CN102055628B - Method of available bandwidth measurement and close link circuit positioning based on single pack column - Google Patents

Abstract

The invention relates to a method for simultaneously measuring available bandwidth and locating tight links based on a single packet sequence, comprising the following steps: a sending end sends a specific detection packet sequence to a destination node; the sending end calculates the detection packet sequence according to the ICMP packet returned by the destination node The extension after each link and the round-trip delay (RTT) measured by each probe packet; then, the sender locates the tight link according to the extension of the packet column, and determines the time to maintain a constant round-trip delay. For the corresponding speed reduction packet, calculate the average sending rate of the packet row when the speed reduction packet is sent as the measurement value of the available bandwidth of the path. The invention can locate the tight link and measure the available bandwidth value by sending a detection packet sequence once, and has the advantages of high measurement precision, fast measurement speed, low intrusion degree, strong robustness and the like.

Description

Simultaneous measurement of available bandwidth and location of tight links based on single-packet column

technical field

The invention relates to the field of network performance measurement, in particular to a method for simultaneously measuring available bandwidth and locating tight links based on a single packet sequence.

Background technique

Available bandwidth refers to the maximum data transfer rate that a link can provide to other streams without affecting the transfer rate of background streams. The link with the smallest available bandwidth in each link is called the tight link of this path (Tight Link). The available bandwidth of the tight link is also the available bandwidth of this path. End-to-end path available bandwidth is an important parameter to dynamically describe the transmission capacity of a network path, and it can effectively evaluate the actual carrying capacity and performance of a network path. Available bandwidth and tight link status reflect network performance status, which is of great significance for monitoring network performance, diagnosing network operation status, and optimizing network performance. Available bandwidth measurement and tight link location are of great significance to applications such as congestion control, routing selection, streaming media applications, dynamic selection of servers, and quality of service (QoS) verification.

According to different detection methods, the current available bandwidth measurement methods can be divided into packet pair method, packet column method and model-based available bandwidth measurement method. Packet Pairs technology (Packet Pairs) uses the change of the time interval (DispersionTime) of the packet pair during transmission to estimate the available bandwidth. Packet technology has derived many algorithms and tools, such as Spruce, IGI, Delphi, LinkPPQ, etc. The measurement of available bandwidth by packet pair technology is based on the assumption that the tight link and the narrow link are the same link. If this assumption is not true, the measurement error may be large. Packet column technology obtains the available bandwidth value by sending one or more columns of probe packets. Its principle is to send data packets through the self-load method to congest the path, and estimate the available bandwidth value by analyzing the one-way delay change of the probe packets in the packet column. Typical methods using package technology include: Pathload, PathChirp, PTR, YAZ, abget, etc. The robustness and stability of measuring the available bandwidth of the package-column technology are relatively strong, but the measurement load is often relatively large. The model-based measurement method is also an important direction in the field of available bandwidth measurement. The basic idea is to simplify the modeling of the complex network, collect path information by sending a small amount of probe flow, and then combine the model to estimate the available bandwidth of the path. This method has no impact on the network itself. The state and background flow have little influence, but when the model cannot accurately describe the flow characteristics, the measurement accuracy will be difficult to guarantee.

Currently, well-known tight link location methods include BFind, DRPS, STAB, and Pathneck. The BFind method is to cause network congestion by continuously sending UDP data streams, and calculate the location of tight links from the round-trip time delay (RTT) of each link measured by traceroute. However, the BFind method generates a large number of payload packets during a positioning process, and its intrusion cannot be ignored. The DRPS method combines the nature of the periodic flow of data packets with the principle of data packet baffles, controls link congestion by changing the switching time of the rate, and uses congestion identification technology at the receiving end to locate tight links. Its disadvantage is that it depends on available Bandwidth measurement, so that its positioning accuracy is directly affected by the available bandwidth measurement accuracy. The STAB method performs tight link location by measuring the available bandwidth of the subpath, but it uses the pathChirp algorithm to estimate the available bandwidth of the subpath, and the measurement accuracy is not high and the robustness is insufficient, thus affecting the tight link location accuracy. The Pathneck method uses the Recursive Packet Column (RPT) technology to determine the position of the tight link. It does not need to know the available bandwidth of the sub-path or the entire path to obtain more accurate positioning results. However, it is based on the ICMP protocol. If the node does not enable the ICMP function, it cannot locate tight links.

In addition, there is Pathtrait, a method for measuring available bandwidth based on tight link location. It first locates tight links, and then measures the link available bandwidth at tight links. Pathtrait, like Pathneck, requires nodes on the path to enable the ICMP function. Its available bandwidth measurement depends on the tight link location results. If the tight link location is wrong, the available bandwidth measurement results will be directly affected.

Although there are many methods for locating tight links and methods for measuring available bandwidth, only STAB can simultaneously obtain available bandwidth measurement values during the process of locating tight links. However, because STAB uses the pathChirp algorithm to estimate the available bandwidth of sub-paths, the measurement accuracy is not high and the robustness is not high enough, resulting in low measurement accuracy of available bandwidth and tight link positioning accuracy, and STAB needs to send detection packets multiple times to complete a measurement. The measurement load is large and the measurement time is long.

Contents of the invention

The purpose of the present invention is to propose a method for simultaneously measuring available bandwidth and locating tight links based on a single packet sequence.

In order to achieve the above object, the method for simultaneously performing available bandwidth measurement and tight link location based on a single packet sequence proposed by the present invention includes: 1) the sending end sends multiple TTLs (Time to live) in a back-to-back manner based on the length of the path to be measured The first positioning packet whose domain value gradually increases, wherein the maximum TTL domain value is not less than the number n of nodes of the link to be tested; 2) after the sending end sends the last first positioning packet, it continues to send in a back-to-back manner Multiple payload packets whose TTL field value is not less than n, and the destination port of each payload packet is set as an unreachable port, so as to trigger the destination node to feed back a DU (ICMPdestination-unreachable) packet; After a load packet, continue to send a plurality of second positioning packets with gradually decreasing TTL domain values in a back-to-back manner, wherein the TTL domain values of each second positioning packet are respectively identical to the TTL domain values of a first positioning packet; 4) After sending the last second positioning packet, the sender sends multiple speed-down packets with a TTL domain value of n at a decreasing rate, and the destination port of the speed-down packet is set as an unreachable port, so as to also trigger the destination port. The node side feeds back the DU packet; 5) the sending end receives each TE (ICMP time-exceeded) packet fed back by each node of the link to be tested based on the corresponding first positioning packet and the second positioning packet and the TE (ICMP time-exceeded) packet fed back by the destination node DU packets fed back by the end based on each load packet and deceleration packet; 6) The sending end determines the packet sequence based on the receiving time of the TE packets corresponding to the first positioning packet and the second positioning packet with the same TTL field value received. The extension status of the corresponding link, and then locate the tight link according to the predetermined threshold; 7) The sending end is based on the sending time of sending each load packet and deceleration packet, and receiving the DU packet corresponding to each load packet and deceleration packet time, calculate the measured round-trip delay of each packet; 8) the sending end determines the slow-down packet corresponding to the constant round-trip delay based on the measured round-trip delay of each packet; 9) the sending end calculates the determined The average send rate of the packet column when the slow packet is sent is used as a measure of the available bandwidth.

To sum up, the method for simultaneously performing available bandwidth measurement and tight link location based on a single packet sequence in the present invention can realize tight link location and available bandwidth measurement only by sending one packet sequence.

Description of drawings

FIG. 1 is a flowchart of a method for simultaneously performing available bandwidth measurement and tight link location based on a single packet sequence according to the present invention.

FIG. 2 is a schematic diagram of packets sent by the method for simultaneously performing available bandwidth measurement and tight link location based on a single packet sequence according to the present invention.

FIG. 3 is a schematic diagram of the round-trip time delay calculated by the method of simultaneously performing available bandwidth measurement and tight link location based on a single packet sequence according to the present invention.

Figure 4 is the topological diagram of the constant background flow experiment used in the simulation experiment.

Figure 5 is a topology diagram of the burst background flow experiment used in the simulation experiment.

Detailed ways

Referring to Fig. 1, the method for simultaneously performing available bandwidth measurement and tight link location based on a single packet column according to the present invention includes the following steps:

First, in step S11, the sending end sends a plurality of first positioning packets with gradually increasing TTL field values in a back-to-back manner based on the length of the link to be tested, wherein the largest TTL field value is not less than the node of the link to be tested number n. For example, as shown in FIG. 2 , the sending end sends n first positioning packets whose TTL field values gradually increase from 1 to n in a back-to-back manner. In order to reduce the tight link positioning error, the size of the first positioning packet should be as small as possible.

Next, in step S12, after sending the last first positioning packet, the sending end continues to send multiple payload packets with a TTL domain value not less than n in a back-to-back manner, and the destination port of each payload packet is set as an unreachable port , so as to trigger the destination node to feed back the DU packet. As shown in FIG. 2 , after sending n first positioning packets, the sending end immediately sends d payload packets, and the TTL field value of each payload packet is n.

Then, in step S13, after sending the last load packet, the sending end continues to send a plurality of second positioning packets with gradually decreasing TTL domain values in a back-to-back manner; wherein, the TTL domain values of each second positioning packets are respectively the same as The TTL field value of a first positioning packet is the same. As shown in FIG. 2 , the sending end sends n second positioning packets whose TTL domain values gradually decrease from n to 1 in a back-to-back manner, and the second positioning packets use the same size as the first positioning packets.

Next, in step S14, after sending the last second positioning packet, the sender sends a plurality of speed-down packets with a TTL domain value of n at a decreasing rate, and the destination port of the speed-down packet is set as unreachable port, so as to also trigger the feedback of DU packets from the destination node. As shown in Figure 2, for example, when the sender sends the speed-down packet, it gradually reduces the average sending rate of the entire detection packet column exponentially. For example, when sending the i-th speed-down packet, the average sending rate Ri of the packet column is α (α>1) times the average sending rate R _i+1 of the packet column when +1 speed-down packet is added.

Next, in step S15, the sending end receives each TE packet fed back by each node of the link to be tested based on the corresponding first positioning packet and the second positioning packet, and the destination node side based on each load packet and speed reduction The DU package that the package feeds back. Based on the characteristic that the TTL field value of each packet sent by the sender will decrease by 1 every time it passes through an intermediate node, each intermediate node of the detection packet passing through the path to be tested will have a first positioning packet and a The second positioning packet is discarded because the TTL domain value is reduced to 0, so that the intermediate node feeds back 2 TE packets to the sender, and then the sender will receive 2n TE packets returned from the intermediate node. Since the TTL domain value of each load packet and speed reduction packet is not less than n, and the destination port is unreachable, when each load packet and speed reduction packet arrives at the destination node, it will trigger the destination node to feed back a DU (destination-unreachable) packet . Therefore, the sending end will successively receive the DU packets fed back by the destination node.

Next, in step S16, based on the receiving time of the TE packets corresponding to the first positioning packet and the second positioning packet with the same TTL field value, the sending end determines the extension status of the packet column after passing through the corresponding link, and then according to the predetermined Threshold locates tight links. For example, the sending end will receive 2n TE packets returned from the intermediate node, and the 2 TE packets arriving successively from the same node (that is, the TE packet fed back according to the first positioning packet and the TE packet fed back according to the second positioning packet TE packet) is the length of the packet sequence after the packet sequence passes through the previous hop path of the node. According to the n pairs of TE packets returned by n nodes, the packet sequence length after the packet sequence passes through each link can be obtained. Knowing the extension of the packet after passing through each link, the sending end then locates the last hop link whose extension exceeds a predetermined threshold as the tight link of the path to be tested.

Next, in step S17, the sending end calculates the measured round-trip delay of each packet based on the sending time of each load packet and the speed reduction packet and the receiving time of the DU packet corresponding to each load packet and the speed reduction packet. For example, the sender sends the i-th payload packet at 12:30:00, and receives the DU packet fed back by the destination node based on the i-th payload packet at 12:30:05, so the sender can calculate the i-th payload The round-trip delay of the packet is: 12:30:05-12:30:00=5 seconds.

Next, in step S18 , the sending end determines, according to the measured round-trip delay of each packet, the deceleration packet corresponding to the round-trip delay that starts to be kept constant. Since the way of sending packets is to send some load packets back-to-back first, these packets keep piling up on the tight link, and the queuing delay continues to increase, so that the queuing delay of the deceleration packets sent later will also gradually increase, thus the round-trip delay also gradually increased. When the average sending rate of the packet column drops to a certain level, the accumulation of data packets in the buffer at the tight link is gradually relieved, and the queuing delay of the detection packet is gradually reduced, so the round-trip delay is also gradually reduced. When the first detection packet arrives at the tight link, the congestion is just eliminated, then the round-trip delay measured by each detection packet starting from this packet will remain unchanged, and the round-trip delay measured by each packet in the packet column changes as shown in Figure 3 Show. From the above analysis, it can be seen that based on the calculated changes in each round-trip delay, the sending end can determine the deceleration packet corresponding to the initial constant round-trip delay. For example, the calculated round-trip delay basically tends to be the same starting from the kth speed-down packet, so the sender can determine the k-th speed-down packet as the speed-down packet corresponding to the initial constant round-trip delay.

Next, in step S19, based on the determined deceleration packet, the sending end calculates the average rate of the packet train when the packet is sent as the measured value of the available bandwidth. The formula for calculating the available bandwidth value avbw is as follows:

$\mathrm{<span\; class="notranslate">\; avbw</span>}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="S"\; filenames="HDA0000043458530000021.png"\; state="\{\{state\}\}">S</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; m</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,2,3</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span>$

Among them, S ₁ is the size of the load packet and the speed reduction packet, S ₂ is the size of the positioning packet, d is the number of load packets, k is the sequence number of the speed reduction packet whose round-trip delay is reduced to a constant, and t is the number of the speed reduction packet from the beginning of sending The time required from the first load packet to sending the kth deceleration packet, m is the number of first positioning packets arriving at the tight link (the number of second positioning packets arriving at the tight link is also m ).

It should be noted that those skilled in the art should understand that the order of the above steps is not limited to what is shown, in fact, step S16 may also be performed after step S17, or may be performed after step S18, and so on.

1: Experimental verification

In order to verify the measurement performance of the present invention, a simulation experiment is carried out using NS2 which is widely used in the field of network research. Since this method is an end-to-end measurement tool, an experimental topology with only one linear backbone path is adopted. Each router in the topology adopts the first-in-first-out scheduling strategy, and the buffer adopts the end-of-queue discarding strategy. All links in the topology are bidirectional symmetrical links with a capacity of 100Mbps.

1 Constant background flow case

The experimental topology is shown in Figure 4. The path P from the sending end Snd to the receiving end Rcv passes through routers n1, n2, n3...n8 in turn, and the background flow 1 is a constant bit rate background flow running through n1, n2...n8. It is a separate background stream; background stream 2 and background stream 3 are constant bit rate background streams that only pass through L4 and L7 respectively, and they are all composed of multiple constant background streams. During the experiment, by setting each background stream Flow generation and end time to control the flow size change on L4 and L7.

1.1 A tight link situation

In the topology shown in Figure 4, set the traffic size of background flow 1 to 30Mbps, set the size of background flow 2 to 10Mbps, set the size of background flow 3 to 40Mbps for 0-10 seconds, and set the size of background flow 3 for 10-20 seconds The size becomes 30Mbps, and the size of the background stream 3 becomes 20Mbps for 20-30 seconds, and the packet size is 1000bytes. After this setting, the traffic size on L4 is always 40Mbps, and the traffic size on L7 is gradually close to but always larger than the traffic size on L4. During the whole measurement process, the tight link is always L7, but as the traffic on L7 gradually approaches the traffic on L4 Traffic, the difficulty of locating tight links is gradually increasing.

The experimental results are shown in Table 1. The accuracy of tight link location and available bandwidth measurement is very high. During the 22 measurements, the location of the tight link is all correct; the measurement error of the available bandwidth is very small, the maximum relative error is only 2.18%, and the average relative error is 0.86%. Experimental results show that the method of the present invention has high accuracy and strong robustness in tight link location and available bandwidth measurement under the condition of constant background flow and single tight link.

Table 1: Experimental results of a single tight link

1.2 Two tight links

In the topology shown in Figure 4, set the size of background stream 1 to be 30Mbps, the size of background streams 2 and 3 for 0-10 seconds are both 30Mbps, and the sizes of background streams 2 and 3 for 10-20 seconds are both changed to 20Mbps. After 20-30 seconds, the sizes of background streams 2 and 3 both change to 10Mbps, and the packet size is both 1000bytes. During the entire measurement process, the traffic sizes on L4 and L7 are the same and larger than those on other links, that is, both L4 and L7 are tight links. However, pathrader only gives one link as the positioning result each time. According to its algorithm idea, the first tight link should be given as the positioning result, that is, L4 should be given as the correct positioning result. In the experiment, the traffic on the two tight links is set to gradually decrease so that the traffic size of their adjacent links gradually approaches, thereby gradually increasing the difficulty of locating the tight link and more effectively verifying the performance of pathrader.

The experimental results are shown in Table 2. In the course of 22 experiments, there were 2 tight link positioning errors. The result of these two wrong positionings was the second tight link L7, and the relative positioning error was 9.09%, showing a very good performance. Positioning effect; the maximum relative error of available bandwidth measurement is 2.01%, and the average relative error is 0.82%, showing high accuracy and strong robustness.

Table 2 Experimental results of double tight links

2Sudden background flow situation

The experimental topology is shown in Figure 5. The background flow on link L4 is composed of 100 flows conforming to the Pareto distribution, and the background flow on link L7 is composed of 200 flows conforming to the Pareto distribution to simulate a burst The packet size of the Pareto flow is set as follows: the data packets with a packet size of 40/550/1500bytes account for 50%, 40%, and 10% of the load respectively, and the Pareto distribution shape parameter α=1.9. This is because multiple streams that obey the Pareto distribution with parameter α < 2 show self-similar characteristics after aggregation. The packet transmission rates on L4 and L7 are 20Mbps and 40Mbps respectively. This setting is for the obvious difference in traffic size between L4 and L7 to clarify the location of tight links. Otherwise, the sudden change of background traffic on L4 and L7 may cause tight links. The link position changes frequently and it is difficult to determine the tight link position at the time of measurement.

The measurement results are shown in Table 3. During the 22 measurements, there was one positioning error, and the positioning error rate was 4.55%. The maximum relative error of the available bandwidth measurement was 11.2%, and the average measurement relative error was 3.13%. The relative errors are all less than 10%. It shows that under the background of bursts, pathrader still shows high measurement accuracy and strong robustness for tight link location and available bandwidth measurement.

Table 3 Experimental results of burst background flow

2. Performance analysis:

1. Measurement accuracy

The method of the present invention locates the tight link by finding the link passed when the last packet sequence extension exceeds a certain threshold as the tight link. Since the extension of a packet row after passing through a link is affected by the input rate of the packet row on the link, the background flow rate on the link, and the link capacity, the extension of the packet row cannot be used as an indicator for judging a tight link. Therefore, theoretically, there is a certain error in locating tight links, and the selection of the judgment threshold will also affect the positioning accuracy.

Since the sending rate of the packet column in the present invention shows a series of continuously decreasing discrete values, when the available bandwidth value to be measured is between two adjacent values, a theoretical error will inevitably occur. The theoretical error is affected by the setting of the decrement factor, the smaller the decrement factor, the smaller the theoretical error.

In the actual measurement process, the round-trip delay of the detection packet is affected by the flow size on the reverse path, so that errors may occur in determining the available bandwidth based on the packets whose round-trip delay begins to remain constant, but the measurement time of the present invention is very short , it is very unlikely that the magnitude of the flow on the reverse path will change greatly in a short period of time, so the impact of changes in the reverse background flow on the measurement accuracy will be relatively small.

2. Measurement time

The time for the present invention to complete a measurement (including tight link location and available bandwidth measurement) is from the time when the packet column is sent to the time when a certain ICMP packet is received to determine that the round-trip delay is kept constant. The measurement time of the present invention is affected by the descending speed of the sending rate of the packet train and the round-trip delay of the detection packet. Among them, the drop speed of the sending rate of packets is affected by the settings of the initial drop rate and the decrease factor. Generally speaking, since the invention can complete the measurement by sending a detection packet sequence once, the measurement time is very short.

3. Measure the load

Measurement load refers to the amount of load generated during a measurement. The present invention sends a packet row for one measurement, and the actual load of the packet row is the measurement load. The measurement load of the present invention is related to the number of load packets, the number of positioning packets, the initial descending rate, the decrement factor and the value of available bandwidth to be measured. By setting reasonable parameters, the measurement load of the present invention can be controlled within a small range.

4. Robustness

Since the time for the present invention to locate tight links is very short, the possibility of background flow changes on the path is very small within a very short time, so the present invention is less affected by sudden background flows and has strong robustness. It can be seen from Fig. 3 that the overall change trend of the round-trip delay value of the packet column is to rise first and then fall to a constant value. In the process, if it is affected by the sudden background flow, the rising or falling trends will be different. If the traffic of the background flow increases, the queuing of the detection flow will intensify, and the decrease of the round-trip delay will slow down, then the sequence number of the packet whose round-trip delay begins to remain constant will increase, and the measured available bandwidth value will decrease; on the contrary, if the traffic of the background flow If the value is reduced, the queuing of the probe flow will be slowed down, and the round-trip delay will decrease faster, and the sequence number of the probe packet will decrease until the round-trip delay drops to a constant value, and the measured available bandwidth value will increase. This ensures that the present invention has high robustness in measuring the available bandwidth under bursty background flows.

In summary, the method for simultaneously performing available bandwidth measurement and tight link location based on a single-packet column in the present invention realizes simultaneous tight link location and available bandwidth measurement at a single end through a single-packet column. The present invention has high measurement accuracy, Fast measurement speed, low intrusion, strong robustness.

The above-mentioned embodiments only illustrate the principles and functions of the present invention, but are not intended to limit the present invention. Anyone skilled in the art can make modifications to the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be listed in the claims.

Claims (2)

1. one kind is carried out the method for available bandwidth measurement and tight link location simultaneously based on single bag row, it is characterized in that comprising:

1) transmitting terminal sends first localization package that a plurality of TTL (Time to live) thresholding progressively increases based on the length of link to be measured with mode back-to-back, and wherein, maximum TTL thresholding is not less than the interstitial content n of said link to be measured;

2) transmitting terminal is after sending last first localization package; Continue to send the load bag that a plurality of TTL thresholdings are not less than n with mode back-to-back; And the destination interface of each load bag is set to unreachable port, feeds back to DU (ICMP destination-unreachable) bag so that can trigger the destination node end;

3) transmitting terminal has been after having sent last load bag, continues to send second localization package that a plurality of TTL thresholdings are progressively successively decreased with mode back-to-back again, and wherein, the TTL thresholding with one first localization package is corresponding identical respectively for the TTL thresholding of each second localization package;

4) transmitting terminal has been after having sent last second localization package, and again with the speed of the successively decreasing reduction of speed bag that to send a plurality of TTL thresholdings be n, and the destination interface of reduction of speed bag is set to unreachable port, feeds back to the DU bag so that also can trigger the destination node end;

5) the transmitting terminal DU bag that receives each TE (ICMP time-exceeded) bag of feeding back to based on corresponding first localization package and second localization package by each node of said link to be measured and feed back in each load bag and reduction of speed bag by the destination node end group;

6) time of reception of first localization package that equates based on the TTL thresholding that receives of transmitting terminal and each self-corresponding TE bag of second localization package is confirmed the extension situation of bag row process respective link, and then locatees tight link according to predetermined threshold; Be that transmitting terminal will receive 2n TE bag that returns from intermediate node; And 2 TE bags that arrive from the priority of same node; Be that the bag row are through the bag row length behind this node previous dive path promptly according to the TE bag of first localization package feedback with according to the time interval that the TE that second localization package feeds back to wraps; The n that returns according to n node can obtain wrapping the bag row length that is listed as through behind every section link to the TE bag; Promptly know the extension situation after the bag row are through every link, subsequently, transmitting terminal will extend and orientate the tight link in path to be measured above the final jump link of predetermined threshold as;

7) transmitting terminal calculates and respectively wraps measured round-trip delay based on the transmitting time of sending each load bag and reduction of speed bag and receive each load bag and the time of reception of the corresponding DU bag of reduction of speed bag;

8) transmitting terminal confirms to begin the pairing reduction of speed bag of round-trip delay that keeps constant based on the measured round-trip delay of each bag;

9) transmitting terminal calculates the average transmission rate that determined reduction of speed encapsulates bag row when sending, and with the measured value of said average transmission rate as available bandwidth;

According to computes available bandwidth value avbw:

$\mathrm{avbw}=\frac{(d+k)\&CenterDot;S_1+2m{S}_2}{t},k=\mathrm{1,2,3}...$

Wherein, S ₁Be the size of load bag and reduction of speed bag, S ₂Be the size of localization package, d is the number of load bag, and k is the sequence number that round-trip delay drops to constant reduction of speed bag, and t sends k needed time of reduction of speed bag for beginning to send to from the bag row, and m is the number that arrives first localization package at tight link place.

2. as claimed in claim 1ly carry out the method for available bandwidth measurement and tight link location simultaneously based on single bag row, it is characterized in that: the transmission rate of reduction of speed bag is successively decreased with exponential form.

Images