WO2016149866A1 - 端对端路径中可用带宽测量方法及装置 - Google Patents

端对端路径中可用带宽测量方法及装置 Download PDF

Info

Publication number
WO2016149866A1
WO2016149866A1 PCT/CN2015/074709 CN2015074709W WO2016149866A1 WO 2016149866 A1 WO2016149866 A1 WO 2016149866A1 CN 2015074709 W CN2015074709 W CN 2015074709W WO 2016149866 A1 WO2016149866 A1 WO 2016149866A1
Authority
WO
WIPO (PCT)
Prior art keywords
rate
delay
ratio
path
probe
Prior art date
Application number
PCT/CN2015/074709
Other languages
English (en)
French (fr)
Inventor
林蓉平
贺行遒
王晟
邹成钢
丁玉军
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to PCT/CN2015/074709 priority Critical patent/WO2016149866A1/zh
Priority to CN201580073719.0A priority patent/CN107210936B/zh
Publication of WO2016149866A1 publication Critical patent/WO2016149866A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00Arrangements for monitoring or testing data switching networks

Definitions

  • the present invention relates to communication technologies, and in particular, to a method and apparatus for measuring available bandwidth in an end-to-end path. Background technique
  • Bandwidth is an important resource in the Internet. It refers to the total amount of data that a link or network path can transmit in a unit of time.
  • the available bandwidth of the network path is a key parameter of Quality of Service (QoS) management, traffic engineering, and congestion control.
  • QoS Quality of Service
  • the end-to-end available bandwidth measurement can be divided into active measurement and passive measurement according to whether or not the test data packet is sent during the measurement process.
  • the active measurement is to actively generate the detection traffic at the end-to-end path sending end, inject the detection traffic into the path, and analyze the available bandwidth of the path according to the transmission condition of the detection flow;
  • a measurement device is installed on the link or device (router, switch, etc.) to monitor the traffic of the path, and the available bandwidth of the path is obtained according to the monitored data. In comparison, the measurement of active measurements is more accurate and requires no additional measurement equipment.
  • a rate-based measurement algorithm requires a higher detection packet transmission rate, which causes a network comparison.
  • measurement algorithm based on inter-packet interval assumes that the bottleneck link (the link with the smallest available bandwidth) and the narrow link (the link with the smallest capacity) in the end-to-end path are the same link, the assumption of the network environment Too harsh; the probability-based measurement algorithm can calculate the available bandwidth by sending a low-rate probe packet. Therefore, the probability-based measurement algorithm is often used in the prior art to calculate the available bandwidth.
  • the probability-based measurement algorithm used in the prior art mainly has an ABEST (Available Bandwidth ESTimator) algorithm.
  • ABEST Automatic Bandwidth ESTimator
  • a certain number of probe packets are transmitted at a certain rate in each round of measurement, assuming that the delay is minimum.
  • the probe packet does not experience the queuing delay when passing the entire path, and then assumes that the probe packets with delays close to the minimum delay do not experience the queuing delay, and assume that the delays of these probe packets are identical, that is, equal to the transmission time.
  • Delay The sum of the propagation delay and the processing delay. However, in the actual process, the delay is introduced in the transmission process of each probe packet. The difference between the assumptions in the prior art and the actual situation of the network is large, resulting in a large error in the measurement result. . Summary of the invention
  • the embodiment of the invention provides a method and a device for measuring the available bandwidth in the end-to-end path, which are used to solve the problem that the available bandwidth measured in the prior art and the actual available bandwidth are too large.
  • a first aspect of the embodiments of the present invention provides a method for measuring an available bandwidth in an end-to-end path, where the end-to-end path includes a transmitting end device and a receiving end device, and the method includes:
  • the ratio of the path utilization corresponding to the first rate to the path utilization corresponding to the second rate is greater than the ratio of the path utilization corresponding to the second rate to the difference of the path utilization corresponding to the third rate as the first ratio, a ratio of a difference between the first rate and the second rate to a difference between the second rate and the third rate as a second ratio, and comparing the first ratio to the second ratio Size, get the comparison result;
  • the determining, by using the comparison result, a preset delay threshold value to obtain a target threshold value includes:
  • the preset threshold is adjusted to a large value to obtain a target threshold.
  • the determining, by using the comparison result, a preset delay threshold value to obtain a target Threshold values including:
  • the preset threshold value is adjusted according to the comparison result by a preset length to obtain a target threshold value.
  • the preset length is equal to The path utilization rate corresponding to the third rate, (for the path utilization rate corresponding to the second rate, ⁇ is the utilization rate corresponding to the first rate, and the third is the third rate, The second rate, the 1 is the first rate, and the parameter is selected for the length.
  • the fourth aspect in the first aspect the obtaining, by the probe packet that does not have a queuing delay, the available bandwidth of the end-to-end path, including:
  • the obtaining, by the probe packet that is not queuing delay, obtaining a final path of the end-to-end path Utilization including:
  • the final path utilization rate the number of probe packets that have not been queued delay / the total number of probe packets received by the receiver device.
  • the method further includes:
  • the canceling a clock drift in a sending delay of each of the probe packets includes: according to a sending rate And grouping the probe packets received by the receiving end, and sorting all the probe packets at each rate according to a transmission delay from small to large, and obtaining a delay curve of all the probe packets at each rate; The delay profile of all probe packets, the clock drift value is estimated; the transmission delay of each probe packet is subtracted from the clock drift value, as each of the probes The transmission delay after the packet is updated.
  • the method further includes:
  • a second aspect of the embodiments of the present invention provides an apparatus for measuring bandwidth available in an end-to-end path, where the end-to-end path includes a transmitting end device and a receiving end device, and the apparatus includes:
  • a delay acquiring module configured to acquire a sending delay of each probe packet received by the receiving device, where the probe packet is sent by the sending device at different rates
  • a calculation module configured to calculate a path utilization rate corresponding to different rates according to a preset delay threshold value and a detection packet sent at different rates;
  • a comparison module configured to use, as a ratio of a path utilization ratio corresponding to the first rate to a path utilization ratio corresponding to the second rate, a ratio of a path utilization ratio corresponding to the second rate to a path utilization ratio corresponding to the third rate a first ratio, comparing a difference between the first rate and the second rate to a ratio of a difference between the second rate and the third rate as a second ratio, and comparing the first ratio with the Describe the size of the second ratio, and obtain a comparison result;
  • an adjustment module configured to adjust a size of the preset delay threshold according to the comparison result, to obtain a target threshold, and delay a delay of all probe packets received by the receiving device by the target threshold The value of the probe packet as a probe packet that has not been queued for delay;
  • An available bandwidth obtaining module is configured to obtain an available bandwidth of the end-end path according to the probe packet that has not been queued.
  • the adjusting module is specifically configured to adjust the preset threshold to a small value if the first ratio is greater than a second ratio Obtaining a target threshold; if the first ratio is less than the second ratio, then the preset threshold is adjusted to a major Integrity, get the target threshold.
  • the adjusting module is specifically configured to adjust a preset threshold value according to the comparison result. Set the length to get the target threshold.
  • the preset length is equal to The path utilization corresponding to the third rate, (the path utilization rate corresponding to the second rate, the utilization rate corresponding to the first rate, and the third is the third rate, Referring to the second rate, the 1 is the first rate, and the parameter is selected for the length.
  • the fourth aspect in the second aspect is configured to acquire a final path utilization rate of the end-to-end path according to the probe packet that has not been queued; and obtain the The available bandwidth of the end-to-end path.
  • the available bandwidth acquiring module acquires the end-to-end according to the probe packet that has not been queued delay
  • the device further includes: a clock drift cancellation module
  • the clock drift elimination module is configured to eliminate clock drift in a transmission delay of each of the probe packets.
  • the clock drift cancellation module is specifically configured to: use the detection packet received by the receiving end according to a sending rate. Packets, all the probe packets at each rate are sorted according to the transmission delay from small to large, and the delay curves of all the probe packets at each rate are obtained; according to the delay curves of all the probe packets at each rate, the delay curve is estimated. The clock drift value is obtained by subtracting the clock drift value from the transmission delay of each probe packet as the update delay of each of the probe packets.
  • the device is further included Includes:
  • the preset module is configured to group the probe packets received by the receiving end according to the sending rate, and sort all the probe packets at each rate according to the sending delay from small to large, and obtain all the probe packets at each rate. a delay curve; obtaining an inflection point in a delay curve of all the detection packets at each rate; and obtaining an initial preset delay threshold according to the inflection point in the delay curve.
  • the device further includes:
  • a third aspect of the embodiments of the present invention provides an available bandwidth measurement apparatus in an end-to-end path, where the end-to-end path includes a sender device and a receiver device, a memory, and a processor, where the memory is used to store an instruction.
  • the processor calls an instruction in the memory;
  • the processor is configured to acquire a sending delay of each probe packet received by the receiving device, where the probe packet is sent by the sending device at different rates; according to a preset delay threshold And the probe packet sent at different rates calculates a path utilization rate corresponding to the different rate; the path utilization corresponding to the first rate corresponds to the path utilization ratio corresponding to the second rate, and the path utilization ratio corresponding to the second rate and the third a ratio of a difference in path utilization corresponding to the rate as a first ratio, and a ratio of a difference between the first rate and the second rate to a difference between the second rate and the third rate is used as a first a second ratio, and comparing the size of the first ratio with the second ratio, obtaining a comparison result; adjusting a size of the preset delay threshold according to the comparison result to obtain a target threshold, a probe packet whose delay is less than the target threshold in all probe packets received by the receiving device as a probe packet that has not undergone queuing delay; according to the probe packet that has not been queued
  • the processor is configured to adjust the preset threshold to a small value if the first ratio is greater than a second ratio And acquiring a target threshold; if the first ratio is less than the second ratio, adjusting the preset threshold to a large value to obtain a target threshold.
  • the processor is configured to adjust a preset threshold value according to the comparison result. Set the length to get the target threshold.
  • the preset length is equal to ⁇ >- r 2 ⁇ r i , where is the path utilization corresponding to the third rate, (for the path utilization corresponding to the second rate, ⁇ is The utilization rate corresponding to the first rate, the third is the third rate, the second rate, the ⁇ i is the first rate, and the parameter is selected for the length.
  • the processor is configured to acquire the end according to the detection packet that is not queuing delay Final path utilization of the peer path; obtaining the available bandwidth of the end-to-end path according to the final path utilization.
  • the final path utilization rate the number of the detection packets that are not queuing delay, the receiving The total number of probe packets received by the end device.
  • the processor is further configured to: after acquiring a transmission delay of each probe packet received by the receiving device, The clock drift in the transmission delay of each probe packet.
  • the processor is configured to group the detection packets received by the receiving end according to a sending rate, All the probe packets at each rate are sorted according to the transmission delay from small to large, and the delay curves of all the probe packets at each rate are obtained.
  • the clock drift is estimated according to the delay curves of all the probe packets at each rate. The value is obtained by subtracting the clock drift value from the transmission delay of each probe packet as the transmission delay after the update of each probe packet.
  • the processor is further configured to group the detection packets received by the receiving end according to a sending rate, and perform all the detections at each rate.
  • the packets are sorted according to the transmission delay from small to large, and the delay curves of all the detection packets at each rate are obtained; the inflection points in the delay curves of all the detection packets at each rate are obtained; according to the inflection points in the delay curve Get the initial preset delay threshold.
  • the processor is further configured to calculate a different rate corresponding to the probe packet sent according to the preset delay threshold value and the different rate After the path utilization, if the difference between the path utilization rates corresponding to the different rates is greater than The preset threshold is used to discard the transmission delay of each probe packet received by the receiving device.
  • the method and device for measuring the available bandwidth in the end-to-end path acquires the transmission delay of each probe packet received by the receiver device, where the probe packet is sent by the sender device at different rates, according to different
  • the rate detection probe packet calculates the path utilization rate corresponding to the different rate, and adjusts the preset delay threshold value by the ratio of the difference between the path utilization rate and the transmission rate, so that the preset time is adjusted.
  • the delay threshold is close to the actual delay, and then the probe packet without the queue delay is determined, and the available bandwidth is calculated so that the available bandwidth obtained is closest to the real available bandwidth.
  • FIG. 1 is a schematic flowchart of Embodiment 1 of a method for measuring available bandwidth in an end-to-end path provided by the present invention
  • Embodiment 2 is a schematic flowchart of Embodiment 2 of a method for measuring available bandwidth in an end-to-end path provided by the present invention
  • Embodiment 3 is a schematic flowchart of Embodiment 3 of a method for measuring available bandwidth in an end-to-end path provided by the present invention
  • FIG. 5 is a schematic flowchart of Embodiment 4 of a method for measuring available bandwidth in an end-to-end path provided by the present invention
  • 6 is an experimental data diagram obtained by using the available bandwidth measurement method in the end-to-end path provided by the present invention.
  • FIG. 7 is a schematic structural diagram of an end-to-end path provided by the present invention.
  • FIG. 8 is a schematic structural diagram of Embodiment 1 of an available bandwidth measuring apparatus in an end-to-end path provided by the present invention.
  • FIG. 9 is a schematic structural diagram of Embodiment 2 of an available bandwidth measurement apparatus in an end-to-end path provided by the present invention
  • 10 is a schematic structural diagram of Embodiment 3 of an available bandwidth measurement apparatus in an end-to-end path provided by the present invention
  • Embodiment 4 is a schematic structural diagram of Embodiment 4 of an available bandwidth measuring apparatus in an end-to-end path provided by the present invention
  • FIG. 12 is a schematic structural diagram of Embodiment 5 of an available bandwidth measuring apparatus in an end-to-end path provided by the present invention. Detailed ways
  • an end-to-end path includes a transmitting end device and a receiving end device, and an execution body of the method may be a transmitting end device. It can also be a receiving device.
  • the method includes:
  • the probe package satisfies the Poisson process.
  • the sending end may randomly generate a preset number of different sending rates, send a certain number of detecting packets at different sending rates, and collect the sending delay of each detecting packet at the receiving end, and the execution body of the method may The transmission delay to each probe packet is obtained at the receiving end device.
  • a delay threshold is preset, that is, the delay experienced by the probe packet that has not undergone the queuing delay from the sender device to the receiver device is the preset delay threshold.
  • the delay threshold calculates the path utilization rate corresponding to different rates.
  • the ratio of the path utilization corresponding to the first rate to the path utilization corresponding to the second rate is greater than the difference between the path utilization corresponding to the second rate and the path utilization corresponding to the third rate.
  • a first ratio comparing a difference between the first rate and the second rate to a ratio of the difference between the second rate and the third rate as a second ratio, and comparing the sizes of the first ratio and the second ratio, and obtaining a comparison result .
  • the path utilization mentioned above takes into account the path utilization after the delay in the transmission of the probe packet.
  • the target threshold is approximate to the true transmission delay of the probe packet that has not undergone the queue delay. It should be noted that, in order to make the obtained target threshold value closest to the real transmission delay of the probe packet that has not been queued, the S 103 and S 104 may be repeatedly executed, that is, each preset may be adjusted. After the threshold is extended, the first preset ratio and the second ratio are returned by using the new preset delay threshold value, and the size is compared, and the preset delay threshold is adjusted according to the comparison size until the preset delay gate is used. The limit is stable, and the preset delay threshold is closest to the true transmission delay of the probe packet that has not been queued.
  • the utilization of the i-th hop link can be expressed as: C ' , where r is the transmission rate of the probe packet, and e ' is the total bandwidth of the ith hop link.
  • C ' the utilization of the i-th hop link
  • r the transmission rate of the probe packet
  • e ' the total bandwidth of the ith hop link.
  • the path utilization of the i-th hop link and the first-order approximation of the transmission rate of the probe packet The relationship can be expressed as: "W ⁇ mi n a, ⁇ + W, where a and b are constants.
  • is the sum of transmission delay, propagation delay and processing delay; the other part is the queuing delay, ⁇ this part is unknown.
  • CDF Distribution Function
  • the probe packet that has not been queued in the actual sense is sent from the probe packet.
  • the delay experienced by the probe packet that has not undergone queuing delay from the sender device to the receiver device is not a fixed value.
  • the probe packet that is statistically unqueued may be from the sender device to the probe device. The delay experienced by the receiving device.
  • the preset delay threshold value ⁇ is continuously adjusted, so that the target threshold value obtained by approaching ⁇ is very close. In order to further calculate the available bandwidth obtained closer to the real available bandwidth.
  • the sending delay of each probe packet received by the receiving device is obtained, where the probe packet is sent by the sending device at different rates, and the path utilization rate corresponding to the different rate is calculated according to the detecting packets sent at different rates respectively. And adjusting the preset delay threshold by the ratio of the difference between the path utilization rate and the transmission rate, so that the preset delay threshold is close to the actual delay, thereby determining that the queue is not queued.
  • the probe packet is extended and the available bandwidth is calculated so that the available bandwidth obtained is closest to the true available bandwidth.
  • the size of the preset delay threshold is adjusted according to the comparison result to obtain a target threshold, which may be specifically:
  • the preset threshold is adjusted to be small to obtain Target threshold; or,
  • the preset threshold is adjusted to be large to obtain the target threshold.
  • the detection packets sent at three rates are selected, and the three rates are: , r 2 , r 3 , where ⁇ r 2 ⁇ r 3 , and the path utilization rates corresponding to the three rates are respectively recorded as: , « (r 2 ), «(r 3 ), calculate the path utilization corresponding to the three rates with the current preset delay threshold t as follows:
  • the description indicates that the current preset delay threshold is ⁇ , and the preset t is adjusted large; in the process of adjusting t, the comparison and the size of the ⁇ can be repeated, that is, each can be adjusted After a t value,
  • the size of the preset threshold is adjusted according to the comparison result to obtain the target threshold. Specifically, the preset threshold is adjusted according to the comparison result to obtain the target threshold. It can be seen that, if the first ratio is greater than the second ratio, the preset threshold is adjusted to be smaller than the preset length; if the first ratio is smaller than the second ratio, the preset threshold is increased. Adjust the preset length. In the specific implementation process, considering * 2 ) - is a monotonically increasing function, if u(r 3 ) - u(r 2 ) r 3 - r 2
  • the setting is larger; if -M ⁇ (the difference between the two is smaller, the difference between ⁇ and ⁇ is smaller, the preset length should be set smaller.
  • the preset length is equal
  • the path utilization rate corresponding to the third rate, (for the path utilization rate corresponding to the second rate, ⁇ ⁇ (for the utilization rate corresponding to the first rate, the ⁇ is the third
  • the rate is 2, the second is the second rate, the ⁇ is the first rate, ⁇ is a parameter for selecting a length, and ⁇ can be set according to a bandwidth of a specific application scenario, which is not limited herein.
  • the foregoing obtaining the end-to-end path according to the detection packet that has not undergone the queuing delay The bandwidth is specifically as follows: The final path utilization of the end-to-end path is obtained according to the probe packet that has not been queued, and the available bandwidth of the end-to-end path is obtained according to the final path utilization.
  • the path utilization rate corresponding to the different rates calculated according to the preset delay threshold has a certain gap from the real value.
  • the target preset delay threshold is obtained, and then The path utilization is calculated again, and the final path utilization obtained by the calculation is close to the real path utilization.
  • the final path utilization the number of probe packets that have not passed the queue delay / the total number of probe packets received by the sink device, "/" indicates the division number.
  • the probe packet with no queue delay here is obtained based on the adjusted target delay threshold.
  • Embodiment 2 is a schematic flowchart of Embodiment 2 of a method for measuring available bandwidth in an end-to-end path provided by the present invention, where the method includes:
  • S20 acquires a transmission delay of each probe packet received by the receiving device.
  • the probe packet is sent by the sending device at different rates.
  • the transmission time of the first probe packets sent by the sender sends the end timestamp is a timestamp at the reception side ⁇ 2, experience ⁇ . After the time, the sender's timestamp increases and the receiver's timestamp increases.
  • the queuing delay experienced by the nth probe packet sent by the sender device is recorded as the time interval between the transmission time of the nth probe packet and the transmission timing of the first probe packet, and the nth The error in the transmission delay experienced by the probe packet is recorded as .
  • the transmission delay measured at the receiver can be expressed as: + - + Lor '
  • the transmission delay measured at the receiver can be expressed as: + ⁇ +t -tt +t and so on.
  • the transmission delay measured at the receiving end can be expressed as: + f delay + — — K , That is: - + H + +C where - is the clock's difference, which is the clock drift.
  • the ratio of the path utilization corresponding to the first rate to the path utilization corresponding to the second rate is greater than the ratio of the path utilization corresponding to the second rate to the path utilization corresponding to the third rate as the first ratio. And comparing a difference between the first rate and the second rate to a ratio of the difference between the second rate and the third rate as a second ratio, and comparing the sizes of the first ratio and the second ratio to obtain a comparison result.
  • S205 Adjust a preset delay threshold value according to the comparison result to obtain a target threshold, and use the detection packet with a delay less than the target threshold in all the detection packets received by the receiving device as the unqueued delay.
  • the target threshold is approximate to the true transmission delay of the probe packet that has not been queued.
  • FIG. 3 is a schematic flowchart of a third embodiment of the method for measuring the available bandwidth in the end-to-end path provided by the present invention, where the clock drift in the transmission delay of each of the foregoing detection packets is eliminated, including:
  • FIG. 4 is a graph of a transmission delay experienced by a probe packet at the same transmission rate, where the abscissa of the delay curve is the sequence number of the probe packet, and the ordinate is the transmission delay experienced by the probe packet.
  • the transmission delay curve is an example. It can be seen that the rise at the beginning is more obvious, the middle is gentle, and the upward trend is more obvious.
  • the most gradual point in the delay curve corresponding to the first rate sent by the transmitting end is marked as ⁇ , and the most gradual point in the delay curve corresponding to the last rate sent by the transmitting end is recorded as ⁇ , :
  • FIG. 5 is a schematic flowchart of Embodiment 4 of a method for measuring available bandwidth in an end-to-end path provided by the present invention.
  • an initial pre-algorithm close to the true value is selected. It is very important to set the delay threshold.
  • the following methods can be used:
  • the delay curve generally starts at a horizontal state and then rises at a certain point.
  • the delay corresponding to the horizontal line is the delay experienced by the probe packet without the queue delay.
  • the abscissa corresponding to the inflection point at the end of the ending segment shall be equal to the total number of packets sent by the transmitting device multiplied by 1 and then subtracted.
  • the inflection point of the delay curve in the real environment may not be too obvious.
  • the point with the slope of k is found as the inflection point in the section where the tail is rising, and k is the preset value, which can be based on the real environment. Various factors are used to determine the k value, which is not limited herein.
  • the path utilization rate corresponding to the different rate is calculated according to the preset delay threshold value and the probe packet sent at different rates, if the path rate corresponding to the different rate is between If the difference is greater than the preset threshold, the transmission delay of each probe packet received by the receiving device is discarded. That is, if the difference between the path utilization rates corresponding to different rates is too large, it is considered that there is a problem with the input of the algorithm, or there is a problem that the introduction error is too large, and the data is discarded as "bad" data, and then reset. Enter the data and re-measure the available bandwidth.
  • the preset threshold may be 10% of the total bandwidth of the minimum link in the end-to-end path, but is not limited thereto, and may be set according to a specific application environment.
  • 6 is an experimental data obtained by using the available bandwidth measurement method in the end-to-end path provided by the present invention, and the total bandwidth of the bottleneck link of the end-to-end path is set to 100 M, and the transmitting end device has 40 different to the receiving end device.
  • the rate of available bandwidth obtained by using the available bandwidth measurement method in the end-to-end path provided by the embodiment of the present invention is as shown in FIG. 6.
  • the estimated value in FIG. 6 is the available bandwidth obtained by the experiment, and the true value is true.
  • the available bandwidth as can be seen from FIG. 6, is that the available bandwidth obtained by the available bandwidth measurement method in the end-to-end path provided by the embodiment of the present invention is very close to the true value.
  • FIG. 7 is a schematic structural diagram of an end-to-end path provided by the present invention.
  • the end-to-end path may include: a probe packet sending end device, a background service sending end device, a first router, a second router, and a receiving end. device.
  • Figure 7 shows two routers as an example, but not limited to this.
  • the sender device described in other parts of the text refers to the probe packet sender device.
  • FIG. 8 is a schematic structural diagram of Embodiment 1 of an available bandwidth measurement apparatus in an end-to-end path provided by the present invention, where the end-to-end path includes a transmitting end device and a receiving end device, and the device may be integrated on the transmitting end device or the receiving end.
  • the device includes: a delay acquisition module 801, a calculation module 802, a comparison module 803, an adjustment module 804, and an available bandwidth acquisition.
  • Module 805 wherein:
  • the delay obtaining module 801 is configured to obtain a sending delay of each probe packet received by the receiving device, where the probe packet is sent by the sending device at different rates.
  • the calculation module 802 is configured to calculate path utilization rates corresponding to different rates according to preset delay thresholds and probe packets sent at different rates.
  • the comparison module 803 is configured to compare a difference between a path utilization rate corresponding to the first rate and a path utilization ratio corresponding to the second rate, and a ratio of a path utilization corresponding to the second rate to a difference between the path utilization corresponding to the third rate. As a first ratio, comparing a difference between the first rate and the second rate to a ratio of a difference between the second rate and the third rate as a second ratio, and comparing the first ratio with The size of the second ratio is obtained as a comparison result.
  • the adjusting module 804 is configured to adjust the size of the preset delay threshold according to the comparison result, to obtain a target threshold, and delay a delay of all the probe packets received by the receiving device by the target gate.
  • the probe packet of the limit is used as a probe packet without a queued delay.
  • the available bandwidth acquisition module 805 is configured to obtain the available bandwidth of the end-to-end path according to the probe packet that has not been queued.
  • the device is used to perform the foregoing method embodiments, and the technical effects and implementation principles thereof are similar, and are not described herein again.
  • the adjusting module 804 is specifically configured to: when the first ratio is greater than the second ratio, adjust the preset threshold to a small value to obtain a target threshold; if the first ratio is smaller than the second The ratio is adjusted to a large threshold to obtain a target threshold.
  • the adjustment module 804 is specifically configured to adjust a preset threshold value according to the comparison result by a preset length to obtain a target threshold.
  • the preset length is equal to The path utilization rate corresponding to the third rate, the path utilization rate corresponding to the second rate, the utilization rate corresponding to the first rate, and the third is the third rate,
  • the ⁇ is the first rate, and ⁇ is the length of the selection parameter.
  • the available bandwidth acquisition module 805 is configured to obtain the final path utilization rate of the end-to-end path according to the detection packet that has not been queued; and obtain the end according to the final path utilization rate. The available bandwidth of the peer path.
  • the available bandwidth acquisition module 805 obtains the final path utilization rate of the end-to-end path according to the probe packet that has not been queued, specifically: the final path utilization rate: when the queue is not queued.
  • FIG. 9 is a schematic structural diagram of Embodiment 2 of an available bandwidth measurement apparatus in an end-to-end path according to the present invention.
  • the apparatus further includes: a clock drift elimination module 901, configured to eliminate each of the detection packets. The clock drift in the transmit delay.
  • the clock drift elimination module 901 is specifically configured to group the probe packets received by the receiver according to the transmission rate, and sort all the probe packets at each rate according to the transmission delay from small to large, and obtain all the probe packets at each rate. a delay curve; estimating a clock drift value according to a delay curve of all the probe packets at each rate; subtracting the clock drift value from the transmission delay of each probe packet as each of the probes The transmission delay after the packet is updated.
  • FIG. 10 is a schematic structural diagram of Embodiment 3 of an available bandwidth measurement apparatus in an end-to-end path provided by the present invention.
  • the apparatus further includes: a preset module 110, configured to send the receiving end according to a sending rate.
  • the received probe packet group, and all the probe packets at each rate are sorted according to the transmission delay from small to large, and the delay curves of all the probe packets at each rate are obtained; when all the probe packets at each rate are obtained An inflection point in the curve; an initial preset delay threshold is obtained according to the inflection point in the delay curve.
  • FIG. 11 is a schematic structural diagram of Embodiment 4 of an available bandwidth measurement apparatus in an end-to-end path provided by the present invention.
  • the apparatus further includes: a discarding module 111, configured to perform a threshold according to a preset delay And after the probe packets sent at different rates are used to calculate the path utilization rate corresponding to the different rates, if the difference between the path utilization rates corresponding to the different rates is greater than a preset threshold, each of the received devices is discarded. The transmission delay of the probe packet.
  • Embodiment 5 is a schematic structural diagram of Embodiment 5 of an available bandwidth measurement apparatus in an end-to-end path provided by the present invention, where the end-to-end path includes a transmitting end device and a receiving end device, and the device can be integrated.
  • the device includes: a memory 121 and a processor 122, wherein: the memory 121 is configured to store an instruction, and the processor 122 calls an instruction in the memory to execute:
  • the device is used to perform the foregoing method embodiments, and the technical effects and implementation principles thereof are similar, and are not described herein again.
  • the processor 122 is configured to: when the first ratio is greater than the second ratio, adjust the preset threshold to a small target to obtain a target threshold; if the first ratio is smaller than the second ratio, The preset threshold is adjusted to a large value to obtain a target threshold.
  • the processor 122 adjusts the preset threshold value by a preset length according to the comparison result to obtain a target threshold value.
  • the preset length is equal to The path utilization rate corresponding to the third rate, the path utilization rate corresponding to the second rate, the utilization rate corresponding to the first rate, the ⁇ is the third rate, ⁇ for the second rate, The i is the first rate, and ⁇ is a length selection parameter.
  • the processor 122 is configured to obtain a final path utilization rate of the end-to-end path according to the probe packet that has not been queued; and obtain an available bandwidth of the end-to-end path according to the final path utilization.
  • the final path utilization rate the number of probe packets that have not been queued delay / the total number of probe packets received by the receiver device.
  • the processor 122 is further configured to cancel the clock drift in the transmission delay of each of the probe packets after acquiring the transmission delay of each probe packet received by the receiver device.
  • the processor 122 is configured to: cancel the clock drift in the transmission delay of each of the probe packets, specifically: group the probe packets received by the receiver according to the transmission rate, and send all the probe packets at each rate according to the transmission rate.
  • the delay is from small to large, and the delay curve of all the probe packets at each rate is obtained; the clock drift value is estimated according to the delay curve of all the probe packets at each rate; the transmission delay of each probe packet is The clock drift value is subtracted as the updated transmission delay of each of the probe packets.
  • the processor 122 is further configured to group the probe packets received by the receiving end according to the sending rate, and sort all the probe packets at each rate according to the sending delay from small to large, and obtain the time of all the probe packets at each rate. Obtaining an inflection point in a delay curve of all the detection packets at each rate; and obtaining an initial preset delay threshold according to the inflection point in the delay curve.
  • the processor 122 is further configured to: after calculating path utilization rates corresponding to different rates according to the probe packets sent according to the preset delay threshold and the different rates, if the difference between the path utilization rates corresponding to the different rates If the threshold is greater than the preset threshold, the transmission delay of each probe packet received by the receiving device is discarded.
  • the foregoing steps include the steps of the foregoing method embodiments; and the foregoing storage medium includes: a ROM, a RAM, a magnetic disk, or an optical disk, and the like, which can store various program codes. Quality.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

本发明提供一种端对端路径中可用带宽测量方法及装置,该方法根据不同速率发送的探测包计算不同速率对应的路径利用率,并通过路径利用率之间的差值与发送速率之间的差值的比值调整预设时延门限值,以使预设时延门限接近实际时延,进而确定未经过排队时延的探测包,并计算可用带宽,以使获取的可用带宽接近真实的可用带宽。

Description

端对端路径中可用带宽测量方法及装置 技术领域
本发明涉及通信技术, 尤其涉及一种端对端路径中可用带宽测量方法及 装置。 背景技术
带宽是互联网 (Internet) 中的一种重要资源, 指代一条链路或者网络路 径在单位时间内能够传输的数据总量。 网络路径的可用带宽是服务质量保证 (Quality of Service, 简称 QoS ) 管理、 流量工程和拥塞控制的关键参数。 目 前, 端对端可用带宽的测量可根据测量过程中有无测试数据包的发送, 分为 主动测量和被动测量。 其中, 主动测量是在端对端的路径发送端有目的地主 动产生探测流量, 将探测流量注入到路径中, 并根据该探测流量的传送情况 来分析路径的可用带宽; 被动测量时在路径上的链路或设备 (路由器、 交换 机等) 上安装测量设备来监测路径的流量, 根据监测的数据分析获得路径的 可用带宽。 相较而言, 主动测量的测量结果更为准确, 且无需额外的测量设 备。
目前, 上述主动测量的算法主要有三类: 基于速率的测量算法、 基于包 间间隔的测量算法以及基于概率的测量算法, 其中, 基于速率的测量算法需 要较高地探测包发送速率, 会给网络造成较大的影响; 基于包间间隔的测量 算法假设端到端路径中的瓶颈链路 (可用带宽最小的链路) 和窄链路 (容量 最小的链路) 是同一条链路, 对网络环境的假设过于苛刻; 而基于概率的测 量算法可以通过发送速率很低的探测包来计算可用带宽, 因此, 现有技术中 常采用基于概率的测量算法来计算可用带宽。
现有技术中使用的基于概率的测量算法主要有 ABEST ( Available Bandwidth ESTimator)算法, 采用 ABEST算法获取可用带宽时, 在每一轮测 量中以一定速率发送一定数目的探测包, 假设其中时延最小的探测包在经过 整个路径时都没有经历排队时延, 然后假设时延接近最小时延的探测包也都 没有经历排队时延, 并假设这些探测包的时延完全相同, 即都等于传输时延、 传播时延和处理时延的和, 但是, 实际过程中, 每个探测包的发送过程中都 会引入时延, 现有技术中的假设与网络实际情况差距较大, 导致测量结果的 误差很大。 发明内容
本发明实施例提供一种端对端路径中可用带宽测量方法及装置, 用于解 决现有技术中测量出的可用带宽与实际可用带宽误差过大的问题。
本发明实施例第一方面提供一种端对端路径中可用带宽测量方法, 所述 端对端路径中包括发送端设备和接收端设备, 该方法包括:
获取所述接收端设备所接收的每一个探测包的发送时延, 其中, 所述探 测包由所述发送端设备以不同速率发出;
根据预设时延门限值以及不同速率下发送的探测包计算不同速率对应的 路径利用率;
将第一速率对应的路径利用率与第二速率对应的路径利用率的差值比第 二速率对应的路径利用率与第三速率对应的路径利用率的差值的比值作为第 一比值, 将所述第一速率与所述第二速率的差值比所述第二速率与所述第三 速率的差值的比值作为第二比值, 并比较所述第一比值与所述第二比值的大 小, 获取比较结果;
根据所述比较结果调整所述预设时延门限值的大小, 以获取目标门限值, 将所述接收端设备接收的所有探测包中时延小于所述目标门限值的探测包作 为未经过排队时延的探测包;
根据所述未经过排队时延的探测包获取所述端对端路径的可用带宽。 结合第一方面, 在第一方面的第一种可能的实施方式中, 所述根据所述 比较结果调整预设时延门限值的大小, 以获取目标门限值, 包括:
若所述第一比值大于第二比值, 则将所述预设门限值向小调整, 获取目 标门限值;
若所述第一比值小于第二比值, 则将所述预设门限值向大调整, 获取目 标门限值。
结合第一方面的第一种可能的实施方式, 在第一方面的第二种可能的实 施方式中, 所述根据所述比较结果调整预设时延门限值的大小, 以获取目标 门限值, 包括:
根据所述比较结果将预设门限值的大小调整预设歩长, 以获取目标门限 值。
结合第一方面的第二种可能的实施方式, 在第一方面的第三种可能的实
施方式中, 所述预设歩长等于
Figure imgf000005_0001
, 其中, 为所述第 三速率对应的路径利用率, ( 为所述第二速率对应的路径利用率, ^为 所述第一速率对应的利用率, 所述 3为所述第三速率, 所述 为所述第二 速率, 所述 1为所述第一速率, 《为歩长选取参数。 结合第一方面至第一方面的第三种可能的实施方式, 在第一方面的第四 种可能的实施方式中, 所述根据所述未经过排队时延的探测包获取所述端对 端路径的可用带宽, 包括:
根据所述未经过排队时延的探测包获取所述端对端路径的最终路径利用 率;
根据所述最终路径利用率获取所述端对端路径的可用带宽。
结合第一方面的第四种可能的实施方式, 在第一方面的第五种可能的实 施方式中, 所述根据所述未经过排队时延的探测包获取所述端对端路径的最 终路径利用率, 包括:
所述最终路径利用率 =所述未经过排队时延的探测包的数量 /所述接收端 设备所接收的探测包总数。
结合第一方面, 在第一方面的第六种可能的实施方式中, 所述获取所述 接收端设备所接收的每一个探测包的发送时延之后, 还包括:
消除所述每一个探测包的发送时延中的时钟漂移。
结合第一方面的第六种可能的实施方式, 在第一方面的第七种可能的实 施方式中, 所述消除所述每一个探测包的发送时延中的时钟漂移, 包括: 按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有 探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 根据所述每个速率下所有探测包的时延曲线, 估算出时钟漂移值; 将每一个探测包的发送时延都减去所述时钟漂移值, 作为所述每一个探 测包更新后的发送时延。
结合第一方面, 在第一方面的第八种可能的实施方式中, 所述方法还包 括:
按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有 探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 获取所述每个速率下所有探测包的时延曲线中的拐点;
根据所述时延曲线中的拐点获取初始预设时延门限值。
结合第一方面, 在第一方面的第九种可能的实施方式中, 所述分别根据 不同速率发送的探测包计算不同速率对应的路径利用率之后, 还包括:
若所述不同速率对应的路径利用率之间的差值大于预设阈值, 则丢弃所 述接收端设备所接收的每一个探测包的发送时延。
本发明实施例第二方面提供一种端对端路径中可用带宽测量装置, 所述 端对端路径中包括发送端设备和接收端设备, 所述装置包括:
时延获取模块, 用于获取所述接收端设备所接收的每一个探测包的发送 时延, 其中, 所述探测包由所述发送端设备以不同速率发出;
计算模块, 用于根据预设时延门限值以及不同速率下发送的探测包计算 不同速率对应的路径利用率;
比较模块, 用于将第一速率对应的路径利用率与第二速率对应的路径利 用率的差值比第二速率对应的路径利用率与第三速率对应的路径利用率的差 值的比值作为第一比值, 将所述第一速率与所述第二速率的差值比所述第二 速率与所述第三速率的差值的比值作为第二比值, 并比较所述第一比值与所 述第二比值的大小, 获取比较结果;
调整模块, 用于根据所述比较结果调整所述预设时延门限值的大小, 以 获取目标门限值, 将所述接收端设备接收的所有探测包中时延小于所述目标 门限值的探测包作为未经过排队时延的探测包;
可用带宽获取模块,用于根据所述未经过排队时延的探测包获取所述端 对端路径的可用带宽。
结合第二方面,在第二方面的第一种可能的实施方式中,所述调整模块, 具体用于若所述第一比值大于第二比值, 则将所述预设门限值向小调整, 获 取目标门限值; 若所述第一比值小于第二比值, 则将所述预设门限值向大调 整, 获取目标门限值。
结合第二方面的第一种可能的实施方式, 在第二方面的第二种可能的实 施方式中, 所述调整模块, 具体用于根据所述比较结果将预设门限值的大小 调整预设歩长, 以获取目标门限值。
结合第二方面的第二种可能的实施方式, 在第二方面的第三种可能的实
施方式中, 所述预设歩长等于
Figure imgf000007_0001
, 其中, 为所述第 三速率对应的路径利用率, ( 为所述第二速率对应的路径利用率, 为 所述第一速率对应的利用率, 所述 3为所述第三速率, 所述 为所述第二 速率, 所述 1为所述第一速率, 《为歩长选取参数。 结合第二方面至第二方面的第三种可能的实施方式, 在第二方面的第四 种可能的实施方式中,所述可用带宽获取模块, 具体用于根据所述未经过排 队时延的探测包获取所述端对端路径的最终路径利用率; 根据所述最终路径 利用率获取所述端对端路径的可用带宽。
结合第二方面的第四种可能的实施方式, 在第二方面的第五种可能的实 施方式中,所述可用带宽获取模块根据所述未经过排队时延的探测包获取所 述端对端路径的最终路径利用率, 具体为: 所述最终路径利用率 =所述未经过 排队时延的探测包的数量 /所述接收端设备所接收的探测包总数。
结合第二方面, 在第二方面的第六种可能的实施方式中, 所述装置还包 括: 时钟漂移消除模块;
所述时钟漂移消除模块,用于消除所述每一个探测包的发送时延中的时 钟漂移。
结合第二方面的第六种可能的实施方式, 在第二方面的第七种可能的实 施方式中, 所述时钟漂移消除模块, 具体用于按照发送速率将所述接收端所 接收的探测包分组, 将每个速率下的所有探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 根据所述每个速率下所有探测包的 时延曲线, 估算出时钟漂移值; 将每一个探测包的发送时延都减去所述时钟 漂移值, 作为所述每一个探测包更新后的发送时延。
结合第二方面, 在第二方面的第八种可能的实施方式中, 所述装置还包 括:
预设模块;
所述预设模块,具体用于按照发送速率将所述接收端所接收的探测包分 组, 将每个速率下的所有探测包按发送时延从小到大排序, 获取每个速率下 所有探测包的时延曲线; 获取所述每个速率下所有探测包的时延曲线中的拐 点; 根据所述时延曲线中的拐点获取初始预设时延门限值。
结合第二方面, 在第二方面的第九种可能的实施方式中, 所述装置还包 括:
本发明实施例第三方面提供一种端对端路径中可用带宽测量装置, 所述 端对端路径中包括发送端设备和接收端设备, 存储器和处理器, 所述存储器 用于存储指令, 所述处理器调用所述存储器中的指令;
所述处理器, 用于获取所述接收端设备所接收的每一个探测包的发送时 延, 其中, 所述探测包由所述发送端设备以不同速率发出; 根据预设时延门 限值以及不同速率下发送的探测包计算不同速率对应的路径利用率; 将第一 速率对应的路径利用率与第二速率对应的路径利用率的差值比第二速率对应 的路径利用率与第三速率对应的路径利用率的差值的比值作为第一比值, 将 所述第一速率与所述第二速率的差值比所述第二速率与所述第三速率的差值 的比值作为第二比值, 并比较所述第一比值与所述第二比值的大小, 获取比 较结果; 根据所述比较结果调整所述预设时延门限值的大小, 以获取目标门 限值, 将所述接收端设备接收的所有探测包中时延小于所述目标门限值的探 测包作为未经过排队时延的探测包; 根据所述未经过排队时延的探测包获取 所述端对端路径的可用带宽。
结合第三方面, 在第三方面的第一种可能的实施方式中, 所述处理器, 具体用于若所述第一比值大于第二比值, 则将所述预设门限值向小调整, 获 取目标门限值; 若所述第一比值小于第二比值, 则将所述预设门限值向大调 整, 获取目标门限值。
结合第三方面的第一种可能的实施方式, 在第三方面的第二种可能的实 施方式中, 所述处理器, 具体用于根据所述比较结果将预设门限值的大小调 整预设歩长, 以获取目标门限值。
结合第三方面的第二种可能的实施方式, 在第三方面的第三种可能的实 施方式中, 所述预设歩长等于 ^>— r2 ~ ri , 其中, 为所述第 三速率对应的路径利用率, ( 为所述第二速率对应的路径利用率, ^为 所述第一速率对应的利用率, 所述 3为所述第三速率, 所述 为所述第二 速率, 所述^ i为所述第一速率, 《为歩长选取参数。 结合第三方面至第三方面的第三种可能的实施方式, 在第三方面的第四 种可能的实施方式中, 所述处理器, 具体用于根据所述未经过排队时延的探 测包获取所述端对端路径的最终路径利用率; 根据所述最终路径利用率获取 所述端对端路径的可用带宽。
结合第三方面的第四种可能的实施方式, 在第三方面的第五种可能的实 施方式中, 所述最终路径利用率 =所述未经过排队时延的探测包的数量 /所述 接收端设备所接收的探测包总数。
结合第三方面, 在第三方面的第六种可能的实施方式中, 所述处理器, 还用于在获取所述接收端设备所接收的每一个探测包的发送时延之后, 消除 所述每一个探测包的发送时延中的时钟漂移。
结合第三方面的第六种可能的实施方式, 在第三方面的第七种可能的实 施方式中, 所述处理器, 具体用于按照发送速率将所述接收端所接收的探测 包分组, 将每个速率下的所有探测包按发送时延从小到大排序, 获取每个速 率下所有探测包的时延曲线; 根据所述每个速率下所有探测包的时延曲线, 估算出时钟漂移值; 将每一个探测包的发送时延都减去所述时钟漂移值, 作 为所述每一个探测包更新后的发送时延。
结合第三方面, 在第三方面的第八种可能的实施方式中, 所述处理器, 还用于按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所 有探测包按发送时延从小到大排序,获取每个速率下所有探测包的时延曲线; 获取所述每个速率下所有探测包的时延曲线中的拐点; 根据所述时延曲线中 的拐点获取初始预设时延门限值。
结合第三方面, 在第三方面的第九种可能的实施方式中, 所述处理器, 还用于在在根据预设时延门限值以及不同速率下发送的探测包计算不同速率 对应的路径利用率之后, 若所述不同速率对应的路径利用率之间的差值大于 预设阈值, 则丢弃所述接收端设备所接收的每一个探测包的发送时延。
本发明实施例提供的端对端路径中可用带宽测量方法及装置, 获取接收 端设备所接收的每一个探测包的发送时延, 其中, 探测包由发送端设备以不 同速率发出, 分别根据不同速率发送的探测包计算不同速率对应的路径利用 率, 并通过路径利用率之间的差值与发送速率之间的差值的比值, 来调整预 设时延门限值, 以使预设时延门限接近实际时延, 进而确定未经过排队时延 的探测包, 并计算可用带宽, 以使获取的可用带宽最接近真实的可用带宽。 附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案, 下面将对 实施例或现有技术描述中所需要使用的附图作一简单地介绍, 显而易见 地, 下面描述中的附图是本发明的一些实施例, 对于本领域普通技术人员 来讲, 在不付出创造性劳动性的前提下, 还可以根据这些附图获得其他的 附图。
图 1 为本发明提供的端对端路径中可用带宽测量方法实施例一的流程示 意图;
图 2为本发明提供的端对端路径中可用带宽测量方法实施例二的流程示 意图;
图 3为本发明提供的端对端路径中可用带宽测量方法实施例三的流程示 意图;
图 4为同一发送速率下的探测包所经历的发送时延曲线图;
图 5为本发明提供的端对端路径中可用带宽测量方法实施例四的流程示 意图;
图 6为采用本发明提供的端对端路径中可用带宽测量方法获取的实验数 据图;
图 7为本发明提供的端对端路径的结构示意图;
图 8为本发明提供的端对端路径中可用带宽测量装置实施例一的结构示 意图;
图 9为本发明提供的端对端路径中可用带宽测量装置实施例二的结构示 意图; 图 10为本发明提供的端对端路径中可用带宽测量装置实施例三的结构示 意图;
图 11为本发明提供的端对端路径中可用带宽测量装置实施例四的结构示 意图;
图 12为本发明提供的端对端路径中可用带宽测量装置实施例五的结构示 意图。 具体实施方式
为使本发明实施例的目的、 技术方案和优点更加清楚, 下面将结合本 发明实施例中的附图, 对本发明实施例中的技术方案进行清楚、 完整地描 述, 显然,所描述的实施例是本发明一部分实施例, 而不是全部的实施例。 基于本发明中的实施例, 本领域普通技术人员在没有作出创造性劳动前提 下所获得的所有其他实施例, 都属于本发明保护的范围。
图 1为本发明提供的端对端路径中可用带宽测量方法实施例一的流程示 意图, 一般地, 端对端路径中包括发送端设备和接收端设备, 该方法的执行 主体可以是发送端设备也可以是接收端设备。
该方法包括:
5101、 获取接收端设备所接收的每一个探测包的发送时延。 其中, 探测 包由发送端设备以不同速率发出。
一般地, 该探测包满足泊松过程。
需要说明的是, 发送端可以随机生成预设个数的不同发送速率, 以不同 的发送速率发送一定数量的探测包,在接收端统计每一个探测包的发送时延, 该方法的执行主体可以在接收端设备处获得到每一个探测包的发送时延。
5102、 根据预设时延门限以及不同速率下发送的探测包计算不同速率对 应的路径利用率。
这里会先预设一个时延门限值, 即假设未经过排队时延的探测包从发送 端设备到接收端设备所经历的时延为该预设时延门限值, 采用该预设时延门 限计算不同速率对应的路径利用率。
5103、 将第一速率对应的路径利用率与第二速率对应的路径利用率的差 值比第二速率对应的路径利用率与第三速率对应的路径利用率的差值的比值 作为第一比值, 将第一速率与第二速率的差值比上述第二速率与第三速率的 差值的比值作为第二比值, 并比较第一比值与第二比值的大小, 获取比较结 果。
上述路径利用率是考虑了探测包传输过程中时延后的路径利用率。
S 104、 根据比较结果调整上述预设时延门限值的大小, 以获取目标门限 值, 将上述接收端设备接收的所有探测包中时延小于目标门限值的探测包作 为未经过排队时延的探测包。
其中,上述目标门限值近似于未经过排队时延的探测包的真实发送时延。 需要说明的是, 为了使得获取到的目标门限值与未经过排队时延的探测 包的真实发送时延最接近, 可以多次重复执行 S 103和 S 104, 即可以每调整 一次预设时延门限值, 就采用新的预设时延门限值返回再计算第一比值和第 二比值, 并比较大小, 根据比较大小再调整预设时延门限值, 直到预设时延 门限值趋于稳定, 此时预设时延门限值最接近未经过排队时延的探测包的真 实发送时延。
S 105、根据上述未经排队时延的探测包获取上述端对端路径的可用带宽。 一般地, 定义端对端路径中第 i跳链路的利用率为: Μ' = 1 + π', 其中 为 该第 i跳链路发送队列为空的概率。 如果发送端设备发送的探测包满足泊松 过程, 由于泊松过程的 PASTA ( Poisson Arrivals See Time Averages ) 特性, r
u(r)≡ min(l, +一 )
第 i跳链路的利用率可以表示为: C' , 其中, r为探测包的 发送速率, e '为第 i跳链路的总带宽。 对于包含多条链路的端对端路径, 假设路径不发生拥塞, 即探测包的发 送过程中不存在时延, 则第 i跳链路的路径利用率与探测包的发送速率的一 阶近似关系可以表示为: "W≡mina,^ + W, 其中, ab都是常数。
进一歩地, 考虑探测包从发送端设备到接收端设备所经历的时延, 假设 探测包从发送端设备到接收端设备所经历的总时延记为 , 包括两部分: 其中一部分是固定值 ^为传输时延、 传播时延和处理时延的总和; 另 一部分是排队时延 , ^这一部分是未知的。
具体实现过程中, 近似地假设发送端设备发送的探测包与背景业务包都 满足排队论中的 MM/1 模型, 则排队时延 的累积分布函数 (Cumulative
P{td < t) = W{t) = \ -—e-(M-X)t
Distribution Function , 简称 CDF ) 为: , 进一
P(td≤t) = P(tt + tq≤t) = P(tq≤t— tt) = l— Ae- " 歩可以得到: , 因此 d ^ , 此处 t为一个变量, 其中, 从发送端设备到接收端设 备所经历的总时延大于 t 的探测包数目占发送端设备所发送的总探测包数目 的比例应接近于 P(¾ > 0。 由路径利用率的定义可知, 路径利用率 M也等于探测包从发送端设备到 接收端设备所经历的时延需要经历排队时延的概率, 具体地, 假设实际意义 上未经过排队时延的探测包从发送端设备到接收端设备所经历的时延为 Λ 理想条件下, 即如果探测包发送过程中本来就不存在排队时延, 那么 但是, 实际情况下, u=P td > 。 需要说明的是, 实际上未经过排队时延 的探测包从发送端设备到接收端设备所经历的时延也并不是固定值, 这里的 可以是统计意义上未经过排队时延的探测包从发送端设备到接收端设备所 经历的时延。
初始状态下的预设时延门限值 ί与 有一定差距,本发明实施例中就是 要通过不断调整预设时延门限值 ί,使 ί接近 Λ即获取的目标门限值与 非 常接近, 以便进一歩计算获取的可用带宽更接近真实的可用带宽。
本实施例中, 获取接收端设备所接收的每一个探测包的发送时延, 其中, 探测包由发送端设备以不同速率发出, 分别根据不同速率发送的探测包计算 不同速率对应的路径利用率, 并通过路径利用率之间的差值与发送速率之间 的差值的比值, 来调整预设时延门限值, 以使预设时延门限接近实际时延, 进而确定未经过排队时延的探测包, 并计算可用带宽, 以使获取的可用带宽 最接近真实的可用带宽。
在上述实施例的基础上, 上述根据上述比较结果调整预设时延门限值的 大小, 以获取目标门限值, 具体可以为:
若上述第一比值大于第二比值, 则将上述预设门限值向小调整, 以获取 目标门限值; 或者,
若上述第一比值小于第二比值, 则将上述预设门限值向大调整, 以获取 目标门限值。
具体地, 选取三种速率下发出的探测包, 记这三种速率为: 、 r2、 r3 , 其中 < r2 < r3,这三种速率对应的路径利用率分别记为: 、 «(r2)、 «(r3), 以当前预设时延门限值 t计算这三种速率对应的路径利用率如下:
由上可推得: u(r3)-u(r2) _ μ μ ― ^ )— 卜
ii( 2)— ύ ι) r2e2 t~t,} -
Figure imgf000014_0001
, 可知, /(ί)为单调递增函数, 因此:
(1) ί=ί'时, Uir^ui^), U{r2)=u{r2), U(r3)=u(r3), 则 u(r3) - u(r2) r3 - r2
f( = -
M(r2)— r2-rx
(2) '时, >^~^ r3~r2
(3) ί< '时, /(0<
~rx 综上, 通过比较 M 3)u^和
^) n的大小, 在 大于
Μ(Γ2) - «( r2一 rx u(r2)一 «(^) r2一 rx 时,说明当前的预设时延门限 ί > ,于是将 t调小;在^^ ^小于^^
M(r2) - r2 - rx 时, 说明说明当前的预设时延门限 < , 预设将 t调大; 可以在调整 t的 过程中,重复比较 和 ϋ的大小,即每可以调整一次 t值之后,
M(r2) - r2 - rx
再进行比较, 不断地进行修正, 直到 t趋于稳定, 说明此时 t值接近 ί', 取 当前值作为目标门限。
进一歩地, 根据上述比较结果调整预设门限值的大小, 以获取目标门 限值, 具体为: 根据上述比较结果将预设门限的大小调整预设歩长, 以获 取目标门限值。 由上可知, 若上述第一比值大于第二比值, 则将上述预设门 限值向小调整预设歩长; 若上述第一比值小于第二比值, 则将上述预设门限 值向大调整预设歩长。 具体实施过程中, 考虑到 * 2 )— 为单调递增函数, 如果 u(r3) - u(r2) r3 - r2
M~(r2) - ^^与^ - 的差值较大, 说明 与 '也相差较远, 则预设歩长应该 u(r3) - u(r2) r3 - r2
设置的较大; 如果 — M~( 与 的差值较小, 说明 ί与 也相差较小, 则预设歩长应该设置的较小。 基于上述内容, 一种实施例中, 预设歩长等于
Figure imgf000015_0001
, 其中, 为所述第三速率对应的路径利用率, ( 为所述第二速率对应的 路径利用率, Μ~( 为所述第一速率对应的利用率, 所述 ^为所述第三速率, 所述 2为所述第二速率, 所述^ i为所述第一速率, ^为歩长选取参数, ^可 以根据具体应用场景的带宽进行设置, 在此不作限制。
另一实施例中,上述根据未经过排队时延的探测包获取端对端路径的可 用带宽, 具体为: 根据未经过排队时延的探测包获取所述端对端路径的最终 路径利用率, 并根据该最终路径利用率获取上述端对端路径的可用带宽。
需要说明的是, 之前根据预设时延门限值所计算的不同速率对应的路径 利用率跟真实值有一定差距, 在不断调整的过程中, 获取了目标预设时延门 限值, 然后再计算路径利用率, 此时计算获取的最终路径利用率接近真实路 径利用率。
一种实施例中, 最终路径利用率=未经过排队时延的探测包的数量 /接收 端设备所接收的探测包总数, "/"表示除号。 此处的未经过排队时延的探测 包, 是根据调整后的目标时延门限值获取的。
图 2为本发明提供的端对端路径中可用带宽测量方法实施例二的流程示 意图, 该方法包括:
S20 获取接收端设备所接收的每一个探测包的发送时延。 其中, 探测 包由发送端设备以不同速率发出。
S202、 消除上述每一个探测包的发送时延中的时钟漂移。
端对端路径中发送端设备和接收端设备之间可能存在时钟漂移, 即经过 同样一段时间, 发送端设备与接收端设备上时间戳的增量不完全相同。
具体地, 将发送端设备发送的第一个探测包的发送时刻发送端的时间戳 记为 接收端的时间戳记为 ί2, 经历 ί。时间后, 发送端时间戳增加 , 接收端时间戳增加 。, 将发送端设备发送的第 η 个探测包所经历的排队 时延记为 , 将第 η个探测包的发送时刻与第一个探测包的发送时刻的 时间间隔记为 ^, 将第 n个探测包所经历的发送时延中的误差记为 。 则对于第一个探测包, 在接收端测量到的发送时延可以表示为: + - + Lor ' 对于第二个探测包, 在接收端测量到的发送时延可以 表示为: + ^ +t -t t +t 以此类推, 对于第 n个探测包, 在接收端测量到的发送时延可以表示为: + fdelay + — — K , 也即: - + H + +C 其中, - 为时钟同歩差值, 为时钟漂移。
S203、 根据预设时延门限以及不同速率下发送的探测包计算不同速率对 应的路径利用率。
S204、 将第一速率对应的路径利用率与第二速率对应的路径利用率的差 值比第二速率对应的路径利用率与第三速率对应的路径利用率的差值的比值 作为第一比值, 将第一速率与第二速率的差值比上述第二速率与第三速率的 差值的比值作为第二比值, 并比较第一比值与第二比值的大小, 获取比较结 果。
S205、 根据比较结果调整预设时延门限值的大小, 以获取目标门限值, 将上述接收端设备接收的所有探测包中时延小于目标门限值的探测包作为未 经过排队时延的探测包。 其中, 上述目标门限值近似于未经过排队时延的探 测包的真实发送时延。
S206、根据上述未经排队时延的探测包获取上述端对端路径的可用带宽。 图 3为本发明提供的端对端路径中可用带宽测量方法实施例三的流程示 意图, 上述消除上述每一个探测包的发送时延中的时钟漂移, 包括:
5301、 按照发送速率将所述接收端所接收的探测包分组, 将每个速率下 的所有探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延 曲线。
参照图 4, 图 4为同一发送速率下的探测包所经历的发送时延曲线图, 该时延曲线的横坐标为探测包的排列序号, 纵坐标为探测包所经历的发送时 延, 以该发送时延曲线图为例, 可以看出: 开始处一段上升较为明显, 中间 一段平缓, 结尾一段上升趋势更为明显。
5302、根据所述每个速率下所有探测包的时延曲线,估算出时钟漂移值。 上述发送时延曲线中中间平缓部分的时延最接近未经过排队时延的探测 包所经历的发送时延,即^ = 0。时延曲线中最平缓的点近似于误差平均值, 即该点 = 0且 ί = (ί )。
将发送端所发送的第一个速率对应的时延曲线中最平缓的点记为^ ^, 将发送端所发送的最后一个速率对应的时延曲线中最平缓的点记为 ί^, 则:
tfirst + (φ - φ、ί= + E( terro
ast
Figure imgf000018_0001
+ (φ - φ)ίι + Ε( terro
进一歩地, 可推到得:
W , 即可以得到时
Figure imgf000018_0002
钟漂移( - )。
一般认为时延曲线中最平缓的点体现误差平均值。
S303、 将每一个探测包的发送时延都减去所述时钟漂移值, 作为所述每 一个探测包更新后的发送时延。
图 5为本发明提供的端对端路径中可用带宽测量方法实施例四的流程示 意图, 为了避免预设时延门限值的调整过程中出现大的变动, 选择一个与真 实值接近的初始预设时延门限值非常重要, 具体确定初始预设时延门限值可 以采用下述方法:
5501、 按照发送速率将所述接收端所接收的探测包分组, 将每个速率下 的所有探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延 曲线。 具体可参照图 4所示的曲线。
5502、 获取所述每个速率下所有探测包的时延曲线中的拐点。
理想状态下, 时延曲线一开始一般都呈一段水平状态, 然后会在某一点 上升, 这一段水平线对应的时延为未经过排队时延的探测包所经历的时延。 结尾段上升处拐点对应的横坐标应等于发送端设备发送的总包数乘以 1再减 去路径利用率, 真实环境下时延曲线的拐点有可能不会太明显, 一般在尾部 上升趋势明显的一段找到斜率为 k的点作为拐点, 其中 k为预设值, 可以根 据真实环境下的各种因素来确定 k值, 在此不作限制。
S503、 根据所述时延曲线中的拐点获取初始预设时延门限值。 找到该拐 点后, 即将该拐点对应的时延作为初始的预设时延门限值。
需要说明的是, 在上述实施例的基础上, 根据预设时延门限值以及不同 速率下发送的探测包计算不同速率对应的路径利用率之后, 若上述不同速率 对应的路径利用率之间的差值大于预设阈值, 则丢弃上述接收端设备所接收 的每一个探测包的发送时延。 即如果不同速率对应的路径利用率之间的差值 过大, 则认为算法的输入出现问题, 或存在引入误差过大等问题, 将这一组 数据为 "坏"数据丢弃, 然后重新设定输入数据, 进而重新进行可用带宽的 测量。 其中, 预设阈值可以取上述端对端路径中最小链路总带宽的 10%, 但 并不以此为限, 可以根据具体应用环境设置。 图 6为采用本发明提供的端对端路径中可用带宽测量方法获取的实验数 据图,将端对端路径的瓶颈链路的总带宽设为 100M,发送端设备向接收端设 备以 40个不同速率发送 8000个探测包, 采用本发明实施例提供的端对端路 径中可用带宽测量方法得到的可用带宽如图 6所示, 图 6中的估计值为实验 获得的可用带宽, 真实值为真实的可用带宽, 从图 6可以看出, 采用本发明 实施例提供的端对端路径中可用带宽测量方法得到的可用带宽与真实值非常 接近。
图 7为本发明提供的端对端路径的结构示意图, 如图 7所示, 端对端路 径可以包括: 探测包发送端设备、 背景业务发送端设备、 第一路由器、 第二 路由器以及接收端设备。 图 7中以包含两个路由器为例, 但并不以此为限。 文中其它部分记载的发送端设备指的是探测包发送端设备。
图 8为本发明提供的端对端路径中可用带宽测量装置实施例一的结构示 意图, 所述端对端路径中包括发送端设备和接收端设备, 该装置可以集成在 发送端设备或接收端设备中的其中一个, 如图 8所示, 该装置包括: 时延获 取模块 801、 计算模块 802、 比较模块 803、 调整模块 804以及可用带宽获取 模块 805, 其中:
时延获取模块 801, 用于获取所述接收端设备所接收的每一个探测包的 发送时延, 其中, 所述探测包由所述发送端设备以不同速率发出。
计算模块 802, 用于根据预设时延门限值以及不同速率下发送的探测包 计算不同速率对应的路径利用率。
比较模块 803, 用于将第一速率对应的路径利用率与第二速率对应的路 径利用率的差值比第二速率对应的路径利用率与第三速率对应的路径利用率 的差值的比值作为第一比值, 将所述第一速率与所述第二速率的差值比所述 第二速率与所述第三速率的差值的比值作为第二比值, 并比较所述第一比值 与所述第二比值的大小, 获取比较结果。
调整模块 804, 用于根据所述比较结果调整所述预设时延门限值的大小, 以获取目标门限值, 将所述接收端设备接收的所有探测包中时延小于所述目 标门限值的探测包作为未经过排队时延的探测包。
可用带宽获取模块 805, 用于根据所述未经过排队时延的探测包获取所 述端对端路径的可用带宽。
该装置用于执行前述方法实施例, 其技术效果和实现原理类似, 在此不 再赘述。
进一歩地, 调整模块 804, 具体用于若所述第一比值大于第二比值, 则 将所述预设门限值向小调整, 获取目标门限值; 若所述第一比值小于第二比 值, 则将所述预设门限值向大调整, 获取目标门限值。
更具体地, 调整模块 804, 具体用于根据所述比较结果将预设门限值的 大小调整预设歩长, 以获取目标门限值。 其中, 所述预设歩长等于
Figure imgf000020_0001
, 其中, 为所述第 三速率对应的路径利用率, 为所述第二速率对应的路径利用率, 为 所述第一速率对应的利用率, 所述 3为所述第三速率, 所述 为所述第二 速率, 所述^ i为所述第一速率, ^为歩长选取参数。
可用带宽获取模块 805,具体用于根据所述未经过排队时延的探测包获 取所述端对端路径的最终路径利用率; 根据所述最终路径利用率获取所述端 对端路径的可用带宽。
更具体地, 可用带宽获取模块 805, 根据所述未经过排队时延的探测包 获取所述端对端路径的最终路径利用率, 具体为: 所述最终路径利用率 =所述 未经过排队时延的探测包的数量 /所述接收端设备所接收的探测包总数。
图 9为本发明提供的端对端路径中可用带宽测量装置实施例二的结构示 意图, 在图 8的基础上, 该装置还包括: 时钟漂移消除模块 901, 用于消除 所述每一个探测包的发送时延中的时钟漂移。
时钟漂移消除模块 901, 具体用于按照发送速率将所述接收端所接收的 探测包分组, 将每个速率下的所有探测包按发送时延从小到大排序, 获取每 个速率下所有探测包的时延曲线; 根据所述每个速率下所有探测包的时延曲 线, 估算出时钟漂移值; 将每一个探测包的发送时延都减去所述时钟漂移值, 作为所述每一个探测包更新后的发送时延。
图 10 为本发明提供的端对端路径中可用带宽测量装置实施例三的结构 示意图, 在图 8的基础上, 该装置还包括: 预设模块 110, 用于按照发送速 率将所述接收端所接收的探测包分组, 将每个速率下的所有探测包按发送时 延从小到大排序, 获取每个速率下所有探测包的时延曲线; 获取所述每个速 率下所有探测包的时延曲线中的拐点; 根据所述时延曲线中的拐点获取初始 预设时延门限值。
找到该拐点后, 即将该拐点对应的时延作为初始的预设时延门限值。 图 11 为本发明提供的端对端路径中可用带宽测量装置实施例四的结构 示意图, 在图 8的基础上, 该装置还包括: 丢弃模块 111, 用于在根据预设 时延门限值以及不同速率下发送的探测包计算不同速率对应的路径利用率之 后, 若所述不同速率对应的路径利用率之间的差值大于预设阈值, 则丢弃所 述接收端设备所接收的每一个探测包的发送时延。
图 12 为本发明提供的端对端路径中可用带宽测量装置实施例五的结构 示意图, 所述端对端路径中包括发送端设备和接收端设备, 该装置可以集成 在发送端设备或接收端设备中的其中一个, 该装置包括: 存储器 121和处理 器 122, 其中: 存储器 121, 用于存储指令; 处理器 122, 调用所述存储器中 的指令, 用于执行:
获取所述接收端设备所接收的每一个探测包的发送时延, 其中, 所述探 测包由所述发送端设备以不同速率发出; 根据预设时延门限值以及不同速率 下发送的探测包计算不同速率对应的路径利用率; 将第一速率对应的路径利 用率与第二速率对应的路径利用率的差值比第二速率对应的路径利用率与第 三速率对应的路径利用率的差值的比值作为第一比值, 将所述第一速率与所 述第二速率的差值比所述第二速率与所述第三速率的差值的比值作为第二比 值, 并比较所述第一比值与所述第二比值的大小, 获取比较结果; 根据所述 比较结果调整所述预设时延门限值的大小, 以获取目标门限值, 将所述接收 端设备接收的所有探测包中时延小于所述目标门限值的探测包作为未经过排 队时延的探测包; 根据所述未经过排队时延的探测包获取所述端对端路径的 可用带宽。
该装置用于执行前述方法实施例, 其技术效果和实现原理类似, 在此不 再赘述。
处理器 122, 具体用于若所述第一比值大于第二比值, 则将所述预设门 限值向小调整, 获取目标门限值; 若所述第一比值小于第二比值, 则将所述 预设门限值向大调整, 获取目标门限值。
更具体地, 处理器 122, 根据所述比较结果将预设门限值的大小调整预 设歩长, 以获取目标门限值。 所述预设歩长等于
Figure imgf000022_0001
, 其中, 为所述第三速率 对应的路径利用率, 为所述第二速率对应的路径利用率, 为所述第 一速率对应的利用率, 所述 ^为所述第三速率, 所述 ^为所述第二速率, 所述 i为所述第一速率, ^为歩长选取参数。
处理器 122, 具体用于根据所述未经过排队时延的探测包获取所述端对 端路径的最终路径利用率; 根据所述最终路径利用率获取所述端对端路径的 可用带宽。
所述最终路径利用率 =所述未经过排队时延的探测包的数量 /所述接收端 设备所接收的探测包总数。
处理器 122, 还用于在获取所述接收端设备所接收的每一个探测包的发 送时延之后, 消除所述每一个探测包的发送时延中的时钟漂移。
处理器 122, 消除所述每一个探测包的发送时延中的时钟漂移, 具体为: 按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有探测 包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 根据 所述每个速率下所有探测包的时延曲线, 估算出时钟漂移值; 将每一个探测 包的发送时延都减去所述时钟漂移值, 作为所述每一个探测包更新后的发送 时延。
处理器 122, 还用于按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有探测包按发送时延从小到大排序, 获取每个速率下所有 探测包的时延曲线; 获取所述每个速率下所有探测包的时延曲线中的拐点; 根据所述时延曲线中的拐点获取初始预设时延门限值。
处理器 122, 还用于在在根据预设时延门限值以及不同速率下发送的探 测包计算不同速率对应的路径利用率之后, 若所述不同速率对应的路径利用 率之间的差值大于预设阈值, 则丢弃所述接收端设备所接收的每一个探测包 的发送时延。 本领域普通技术人员可以理解: 实现上述方法实施例的全部或部分歩骤 可以通过程序指令相关的硬件来完成, 前述的程序可以存储于一计算机可读 取存储介质中, 该程序在执行时, 执行包括上述方法实施例的歩骤; 而前述 的存储介质包括: ROM、 RAM,磁碟或者光盘等各种可以存储程序代码的介 质。
最后应说明的是: 以上各实施例仅用以说明本发明的技术方案, 而非对 其限制; 尽管参照前述各实施例对本发明进行了详细的说明, 本领域的普通 技术人员应当理解:其依然可以对前述各实施例所记载的技术方案进行修改, 或者对其中部分或者全部技术特征进行等同替换; 而这些修改或者替换, 并 不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims

权 利 要 求 书
1、一种端对端路径中可用带宽测量方法, 所述端对端路径中包括发送端 设备和接收端设备, 其特征在于, 所述方法包括:
获取所述接收端设备所接收的每一个探测包的发送时延, 其中, 所述探 测包由所述发送端设备以不同速率发出;
根据预设时延门限值以及不同速率下发送的探测包计算不同速率对应的 路径利用率;
将第一速率对应的路径利用率与第二速率对应的路径利用率的差值比第 二速率对应的路径利用率与第三速率对应的路径利用率的差值的比值作为第 一比值, 将所述第一速率与所述第二速率的差值比所述第二速率与所述第三 速率的差值的比值作为第二比值, 并比较所述第一比值与所述第二比值的大 小, 获取比较结果;
根据所述比较结果调整所述预设时延门限值的大小, 以获取目标门限值, 将所述接收端设备接收的所有探测包中时延小于所述目标门限值的探测包作 为未经过排队时延的探测包;
根据所述未经过排队时延的探测包获取所述端对端路径的可用带宽。
2、 根据权利要求 1所述的方法, 其特征在于, 所述根据所述比较结果调 整预设时延门限值的大小, 以获取目标门限值, 包括:
若所述第一比值大于第二比值, 则将所述预设门限值向小调整, 获取目 标门限值;
若所述第一比值小于第二比值, 则将所述预设门限值向大调整, 获取目 标门限值。
3、 根据权利要求 2所述的方法, 其特征在于, 所述根据所述比较结果调 整预设时延门限值的大小, 以获取目标门限值, 包括:
根据所述比较结果将预设门限值的大小调整预设歩长, 以获取目标门限 值。
4、 根据权利要求 3 所述的方法, 其特征在于, 所述预设歩长等于
Figure imgf000025_0001
,其中, W 为所述第三速率对应的路径利用率, 为所述第二速率对应的路径利用率, Μ~( 为所述第一速率对应的利用率, 所述 3为所述第三速率, 所述 为所述第二速率, 所述^ i为所述第一速率, «为歩长选取参数。
5、 根据权利要求 1-4任一项所述的方法, 其特征在于, 所述根据所述未 经过排队时延的探测包获取所述端对端路径的可用带宽, 包括:
根据所述未经过排队时延的探测包获取所述端对端路径的最终路径利用 率;
根据所述最终路径利用率获取所述端对端路径的可用带宽。
6、 根据权利要求 5所述的方法, 其特征在于, 所述根据所述未经过排 队时延的探测包获取所述端对端路径的最终路径利用率, 包括:
所述最终路径利用率 =所述未经过排队时延的探测包的数量 /所述接收端 设备所接收的探测包总数。
7、 根据权利要求 1所述的方法, 其特征在于, 所述获取所述接收端设备 所接收的每一个探测包的发送时延之后, 还包括:
消除所述每一个探测包的发送时延中的时钟漂移。
8、 根据权利要求 7所述的方法, 其特征在于, 所述消除所述每一个探测 包的发送时延中的时钟漂移, 包括:
按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有 探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 根据所述每个速率下所有探测包的时延曲线, 估算出时钟漂移值; 将每一个探测包的发送时延都减去所述时钟漂移值, 作为所述每一个探 测包更新后的发送时延。
9、 根据权利要求 1所述的方法, 其特征在于, 还包括:
按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有 探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 获取所述每个速率下所有探测包的时延曲线中的拐点;
根据所述时延曲线中的拐点获取初始预设时延门限值。
10、 根据权利要求 1所述的方法, 其特征在于, 所述分别根据不同速率 发送的探测包计算不同速率对应的路径利用率之后, 还包括:
若所述不同速率对应的路径利用率之间的差值大于预设阈值, 则丢弃所 述接收端设备所接收的每一个探测包的发送时延。
11、 一种端对端路径中可用带宽测量装置, 所述端对端路径中包括发送 端设备和接收端设备, 其特征在于, 所述装置包括:
时延获取模块, 用于获取所述接收端设备所接收的每一个探测包的发送 时延, 其中, 所述探测包由所述发送端设备以不同速率发出;
计算模块, 用于根据预设时延门限值以及不同速率下发送的探测包计算 不同速率对应的路径利用率;
比较模块, 用于将第一速率对应的路径利用率与第二速率对应的路径利 用率的差值比第二速率对应的路径利用率与第三速率对应的路径利用率的差 值的比值作为第一比值, 将所述第一速率与所述第二速率的差值比所述第二 速率与所述第三速率的差值的比值作为第二比值, 并比较所述第一比值与所 述第二比值的大小, 获取比较结果;
调整模块, 用于根据所述比较结果调整所述预设时延门限值的大小, 以 获取目标门限值, 将所述接收端设备接收的所有探测包中时延小于所述目标 门限值的探测包作为未经过排队时延的探测包;
可用带宽获取模块,用于根据所述未经过排队时延的探测包获取所述端 对端路径的可用带宽。
12、 根据权利要求 11 所述的装置, 其特征在于, 所述调整模块, 具 体用于若所述第一比值大于第二比值, 则将所述预设门限值向小调整, 获取 目标门限值; 若所述第一比值小于第二比值, 则将所述预设门限值向大调整, 获取目标门限值。
13、 根据权利要求 12所述的装置, 其特征在于, 所述调整模块, 具体 用于根据所述比较结果将预设门限值的大小调整预设歩长, 以获取目标门限 值。
14、 根据权利要求 13 所述的装置, 其特征在于, 所述预设歩长等于
Figure imgf000027_0001
,其中, 为所述第三速率对应的路径利用率, 为所述第二速率对应的路径利用率, M~( 为所述第一速率对应的利用率, 所述 ^为所述第三速率, 所述 ^为所述第二速率, 所述^ i为所述第一速率, «为歩长选取参数。
15、 根据权利要求 11-14任一项所述的装置, 其特征在于, 所述可用 带宽获取模块, 具体用于根据所述未经过排队时延的探测包获取所述端对端 路径的最终路径利用率; 根据所述最终路径利用率获取所述端对端路径的可 用带宽。
16、 根据权利要求 15所述的装置, 其特征在于, 所述可用带宽获取 模块根据所述未经过排队时延的探测包获取所述端对端路径的最终路径利用 率, 具体为: 所述最终路径利用率 =所述未经过排队时延的探测包的数量 /所 述接收端设备所接收的探测包总数。
17、 根据权利要求 11所述的装置, 其特征在于, 还包括: 时钟漂移消 除模块;
所述时钟漂移消除模块,用于消除所述每一个探测包的发送时延中的时 钟漂移。
18、 根据权利要求 17所述的装置, 其特征在于, 所述时钟漂移消除模 块, 具体用于按照发送速率将所述接收端所接收的探测包分组, 将每个速率 下的所有探测包按发送时延从小到大排序, 获取每个速率下所有探测包的时 延曲线; 根据所述每个速率下所有探测包的时延曲线, 估算出时钟漂移值; 将每一个探测包的发送时延都减去所述时钟漂移值, 作为所述每一个探测包 更新后的发送时延。
19、 根据权利要求 11所述的装置, 其特征在于, 还包括: 预设模块; 所述预设模块, 用于按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有探测包按发送时延从小到大排序, 获取每个速率下所有 探测包的时延曲线; 获取所述每个速率下所有探测包的时延曲线中的拐点; 根据所述时延曲线中的拐点获取初始预设时延门限值。
20、 根据权利要求 11所述的装置, 其特征在于, 还包括: 丢弃模块; 所述丢弃模块,用于在根据预设时延门限值以及不同速率下发送的探测 包计算不同速率对应的路径利用率之后, 若所述不同速率对应的路径利用率 之间的差值大于预设阈值, 则丢弃所述接收端设备所接收的每一个探测包的 发送时延。
21、 一种端对端路径中可用带宽测量装置, 所述端对端路径中包括发送 端设备和接收端设备, 其特征在于, 所述装置包括: 存储器和处理器, 所述 存储器用于存储指令, 所述处理器调用所述存储器中的指令;
所述处理器, 用于获取所述接收端设备所接收的每一个探测包的发送时 延, 其中, 所述探测包由所述发送端设备以不同速率发出; 根据预设时延门 限值以及不同速率下发送的探测包计算不同速率对应的路径利用率; 将第一 速率对应的路径利用率与第二速率对应的路径利用率的差值比第二速率对应 的路径利用率与第三速率对应的路径利用率的差值的比值作为第一比值, 将 所述第一速率与所述第二速率的差值比所述第二速率与所述第三速率的差值 的比值作为第二比值, 并比较所述第一比值与所述第二比值的大小, 获取比 较结果; 根据所述比较结果调整所述预设时延门限值的大小, 以获取目标门 限值, 将所述接收端设备接收的所有探测包中时延小于所述目标门限值的探 测包作为未经过排队时延的探测包; 根据所述未经过排队时延的探测包获取 所述端对端路径的可用带宽。
22、 根据权利要求 21所述的装置, 其特征在于, 所述处理器, 具体用于 若所述第一比值大于第二比值, 则将所述预设门限值向小调整, 获取目标门 限值; 若所述第一比值小于第二比值, 则将所述预设门限值向大调整, 获取 目标门限值。
23、 根据权利要求 22所述的装置, 其特征在于, 所述处理器, 具体用于 根据所述比较结果将预设门限值的大小调整预设歩长, 以获取目标门限值。
24、 根据权利要求 23 所述的装置, 其特征在于, 所述预设歩长等于
Figure imgf000029_0001
,其中, 为所述第三速率对应的路径利用率, 为所述第二速率对应的路径利用率, 为所述第一速率对应的利用率, 所述 ^为所述第三速率, 所述 ^为所述第二速率, 所述^ i为所述第一速率, «为歩长选取参数。
25、 根据权利要求 21-24任一项所述的装置, 其特征在于, 所述处理器, 具体用于根据所述未经过排队时延的探测包获取所述端对端路径的最终路径 利用率; 根据所述最终路径利用率获取所述端对端路径的可用带宽。
26、 根据权利要求 25所述的装置, 其特征在于, 所述最终路径利用率 = 所述未经过排队时延的探测包的数量 /所述接收端设备所接收的探测包总数。
27、 根据权利要求 21所述的装置, 其特征在于, 所述处理器, 还用于在 获取所述接收端设备所接收的每一个探测包的发送时延之后, 消除所述每一 个探测包的发送时延中的时钟漂移。
28、 根据权利要求 27所述的装置, 其特征在于, 所述处理器, 具体用于 按照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有探测 包按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 根据 所述每个速率下所有探测包的时延曲线, 估算出时钟漂移值; 将每一个探测 包的发送时延都减去所述时钟漂移值, 作为所述每一个探测包更新后的发送 时延。
29、 根据权利要求 21所述的装置, 其特征在于, 所述处理器, 还用于按 照发送速率将所述接收端所接收的探测包分组, 将每个速率下的所有探测包 按发送时延从小到大排序, 获取每个速率下所有探测包的时延曲线; 获取所 述每个速率下所有探测包的时延曲线中的拐点; 根据所述时延曲线中的拐点 获取初始预设时延门限值。
30、 根据权利要求 21所述的装置, 其特征在于, 所述处理器, 还用于在 在根据预设时延门限值以及不同速率下发送的探测包计算不同速率对应的路 径利用率之后,若所述不同速率对应的路径利用率之间的差值大于预设阈值, 则丢弃所述接收端设备所接收的每一个探测包的发送时延。
PCT/CN2015/074709 2015-03-20 2015-03-20 端对端路径中可用带宽测量方法及装置 WO2016149866A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2015/074709 WO2016149866A1 (zh) 2015-03-20 2015-03-20 端对端路径中可用带宽测量方法及装置
CN201580073719.0A CN107210936B (zh) 2015-03-20 2015-03-20 端对端路径中可用带宽测量方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2015/074709 WO2016149866A1 (zh) 2015-03-20 2015-03-20 端对端路径中可用带宽测量方法及装置

Publications (1)

Publication Number Publication Date
WO2016149866A1 true WO2016149866A1 (zh) 2016-09-29

Family

ID=56977859

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/074709 WO2016149866A1 (zh) 2015-03-20 2015-03-20 端对端路径中可用带宽测量方法及装置

Country Status (2)

Country Link
CN (1) CN107210936B (zh)
WO (1) WO2016149866A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101026509A (zh) * 2007-02-28 2007-08-29 西南科技大学 一种端到端低可用带宽测量方法
CN101119240A (zh) * 2007-07-05 2008-02-06 中国科学技术大学 一种基于pgm的有效带宽测量方法
US20140160945A1 (en) * 2012-12-06 2014-06-12 Accedian Networks Inc. Using bandwidth measurements to adjust cir and eir on a sub-rate link

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100499526C (zh) * 2007-03-23 2009-06-10 中山大学 一种端到端排队时延测量方法
CN101222380B (zh) * 2007-12-27 2010-07-07 腾讯科技(深圳)有限公司 可用带宽测量方法、系统与装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101026509A (zh) * 2007-02-28 2007-08-29 西南科技大学 一种端到端低可用带宽测量方法
CN101119240A (zh) * 2007-07-05 2008-02-06 中国科学技术大学 一种基于pgm的有效带宽测量方法
US20140160945A1 (en) * 2012-12-06 2014-06-12 Accedian Networks Inc. Using bandwidth measurements to adjust cir and eir on a sub-rate link

Also Published As

Publication number Publication date
CN107210936B (zh) 2019-12-17
CN107210936A (zh) 2017-09-26

Similar Documents

Publication Publication Date Title
CN107624232B (zh) 在通信系统中控制下行链路吞吐量的设备和方法
US8824281B2 (en) Network-friendly transmission control protocol (TCP) methods, apparatus and articles of manufacture
JP4738148B2 (ja) パケット・ネットワーク上の利用可能な帯域幅を推定する方法および装置
US8004989B2 (en) Controlling a window for congestion control in a network
US8665745B2 (en) Method and device for determining a jitter buffer level
CN104581422B (zh) 一种用于处理网络数据传输的方法和装置
CN104618258B (zh) 一种数据传输速率的控制方法
CN106713166B (zh) 确定拥塞窗口大小的方法及装置
JP2007520957A (ja) パケットベース・ネットワークのエンドツーエンド・パスを特徴付ける方法およびそのための装置
CA2915487C (en) Method for performing a bandwidth test for comminications from a first network station to a second network station of a communication network and corresponding apparatuses for performing the method steps and corresponding computer program products
WO2011006372A1 (zh) 一种端到端可用带宽测量方法及系统
US9246829B2 (en) Utilizing latency control to alleviate bufferbloat
US8345570B2 (en) Network impairment metrics for timing over packet
JP5427707B2 (ja) データ伝送システム及び方法
US10079746B2 (en) System and method for testing data channel bandwidth, packet loss, latency and jitter
US9769695B2 (en) Adaptive quality of service for wide area network transport
CN104579582B (zh) 一种基于通信网络的高质量语音传输方法
KR20110082917A (ko) 통신 시스템에서 혼잡 제어를 수행하는 장치 및 방법
WO2016149866A1 (zh) 端对端路径中可用带宽测量方法及装置
JP5169441B2 (ja) ネットワーク品質測定システム、ネットワーク品質測定方法及びネットワーク品質測定用プログラム
US20210344581A1 (en) Available bandwidth estimation based on packet skipping in one-way delay (owd) based analysis
Chen et al. Congestion-aware mac layer adaptation to improve video teleconferencing over wi-fi
Novak et al. Auto-tuning active queue management
Liu et al. Uplink delay variation compensation in queue length estimation over mobile data networks
Fares et al. A new algorithm for controlling a buffer with time-varying arrival rate

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15885805

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15885805

Country of ref document: EP

Kind code of ref document: A1