KR20060074411A - Method for end-to-end measurement of available bandwidth between two network nodes - Google Patents

Abstract

The present invention relates to a technique for measuring the transmission bandwidth of a digital network over which packet communication takes place. According to the present invention, an end-to-end of a packet network is transmitted based on a single hop model based on a single hop model from a series of probe packets having an initial gap at a transmitter and received from a series of probe packets received at a receiver. -end) An improved method of measuring bandwidth and available bandwidth is provided. According to the present invention, an ordered pair g _i ² , g _o ² of an initial gap and an output gap where the difference between the output gap and the initial gap becomes substantially zero, and another initial gap, an ordered pair of output gap g _i ^1, using ¹ g _o) calculating a bandwidth, and the amount of the competing traffic bottleneck link, and calculates a point-to-point (end-to-end) the bandwidth and available bandwidth therefrom. In addition, the difference between the initial gap and the output gap varies linearly with the bandwidth of the bottleneck link and the amount of contention traffic so that the initial gap value can be predicted to substantially zero the difference between the output gap and the initial gap value. By this, a fast decision algorithm is presented.

Packet network, single-hop, packet train, bandwidth measurement, end-to-end

Description

Method for end-to-end measurement of available bandwidth between two network nodes}

1 is a diagram illustrating an embodiment of a configuration of network nodes that are an object of measuring available bandwidth by applying the present invention.

FIG. 2 is a diagram illustrating the concept of a gap in a packet train, which is a premise of the proposed single-hop gap model for measuring available bandwidth.

3 is a diagram illustrating a relationship between an initial gap of a packet train and an output gap and an initial gap.

4 is a block diagram showing the overall configuration of an apparatus for measuring bandwidth according to an embodiment of the present invention.

5 is a flowchart schematically illustrating a bandwidth measuring method according to an embodiment of the present invention.

<Description of the symbols for the main parts of the drawings>

500: transmitter 510: probe packet receiver

520: Observation information receiving unit 530: Output gap measuring unit

540: initial gap allocation unit 560: bandwidth measurement unit

580: investigation packet transmitter 590: receiver

The present invention relates to a technique for measuring the transmission bandwidth of a digital network over which packet communication takes place.

As the network becomes more advanced, new application services such as multimedia transmission, peer-to-peer file sharing, and multicast, as well as existing best-effort policy-based data transmission, are emerging. -to-end Various engineering issues such as quality-of-service (QoS), congestion control, service level agreement (SLA) verification, and traffic engineering have been raised.

In solving this problem, the measurement of available bandwidth can help by selecting a link that can guarantee the bandwidth or by constructing an efficient file transfer tree. This problem of measuring available bandwidth is becoming an important field as a network is formed through numerous Internet Service Providers (ISPs), a Broadband Convergence Network (BcN) is realized, and the network structure is complicated.

There are two methods of measuring the available bandwidth of a conventional network, a method of directly obtaining and analyzing information using a protocol such as SNMP from a router and a method of collecting information from a user side. Tools for processing and analyzing information from the network include tools such as the Multi Router Traffic Grapher (MRTG) and the Round Robin Database (RRD). These methods have the advantage of obtaining accurate data because statistical analysis can be performed through the data obtained from the router by the network administrator. However, this information is limited to network administrators because it requires direct access to the router, and it is difficult to apply in a network composed of various ISPs. Also, in case of file sharing between each other, it is important for individual users to collect information because they configure file sharing tree among individual users without the help of a separate central administrator. Therefore, there is a need for a method for collecting network path information from end to end.

1 is a diagram illustrating an embodiment of a configuration of network nodes that are an object of measuring available bandwidth by applying the present invention. Two terminals of the host 1 (100) and the host 2 (195) are connected to a network through an access network (115, 180). In the path between the two hosts, there are several routers belonging to different ISPs and links connecting each router. The available bandwidth of the link with the smallest available bandwidth among the links on the path is defined as the available bandwidth of the end-to-end path.

 The terminal may send an SNMP query to each router to measure the available bandwidth. When the host 1 100 sends an SNMP query to the neighboring router 110, the router sends the information of the link 120 transmitted by the router 1 to the host 1 100. In this way, Host 1 knows the available bandwidth information of the links on the end-to-end path. However, since Host 1 is not authorized to receive information from the router 135 belonging to another AS (Autonomous System) 150, it is not possible to measure the available bandwidth of all links on the end-to-end path by sending SNMP queries. none.

 Another way to measure end-to-end available bandwidth is to use a packet train. By sending probe packets from Host 1 (100) to Host 2, the available bandwidth is estimated by analyzing the interval at which packets arrive. This method does not require the information of the intermediate router, and measures the available bandwidth using only the information of the terminals, so that the available bandwidth can be measured even if the hosts belong to different ASs.

In the early stages, the research began with measuring bottleneck bandwidth on an end-to-end basis and is now expanding to measuring available bandwidth and analyzing bottleneck links. In particular, the field of measuring the available bandwidth has been actively studied to date.

The basic concepts used in the above end-to-end bandwidth and available bandwidth measurement techniques can be defined as follows. The network path L is i = 1,... It consists of N individual links. Each link i has a maximum bit rate per second that can be transmitted. This is called Ci bps. At this time, the link having the lowest data rate on the path L is defined as the bottleneck link, and this is defined as the bandwidth of the path L, and is expressed by the following equation.

CL = min {Ci} (1)

In addition, the sum of transmission rates of other flows present at time t for link i is defined as the amount Bi (t) (0 ≦ Bi (t) ≦ Ci) of the contention flow. Where the link utilization ui (t, T) (0 ≤ ui (t, T) ≤ 1) for time [t, T] of link i is

(2)

(2)

Is defined as

At this time, the available bandwidth Ai (t, T) during the time [t, T] of the link i is

(3)

(3)

Is defined as

Where the available bandwidth AL (t, T) for time [t, T] of path L is

A _L (t, T) = min (A _i (t, T)) (4)

Can be defined as

Many methods have been proposed to measure the bandwidth and available bandwidth on the path, and in particular, a packet pair or a packet train is mainly used. Such a technique is to transmit the same length packet at regular intervals at the transmitter side and to measure the changed interval at the receiver side to measure the bandwidth or available bandwidth on the path. In particular, in order to measure the available bandwidth, many methods are used to model the network as a single hop having a bottleneck link or to statistically analyze the changed interval measured by the receiver at histogram analysis, regression analysis, and the like.

In this regard, the techniques for measuring bandwidth on a end-to-end basis and techniques for measuring available bandwidth can be classified as follows. First, bandwidth measurement techniques include bprobe, Packet Bunch Modes (PBM), Asymptotic Dispersion Rate (ADR), and Capacity Probe (CapProbe). bprobe is a technique of measuring bandwidth on a path by transmitting a large number of packet pairs having different packet lengths, obtaining a sample having a minimum transmission time, and obtaining a transmission rate per packet length based on the sample. PBM proves that the sample of the packet pair can be multimodal, unlike the conventional techniques, assuming that the sample of the packet pair is unimodal, and to overcome the limitations, PBM transmits different packet trains to The bandwidth on the path was measured by the method that can handle the mode. The ADR proved that the measurement of bandwidth could be measured high or low depending on the network configuration of the network, and proposed a method that can measure the bandwidth by analyzing the frequency of histogram using this. CapProbe is a technique for measuring bandwidth by measuring the distributed interval of samples with the minimum transmission delay of packet pairs in order to reduce the effects of delay in the router's queue when packet pairs are delivered.

These techniques are based on the distributed interval of samples measured at the receiver as it transmits back-to-back without the initial interval of the packet pair at the transmitter. Therefore, for accurate and error-free measurement, it is important to minimize the effects of contention packets on the link between two pairs of packets as much as possible. For this purpose, it is known to use a small packet size.

In addition, the field of measuring available bandwidth has been actively studied to date, and can be divided into two researches based on a single-hop model and self-induced congestion techniques.

The single-hop model has the following characteristics. The single-hop model is a model that models the path on the Internet as a link with a bottleneck link, taking into consideration that the bottleneck link has the greatest effect on the transmission rate of the packet being transmitted. FIG. 2 is a diagram illustrating the concept of a gap in a packet train, which is a premise of the proposed single-hop gap model for measuring available bandwidth. As shown in FIG. 2, the two survey packets 210 arrive at the bottleneck router 220 at regular intervals 200. The bottleneck router also has other traffic 240 in addition to the probe packet. In the bottleneck router's queue, probe packets and packets of other traffic are queued together and processed in order. The interval 230 between packets after both survey packets have been processed is different from the initial interval 200 due to the influence of other traffic. Based on the change in the two intervals, the amount of traffic other than the probe packet passing through the bottleneck router is estimated. The difference between the bandwidth of the bottleneck router and the traffic passing through the bottleneck router is the available bandwidth.

 When a research packet passes a path between two terminals, it passes through several routers, and the interval between the packets can be expected to be most affected by the bottleneck router. Based on these expectations, the path between two terminals can be thought of as a single hop model consisting of one bottleneck link.

Studies based on single-hop models include Delphi, Initial Gap Increasing (IGI) / Packet Transmission Rate (PTR), and Spruce. The techniques based on the single-hop gap model above model the network as a single hop with a bottleneck, and the output interval observed at the receiver when transmitting a packet pair or packet train with a constant initial gap. It is a technique to determine the transmission rate as the available bandwidth when the initial interval and the output interval are the same by measuring the change in the (output gap). Delphi is a technique for measuring the amount of competitive traffic based on Multifractal Wavelet Modal (MWM). IGI / PTR is a technique for modeling the network as a single hop model for the first time and measuring the available bandwidth by analyzing the output gap where the input gap is changed by contention traffic when packet pairs are transmitted. Spruce is a technique for sampling competing traffic with random properties by having a Poisson distribution in transmitting multiple packet pairs.

This technique has the advantage of quickly finding available bandwidth using fewer packets. However, since the available bandwidth is measured using the initially measured bottleneck bandwidth, an error may occur if the bottleneck bandwidth is not accurately measured.

Self-induced congestion techniques include Train of Packet Pair (TOPP), PathChirp, and Pathload. Such congestion induction techniques measure the available bandwidth by observing the moment when the observed speed of the receiving side becomes larger by the queuing than the speed of the transmitting packet to find the available bandwidth while increasing the transmission speed of the transmitting packet. TOPP defines several packet pairs with regular intervals as trains of packet pairs in order not to put a heavy load on the network temporarily, and outputs through multiple transmissions of trains of packet pairs while gradually reducing the initial gap of packet pairs. It is a technique to measure the rate of transmission through the point of congestion by measuring the change of gap. PathChirp reduces the initial gap at a constant rate within one packet train and defines it as a packet chirp. By sending these packet chirps several times, we find a pair of packets in which the output gap begins to increase rapidly within the packet chirp, and determine that the transmission rate at this time causes congestion on the network path and finds available bandwidth. Pathload considers that when the sender transmits more than the available bandwidth, the delay in the router's queue increases, resulting in an overall increase in the delay of the entire packet train, thus changing the initial rate in a manner similar to binary tree search. This technique finds a transmission rate in which the delay of the train increases as a whole.

These techniques have the advantage of finding the available bandwidth relatively accurately regardless of the location of bottlenecks in the path. However, because it causes congestion, it has a disadvantage of putting a lot of load on the network and taking a lot of time to measure.

  It is an object of the present invention to provide a method for quickly and accurately measuring the amount of bandwidth, available bandwidth, and contention traffic on a path end-to-end in a packet network. More specifically, an object of the present invention is to provide a bandwidth measuring method of a packet network capable of accurately measuring bandwidth regardless of the amount of contention traffic while simultaneously measuring the available bandwidth.                         

Furthermore, an object of the present invention is to reduce the influence of a survey packet applied to a target network on the network, thereby reducing the load on the network and obtaining a measurement value well reflecting a rapidly changing network situation.

According to an aspect of the present invention for achieving the above object, a new method for measuring the available bandwidth by separating the bottleneck bandwidth and the competitive bandwidth in measuring the bottleneck bandwidth and available bandwidth (Decoupling Capacity estimation and Initial Gap of the packet train) is presented. This allows accurate bandwidth measurements while accurately measuring available bandwidth, regardless of the amount of competitive traffic.

According to another aspect of the present invention, the existing methods send a large number of survey packet pairs or packet trains to find the proper initial gap of the packet pairs, which incurs additional load on the network and takes a long time to find available bandwidth. Fast Converging algorithm is proposed to solve this problem. According to this, it is possible to quickly find an appropriate initial gap by measuring the amount of change based on the characteristic that the output gap of the initial gap of the packet pair varies linearly according to the bandwidth of the bottleneck link and the amount of contention traffic. This reduces the number of packet trains needed to find available bandwidth, reducing the additional load on the network and the time needed to measure bandwidth.

According to an aspect of the present invention, a method for measuring bandwidth is based on a single hop model based on a single hop model from information observed in a series of survey packets received at a transmitter and transmitting a series of probe packets having an initial gap. A method of measuring the end-to-end bandwidth and the available bandwidth of a, and by monitoring the output gap value included in the observed information while changing the initial gap value, the difference between the output gap and the initial gap is Bandwidth of the bottleneck link using an ordered pair of initial gap and output gap (g _i ² , g _o ² ) that is substantially zero, and another ordered pair of output gap, output gap (g _i ¹ , g _o ¹ ) And calculating the amount of contention traffic, from which end-to-end bandwidth and available bandwidth are calculated.

In this case, according to an additional aspect of the present invention, the modified incremental initial gap has a difference between the initial gap and the output gap by using a characteristic in which the difference between the initial gap and the output gap varies linearly according to the bandwidth of the bottleneck link and the amount of contention traffic. It is determined by predicting an initial gap value of zero.

The foregoing and further aspects of the present invention will become more apparent later through the preferred embodiments described with reference to the accompanying drawings. Hereinafter, the present invention will be described in detail such that those skilled in the art can easily understand and reproduce the present invention.

4 is a block diagram showing the overall configuration of an apparatus for measuring bandwidth according to an embodiment of the present invention. As shown, an end-to-end of a packet network is transmitted based on a single hop model from a series of probe packets having an initial gap and from the information observed in a series of probe packets received at the transmitting end. end) The network bandwidth measuring apparatus for estimating the bandwidth and the available bandwidth receives the transmitter 500 for transmitting a series of probe packets and the probe packets transmitted from the transmitter 500 at the other end of the network to generate observation information. It is composed of a receiver 590 for transmitting this to the transmitter 500 again.

In one embodiment, the transmitter 500 receives the survey packet transmitter 580 for transmitting a series of probe packets having a specified initial gap through the access network, and receives observation information of the probe packets received at the other end of the network. An initial gap allocator 520 for calculating and providing a modified initial gap from the observation information receiver 520, the output gap information included in the observation information, and providing it to the probe packet transmitter 580, and the initial gap And a bandwidth measuring unit 560 for estimating an end-to-end bandwidth and an available bandwidth using a plurality of pairs of output gaps included in the observation information.

The survey packets have an initial gap as shown in FIG. 2 and are injected into the network shown in FIG. 1 with the destination 590 host as the destination through the access point. The initial gap is initially set to an appropriately small value and then incrementally incremented to find the optimal value.

The survey packet receiver 510 of the receiver 590 at the counterpart terminal receives these probe packets, and the output gap measurer 530 measures the output gap in the series of these probe packets. The observation information transmitter 550 transmits the information of the survey packets including the measured output gap to the transmitter 500. The observation information receiver 520 of the transmitter 500 receives the observation information.

Hereinafter, the operation of the bandwidth measuring unit 560 will be described. In one embodiment, the bandwidth measurement unit analyzes the observation information, the ordered pair of the initial gap and the output gap (g _i ² , g _o ² ), the difference between the output gap and the initial gap is substantially zero, and another one Calculate the bottleneck link bandwidth and the amount of contention traffic, using the initial gap and the ordered pair of output gaps (g _i ¹ , g _o ¹ ), from which end-to-end bandwidth and available bandwidth are calculated. do. That is, in the single-hop gap model, the gap is increased by contention traffic on the bottleneck link when a packet pair is delivered. This model models the path of a network with a bottleneck link by shrinking one bottleneck link and analyzes how the gap of the packet pair transmitted from the sender increases in the receiver. If the packet pair is affected by contention traffic on the bottleneck link, the gap increases, where the increased output gap go for the initial gap gi is

(5)

(5)

Where p is the length of the probe packet, Bi is the transmission rate of the competing traffic on the bottleneck link, and C is the bandwidth of the bottleneck link. To obtain a sample of a packet pair that satisfies these characteristics, the bandwidth of the link is measured by transmitting multiple packet pairs or by transmitting a packet train. In the case of a packet train, N-1 packet pairs are configured when N packets are transmitted by successive packet pairs. Samples obtained by sending two packet trains with different initial gaps (g ¹ _i , g ¹ _o ), (g ² _i , g ² _o ) can be used to calculate the bandwidth of the bottleneck link and the amount of contention traffic in equation (5). Equation 5 of packet contention traffic present in g _o two bandwidth C and a bottleneck link in the bottleneck link variables except that measured in the outbound packet length p and the initial gap g _i determined at the reception side when casting a train in Solving the equation for both Bi is the same as (6).

(6)

(6)

This value can be applied to Equation (3) to obtain the bandwidth C of the link L and the available bandwidth AL = C-Bi for the time corresponding to the length of the packet train.

Next, the operation of the initial gap allocation unit 540 will be described. In one embodiment, the initial gap allocator uses an attribute in which the difference between the initial gap and the output gap varies linearly according to the bandwidth of the bottleneck link and the amount of contention traffic, so that the initial gap and the output gap become zero. It provides a modified incremental initial gap that is determined by predicting the value. That is, except for the first case in which a back-to-back packet train is transmitted as shown in FIG. 3, the characteristics of the difference between the output gap and the initial gap for the initial gap of the packet pair are competing with the bandwidth of the bottleneck link. It changes linearly with the amount of traffic. Therefore, you can use it to find the point where the output gap and the initial gap are the same. As such, the relationship between the output gap and the initial gap according to the initial gap is expressed by the equation (7).

(7)

(7)

Where g ¹ _diff and g ² _diff are g ¹ _o -g ¹ _i and g ² _o -g ² _i , respectively, and are the difference between the output gap and the initial gap of the first sample and the second sample, and the constant K is a subsequent packet. The difference between the output and receive gaps obtained when the train is transmitted. In equation (7) we quickly predict g _i with g _o -g _i being zero. This is simply implemented by obtaining a gi value of 0 on the right side of Equation (7). From the obtained g _i , two samples reflecting well the content of the competitive traffic are obtained and used in the algorithms of Eqs. (6) and (3) to calculate the bandwidth, available bandwidth, and the amount of contention traffic.

Application of the method according to the present invention has the advantage of quickly finding an initial gap having the same initial gap and output gap using a small number of irradiation packets, and obtaining a desired sample.

Hereinafter, an embodiment of a bandwidth measuring method according to the present invention will be described. According to an embodiment of the present invention, a bandwidth measuring method includes transmitting a series of probe packets having an initial gap and obtaining an output gap, and comparing an output gap and an initial gap, where the difference between the initial gap and the output gap is determined by the bottleneck link. A fast convergence step that predicts an initial gap value where the difference between the initial gap and the output gap becomes zero by using a characteristic that varies linearly with the bandwidth and the amount of contention traffic, and transmits a new probe packet having this gap. If the difference between the gap and the output gap becomes substantially zero, a bandwidth calculation step of estimating the end-to-end bandwidth and the available bandwidth is performed by using the initial pair and the ordered pair of the output gap and previous order pair values. It is configured to include.

5 is a flowchart schematically illustrating a bandwidth measuring method according to an embodiment of the present invention. When the measurement starts, the first probe packets are sent first, in which case the initial gap is set to 0 to transmit subsequent packets, and the measured output gap is sent by the receiver to the sender through the response packet (step 600). ). This process is implemented by a function called send_pkt (gi). Here we initialize a variable called phase to count the number of transmissions of probe packets. At this time, the sender compares the output gap received from the receiver with the initial gap (step 605), and calculates the transmission rate at the time when the output gap and the initial gap are the same as available bandwidth. The function avail_bwth () calculates the available bandwidth. The available bandwidth is then obtained by calculating the bytes of probe packets transmitted during the entire output gap and ending the measurement.

Meanwhile, if the initial gap and the output gap are different from each other in step 605, the initial gap is set using the initially measured output gap go (step 610). In this embodiment, the initial gap is set to go / 2 and the gap_diff variable corresponding to the future increment is set to go / 8. This value can be adjusted as needed. Resend probe packets with the set initial gap and increase the phase value by one. Obtaining information about the output gap contained in the received response packet again, comparing the initial gap and the output gap previously (step 615) and measuring the available bandwidth using the measured output gap if the initial gap and the output gap are equal. Transition to step 630.

If the initial gap and the output gap are different, the number of transmissions is compared in step 620, and if the phase is 2, two initial samples are obtained, and thus, steps 635 of rapidly predicting the initial gap are performed. Here, FC (), which is a fast convergence algorithm, is performed through two samples based on Equation (7). The probe then transmits the survey packets using the predicted initial gap and then transitions to step 615 of comparing the output gap with the reception gap using the output gap obtained from the receiver.

If the number of transmissions is not 2, the initial gap is increased by the gap_diff, which is an increase obtained by transmitting the initial packet, the counter phase is increased, the probe packet is additionally transmitted, the response packet is received, and the initial gap and the output are received. Transition to step 615 for comparing the gaps.

If the initial gap and the output gap are not equal to each other, the process continues to increase the initial gap by the initial gain and transmits the probe packet and repeats the process of comparing the output gap through the response packet. In this case, if the initial gap and the output gap are the same, the function cal_availbwdth () applying the equation (6) using the output gap measured in step 630 is used to measure the amount of bandwidth and contention traffic and subtract the amount of contention traffic from the bandwidth. Measure the amount of bandwidth. When all measurements are finished, the measurement process is finished.

Although the present invention has been described with reference to a method and an apparatus made at a transmitting end of an investigation packet, the same may be applied to a case made at a receiving end. The present invention encompasses such obvious modifications.

   As described above, the present invention provides a new technique for measuring the amount of bandwidth, available bandwidth, and contention traffic in an end-to-end manner using a packet pair or packet train on a packet network. A new technique is proposed to speed up the measurement by reducing the number of packet trains required for the system.

Through this, it is possible to measure the available bandwidth on various network paths and measure the available bandwidth quickly to cope with the rapidly changing internet path situation. The end-to-end bandwidth measurement can be used to prevent network problems by preventing the packet pairs or packet trains used to measure the available bandwidth from putting a heavy load on the network.

Although the present invention has been described with reference to the embodiments described with reference to the accompanying drawings, it is not limited thereto, and should be interpreted by the claims intended to cover various obvious modifications.

Claims (14)

End-to-end bandwidth of the packet network based on a single hop model from a series of probe packets with an initial gap sent from the transmitter and received from the receiver packets received at the receiver And a method of measuring available bandwidth, the method comprising: By monitoring the output gap values included in the observed information while varying the initial gap values, the ordered pairs of the initial gap and the output gap (g _i ² , g _o ² ) and the difference between the output gap and the initial gap become substantially zero and Using another initial gap, an ordered pair of output gaps (g _i ¹ , g _o ¹ ), calculate the bandwidth of the bottleneck link and the amount of contention traffic, from which end-to-end bandwidth and Bandwidth measurement method characterized in calculating the available bandwidth. The method of claim 1, wherein the bandwidth measurement method that is characterized in that calculated from the amount (B _i) is the following formula and competing traffic bandwidth (C) of the bottleneck link.

2. The method of claim 1, wherein in determining a value of an initial gap where the difference between the output gap and the initial gap is substantially zero, the difference between the initial gap and the output gap is linear depending on the bandwidth of the bottleneck link and the amount of competitive traffic. Bandwidth measurement method, characterized in that determined by the prediction using a characteristic that changes to. 4. The method of claim 3, wherein the initial gap value is determined such that the equation below is zero.

Where g ¹ _diff = g ¹ _o -g ¹ _i , g ² _diff = g ² _o -g ² _i , K {Difference between output gap and reception gap obtained when transmitting consecutive packet trains} A computer-readable recording medium having recorded thereon a method according to any one of claims 1 to 4. End-to-end bandwidth and available bandwidth of the packet network, based on the single hop model of sending a series of probe packets with an initial gap and from the information observed in a series of probe packets received at the transmitter. A network bandwidth measurement apparatus for estimating bandwidth, the apparatus comprising: A survey packet transmitter for transmitting a series of survey packets having a designated initial gap through an access network; An observation information receiving unit which receives observation information of survey packets received at the other end of the network; An initial gap allocator configured to calculate a modified initial gap from the output gap information included in the observation information and provide the initial gap to the probe packet transmitter; By analyzing the observation information, an ordered pair g _i ² , g _o ² of an initial gap and an output gap where the difference between the output gap and the initial gap becomes substantially zero, and another pair of initial gaps and output gaps ( a bandwidth measurement unit for calculating the bandwidth of the bottleneck link and the amount of contention traffic using g _i ¹ , g _o ¹ ), and calculating an end-to-end bandwidth and an available bandwidth therefrom; Bandwidth measurement apparatus comprising a. 7. The method of claim 6, wherein the bandwidth measuring unit calculates the bandwidth C of the bottleneck link and the amount of contention traffic _Bi by the following formula.

The method of claim 7, wherein the initial gap allocation unit Modified increment determined by estimating the initial gap value where the difference between the initial gap and the output gap becomes zero using the characteristic that the difference between the initial gap and the output gap varies linearly with the bandwidth of the bottleneck link and the amount of contention traffic. Bandwidth measuring device, characterized in that to provide an initial gap. 10. The method of claim 8, wherein the initial gap allocator determines an initial gap value such that the following equation becomes 0 as a next initial gap value.

Where g ¹ _diff = g ¹ _o -g ¹ _i , g ² _diff = g ² _o -g ² _i , K {Difference between output gap and reception gap obtained when transmitting consecutive packet trains} A method for measuring the end-to-end bandwidth and available bandwidth of a packet network based on a single hop model, the method comprising: a) sending a series of survey packets having an initial gap and obtaining an output gap; c) If the difference between the output gap and the initial gap is different, the difference between the initial gap and the output gap is zero, using the characteristic that the difference between the initial gap and the output gap varies linearly with the bandwidth of the bottleneck link and the amount of contention traffic. A fast convergence step of predicting an initial gap value to be transmitted and transmitting a new survey packet having this gap; d) When the difference between the initial gap and the output gap becomes substantially zero, the bandwidth for estimating the end-to-end bandwidth and the available bandwidth by using the initial pair and the previous order pair values of the output gap and the output gap. Output stage; Bandwidth measurement method comprising a. The method of claim 10, wherein after step a) b) transmitting a survey packet having a new initial gap determined according to the output gap to obtain an output gap value. 12. The method of claim 11, wherein the fast convergence step uses an ordered pair of initial gap and output gap values of the first probe packet of step a) and the second probe packet of step b) so that the following equation becomes zero: Is determined as a next initial gap value.

Where g ¹ _diff = g ¹ _o -g ¹ _i , g ² _diff = g ² _o -g ² _i , K {Difference between output gap and reception gap obtained when transmitting consecutive packet trains} 12. The method of claim 11 wherein step b) determines the initial gap with a value starting from half the output gap and incrementing the fraction of the output gap. The method of claim 13, wherein in the step d), when the difference between the initial gap and the output gap does not become substantially zero, the initial gap is determined by incrementing the fraction of the output gap. .

Images