addition, the transmission rate is determined, and when the priority is low, the transmission rate is determined so that the amount of decrease in the transmission rate is larger than when the priority is high, and the transmission rate is increased. When the priority is high, the transmission rate is determined so that the increase in the transmission rate is larger than when the priority is low, and when the priority is low, the priority is When the transmission rate is determined and the transmission rate is decreased so that the increase amount of the transmission rate is smaller than when the transmission rate is high, the RTT is large when the RTT is small. The transmission rate is determined so that the decrease amount of the transmission rate is small, and when the RTT is small, the transmission rate is set so that the decrease amount of the transmission rate is larger than when the RTT is large. determined, in increasing the transmission rate, the case RTT is large, the RTT Ri name large increase of the transmission rate than small, if the RTT is small, than when the RTT is large Ri also an small increase of the transmission rate, further wherein when the priority is the same, increase of the transmission rate to be constant at a predetermined period, to determine the transmission rate, it is characterized.

The transmission device according to the present invention further includes a packet loss detection unit that detects occurrence of packet loss based on an acknowledgment number included in the response received by the reception unit, and the transmission rate determination unit includes: When the occurrence of a packet loss is detected by the packet loss detection unit, or when the occurrence of the packet loss is not detected and the RTT measured by the RTT measurement unit is increasing, the path of the network A path state determination unit that determines that congestion has occurred, the occurrence of the packet loss has not been detected, and the RTT has decreased, determines that no congestion has occurred in the network path; When it is determined by the route state determination means that congestion has occurred in the route of the network, the amount of data is decreased from the previously calculated data amount. And when the priority is high, the amount of decrease is smaller than when it is low, when the priority is low, the amount of decrease is larger than when it is high, and when the RTT is large, it is smaller than when it is small When the RTT is small, the new data amount is calculated so that the decrease amount becomes larger than when the RTT is large, and it is determined that the network path is not congested. The amount of increase is larger than the previously calculated data amount, and when the priority is high, the amount of increase is larger than when the priority is low, and when the priority is low, the amount of increase is smaller than when it is high. and the RTT is Ri the name large increase than when time is small or large, RTT Ri is smaller increase than when the larger is smaller, further wherein priority at the same time, the As the increment regardless serial RTT becomes constant at a predetermined period, the data amount calculation means for calculating an amount of new data, the new data amount calculated by the data amount calculating means, by the receiving unit Compared with the maximum acceptable data amount included in the received response, and when the new data amount is less than or equal to the maximum acceptable data amount, the new data amount is determined as a packet transmission amount, and the new data A packet transmission amount determination means for determining the maximum acceptable data amount as a packet transmission amount when the amount is larger than the maximum acceptable data amount; and a packet transmission amount determined by the packet transmission amount determination means. And controlling the transmission rate.

  Further, the transmitting device according to the present invention further inputs a destination address of a receiving device to which the data packet is transmitted, a priority holding unit that holds a priority corresponding to the destination address, and a destination address, A priority generation unit that reads out a priority corresponding to the destination address from a priority holding unit and outputs the priority to the transmission rate determination unit, and the transmission rate determination unit includes: The output priority is input and the transmission rate is determined.

The transmission apparatus according to the present invention increases the transmission rate when the transmission rate determining unit determines that the RTT has increased before detecting occurrence of packet loss as a result of increasing the transmission rate. It is characterized by suppressing .

The transmission apparatus according to the present invention further includes a stay packet count calculation unit, wherein the stay packet count calculation unit has an RTT measured by the RTT measurement unit and a transmission rate determined by the transmission rate determination unit. Based on the amount of data packets transmitted, the number of staying packets staying in the network is calculated, and the transmission rate determining unit calculates the response received by the receiving unit and the staying packet number calculating unit. The congestion state of the network is determined based on the number of staying packets .

Furthermore, a transmission method according to the present invention includes a step of transmitting a data packet to a receiving apparatus while controlling a transmission rate according to a congestion state of the network, and a step of transmitting the data packet and a response to the transmitted data packet. A step of receiving, a step of measuring an RTT (Round Trip Time) based on a transmission time of the transmitted data packet and a reception time of the received response, the received response and the measured RTT Determining a congestion state of the network based on the network, and determining a transmission rate for avoiding the congestion based on the determined congestion state , priority, and the measured RTT , When the step of determining the transmission rate reduces the transmission rate. When the priority is high, the transmission rate is determined such that the amount of decrease in the transmission rate is smaller than when the priority is low, and when the priority is low, the priority is higher than when the priority is high. When the priority is high when the transmission rate is increased and the transmission rate is increased, the transmission rate is increased more than when the priority is low. Determining the transmission rate so that the amount increases, and determining the transmission rate such that, when the priority is low, the increase in the transmission rate is smaller than when the priority is high , and When the RTT is large when decreasing the transmission rate, the transmission rate is determined so that the amount of decrease in the transmission rate is smaller than when the RTT is small, and the RTT is determined. When the RTT is large, the RTT is small when the RTT is large when the transmission rate is increased and the transmission rate is increased so that the decrease amount of the transmission rate is larger than when the RTT is large. When the RTT is small, the increase in the transmission rate is smaller than when the RTT is large. Furthermore, when the priority is the same, the transmission rate increases. First, the transmission rate is determined so that an increase amount of the transmission rate is constant in a predetermined period .

  Furthermore, a transmission program according to the present invention causes a computer to function as the transmission device.

  According to the present invention, since the priority reflecting the user's intention is taken into account when calculating the packet transmission amount, data transmission according to the user's intention can be realized. As a result, the existing network is used as it is, and it is only necessary to change the congestion control method without changing the procedure of the TCP method. Therefore, data transmission according to the priority can be realized by a simple method. Also, according to the present invention, since RTT is taken into account when calculating the packet transmission amount, data transmission according to the user's intention can be realized regardless of the size of RTT.

It is a figure which shows the structure of the whole system containing the transmission apparatus of Example 1. FIG. 1 is a block diagram illustrating a configuration of a transmission device according to Embodiment 1. FIG. It is a block diagram which shows the structure of a transmission rate determination part. It is a flowchart which shows the process of congestion control by a transmitter. 6 is a graph illustrating changes in CWND in Example 1. FIG. 6 is a diagram illustrating a configuration of an entire system including a transmission device according to a second embodiment. FIG. 10 is a diagram illustrating a configuration of an entire system including a transmission device according to a third embodiment. FIG. 10 is a diagram illustrating a configuration of an entire system including a transmission device according to a fourth embodiment. FIG. 10 is a diagram illustrating components included in a transmission device according to a fifth embodiment. It is a graph explaining the change of CWND in the conventional NewReno system. It is a graph explaining the change of CWND by the difference in RTT in the conventional NewReno system.

  The best mode for carrying out the present invention will be described below with reference to the drawings. The first embodiment (Example 1) assumes a system in which data is uploaded from a plurality of transmission devices to one reception device, and is an example in which the transmission rate is adjusted according to the priority directly input from the outside. is there. The second embodiment (Example 2) assumes a system in which data is uploaded from a plurality of transmission devices to one reception device, so that in addition to following the priority, it is not affected by RTT. This is an example of adjusting the transmission rate. The third embodiment (Example 3) assumes a system that downloads data from one transmitter to a plurality of receivers, and corresponds to Example 1, according to the priority directly input from the outside. This is an example of adjusting the transmission rate. The fourth embodiment (Example 4) assumes a system that downloads data from one transmitting device to a plurality of receiving devices, and corresponds to Example 2, in addition to following priority, RTT This is an example in which the transmission rate is adjusted so as not to be affected by. The fifth embodiment (Example 5) assumes a system for uploading or downloading data, and does not directly input the priority from the outside, but generates the priority based on the destination address indicating the receiving device. This is an example. The sixth embodiment (Example 6) assumes a system for uploading or downloading data, and is an example in which the transmission rate is adjusted so as not to be affected by RTT.

  First, Example 1 will be described in detail. The first embodiment assumes a system in which data is uploaded from a plurality of transmission apparatuses to one reception apparatus, and is an example in which the transmission rate is adjusted according to the priority directly input from the outside.

  FIG. 1 is a diagram illustrating a configuration of the entire system including the transmission apparatus according to the first embodiment. The system according to the first embodiment includes a plurality of transmission apparatuses 1-1, 1-2, 1-3,... (For example, a plurality of clients) and one reception apparatus 2 (for example, one server). ), And a plurality of transmitting apparatuses 1-1, 1-2, 1-3,... Are assumed to upload data to be transmitted simultaneously to one receiving apparatus 2. . The transmission apparatuses 1-1, 1-2, 1-3,... And the reception apparatus 2 are connected by a network path 3.

  Each of the transmission devices 1-1, 1-2, 1-3,... Transmits data to the reception device 2 via the shared network path 3. For this reason, in order to transmit data simultaneously, it is necessary to share a band. In conventional data transmission by TCP, the transmission rate is controlled so that each transmission device uses the bandwidth fairly by the congestion control function provided in the transmission device.

  However, depending on the content of each transmission, even if there is a user's request to transmit file data of high importance or urgency to the receiving device 2 at a higher speed, data can be transmitted with priority. could not. Therefore, in the first embodiment, data transmission according to the priority desired by the user is realized. The same applies to Examples 2 to 5.

(Configuration of transmitter)
Next, the configuration of transmitting apparatuses 1-1, 1-2, 1-3,... Shown in FIG. FIG. 2 is a block diagram illustrating a configuration of the transmission device 1 according to the first embodiment (hereinafter, the transmission devices 1-1, 1-2, 1-3,... Are collectively referred to as the transmission device 1). The transmission device 1 includes a transmission unit 10, a reception unit 11, an RTT (round trip time) measurement unit 12, a packet loss detection unit 13, an RWND (reception window size) detection unit 14, and a transmission rate determination unit 15.

  The transmission unit 10 receives transmission data, inputs a packet transmission amount from the transmission rate determination unit 15, generates a packet (data packet) by adding a TCP header to the transmission data, and receives a packet of the packet transmission amount Transmit to device 2. Further, the transmission unit 10 acquires the time when the packet was transmitted from a built-in timer (not shown), and outputs the transmission time to the RTT measurement unit 12.

  Here, the receiving device 2 receives the packet transmitted by the transmitting device 1, and determines that the received packet is a packet of the packet number that should be received (when the packet is determined to be normal), The packet number included in the packet is set as an acknowledgment number, and an ACK (ACK packet) including the acknowledgment number is generated and transmitted to the transmission apparatus 1. On the other hand, the receiving device 2 receives the packet transmitted by the transmitting device 1 and receives the packet when it is determined that the received packet is not the packet of the packet number that should be received (when the packet is determined not to be normal). The packet number included in the power packet is set as the confirmation response number, and an ACK including the confirmation response number is generated and transmitted to the transmission apparatus 1. Therefore, if the reception device 2 continuously determines that the packet is not normal, the reception device 2 continuously transmits ACKs including the same acknowledgment number to the transmission device 1.

  The ACK transmitted from the receiving apparatus 2 to the transmitting apparatus 1 is a response to the packet transmitted from the transmitting apparatus 1 to the receiving apparatus 2. When generating the ACK, the receiving device 2 describes RWND indicating the maximum acceptable data amount of the receiving device 2 in the TCP header, and transmits the ACK with the TCP header added to the transmitting device 1. That is, the ACK received by the transmission device 1 includes the confirmation response number and the RWND. RWND is a reception window size indicating the maximum amount of data that can be received by the receiving device 2 from the transmitting device 1.

  The receiving unit 11 receives an ACK including the acknowledgment number and RWND from the receiving device 2, acquires the time when the ACK is received from a built-in timer (not shown), and outputs the received time to the RTT measuring unit 12 To do. In addition, the reception unit 11 outputs the received ACK to the packet loss detection unit 13 and the RWND detection unit 14.

  The RTT measurement unit 12 receives and stores the packet transmission time from the transmission unit 10, and inputs the ACK reception time for the packet from the reception unit 11. The RTT measuring unit 12 measures the RTT by subtracting the transmission time from the reception time, and outputs the RTT that is the subtraction result to the transmission rate determining unit 15.

  The packet loss detection unit 13 receives the ACK from the reception unit 11 and, based on the confirmation response number included in the input ACK, the confirmation response number included in the ACK received and input last time and received and input this time The occurrence of packet loss is detected by comparing with an acknowledgment number included in the ACK. Then, the packet loss detection unit 13 outputs information on whether or not a packet loss has occurred indicating whether or not a packet loss has occurred to the transmission rate determination unit 15. Specifically, the packet loss detection unit 13 determines that the previous confirmation response number and the current confirmation response number are the same, and if the same determination continues for a predetermined number of times, a packet loss has occurred. Is determined. For example, when the predetermined number is set to 2, the packet loss detection unit 13 determines that the confirmation response numbers are the same twice, that is, an ACK including the same confirmation response number. Is determined to have occurred three times, it is determined that a packet loss has occurred. This determination method is called 3DUPACK.

  The RWND detection unit 14 receives the ACK from the reception unit 11, detects the RWND by acquiring the RWND described in the TCP header of the ACK, and outputs the RWND to the transmission rate determination unit 15.

  The transmission rate determination unit 15 inputs the priority desired by the user from the outside, receives RTT from the RTT measurement unit 12, information on the presence / absence of packet loss from the packet loss detection unit 13, and RWND from the RWND detection unit 14. Enter each. Then, the transmission rate determining unit 15 calculates CWND based on the priority, RTT, and packet loss occurrence data, determines the smaller of CWND and RWND as the packet transmission amount, and transmits the packet transmission amount. To the unit 10. Thereby, the transmission rate is determined. Details of the CWND calculation method will be described later. Hereinafter, it is assumed that RWND is a sufficiently large value compared to CWND, and that the increase / decrease in CWND directly affects the packet transmission amount. The packet having the packet transmission amount determined in this way is transmitted from the transmission unit 10.

  Although the transmission rate determination unit 15 inputs priority from the outside, the present invention does not limit the input method. For example, the transmission rate determination unit 15 may input the priority by the user directly performing a key operation on the transmission device 1. The receiving unit 11 may receive the priority from the receiving device 2 at the start of the session, and the transmission rate determining unit 15 may input the priority from the receiving unit 11. Furthermore, the transmission rate determination unit 15 may input the changed priority during packet transmission.

(Configuration of transmission rate determination unit)
Next, the configuration of the transmission rate determination unit 15 shown in FIG. 2 will be described. FIG. 3 is a block diagram showing a configuration of the transmission rate determination unit 15. The transmission rate determination unit 15 includes a path state determination unit 20, a parameter setting unit 21, a CWND calculation unit 22, and a packet transmission amount determination unit 23.

The route state determination means 20 inputs RTT from the RTT measurement unit 12 and information on whether or not packet loss has occurred from the packet loss detection unit 13, and the congestion state of the network route 3 according to the following (a) to (c) And the determination result of CWND maintenance, CWND decrease or CWND increase is generated according to the congestion state, and the determination result is output to the parameter setting means 21 and the CWND calculation means 22.
(A) When packet loss has occurred, it is determined that congestion has occurred in the bandwidth, and a determination result of CWND reduction is generated.
(B) When no packet loss occurs and RTT increases, it is determined that congestion has occurred in the band, and a determination result of CWND maintenance or CWND reduction is generated.
(C) If no packet loss occurs and there is no change in RTT or RTT is reduced, it is determined that no congestion has occurred in the bandwidth, and a determination result of CWND increase is generated.

  Here, the path state determination means 20 compares the RTT input this time with the RTT input last time, and determines the change state of the RTT. That is, the route state determination unit 20 determines that the RTT is increasing, there is no change in the RTT, or the RTT is decreasing based on the RTT. The path state determination unit 20 may determine the RTT change state by calculating the RTT change amount and comparing the RTT change amount with a preset threshold value. For example, the RTT change amount is compared with two preset threshold values. If the RTT change amount is larger than the first threshold value, it is determined that the RTT has increased, the RTT change amount is equal to or less than the first threshold value, and the second threshold value. When it is above, it is determined that there is no change in the RTT, and when the RTT change amount is smaller than the second threshold, it is determined that the RTT is decreasing. Further, when the path state determination unit 20 determines that there is no packet loss and the RTT has increased, if the RTT change amount is larger than a preset threshold value, the path state determination unit 20 generates a determination result of CWND maintenance, If the RTT change amount is equal to or less than a preset threshold value, a determination result of CWND reduction is generated.

  The parameter setting unit 21 inputs priority from the outside, and also inputs a determination result (CWND maintenance, CWND decrease or CWND increase) from the route state determination unit 20, and sets the parameter α based on the determination result and priority. The parameter α is output to the CWND calculation means 22.

  Specifically, when the determination result is CWND maintenance, the parameter setting unit 21 sets the parameter α = 1 regardless of the priority. Further, the parameter setting means 21 sets the parameter α in the range of 0 <α <1 so as to be proportional to the priority when the determination result is CWND decrease. For example, when the priority is high, the CWND calculated by the CWND calculating unit 22 to be described later is not reduced so much, that is, in a range of 0.5 <α <1 so that the reduction rate is lower than ½. Set the parameter α. When the priority is low, the parameter α is set in a range of 0 <α <0.5 so that CWND is sufficiently reduced, that is, the reduction rate is higher than ½. When the priority is medium, the parameter α of α = 0.5 is set so that the CWND reduction rate is halved.

  Also, the parameter setting means 21 sets the parameter α so as to be proportional to the priority when the determination result is an increase in CWND. For example, when the priority is high, the parameter α is set in a range of α> 1 so that the increase amount of CWND is larger than 1 / CWND. When the priority is low, the parameter α is set in a range of 0 <α <1 so that the increase amount of CWND is smaller than 1 / CWND. When the priority is medium, the parameter α with α = 1 is set so that the increase amount of CWND becomes 1 / CWND.

  The CWND calculation means 22 inputs a determination result (CWND maintenance, CWND decrease or CWND increase) from the path state determination means 20, and also inputs a parameter α from the parameter setting means 21, calculates CWND by the following formula, and CWND Is output to the packet transmission amount determination means 23.

When the determination result is CWND maintenance or CWND decrease, CWND calculation means 22 calculates CWND by the following formula.
CWND = α × CWND (3)
Here, as described above, when the determination result is CWND maintenance, the parameter α = 1, and when the determination result is CWND decrease, the parameter α is a value in the range of 0 <α <1 proportional to the priority. Therefore, when the determination result is CWND maintenance, CWND calculation means 22 calculates the same CWND as the previous time regardless of the priority. Also, the CWND calculation means 22 is such that when the determination result is CWND decrease and the priority is high, the value is smaller than the previous CWND and larger than the value obtained by multiplying the previous CWND by 1/2. Calculate a new CWND. In this case, CWND decreases to a relatively large value compared to the value calculated by the conventional New Reno method. In addition, the CWND calculation unit 22 newly sets the CWND calculation unit 22 so that when the determination result is CWND decrease and the priority is low, the value is smaller than a value obtained by multiplying the previous CWND by 1/2 and larger than 0. The correct CWND is calculated. In this case, CWND decreases to a relatively small value compared to the value calculated by the conventional New Reno method.

On the other hand, when the determination result is an increase in CWND, the CWND calculation unit 22 calculates CWND by the following equation.
CWND = CWND + α × (1 / CWND) (4)
Here, the parameter α is a value proportional to the priority (α> 1 when the priority is high, and 0 <α <1 when the priority is low). Therefore, when the priority is high, the CWND calculation unit 22 calculates a new CWND so that the increase is larger than the previous CWND and the increase amount is larger than the reciprocal of the previous CWND. In this case, CWND increases to a relatively large value compared to the value calculated by the conventional New Reno method. In addition, when the priority is low, the CWND calculation unit 22 calculates a new CWND so that it is larger than the previous CWND and the increase amount is smaller than the reciprocal of the previous CWND. In this case, CWND increases to a relatively small value compared to the value calculated by the conventional New Reno method.

  The packet transmission amount determination means 23 receives CWND from the CWND calculation means 22 and also receives RWND from the RWND detection unit 14, determines the smaller of CWND and RWND as the packet transmission amount, and transmits the packet transmission amount. To the unit 10.

  As described above, the transmission rate determination unit 15 inputs the priority, RTT, and packet loss occurrence data. (A) When packet loss occurs, the transmission rate determination unit 15 determines that the bandwidth is congested, and CWND. A decrease determination result is generated, and a new CWND is calculated within a range smaller than the previous CWND so that the CWND becomes larger when the priority is high than when the priority is low. Further, (b) when no packet loss occurs and RTT increases, it is determined that congestion has occurred, and a determination result of CWND maintenance or CWND reduction is generated. And when the determination result of CWND maintenance is produced | generated, CWND is calculated so that it may become the same value as last time. Further, when the determination result of CWND reduction is generated, a new CWND is calculated in a range smaller than the previous CWND so that the CWND is larger when the priority is higher than when the priority is low. Also, (c) When no packet loss occurs and there is no change in RTT or RTT is reduced, it is determined that congestion has not occurred, a determination result of CWND increase is generated, and the priority is high Calculates a new CWND in a range larger than the previous CWND so that the CWND is larger than when the priority is low. Thereby, the transmission apparatus 1 can control to transmit a packet at a relatively high transmission rate when the priority is high, and to transmit a packet at a relatively low transmission rate when the priority is low. . The CWND calculation method is for giving a relative priority to the transmission by the New Reno method, but the increase and decrease amount of the CWND is also used in the transmission by another method different from the New Reno method. By adjusting relative to, it is possible to give priority as in the New Reno method.

(Transmission device processing)
Next, congestion control processing by the transmission apparatus 1 will be described. FIG. 4 is a flowchart illustrating the congestion control process performed by the transmission apparatus 1. As described above, the transmission device 1 receives the ACK from the reception device 2, adjusts the packet transmission amount by increasing or decreasing the CWND for each ACK.

  The transmission unit 10 transmits the packet with the packet transmission amount determined in step S408 described later (step S401). The receiving unit 11 waits for an ACK that is a response to the packet transmitted by the transmitting unit 10, and receives the ACK (step S402).

  Here, the RTT measurement unit 12 subtracts the packet transmission time from the ACK reception time, measures the RTT, and outputs the RTT to the transmission rate determination unit 15. Based on the acknowledgment number included in the ACK, the packet loss detector 13 compares the acknowledgment number included in the previously received ACK with the acknowledgment number included in the ACK received this time, and the acknowledgment number is predetermined. When it is determined that the packet loss is the same several times, the occurrence of packet loss is detected, and when it is determined that the confirmation response number is continuous, the occurrence of packet loss is detected and the presence / absence of packet loss is determined as a transmission rate determination unit. 15 is output. Also, the RWND detection unit 14 acquires the RWND included in the ACK and outputs the RWND to the transmission rate determination unit 15.

  The transmission rate determining unit 15 determines whether or not a packet loss has occurred based on the information on whether or not a packet loss has been input from the packet loss detecting unit 13 (step S403). When it is determined that packet loss has occurred (step S403: packet loss is present), it is determined that congestion has occurred in the bandwidth, and a determination result of CWND reduction is generated as in (a) above, and according to priority The parameter α is set to perform the CWND reduction process, and the CWND is calculated using the equation (3) (step S405). On the other hand, when it is determined that no packet loss has occurred (step S403: no packet loss), comparison processing is performed using the RTT input from the RTT measurement unit 12 (step S404).

  If the transmission rate determining unit 15 determines in step S404 that the RTT has increased (step S404: RTT increase), the transmission rate determining unit 15 determines that congestion has occurred in the band, and sets the RTT change amount to a preset threshold value. To generate a determination result of CWND maintenance or CWND decrease, set the parameter α according to the priority, perform CWND maintenance or CWND decrease processing, and calculate CWND using the equation (3) ( Step S405). On the other hand, if it is determined that there is no change in RTT or RTT has decreased (step S404: no change or RTT decrease), it is determined that no congestion has occurred in the band, and parameter α is set according to the priority. A CWND increase process is performed, and the CWND is calculated using the equation (4) (step S406).

  Then, when the transmission apparatus 1 shifts from step S405 or step S406 to end the process (step S407: Y), the congestion control process ends. On the other hand, when the process is not terminated (step S407: N), the transmission rate determination unit 15 compares the CWND calculated in step S405 or step S406 with the RWND input from the RWND detection unit 14, and the CWND is equal to or less than RWND. In this case, CWND is determined as the packet transmission amount, and when CWND is larger than RWND, RWND is determined as the packet transmission amount and output to the transmission unit 10 (step S408). In step S401, the transmission unit 10 transmits a packet having the packet transmission amount determined in step S408 to the reception device 2, and in step S402, the reception unit 11 waits for an ACK transmitted from the reception device 2. . In this way, the processing from step S401 to step S408 is repeated.

  2, 3, and 4, the RTT measurement unit 12 measures the RTT, and the transmission rate determination unit 15 determines whether or not congestion has occurred based on the RTT (step of FIG. 4). S404), it may be determined using the number of staying packets instead of RTT. That is, the transmission apparatus 1 may calculate the number of packets staying in the network path 3 and determine whether congestion has occurred based on the number of staying packets.

Specifically, the transmission apparatus 1 includes a stay packet count calculation unit. The stay packet count calculation unit inputs CWND from the transmission rate determination unit 15 and RTT from the RTT measurement unit 12, and has been measured so far. The minimum RTT among the input RTTs is set to RTTmin, the currently measured RTT input is set to RTTnow, and the number of staying packets δ, which is the number of packets staying in the network path 3, according to the following equation: And the number of staying packets δ is output to the transmission rate determining unit 15.
δ = (CWND / RTTmin−CWND / RTTnow) × RTTmin
... (5)
Then, instead of inputting the RTT from the RTT measuring unit 12, the transmission rate determining unit 15 inputs the staying packet number δ from the staying packet number calculating unit, and in the same manner as the RTT comparison process in step S404 in FIG. Comparison processing of the number of staying packets λ is performed to determine the occurrence of congestion. In this case, the transmission rate determination unit 15 may determine the occurrence of congestion by calculating a change amount of the staying packet number λ and comparing the change amount with a preset threshold value. When it is determined that the bandwidth is congested, the transmission rate determination unit 15 generates a determination result of CWND maintenance or CWND reduction, and proceeds to step S405 to perform CWND maintenance or CWND reduction processing. On the other hand, if the transmission rate determining unit 15 determines that congestion has not occurred in the band, the transmission rate determining unit 15 generates a CWND increase determination result, proceeds to step S406, and performs a CWND increase process.

(simulation result)
FIG. 5 is a graph illustrating changes in CWND in the first embodiment. This graph shows the results obtained by simulation. FIG. 5 is a graph showing a change in CWND calculated at a high priority and a change in CWND calculated at a low priority. Further, the CWND calculated by the conventional NewReno method is shown in a graph. Changes are also shown in the graph. From FIG. 5, it can be seen that when the priority is high, CWND increases faster and the amount of decrease in CWND when packet loss occurs is small. It can also be seen that when the priority is low, CWND increases more slowly, and the amount of decrease in CWND is greater when packet loss occurs.

  Further, from FIG. 5, the average value of CWND when the priority is high is larger than the average value of CWND according to the conventional New Reno method, and the average value of CWND when the priority is low is the conventional New Reno method. It can be seen that it is smaller than the average value of CWND.

  As described above, according to the transmission apparatus 1 of the first embodiment, when data is uploaded from a plurality of transmission apparatuses 1-1, 1-2, 1-3,... To one reception apparatus 2. If it is determined that packet loss has occurred, or if packet loss has not occurred and RTT has increased, it is determined that congestion has occurred in the bandwidth, and priority When C is high, the new CWND is calculated within a range smaller than the previous CWND so that it is larger than the CWND according to the conventional method, and when the priority is low, it is smaller than the CWND according to the conventional method. And adjusted the transmission rate. Further, when it is determined that no packet loss has occurred and there is no change in RTT or RTT has decreased, it is determined that there is no congestion in the bandwidth, and when the priority is high, When the priority is low so as to be larger than the CWND based on the scheme, a new CWND is calculated within a range larger than the previous CWND so as to be smaller than the CWND based on the conventional scheme, and the transmission rate is adjusted. I did it. As a result, since the router arranged in the existing network path 3 can be used as it is, only the congestion control system is changed without changing the network path 3 and without changing the procedure of the TCP system. The data transmission according to the priority desired by the user can be realized by a simple method.

  Further, according to the transmission apparatus 1 of the first embodiment, when it is determined in step S404 illustrated in FIG. 4 that the RTT has increased, it is determined that the bandwidth is congested, and the CWND maintenance process or When CWND reduction processing is performed and it is determined that there is no change in RTT or RTT is decreasing, it is determined that there is no congestion in the bandwidth, and CWND increase processing is performed.

  Here, when the transmission rate determination unit 15 of the transmission device 1 increases CWND abruptly to increase the transmission rate, when the transmission rate exceeds the bandwidth of the network path 3, the transmission rate is determined continuously in the network path 3. Packets are discarded. Then, since the period in which the ACK is not transmitted by the reception device 2 becomes long, the reception unit 11 of the transmission device 1 cannot receive the ACK corresponding to the packet transmitted from the transmission unit 10 and performs a timeout process. Do. Then, the transmission rate determination unit 15 resets CWND to a small value (for example, an initial value 1) with a timeout. That is, when the transmission rate determining unit 15 suddenly increases CWND and the receiving unit 11 detects a timeout, the transmission rate determining unit 15 immediately resets CWND to a small value. For this reason, the transmission rate determination unit 15 cannot increase the average value of CWND, although it is necessary to increase CWND. Therefore, in order to adjust the transmission rate by appropriately increasing CWND, it is necessary for the transmission rate determination unit 15 to suppress the increase amount of CWND in the vicinity of the upper limit of the band so that packets are not continuously discarded in the network path 3. There is.

  In the first embodiment, the process of step S404 is performed in order to suppress an increase in CWND near the upper limit of the band. Specifically, the transmission rate determination unit 15 determines an increase in RTT in step S404 in association with a rapid increase in CWND. Thereby, it is possible to detect a sign of bandwidth congestion in the network path 3 before packet loss occurs. In this case, the transmission rate determining unit 15 performs a CWND maintenance process or a CWND reduction process in step S405. As a result, an increase in CWND leading to occurrence of packet loss can be suppressed. As described above, since the increase in CWND can be suppressed, it is possible to prevent the occurrence of continuous packet loss and to avoid the timeout process due to the non-reception of ACK.

As a modification of the first embodiment, the transmission device 1 may not include the RTT measurement unit 12. The transmission apparatus 1 according to the first embodiment includes the RTT measurement unit 12, and the transmission rate determination unit 15 determines in step S403 in FIG. 4 that no packet loss has occurred (step S403: Whether there is no packet loss ) is determined whether the RTT is measured by the RTT measurement unit 12 (step S404). When it is determined that the RTT has increased, the CWND maintenance or CWND decrease process is performed. When it is determined that the RTT has not changed or the RTT has decreased, the CWND increase process is performed. On the other hand, in the modification of the first embodiment, when the transmission rate determining unit 15 determines in step S403 in FIG. 4 that no packet loss has occurred (step S403: no packet loss ), the transmission rate determining unit 15 The process of increasing CWND is performed without performing the process (step S406).

  As described above, according to the transmission device 1 according to the modification of the first embodiment, when it is determined that a packet loss has occurred, it is determined that congestion has occurred in the bandwidth, and when the priority is high, When the priority is low so as to be larger than the CWND according to the conventional method, a new CWND is calculated within a range smaller than the previous CWND so as to be smaller than the CWND according to the conventional method, and the transmission rate is set. I adjusted it. When it is determined that no packet loss has occurred, it is determined that there is no congestion in the bandwidth, and when the priority is high, the priority is low so that it is larger than CWND according to the conventional method. Has calculated a new CWND within a range larger than the previous CWND so as to be smaller than the CWND according to the conventional method, and adjusts the transmission rate. As a result, the router arranged in the existing network path 3 can be used as it is in the same manner as in the first embodiment. Therefore, the congestion can be achieved without changing the network path 3 and without changing the TCP procedure. By simply changing the control method, data transmission according to the priority desired by the user can be realized by a simple method.

  As described above, with respect to the congestion control method performed by the transmission apparatus 1 according to the first embodiment, a plurality of transmission apparatuses 1-1, 1-2, 1-3,. 2 is described assuming that the system simultaneously transmits data to the transmission apparatus 1-1. However, in this congestion control method, the transmission apparatuses 1-1, 1-2, 1-3,. In this case, the relative transmission rate of each device is controlled according to the priority. Therefore, the congestion control method by the transmission apparatus 1 according to the first embodiment can be applied to a system including a plurality of reception apparatuses 2 in addition to the system illustrated in FIG.

  Next, Example 2 will be described in detail. The second embodiment assumes a system in which data is uploaded from a plurality of transmission apparatuses to one reception apparatus. In addition to following the priority, the transmission rate is adjusted so as not to be affected by RTT. is there.

  FIG. 6 is a diagram illustrating a configuration of the entire system including the transmission apparatus according to the second embodiment. The system according to the second embodiment includes a plurality of transmission apparatuses 1-1 and 1-2 (for example, a plurality of clients) and one reception apparatus 2 (for example, one server). In this system, it is assumed that a plurality of transmission apparatuses 1-1 and 1-2 simultaneously upload data to be transmitted to one reception apparatus 2. The transmitting apparatus 1-1 and the receiving apparatus 2 are connected by a network path 3-1 with RTT = 50 msec and a network path 3-3 with RTT = 50 msec. The transmitting device 1-2 and the receiving device 2 are connected by a network path 3-2 with RTT = 150 msec and a network path 3-3 with RTT = 50 msec.

  When transmitting data to the receiving device 2 while sharing the network path 3-3, each of the transmitting devices 1-1 and 1-2 uses a congestion control method described in the first embodiment to make a user's request. Data transmission according to the priority to be performed can be realized.

However, as shown in FIG. 6, the sum is 100msec of RTT upon transmission of data toward the transmit unit 1-1 to the receiver 2, the data toward the transmission apparatus 1-2 to the receiver 2 The total RTT for transmission is 200 msec. For this reason, the time until the transmission device 1-2 receives the ACK for the packet transmitted toward the reception device 2 is the time until the transmission device 1-1 receives the ACK for the packet transmitted toward the reception device 2. Twice compared to time. Since the CWND increases when an ACK is received, as shown in FIG. 11, the transmission apparatus 1-2 with a large RTT requires more time to increase the CWND than the transmission apparatus 1-1 with a small RTT. Will be required. That is, as the RTT increases, the transmission rate increases at a slower rate. Therefore, even if the same priority is set, the transmission rate varies depending on the size of the RTT.

  Here, depending on each transmission content, depending on the importance or urgency of the transmission content, there is a case where data transmission from the transmission device 1-2 having a large RTT to the reception device 2 is desired to be performed at higher speed than other data transmissions. is there. In addition, data transmission from the transmission device 1-1 having a small RTT to the reception device 2 may be performed at a lower speed than other data transmissions.

  Therefore, in the second embodiment, data transmission according to the priority desired by the user is realized regardless of the size of the RTT without being affected by the RTT.

(Configuration of transmission device, configuration of transmission rate determination unit, processing of transmission device)
Next, the configuration of the transmission devices 1-1 and 1-2 according to the second embodiment illustrated in FIG. 6, the configuration of the transmission rate determination unit 15, and the processing of the transmission devices 1-1 and 1-2 will be described. The configuration of the transmission device 1 according to the second embodiment (hereinafter, the transmission devices 1-1 and 1-2 are collectively referred to as the transmission device 1) is basically the same as the configuration illustrated in FIG. The rate determining unit 15 performs processing different from that in the first embodiment. The configuration of the transmission rate determination unit 15 provided in the transmission apparatus 1 is basically the same as the configuration illustrated in FIG. 3, but the CWND calculation unit 22 performs processing different from that in the first embodiment. The processing of the transmission apparatus 1 according to the second embodiment is basically the same as the processing illustrated in FIG. 4, but the CWND maintenance and CWND reduction processing in step S405 and the CWND increase processing in step S406 are performed. Different from Example 1.

  Referring to FIG. 3, CWND calculation unit 22 according to the second embodiment inputs a determination result (CWND maintenance, CWND decrease or CWND increase) from path state determination unit 20 and parameter α from parameter setting unit 21. Further, RTT is input from the RTT measurement unit 12. Then, the CWND calculation unit 22 calculates CWND by the following formula and outputs the CWND to the packet transmission amount determination unit 23.

When the determination result is CWND maintenance or CWND decrease, the CWND calculation unit 22 calculates CWND by the following equation (step S405 in FIG. 4).
CWND = α × (RTT / RTTorg) × CWND (6)
Here, RRTorg is a preset value that serves as a reference for RTT. As in the first embodiment, the parameter α is a parameter α = 1 when the determination result is CWND maintenance, and a value in the range of 0 <α <1 proportional to the priority when the determination result is CWND decrease. is there.

  Here, since CWND is calculated for each RTT that receives an ACK, when RTT is large, the update rate of CWND is slow, and when RTT is small, the update rate of CWND is fast. According to the second embodiment, the CWND is calculated using the equation (6) in consideration of the RTT. Therefore, when the RTT is large, the decrease amount of the CWND can be suppressed to a large value. Further, when RTT is small, the amount of decrease in CWND can be increased, and CWND can be made smaller than CWND when RTT is large. Therefore, considering that CWND is increased after being decreased, when RTT is large, the update frequency when CWND is increased is less than when RTT is small, and the amount of increase is reduced, so CWND is decreased. By suppressing the amount of decrease, the same CWND as when the RTT is small can be obtained at an early stage when increasing the CWND.

On the other hand, when the determination result is an increase in CWND, the CWND calculation unit 22 calculates CWND by the following equation (step S406 in FIG. 4).
CWND = CWND + α × (RTT / RTTorg) × (1 / CWND)
... (7)
As in the first embodiment, the parameter α is a value proportional to the priority (α> 1 when the priority is high, and 0 <α <1 when the priority is low).

  As described above, CWND is calculated for each RTT that receives an ACK. Therefore, when RTT is large, the update rate of CWND is slow, and when RTT is small, the update rate of CWND is fast. Therefore, when the determination result is an increase in CWND, the CWND calculation means 22 can set an increase rate equivalent to the CWND increase rate at the RTRT as a reference regardless of the RTT. This is because CWND is updated (T / RTT) times in a predetermined period T. Therefore, if the priority is the same, the increase amount of CWND is α × (RTT / RTTorg) × (1 / CWND) × (T / RTT) = α × (1 / RTTorg) × (1 / CWND) × T, and the increase in CWND is constant without depending on the value of RTT. That is, as shown in FIG. 11, when the RTT is large, the CWND update rate has been slow since the CWND update rate has been slow. Also, when the RTT is small, conventionally, the CWND update rate is fast, so the CWND increase rate is fast. According to the second embodiment, the CWND is calculated using the equation (7) in consideration of the RTT. Therefore, the increase rate of the CWND can be matched with the increase rate of the RRTorg according to the size of the RTT.

  As described above, according to the transmission device 1 of the second embodiment, as in the first embodiment, when uploading data from a plurality of transmission devices 1-1 and 1-2 to one reception device 2, It is determined whether or not the bandwidth is congested according to the occurrence of packet loss and the increase in RTT. When the priority is high, the priority is set so as to be larger than CWND according to the conventional method. When it is low, CWND is calculated to be smaller than CWND according to the conventional method, and when it is determined that congestion occurs in the bandwidth, a new CWND is calculated within a range smaller than the previous CWND, When it is determined that there is no congestion in the bandwidth, a new CWND is calculated within a range that is larger than the previous CWND, and the transmission rate is adjusted. Thereby, similarly to the first embodiment, data transmission according to the priority desired by the user can be realized by a simple method. Similarly to the first embodiment, the processing in step S404 illustrated in FIG. 4 detects signs of bandwidth congestion in the network paths 3-1, 3-2, and 3-3 before packet loss occurs. Since the increase in CWND can be suppressed, it is possible to prevent the occurrence of continuous packet loss and to avoid the timeout process due to the non-reception of ACK.

  Also, according to the transmission apparatus 1 of the second embodiment, when performing CWND maintenance and CWND decrease processing, CWND is calculated using Equation (6) considering RTT, and when CWND increase processing is performed, RTT is calculated. Since the CWND is calculated using the considered equation (7), the increase rate of the CWND can be matched with the increase rate of the reference RTTorg according to the size of the RTT. Therefore, regardless of the size of RTT, the transmission rate can be adjusted so that the average value (predetermined packet transmission amount) of the packet transmission amount in a predetermined time is equal. That is, data transmission according to the priority desired by the user can be realized by a simple method regardless of the size of the RTT without being affected by the RTT.

  As described above, regarding the congestion control method performed by the transmission apparatus 1 according to the second embodiment, a plurality of transmission apparatuses 1-1 and 1-2 simultaneously transmit data to one reception apparatus 2 as in the system illustrated in FIG. However, in this congestion control method, the transmission apparatuses 1-1 and 1-2 and the reception apparatus 2 share a bandwidth and pass through network paths 3-1 and 3-2 having different RTTs. When transmitting data, the relative transmission speed of each device is controlled according to the priority without being affected by RTT. Therefore, the congestion control method by the transmission apparatus 1 according to the second embodiment can be applied to a system including a plurality of reception apparatuses 2 in addition to the system illustrated in FIG.

  Next, Example 3 will be described in detail. The third embodiment assumes a system in which data is downloaded from one transmission device to a plurality of reception devices, and corresponds to the first embodiment and is an example in which the transmission rate is adjusted according to the priority directly input from the outside. is there.

  FIG. 7 is a diagram illustrating a configuration of the entire system including the transmission apparatus according to the third embodiment. The system according to the third embodiment includes one transmission device 1 (for example, one server) and a plurality of reception devices 2-1, 2-2, 2-3,... (For example, a plurality of clients). ), And a system that assumes download of data that one transmitter 1 transmits simultaneously to a plurality of receivers 2-1, 2-2, 2-3,. . The transmission device 1 and the reception devices 2-1, 2-2, 2-3,... Are connected by a network path 3.

  The transmission device 1 transmits data to the reception devices 2-1, 2-2, 2-3,... Via the shared network path 3. For this reason, in order to transmit data simultaneously, it is necessary to share a band. In the conventional data transmission by TCP, the transmission rate is controlled so that the transmission apparatus uses the bandwidth fairly by the congestion control function provided in the transmission apparatus.

  However, depending on the content of each transmission, there is a user's request to transmit file data of high importance or urgency to the receiving devices 2-1, 2-2, 2-3,. However, data could not be transmitted with priorities. Thus, in the third embodiment, data transmission according to the priority desired by the user is realized.

(Configuration of transmission device, configuration of transmission rate determination unit, processing of transmission device)
The configuration of the transmission device 1 according to the third embodiment illustrated in FIG. 7, the configuration of the transmission rate determination unit 15 included in the transmission device 1, and the processing of the transmission device 1 are the same as those of the transmission device 1 according to the first embodiment illustrated in FIG. The configuration is the same as the configuration of the transmission rate determination unit 15 shown in FIG. 3 and the processing of the transmission apparatus 1 shown in FIG.

  As described above, according to the transmission device 1 of the third embodiment, when data is downloaded from one transmission device 1 to a plurality of reception devices 2-1, 2-2, 2-3,. In addition, with the same configuration and processing as those of the transmission device 1 according to the first embodiment, data transmission according to the priority desired by the user can be realized by a simple method as in the first embodiment.

As a modification of the third embodiment, the transmission device 1 may not include the RTT measurement unit 12 as in the modification of the first embodiment. That is, if the transmission rate determining unit 15 determines in step S403 in FIG. 4 that no packet loss has occurred (step S403: no packet loss ), the transmission rate determining unit 15 performs the CWND increase process without performing the process in step S404. Is performed (step S406).

  Next, Example 4 will be described in detail. The fourth embodiment assumes a system that downloads data from one transmission device to a plurality of reception devices, and corresponds to the second embodiment, and is not affected by the RTT in addition to following the priority. In this example, the transmission rate is adjusted.

  FIG. 8 is a diagram illustrating a configuration of the entire system including the transmission apparatus according to the fourth embodiment. The system according to the fourth embodiment includes one transmission device 1 (for example, one server) and a plurality of reception devices 2-1 and 2-2 (for example, a plurality of clients). In this system, it is assumed that one transmitter 1 downloads data simultaneously transmitted to a plurality of receivers 2-1, 2-2. The transmission apparatus 1 and the reception apparatus 2-1 are connected by a network path 3-1 with RTT = 50 msec and a network path 3-2 with RTT = 50 msec. The transmitting apparatus 1 and the receiving apparatus 2-2 are connected by a network path 3-1 with RTT = 50 msec and a network path 3-3 with RTT = 150 msec.

  When the transmission apparatus 1 transmits data to the reception apparatus 2 while sharing the network path 3-1, the data according to the priority desired by the user is transmitted by the congestion control method described in the third embodiment. Transmission can be realized.

However, as shown in FIG. 8, the sum of the RTT in transmitting the data toward the transmit device 1 to the receiving device 2-1 is 100 msec, the data toward the transmitter 1 to the receiver 2-2 The total RTT for transmission is 200 msec. For this reason, the time until the transmission device 1 receives an ACK for the packet transmitted toward the reception device 2-2 is the time until the transmission device 1 receives an ACK for the packet transmitted toward the reception device 2-1. Twice compared to time. Since CWND increases when an ACK for a transmitted packet is received, the increase rate of the transmission rate becomes slower as RTT increases, and even if the same priority is set, it depends on the difference in size of RTT. , It becomes a different transmission rate.

  Here, depending on each transmission content, depending on the importance or urgency of the transmission content, data transmission from the transmission device 1 having a large RTT to the reception device 2-2 may be performed at a higher speed than other data transmissions. is there. Further, data transmission from the transmission device 1 having a small RTT to the reception device 2-1 may be performed at a lower speed than other data transmissions.

  Thus, in the fourth embodiment, data transmission according to the priority desired by the user is realized regardless of the size of the RTT without being affected by the RTT.

(Configuration of transmission device, configuration of transmission rate determination unit, processing of transmission device)
The configuration of the transmission device 1 according to the fourth embodiment illustrated in FIG. 8, the configuration of the transmission rate determination unit 15 included in the transmission device 1, and the processing of the transmission device 1 are the same as the configuration of the transmission device 1 according to the second embodiment. This is the same as the configuration of the transmission rate determination unit 15 and the processing of the transmission device 1 according to the second embodiment.

  As described above, according to the transmission device 1 of the fourth embodiment, even when data is downloaded from one transmission device 1 to the plurality of reception devices 2-1 and 2-2, the transmission device according to the second embodiment. As with the second embodiment, the same configuration and processing as in the first embodiment are not affected by the RTT, and the data transmission according to the priority desired by the user is realized by a simple method regardless of the size of the RTT. can do.

  Next, Example 5 will be described in detail. The fifth embodiment is applied to any of the systems according to the first to fourth embodiments and the first and third embodiments, and assumes a system for uploading or downloading data. The priority is directly input from the outside. Instead, the priority is generated based on the destination address indicating the receiving device.

  FIG. 9 is a diagram illustrating components included in the transmission device 1 according to the fifth embodiment. In addition to the configuration illustrated in FIG. 2, the transmission device 1 according to the fifth embodiment includes a priority holding unit 16 and a priority generation unit 17 before the transmission rate determination unit 15. The priority holding unit 16 holds priority information corresponding to the destination address. The priority generation unit 17 inputs a destination address from the outside, reads the priority corresponding to the destination address from the priority holding unit 16, generates the priority, and sets the priority corresponding to the destination address to the transmission rate determination unit 15 is output. The transmission unit 10, the reception unit 11, the RTT measurement unit 12, the packet loss detection unit 13, the RWND detection unit 14, and the transmission rate determination unit 15 are the same as the configuration units illustrated in FIG. Further, the processing of the transmission device 1 is the same as the processing of the first to fourth embodiments described above.

  As described above, according to the transmission device 1 of the fifth embodiment, the priority generation unit 17 can generate priorities so that the priorities are the same as, for example, the conventional New Reno method. Even if there is no input of priority, data transmission can be performed at a transmission rate comparable to that of the conventional method. Further, even if no priority is input from the outside, data transmission according to the priority for each destination address indicating the receiving apparatus can be realized by a simple method.

  Note that, when the transmission device starts data transmission from the transmission device, the destination address is input from the outside and the priority is generated. However, the present invention is not limited to this. For example, when the transmission device receives a transmission start request from the reception device, the transmission device may acquire a destination address in a start session from the reception device and generate a priority.

  Next, Example 6 will be described in detail. In the sixth embodiment, the system of the second embodiment shown in FIG. 6 or the system of the fourth embodiment shown in FIG. 8 is applied, and a system for uploading or downloading data is assumed. In the sixth embodiment, when the priority is not set, or when the same priority is set in a plurality of transmission apparatuses, the transmission rate is adjusted so that the transmission rate is the same regardless of the size of the RTT. It is an example. For example, when the distance between the transmission device and the reception device is long, the RTT is larger than when the distance is short. However, in order to perform data transmission with a transmission amount according to the user's intention, it is necessary to transmit data at the same transmission rate regardless of the size of the RTT. Thus, the sixth embodiment aims to realize data transmission according to the user's intention by a simple method regardless of the size of the RTT without changing the existing network. Hereinafter, a case where priority is not set will be described as an example.

(Configuration of transmission device, configuration of transmission rate determination unit, processing of transmission device)
The configuration of the transmission apparatus 1 according to the sixth embodiment is basically the same as the configuration illustrated in FIG. 2, but the transmission rate determination unit 15 does not input priority. Further, the transmission rate determination unit 15 included in the transmission apparatus 1 according to the sixth embodiment does not include the parameter setting unit 21 illustrated in FIG. 3, and includes a path state determination unit 20, a CWND calculation unit 22, and a packet transmission amount determination unit. 23. The processing of the transmission apparatus 1 according to the sixth embodiment is basically the same as the processing illustrated in FIG. 4, but the CWND maintenance and CWND reduction processing in step S405 and the CWND increase processing in step S406 are performed. Different from Examples 1-5.

  With reference to FIG. 3, the CWND calculation unit 22 according to the sixth embodiment inputs a determination result (CWND maintenance, CWND decrease or CWND increase) from the path state determination unit 20, and inputs an RTT from the RTT measurement unit 12. CWND is calculated by the following formula, and CWND is output to the packet transmission amount determining means 23.

When the determination result is CWND maintenance or CWND decrease, the CWND calculation unit 22 calculates CWND by the following equation (step S405 in FIG. 4).
CWND = (RTT / RTTorg) × CWND (8)
Here, RRTorg is a preset value that serves as a reference for RTT.

  Here, since CWND is calculated for each RTT that receives an ACK, when RTT is large, the update rate of CWND is slow, and when RTT is small, the update rate of CWND is fast. According to the sixth embodiment, CWND is calculated using the equation (8) considering RTT. Therefore, when RTT is large, the amount of decrease in CWND can be suppressed to a large value. Further, when RTT is small, the amount of decrease in CWND can be increased, and CWND can be made smaller than CWND when RTT is large. Therefore, considering that CWND is increased after being decreased, when RTT is large, the update frequency when CWND is increased is less than when RTT is small, and the amount of increase is reduced, so CWND is decreased. By suppressing the amount of decrease, the same CWND as when the RTT is small can be obtained at an early stage when increasing the CWND.

On the other hand, when the determination result is an increase in CWND, the CWND calculation unit 22 calculates CWND by the following equation (step S406 in FIG. 4).
CWND = CWND + (RTT / RTTorg) × (1 / CWND)
... (9)

  As described above, CWND is calculated for each RTT that receives an ACK. Therefore, when RTT is large, the update rate of CWND is slow, and when RTT is small, the update rate of CWND is fast. Therefore, when the determination result is an increase in CWND, the CWND calculation means 22 can set an increase rate equivalent to the CWND increase rate at the RTRT as a reference regardless of the RTT. This is because CWND is updated (T / RTT) times in a predetermined period T. Therefore, if the priority is the same, the increase amount of CWND is (RTT / RTTorg) × (1 / CWND) × (T / RTT). = (1 / RTTorg) × (1 / CWND) × T, and the increase in CWND is constant without depending on the value of RTT. That is, as shown in FIG. 11, when the RTT is large, the CWND update rate has been slow since the CWND update rate has been slow. Also, when the RTT is small, conventionally, the CWND update rate is fast, so the CWND increase rate is fast. According to the sixth embodiment, since CWND is calculated using Equation (9) considering RTT, the increase rate of CWND can be matched to the increase rate of RTTorg according to the size of RTT.

  As described above, according to the transmission device 1 of the sixth embodiment, when performing CWND maintenance and CWND decrease processing, when calculating CWND using Equation (8) considering RTT and performing CWND increase processing Since CWND is calculated using equation (9) considering RTT, the increase rate of CWND can be matched with the increase rate of RTTT as a reference according to the size of RTT. Therefore, regardless of the size of RTT, the transmission rate can be adjusted so that the average value (predetermined packet transmission amount) of the packet transmission amount in a predetermined time is equal. That is, data transmission according to the user's intention can be realized by a simple method regardless of the size of the RTT without being affected by the RTT.

  As the hardware configuration of the transmission apparatus 1 according to the first to sixth embodiments of the present invention, a normal computer can be used. The transmission device 1 is configured by a computer including a volatile storage medium such as a CPU and a RAM, a non-volatile storage medium such as a ROM, an interface, and the like. The functions of the transmission unit 10, the reception unit 11, the RTT measurement unit 12, the packet loss detection unit 13, the RWND detection unit 14, and the transmission rate determination unit 15 included in the transmission device 1 of the first to fourth and sixth embodiments are as follows. Each is realized by causing the CPU to execute a program describing the function. Further, the transmission unit 10, the reception unit 11, the RTT measurement unit 12, the packet loss detection unit 13, the RWND detection unit 14, the transmission rate determination unit 15, the priority holding unit 16, and the priority included in the transmission device 1 of the fifth embodiment. Each function of the degree generation unit 17 is also realized by causing the CPU to execute a program describing these functions. These programs can be stored and distributed in a storage medium such as a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, or the like.

  The present invention has been described with reference to the embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the technical idea thereof. For example, in the above-described embodiment, the New Reno method has been described as an example, but the present invention is not limited to the New Reno method, and can be applied to other methods.

DESCRIPTION OF SYMBOLS 1 Transmission apparatus 2 Reception apparatus 3 Network path | route 10 Transmission part 11 Reception part 12 RTT measurement part 13 Packet loss detection part 14 RWND detection part 15 Transmission rate determination part 16 Priority holding part 17 Priority generation part 20 Path | route state determination means 21 Parameter Setting means 22 CWND calculating means 23 Packet transmission amount determining means

Claims (7)

In the transmission device that transmits the data packet to the reception device while controlling the transmission rate according to the congestion state of the network, and receives a response to the data packet from the reception device,
A receiver for receiving a response to the transmitted data packet;
An RTT measurement unit that measures an RTT (Round Trip Time) based on a transmission time of the data packet and a reception time of a response received by the reception unit;
Based on the response received by the receiving unit and the RTT measured by the RTT measuring unit , the congestion state of the network is determined, and based on the congestion state, the priority, and the RTT measured by the RTT measuring unit. A transmission rate determining unit for determining a transmission rate for avoiding the congestion,
The transmission rate determination unit
When decreasing the transmission rate, when the priority is high, the transmission rate is determined so that the amount of decrease in the transmission rate is smaller than when the priority is low, and the priority is low Determines the transmission rate so that the amount of decrease in the transmission rate is greater than when the priority is high,
When increasing the transmission rate, when the priority is high, the transmission rate is determined so that the increase amount of the transmission rate is larger than when the priority is low, and the priority is low Determines the transmission rate so that the increase in the transmission rate is smaller than when the priority is high, and
When reducing the transmission rate, when the RTT is large, the transmission rate is determined so that the amount of decrease in the transmission rate is smaller than when the RTT is small, and when the RTT is small, Determining the transmission rate such that the amount of decrease in the transmission rate is greater than when the RTT is large;
In increasing the transmission rate, the case RTT is large, the RTT Ri name large increase of the transmission rate than small, if the RTT is small, the transmission than the RTT is large Ri rate increase is small, furthermore, the when the priority is the same as the increase of the transmission rate irrespective of the RTT becomes constant at a predetermined period, to determine the transmission rate, and wherein the Transmitting device. The transmission apparatus according to claim 1 ,
And a packet loss detection unit that detects occurrence of packet loss based on an acknowledgment number included in the response received by the reception unit,
The transmission rate determination unit
When the occurrence of a packet loss is detected by the packet loss detection unit, or when the occurrence of the packet loss is not detected and the RTT measured by the RTT measurement unit is increasing, the path of the network A path state determination unit that determines that congestion has occurred, the occurrence of the packet loss has not been detected, and the RTT has decreased, determines that no congestion has occurred in the network path;
When it is determined by the route state determination means that congestion has occurred in the route of the network, the amount of decrease is smaller than the previously calculated data amount, and when the priority is high, the amount of decrease is smaller than when it is low. Thus, when the priority is low, the amount of decrease is larger than when it is high, and when the RTT is large, the amount of decrease is smaller than when it is small. When the new data amount is calculated so that the decrease amount increases, and it is determined that the network path is not congested, the new data amount is increased more than the previously calculated data amount, and the priority is When the priority is low, the increase amount is larger than when it is low. When the priority is low, the increase amount is smaller than when it is high. When the RTT is large, the increase amount is small. Kiyori even Ri is Na large increase, Ri RTT a smaller increase than when the time small is large, further, the when the priority is the same, the increase being constant in a predetermined time period regardless of the RTT as such, the data amount calculation means for calculating an amount of new data,
The new data amount calculated by the data amount calculation means is compared with the maximum acceptable data amount included in the response received by the receiving unit, and the new data amount is equal to or less than the maximum acceptable data amount. A packet transmission amount determining means for determining the new data amount as a packet transmission amount, and determining the maximum acceptable data amount as a packet transmission amount when the new data amount is larger than the maximum acceptable data amount. And comprising
A transmission apparatus, wherein a transmission rate is controlled by a packet transmission amount determined by the packet transmission amount determination means. The transmission device according to claim 1 or 2 ,
Furthermore, a priority holding unit that holds a destination address of a receiving device to which the data packet is transmitted and a priority corresponding to the destination address;
A priority generation unit that inputs a destination address, reads a priority corresponding to the destination address from the priority holding unit, and outputs the priority to the transmission rate determination unit;
The transmission apparatus, wherein the transmission rate determination unit receives the priority output by the priority generation unit and determines the transmission rate. The transmission device according to claim 2,
The transmission rate determination unit suppresses an increase in the transmission rate when it is determined that the RTT has increased before detecting occurrence of a packet loss as a result of increasing the transmission rate. apparatus. The transmission apparatus according to claim 1 ,
In addition , a retention packet number calculation unit,
The stay packet count calculation unit stays in the network based on the RTT measured by the RTT measurement unit and the amount of data packets transmitted at the transmission rate determined by the transmission rate determination unit. Calculate the number of staying packets,
The transmission rate determining unit, based on the number of accumulated packets calculated by the response and the staying packet number calculating unit received by the receiving unit, determines the congestion state of the network, it sends device according to claim. In a transmission method for transmitting a data packet to a receiving device while controlling a transmission rate according to a congestion state of the network,
Sending a data packet;
Receiving a response to the transmitted data packet;
Measuring RTT (Round Trip Time) based on the transmission time of the transmitted data packet and the reception time of the received response;
Determining a congestion state of the network based on the received response and the measured RTT ;
Determining a transmission rate for avoiding the congestion based on the determined congestion state , priority, and the measured RTT , and
Determining the transmission rate comprises:
When decreasing the transmission rate, when the priority is high, the transmission rate is determined so that the amount of decrease in the transmission rate is smaller than when the priority is low, and the priority is low Determines the transmission rate so that the amount of decrease in the transmission rate is greater than when the priority is high,
When increasing the transmission rate, when the priority is high, the transmission rate is determined so that the increase amount of the transmission rate is larger than when the priority is low, and the priority is low Determines the transmission rate so that the increase in the transmission rate is smaller than when the priority is high , and
When reducing the transmission rate, when the RTT is large, the transmission rate is determined so that the amount of decrease in the transmission rate is smaller than when the RTT is small, and when the RTT is small, Determining the transmission rate such that the amount of decrease in the transmission rate is greater than when the RTT is large;
When increasing the transmission rate, if the RTT is large, the amount of increase in the transmission rate is larger than when the RTT is small, and if the RTT is small, the transmission rate is larger than when the RTT is large. The transmission rate is determined so that the increase rate of the transmission rate is constant in a predetermined period regardless of the RTT when the increase amount of the transmission rate is small and the priority is the same. Method. The transmission program for functioning a computer as a transmission device as described in any one of Claim 1-5 .

Images