JP5923914B2 - Network state estimation apparatus and network state estimation program - Google Patents

Description

  The present invention relates to a network state estimation device and a network state estimation program. For example, in a network composed of nodes such as routers and switches and links such as optical fibers, the bandwidth is divided and each service is virtualized. The present invention can be applied to a network state estimation device and a network state estimation program for estimating the use state of a virtual network when a virtual network that functions as another network is constructed.

  First, a conventional network state estimation technique for grasping the network usage state will be described.

  Non-Patent Document 1 describes an available bandwidth estimation technique called PathLoad. PathLoad was developed by Manish Jain and Constantinos Dovrolisy.

  The technique described in Non-Patent Document 1 inputs a packet sequence having the same transmission interval to the network and observes the arrival interval on the receiving side. If the bandwidth of the packet sequence at the time of input is smaller than the minimum value of available (usable) bandwidth on the network communication path, the transmission interval and the arrival interval are substantially equal. If the bandwidth of the packet sequence at the time of input is larger than the available bandwidth of the network, the arrival interval is larger than the transmission interval.

  Therefore, the technology described in Non-Patent Document 1 estimates the free bandwidth of the network by repeating the operation of shortening the transmission interval if the arrival interval is not narrowed and increasing the transmission interval when the arrival interval is extended. Is. Non-Patent Document 1 also shows other available bandwidth estimation techniques such as the PathChip method and the IGI (Initial Gap Increasing) method, but the point that the available bandwidth is estimated from the arrival interval of packets is the same.

  As another network state estimation technique, there is a technique described in Non-Patent Document 2. The technology described in Non-Patent Document 2 measures quality information including a dispersion value of a delay time of a packet received by the terminal when the terminal performs real-time communication, and the terminal transfers the quality information to the transmission side. Estimate the state of the network from the information.

  As described above, the network state estimation technique can be used in the following specific manner, for example. FIG. 2B is a configuration diagram showing the configuration of a physical network. As shown in FIG. 2 (B), for example, two 100 Mbit / sec links are connected by a 200 Mbit / sec transmission path (for example, used as one link by IEEE 802.3ad LACP). Assume a network composed of router 1 and router 2 capable of forwarding.

  Consider a case where such a network receives a TV conference service with a bandwidth of 10 Mbit / sec, and an on-demand video distribution service with a bandwidth of 10 Mbit / sec.

  Assume that when a subscriber who can use this network tries to hold a video conference, video and audio are disturbed. Possible causes are, for example, that the TV conference device body, TV camera, and microphone are out of order, or that the transmission path and router are congested, resulting in insufficient bandwidth.

  At this time, the available bandwidth between the two sites where the TV conference is performed is measured using the available bandwidth estimation technology. If the available bandwidth is only 5 Mbit / sec, it is estimated that the shortage of bandwidth is the cause. On the other hand, if the usable bandwidth is 30 Mbit / sec, it is considered good to suspect a failure of the TV conference apparatus body, camera, or microphone.

  In this way, the network state estimation technology can be used to estimate the cause of a failure, and in the first place, the available bandwidth is estimated before the video conference, and if the available bandwidth is insufficient, the user gives up the video conference, Alternatively, measures such as shifting the usage time can be taken.

Satoshi Kato, Kenichi Yukimatsu, Hitoshi Hashimoto, “Examination of Performance Evaluation of End-to-End Available Bandwidth Estimation Method and Application to Streaming Communication”, 2005 IPSJ Tohoku Branch Meeting, Lecture Number 8 IETF Technical Standard RFC5093 (RTCP-XR: BT ’s Extended Network Quality RTP Control Protocol Extended Reports)

  By the way, “virtual network (or virtual network)” is used in various meanings. Here, the following cases are shown.

  This network performs policing control according to the type of packet on the routers 1 and 2 in the physical class of FIG. For example, up to 30 Mbit / sec for video conference traffic, up to 50 Mbit / sec for on-demand video distribution, and up to 120 Mbit / sec for all other apps including regular web and email. Suppose you have no settings.

  At this time, the network shown in FIG. 2B can be logically regarded as a virtual network as shown in FIG. That is, as shown in FIG. 2A, the network is a virtual network (TV conference virtual network) 1 that performs a TV conference with a bandwidth of 30 Mbit / sec, and a virtual network that performs on-demand video distribution with a bandwidth of 50 Mbit / sec. (Virtual network for on-demand video distribution) 2 and a virtual network (other application virtual network) 3 having a bandwidth of 120 Mbit / sec and allowing other applications to pass through are configured.

  By using such a virtual network, the following effects can be obtained.

  For example, if the virtual network 1 for video conference is constructed, a maximum of three video conferences can be performed regardless of how crowded other applications are.

  Also, for example, in the case of on-demand video distribution, retransmission control is possible, so it is not as severe as a TV conference. However, when the download speed is slower than the video playback rate, the video is stopped after being played back a little. It is. If the above-mentioned Ondeman virtual rope is constructed, it is possible to view up to 5 programs at the same time, no matter how much communication of other apps is used. The carrier assumes the usage status of each application of the user and sets the bandwidth of the virtual network.

  The problem at the time of estimating an available band with such a virtual rope is demonstrated.

  For example, an Internet Control Message Protocol (Network Control Notification Protocol) packet called ICMP is used as a packet used when measuring the available bandwidth.

  This packet is classified as traffic on the other application virtual network 3 in the above classification. Therefore, the available bandwidth estimated using ICMP is the available bandwidth of the other application virtual network 3.

  For example, in FIG. 2A, a video conference is performed between a video conference terminal 91 having an IP address “192.168.1.10” and a video conference terminal 92 having “192.168.2.10”. At this time, the available bandwidth is estimated using the ICMP packet in advance, and even if a determination result that there is sufficient bandwidth is available, it is the status of the other application virtual network 3 and the virtual Different from the state of the network 1.

  As a result, there is a possibility that the TV conference cannot be performed. That is, the available bandwidth of the virtual network constructed for each service such as the TV conference service or the on-demand video distribution service cannot be estimated.

  The following problems exist as a larger problem. That is a problem caused by power saving control.

  For example, if the bandwidth used for the video conference virtual network 1 is 20 Mbit / sec, the bandwidth used for the on-demand video distribution virtual network 2 is 30 Mbit / sec, and the bandwidth used for the other application virtual network 3 is 41 Mbit / sec, the total traffic The amount is 91 Mbit / sec, which is smaller than the capacity of one link between the router 1 and the router 2.

  The network control function (power saving control function) automatically has a bandwidth of the video conference virtual network 1 of 23 Mbit / sec, a bandwidth of the on-demand video distribution virtual network 2 of 33 Mbit / sec, and the other application virtual network 3. The router 1 and the router 2 sometimes stop the link connecting the IF 12 and IF 22 and perform communication with one link.

  In this way, the power consumption of the interface for one link can be reduced. Further, although not shown in FIGS. 2A and 2B, power consumption of components related to transmission such as an optical amplifier on a transmission path connecting between the router 1 and the router 2 can be reduced.

  Note that the network control function immediately restores the paused interface and link when a new video conference communication is performed, so that the policing setting value is restored, so that there is no disadvantage to the user. To do.

  In such a virtual network, when the available bandwidth is estimated, communication is performed with a single link by power-saving control on the network side even though there is actually a free bandwidth of 100 Mbit / sec or more in total. The measurement value obtained by the band estimation technique may be a value for one link, and the measurement value may be a value smaller than the original value, for example, several Mbit / sec. As a result, it may be impossible to additionally perform a TV conference service or an on-demand video distribution service.

  If the application user can obtain all of the network operation information, for example, if the MIB information of all the nodes such as the router 1 and the router 2 can be obtained, the available bandwidth estimation or the number of services to be used is determined. The need for prospective quality when increased may not be very high.

  However, the network in FIG. 2B is simplified, and a network with only two routers as shown in FIG. 2B is rare. Since the actual network is more complex, (1) information on all nodes cannot be obtained, (2) there is information that cannot be obtained across multiple network operators, and (3) all information Even if it is obtained, it may be difficult to estimate the quality of experience.

  Therefore, the present invention solves the above-described problem, and “uses the network control function to narrow down resources that can be used when the bandwidth usage rate is low for the purpose of power saving control”. An object of the present invention is to provide a network state estimation device and a network state estimation program that can correctly grasp a network state even in a virtual network.

  Specifically, the network state here refers to, for example, (1) an estimated value of available bandwidth, (2) an estimated value of the quality of experience of a service currently being executed, and (3) an additional service connection request. Before, it refers to an estimate of the quality of experience of the added service or services that are already running.

In order to solve this problem, the first aspect of the present invention provides a network state estimation apparatus for estimating the state of one or a plurality of virtual networks constructed by dividing a physical network band into a plurality of (1) provided service A bearer characteristic measuring means for measuring bearer characteristics of packets transmitted intermittently based on any one or a combination of communication protocol, IP address, port number and packet length, or (2) provided Service quality measurement means for measuring the quality of service for each service; (3) measurement information acquisition means for acquiring measurement information obtained by measuring traffic information of transmission packets by one or more node devices existing on the physical network; , from among the (4) bearer characteristics and measurement information measured by the measuring means and service quality measuring means, which is acquired by the measurement information acquisition unit measurement information, The measurement result measured at the time when the measurement packet is intermittently sent by the error characteristic measurement means and the service quality measurement means, and the measurement result measured at the time when the measurement packet is not sent by the measurement information acquisition means. Group creation means for obtaining measurement information that tends to be similar for each packet based on a small difference, and grouping the packets. (5) Packets that are grouped by the group creation means communicate using the same virtual network. And a network state estimation unit that determines that the network state is determined and outputs an estimated value of the network state for each virtual network.

According to a second aspect of the present invention, there is provided a network state estimation program for estimating a state of one or a plurality of virtual networks constructed by dividing a bandwidth of a physical network into a plurality of (1) provided service, communication protocol, IP Bearer characteristic measuring means for measuring bearer characteristics of packets transmitted intermittently based on any or a combination of address, port number and packet length, or (2) quality of service provided Service quality measurement means for measuring each time, (3) measurement information acquisition means for acquiring measurement information obtained by measuring traffic information of transmission packets by one or a plurality of node devices existing on the physical network, and (4) bearer characteristic measurement means and services and measurement information measured by the quality measuring means, from the obtained measurement information by measuring information obtaining means, bearer characteristics meter The difference between the measurement result measured at the time when the measurement packet is intermittently sent by the means and the service quality measurement means and the measurement result measured at the time when the measurement packet is not sent by the measurement information acquisition means based on, group creation means for performing grouping seeking measurement information tends to similar by packet, when communicating by using (5) the same virtual network packets with the same group by the group creating means A network state estimation program that functions as a network state estimation unit that determines and outputs an estimated value of a network state for each virtual network.

  According to the present invention, for the purpose of, for example, power saving control, when the bandwidth utilization rate is low, the network state is correctly grasped even in a virtual network that is operated such that the network control function narrows down available resources. be able to.

It is a block diagram explaining the example of whole structure of the network of embodiment, and the internal structure of a network state estimation apparatus. It is explanatory drawing explaining the structure of a physical network, and the structure of a virtual network. 5 is a flowchart illustrating an operation example of virtual network determination and virtual network state estimation processing by a network state estimation function unit in the embodiment. It is explanatory drawing explaining the determination of a virtual network, and the state estimation process of a virtual network in embodiment. In embodiment, it is explanatory drawing explaining the relationship between the load information provided according to the application, each measured value, and the usable bandwidth estimation value of a virtual network. In embodiment, it is explanatory drawing explaining the method of estimating the state of the present true network based on the information of the present measured value and the past measured value. In embodiment, it is explanatory drawing explaining the operation | movement which calculates | requires the estimated value of the quality of the additional service at the time of adding a service from the present measured value, and the existing service. In an embodiment, it is an image figure explaining a method of estimating a state of a virtual network using a set of a plurality of past measurement values.

(A) Embodiment Hereinafter, a main embodiment of a network state estimation apparatus and a network state estimation program of the present invention will be described in detail with reference to the drawings.

(A-1) Configuration of Embodiment FIG. 1 is a configuration diagram illustrating an example of the overall configuration of a network according to this embodiment and an internal configuration of a network state estimation device.

  In FIG. 1, the network 7 of this embodiment includes a plurality of (four in FIG. 1) routers 1 to 4, a plurality of terminals 5, a server 6, and a network state estimation device 10. In addition, the number of routers 1-4, the terminal 5, the server 6, etc. is not specifically limited.

  The routers 1 to 4 are transfer devices that perform transfer processing by connecting to a user-side network (LAN) terminal 5 and server 6 via a user network interface (UNI). The routers 1 to 4 are connected to each other by a plurality of links, and use these plurality of links to divide the bandwidth for each service to construct one or a plurality of virtual networks. For example, the routers 1 to 4 can apply a transfer device that supports a standardized technique standardized by IEEE 802.3ad LACP (Link Aggregation Control Protocol).

  The terminal 5 is a terminal device used by the user, and receives a service provided by the server 6. The terminal 5 corresponds to, for example, a TV conference terminal, a video viewing terminal, a WEB terminal, or the like.

  The server 6 is a server that provides various services, and corresponds to, for example, a TV conference server, a moving image distribution server, a WEB server, or the like.

  The network state estimation device 10 is connected to a user-side network or routers 1 to 4, measures the communication state of packets exchanged between the terminal 5 and the server 6, and based on the measurement result, the virtual network The usage state of is estimated.

  FIG. 1 shows main functions realized by the network state estimation apparatus 10. The network state estimation device 10 is configured to include, for example, a CPU, ROM, RAM, EEPROM, input / output interface, and the like. The function of the network state estimation apparatus 10 is realized by the CPU executing a network state estimation program stored in the ROM. The network estimation program may be downloaded and installed through a network, and even in that case, the network estimation program functionally has the configuration shown in FIG.

  As shown in FIG. 1, the network state estimation device 10 includes an interface function unit 11, a control function unit 12, an application quality measurement function unit 13, a virtual network bearer characteristic measurement function unit 14, a node measurement information acquisition function unit 15, A load application function unit 16, a measurement information storage function unit 17, and a network state estimation function unit 18 are included.

  The interface function unit 11 is an interface unit connected to a physical network, and is used as an entrance through which an application quality measurement function unit 13 and a virtual network bearer characteristic measurement function unit 14 described later acquire information on the physical network. Specifically, the interface function unit 11 connects to the same UNI as the terminal 5 and the server 6 or connects to a mirror port of a node device such as the routers 1 to 4.

  The application quality measurement function unit 13 measures the quality of each application when executing an application program such as a TV conference, on-demand video reception, or WEB, for example.

  As the quality measured by the application quality measurement function unit 13, for example, either objective quality or bodily sensation quality (subjective quality) or both can be applied. The method for measuring the objective quality and the quality of experience is not particularly limited, and various methods can be widely applied. For example, as an objective quality measurement method for video, ITU-T Recommendation J.I. 247 and ITU-T recommendation J.I. For example, a MOS (Mean Opinion Score) value, a DSCQS (Double Stimulus Continuous Scale) value, or the like can be applied as a video quality measurement method.

  When the service is provided, the terminal 5 of the user measures the objective quality and the quality of experience, and when the network state estimation device 10 can acquire the measurement information from the terminal 5, the terminal 5 may be substituted for it. .

  The virtual network bearer characteristic measurement function unit 14 measures the state of the virtual network of each service.

  The virtual network bearer characteristic measurement function unit 14 can widely apply a conventional available bandwidth estimation technique. In other words, the virtual network bearer characteristic measurement function unit 14 can apply any of the measurement methods of active type, passive type, and inline type.

  The active type usable bandwidth estimation method is a method in which a host sends a measurement packet and estimates the state (available bandwidth) of a virtual network based on measurement results such as a transfer delay and a packet arrival time. The passive type available bandwidth estimation method is a method of monitoring traffic passing through a measurement point in a virtual network, collecting the monitoring information, and estimating the state (available bandwidth) of the virtual network. The in-line type usable bandwidth measurement method is a method in which a packet from the transmission side is transferred as it is by the reception side, and an available bandwidth is estimated based on a transmission interval, a reception interval, or the like on the transmission side.

  For example, the virtual network bearer characteristic measurement function unit 14 includes, as packet information, a provided service (application) type, a protocol to be used (for example, TCP, UDP, etc.), an IP address (source IP address, destination IP address). , Port number (source port number, destination port number), packet length, etc., or a combination of all or part of the packet delay, packet delay variation, packet transmission or reception interval variation, and packet loss Measure the rate and traffic volume.

  Moreover, it is more suitable if the virtual network bearer characteristic measurement function part 14 can measure with respect to the same virtual network with a some measurement method.

  Furthermore, the measurement result measured by the virtual network bearer characteristic measurement function unit 14 may be, for example, a statistical value such as a traffic amount, a delay time distribution, or the like, or the availability of each virtual network derived based on the statistical value. It may be a band value.

  As a measurement method by the virtual network bearer characteristic measurement function unit 14, for example, a method of performing measurement by performing the same connection as that of another general terminal 5 or server 6 through the UNI from the interface function unit 11 can be applied. As another method, output information from a mirror port of a node device such as the routers 1 to 4 may be used, and information about the first several bytes of traffic input to the node device may be used.

  The node measurement information acquisition function unit 15 acquires, for example, quality information measured in the node devices such as the routers 1 to 4 and quality information that can be stored or output by the node device. The quality information measured by the node device is stored in the MIB in the node device and is disclosed to the outside. In this case, the node measurement information acquisition function unit 15 can acquire quality information by communicating with the node device using a protocol such as SNMP (Simple Network Management Protocol). The quality information from the node device is mostly statistical information such as traffic volume and delay time distribution.

  The application-by-application load assigning function unit 16 applies traffic (packets) to the virtual network for each application and applies a load to the transmission path, or transmits a connection request packet to apply the load.

  The measurement information accumulation function unit 17 stores a measurement result measured by the application quality measurement function unit 13 and the virtual network bearer characteristic measurement function unit 14 and quality information acquired by the node measurement information acquisition function unit 15 (not shown). ).

  The measurement information storage function unit 17 also stores information on the load applied to the virtual network by the application-specific load application function unit 16. At this time, the measurement information storage function unit 17 recognizes what measurement change has occurred with respect to the applied network load when the application-specific load application function unit 16 applies a load to the network for each application. In order to be able to do so, for example, the time axis is adjusted, and the measurement result for load application and the measurement information of the node device are stored so as to be able to cope with them.

  The network state estimation function unit 18 estimates the state of the network based on the information stored in the measurement information storage function unit 17 and outputs the estimation result to a display unit (for example, a display). The network state estimation method by the network state estimation function unit 18 will be described in detail in the operation section.

  The control function unit 12 controls and manages the functions of the network state estimation apparatus 10.

(A-2) Operation | movement of embodiment Next, operation | movement of the network state estimation method of this embodiment is demonstrated in detail, referring drawings. In the following, since the state of the virtual network constructed for each service is estimated, the logical virtual network of FIG. 2 will be described as appropriate.

  Prior to the description of the operation, the principle by which the network state estimation apparatus 10 according to the present invention correctly estimates the state of the network will be described.

  First, the reason why the state of the virtual network cannot be estimated correctly in the prior art will be briefly explained again.

  The first reason is that, as described above, when the state of the virtual network is measured using the measurement packet, the measurement packet flows in a virtual network different from the virtual network to be measured, such as a conference TV virtual network. by.

  The second reason is that when the bandwidth of the virtual network to be measured has a margin, the network side narrows the bandwidth and cannot detect it. Originally, the system side automatically expands the bandwidth that has been narrowed down automatically when there is no room in the virtual network bandwidth, but the available bandwidth measurement technology basically aims to not put a load on the network. Measurement is performed with a load that cannot be detected by the side (for example, a load that is recognized as a burst that has occurred by chance).

  Therefore, the network state estimation apparatus 10 according to the present invention estimates the state of the virtual network based on the following principle.

(A-2-1) Principle of virtual network state estimation (1) The network state estimation device 10 obtains information on the quality of the virtual network to be evaluated by measuring the quality for each application.

  Further, there may be a case where the virtual network does not correspond to the application 1: 1, and a plurality of applications are collected in the same virtual network. Therefore, in preparation for such a case, the network state estimation device 10 compares the estimated value of the network state such as the quality measurement value and the available bandwidth for each application, and applications having the same value (or an approximate value) are in the same group. Assuming that the same virtual network is used.

  The virtual network is not prepared for each application, but is prepared for each user group (for example, a company) such as VPN. For example, from the viewpoint of fairness, a heavy user And light users. Therefore, the network state estimation device 10 is not limited to measuring the quality for each application. For example, the network state estimation device 10 measures the quality for each attribute such as the group to which the user belongs and the usage frequency, and the quality of the virtual network for each characteristic. May be obtained.

(2) The network state estimation device 10 does not measure the quality of one specific application, but evaluates the quality of one or more other applications and makes a comprehensive determination.

  For example, if the usable bandwidth of the video conference virtual network 1 is measured as 10 Mbit / sec and the usable bandwidth of the other application virtual network 3 is measured as 100 Mbit / sec, both links are used, and the video conference virtual network is used. It can be seen that the usable bandwidth of the network 1 is really only 10 Mbit / sec.

  However, when the usable bandwidth of the video conference virtual network 1 is 10 Mbit / sec and the usable bandwidth of the other application virtual network 3 is 20 Mbit / sec, both of the links are alive, and the remaining is really only that. One link has dropped due to power-saving control, and it cannot be determined by itself whether it can actually be used more.

  Therefore, the network state estimation function unit 18 performs a virtual network determination and a virtual network state estimation process for each application as follows.

(3) Virtual Network Determination Processing and State Estimation Processing for Each Application FIG. 3 is a flowchart illustrating an operation example of virtual network determination processing and virtual network state estimation processing by the network state estimation function unit 18. FIG. 4 is an explanatory diagram for explaining virtual network determination and virtual network state estimation processing.

  First, before starting operation, the virtual network bearer characteristic measurement function unit 14 uses, for example, an application type, a communication protocol (eg, TCP, etc.), a destination, and the like by various measurement methods such as an active type, a passive type, and an inline type. Measure bearer characteristics of the network such as by IP address and / or destination port number and by packet length.

  At this point, since the virtual network for each application has not been determined, the virtual network bearer characteristic measurement function unit 14 measures the bearer characteristic of the network measured for each application.

  In addition, the application quality measurement function unit 13 measures a quality value for each application based on objective quality and / or quality of experience. Furthermore, the node measurement information acquisition function unit 15 acquires node measurement information measured by the routers 1 to 4 and the like.

  In this way, the measurement information storage function unit 17 uses the application-specific load assignment function unit 16 to measure a large amount of measured values, that is, a virtual network bearer characteristic measurement function, for load information of various application patterns. The measurement values measured by the unit 14 and the application quality measurement function unit 13 and the node measurement information acquired by the node measurement information acquisition function unit 15 are accumulated in the measurement information accumulation function unit 17 (S101).

  In FIG. 3, the network state estimation function unit 18 refers to the bearer characteristic measurement value of the network for each application and the quality measurement value for each application stored in the measurement information storage function unit 17 (S102), and estimates the network state. The function unit 18 detects slight differences between various observation results obtained by a plurality of measurement methods.

  4A shows, for example, a video conference virtual network in which one of the two links of the router 1 and the router 2 is dropped and the available bandwidth is 10 Mbit / sec using the link for one link. 1 and other application virtual network 3 are constructed.

  4A-1 is an example of the measured bearer characteristics of the network in the case of FIG. 4A, and FIG. 4A-2 is an application of the TV conference service in the case of FIG. 4A. It is an example of a bodily sensation quality measurement value. The horizontal axis represents time.

  On the other hand, FIG. 4B shows that, for example, the TV conference virtual network 1 having a possible bandwidth of 10 Mbit / sec is constructed by using one of the two links of the router 1 and the router 2. The case where the virtual network 3 for other applications is constructed using one link is shown.

  4B-1 is an example of the measured bearer characteristics of the network in the case of FIG. 4B, and FIG. 4B-2 is an application of the TV conference service in the case of FIG. 4B. It is an example of a bodily sensation quality measurement value. The horizontal axis represents time.

  For example, as shown in FIG. 4A, when one link is dropped and the video conference virtual network 1 and the other application virtual network 3 use the same link, In this way, there is a strange difference between the measured network bearer characteristics and application quality when all links are in operation. The network state estimation function unit 18 makes a virtual network determination by identifying by machine learning based on this slight difference.

(4) Description of slight difference between a plurality of measurement methods For example, in FIG. 4, the virtual network bearer characteristic measurement function unit 14 measures the bearer characteristic of the network in an active type. That is, the virtual network bearer characteristic measurement function unit 14 sends and measures, for example, an ICMP packet as a measurement packet to the network, as in the conventional case.

  As described above, since the available bandwidth estimated by the measurement packet (ICMP packet) is the available bandwidth of the other application virtual network 3, the usable bandwidth of the TV conference virtual network 1 cannot be measured. Since the measurement packet is output to the network, the user packet traffic is affected.

  For example, as shown in FIG. 4A, when an active measurement is performed when the TV conference virtual network 1 and the other application virtual network 3 share one link, FIG. As shown in A-3) and (A-4), there is a slight difference between the measured value of the bearer characteristic of the network and the measured value of the app experience quality.

  4A-3 and 4A-4, the dotted lines are the measured values in FIGS. 4A-1 and 4A-2, and the solid line is the active-type measured value.

  On the other hand, as shown in FIG. 4B, when the TV conference virtual network 1 and the other application virtual network 3 have different links, the TV conference can be performed even if the measurement packet is sent to the network. The traffic of user packets in the virtual network 1 is not affected.

  Therefore, for example, as shown in FIGS. 4 (B-3) and (B-4), no change is observed in the bearer characteristic measurement value and the app experience quality measurement value of the network.

  As described above, the virtual network bearer characteristic measurement function unit 14 performs active measurement intermittently (or periodically), accumulates the measurement value in the measurement information accumulation function unit 17, and performs the network state estimation function unit. 18 detects the degree of influence of packets (traffic) caused by active measurement values.

  Here, the case where the measurement packet is transmitted to the network by active measurement to detect the above-described difference is illustrated, but the present invention is not limited to active measurement. That is, the application-specific load giving function unit 16 may give a load for each application to the network. For example, even when the application-specific load assignment function unit 16 transmits a video conference packet to the network, the video conference packet is transmitted through the video conference virtual network 1. The link state between the virtual network 1 and the other application virtual network 3 can be detected. However, this method has the possibility of adversely affecting the network in operation, so care must be taken when using it continuously. That is, consideration should be given to immediately stopping the load when the available bandwidth approaches zero.

  Next, the network state estimation function unit 18 performs machine learning on the above-described difference with the past accumulated data of the measurement information storage function unit 17 so that the difference in the measurement value of the active measurement is equal (or approximate). When there is an application that becomes (value), these applications are grouped into one group, and the network state estimation function unit 18 determines that this group is the same virtual network (S103).

  The machine learning method is not particularly limited as long as it is a method capable of obtaining a value equivalent to the above-described slight difference by collating with past measurement values by various measurement methods. A method may be used, or an original calculation model may be used.

  Thereby, a virtual network for each application can be determined. Further, even when a plurality of applications are collected in the same virtual network, the virtual network for each application can be determined by the above method.

  Then, the network state estimation function unit 18 can obtain the measured quality value of the virtual network and the estimated value of the available bandwidth based on the various measured values for each determined virtual network, and displays them on the display unit (for example, Etc.). The virtual network state estimation method will be described in detail below.

(A-2-2) State estimation process of virtual network (S104, S105)
Next, the operation of virtual network state estimation processing by the network state estimation function unit 18 will be described.

  Until it becomes possible to actually estimate the state of the virtual network, it is necessary to acquire and learn response data for the load on the network.

  First, the application-specific load assignment function unit 16 assigns load information to the network in various patterns for each application.

  For example, the application-specific load function unit 16 applies a load by sending packets by changing the transfer rate, transmission time, etc. of video and audio packets in various patterns for each application such as a TV conference.

  Further, for example, the application-specific load function unit 16 executes two services, for example, a TV conference and on-demand video distribution at the same time, and activates services by a plurality of applications at the same time.

  The virtual network bearer characteristic measurement function unit 14 determines, for example, for each application, for each communication protocol (for example, TCP), for each destination IP address and / or each destination port number, for the load of the application-specific load assignment function unit 16, for packet length Measure bearer characteristics of other networks. It is more preferable that the virtual network bearer characteristic measurement function unit 14 can simultaneously measure the same virtual network using a plurality of measurement methods.

  Further, the application quality measurement function unit 13 measures the quality value for each application based on the objective quality and / or the quality of experience with respect to the load of the load application function unit 16 for each application. Furthermore, the node measurement information acquisition function unit 15 acquires node measurement information measured by the routers 1 to 4 and the like with respect to the load of the application-specific load application function unit 16.

  In this way, with respect to load information of various application-specific patterns, a large amount of measurement values, that is, measurement values measured by the virtual network bearer characteristic measurement function unit 14 and the application quality measurement function unit 13, and node measurement information The node measurement information acquired by the acquisition function unit 15 is stored in the measurement information storage function unit 17.

  FIG. 5 is an explanatory diagram for explaining the relationship between the load information and each measured value given for each application, and the available bandwidth estimation value of the virtual network. In FIG. 5, the horizontal axis represents time.

  FIG. 6 is an explanatory diagram for explaining a method of estimating the current true network state based on information on the current measurement value and the past measurement value.

  Hereinafter, a method for estimating the state of the network during operation (a certain time Tn) will be described.

  First, since there is a possibility of adversely affecting other traffic, the control function unit 12 stops the application-specific load provision function unit 16.

  Also, the control function unit 12 stops the function of applying a load to the network for the virtual network bearer characteristic measurement function unit 14, the application quality measurement function unit 13, and the node measurement information acquisition function unit 15.

  For example, the control function unit 12 places a burden on the network such as passive type or inline type measurement and report information such as RTCP-XR (RTP Control Protocol-Extended Report) on the virtual network bearer characteristic measurement function unit 14. Let me measure only with no method.

  In this state, the virtual rope bearer characteristic measurement function unit 14, the node measurement information acquisition function unit 15, and the application quality measurement function unit 13 perform measurement at time Tn. A set of obtained measurement values (network quality information) is Q (Tn) (S201).

  Next, the network state estimation function unit 18 searches the measurement information accumulation function unit 17 for a set Q (Ta) of past measurement values closest to the measurement value set Q (Tn) (S202). The set Q (Ta) is a set of measurement values at the past time Ta. The search can use a search using a machine learning such as a neuro estimation or a search for a set having the smallest total difference of measurement points.

  Then, the network state estimation function unit 18 obtains the measurement value Q (Tz) when the load is further applied by the application-specific load applying function unit 16 from the state where Q (Ta) has been measured in the past, and the measurement information accumulation function unit 17. (S203).

  If the available bandwidth is widened by applying a load, the network state estimation function unit 18 assumes that the link has been dropped by the power-saving control on the network side, and actually flows how much from the past learning value. It retrieves information about whether or not there was a thing and outputs it as the value of the usable bandwidth of the true virtual network.

  On the other hand, if the available bandwidth does not widen and the quality deteriorates, the network state estimation function unit 18 understands that all links are used, and that the measured value at the present time is correct as the available bandwidth. Can be output as a bandwidth value.

(A-2-3) When a new service is requested to be added Next, an operation for estimating the quality assumed when a new service is requested for addition will be described.

  FIG. 7 is an explanatory diagram for explaining an operation for obtaining an estimated value of quality of an additional service and an existing service when a service is added from the current measurement value.

  The application quality measurement function unit 13, the virtual network bearer characteristic measurement function unit 14, and the node measurement information acquisition unit 15 obtain a set Q (Tn) of measurement values at time Tn (S301).

  The network state estimation function unit 18 searches the measurement value set Q (Ta) closest to the measurement value set Q (Tn) at the time tn from the past measurement information stored in the measurement information storage function unit 17. (S302).

  The network state estimation function unit 18 uses the measurement information stored in the measurement information storage function unit 17 to newly request the service A at the time Ta when the past set Q (Ta) is measured. At time Ta + Δ, a set Q (Ta + Δ) of past measurement values indicating the quality of the service that is already being executed or the new service A that has been added is retrieved (S303).

  The network state estimation function unit 18 uses the service A when a new service A is requested to be added at the time Tn + Δ for the measured value of the app experience quality in the set Q (Ta + Δ) of the past measurement values, and is already being executed. The service is output as an estimated value Q (Tn + Δ) (S304).

  Although the above method exemplifies the case of selecting the past measurement value closest to the measurement value at the current time, the method is not limited to this method.

  As another method, the measurement value set Q (Tn) at the current time Tn falls within the measurement value sets Q (Ta), Q (Tb), and Q (Tc) at the past times Ta, Tb, and Tc. If they are close or close to each other, an estimated value can be obtained by linear interpolation or the like.

  FIG. 8 is an image diagram illustrating a method for estimating the state of a virtual network using a set of a plurality of past measurement values. In FIG. 8, the horizontal axis represents an element (measurement value x, for example, bearer characteristic measurement value) of the measurement value set Q, and the vertical axis represents an element (measurement value y, for example, an application) of the measurement value set Q. This is another quality measurement value).

  For example, a set Q (Tn) of measured values at the current time Tn falls within a set of measured values Q (Ta), Q (Tb), and Q (Tc) at past times Ta, Tb, and Tc. The estimated value Q (Tn + Δ) of the network state at the time Tn + Δ is constant among the sets Q (Ta + Δ), Q (Tb + Δ), and Q (Tc + Δ) of the measured values at the past times Ta + Δ, Tb + Δ, and Tc + Δ. It is possible to make transitions with a relationship.

  Therefore, the network state estimation function unit 18 obtains a relationship between the measurement value set Q (Tn) at the current time and the previous measurement value sets Q (Ta), Q (Tb), and Q (Tc). The estimated value Q (Tn + Δ) of the network state at the time Tn + Δ is output using a set of measured values Q (Ta + Δ), Q (Tb + Δ), and Q (Tc + Δ) to reflect a predetermined relationship.

  For example, the positional relationship between the set of measured values Q (Ta), Q (Tb), Q (Tc) of the past three points and the set of measured values Q (Tn) at the current time Tn is the same at time Tn + Δ. Assuming that there is a relationship, the estimated value Q (Tn + Δ) may be obtained using a set of measured values Q (Ta + Δ), Q (Tb + Δ), and Q (Tc + Δ).

  In FIG. 8, the relationship between the set of three past measurement values and the set of measurement values at the current time has been described as an example. However, the set of past measurement values may be two points. Four or more points may be used.

  In the above description, a method for obtaining an estimated value of quality when one new service is added has been described. The quality when M new services are added is estimated. For example, there is no quality degradation even if five 10 Mbit / sec services are added in a virtual network for video conferences. When it is estimated that it will occur, the available bandwidth is estimated to be between 50 Mbit / sec and 60 Mbit / sec and output, and may be used as an available bandwidth estimation method different from the available bandwidth estimation method shown in FIG. it can.

(A-3) Effect of Embodiment As described above, according to the embodiment, even in a network in which a virtual network is used and a bandwidth that can be used for each application is separated, an available bandwidth or a new service request is issued. Service quality and quality of existing services can be estimated.

  In addition, according to the embodiment, when there is a vacant bandwidth, an available bandwidth or a new service request is made even in a network that reduces available resources such as dropping a link for the purpose of power saving. Service quality and quality of existing services can be estimated.

(B) Other Embodiments In the above-described embodiments, the case where a virtual network for each application is determined has been described. However, there are cases where a virtual network is constructed according to the group to which the user belongs and the frequency of use by the user. Therefore, in order to be able to determine the virtual network according to such attributes, for example, the virtual network bearer characteristic measurement function unit 14 performs measurement for each VPN identification information, each source IP address and / or port number, and the like. Then, the network state estimation function unit 18 also determines the virtual network according to the group to which the user belongs and the frequency of use by using the measured values for each VPN identification information, the source IP address and / or the port number, etc. You may do it.

10: Network state estimation device,
11 ... interface function unit, 12 ... control function unit,
13 ... Application quality measurement function unit, 14 ... Virtual network bearer characteristic measurement function unit,
15 ... Node measurement information acquisition function unit, 16 ... Load application function unit by application,
17 ... Measurement information storage function part, 18 ... Network state estimation function part,
1-4 ... router, 5 ... terminal, 6 ... server, 7 ... network.

Claims (5)

In a network state estimation device for estimating the state of one or a plurality of virtual networks constructed by dividing a bandwidth of a physical network into a plurality,
Bearer characteristic measuring means for measuring bearer characteristics of a packet transmitted intermittently based on any one of a provided service, a communication protocol, an IP address, a port number, a packet length, or a combination of all or a part thereof;
A service quality measuring means for measuring the quality of services provided for each service;
Measurement information acquisition means for acquiring measurement information obtained by measuring traffic information of a transmission packet by one or a plurality of node devices existing on the physical network;
From the measurement information measured by the bearer characteristic measurement unit and the service quality measurement unit and the measurement information acquired by the measurement information acquisition unit, intermittently by the bearer characteristic measurement unit and the service quality measurement unit. Measurements that tend to be similar based on the slight difference between the measurement result measured at the time when the measurement packet was sent and the measurement result measured at the time when the measurement packet was not sent by the measurement information acquisition means A group creation means for obtaining information by packet and performing grouping;
Determines that packets with the same group by the group creation means is communicating using the same virtual network, characterized in that it comprises a network state estimation means for outputting an estimated value for each of the network state virtual network A network state estimation device. Load applying means for applying a predetermined load to each of the virtual networks;
For the load information provided by the load applying means in a plurality of patterns, the measurement information measured by the bearer characteristic measuring means and the service quality measuring means and the measurement information acquired by the measurement information acquiring means are accumulated. Measurement information storage means, and
The network state estimation means is
For the measurement information at the current time, search the past measurement information closest to the measurement information at the current time from the measurement information storage means,
From the retrieved past measurement state, the measurement information when the load is applied by the load application unit is searched from the measurement information storage unit,
When searching for measurement information indicating that the available bandwidth has expanded when the load is applied, search for information that could actually flow the packet to the virtual network from the past measurement information of the measurement information storage means. , Output it as true virtual network status information,
The measurement value at the present time is output as true virtual network state information when searching for measurement information that the usable bandwidth does not widen and the quality deteriorates even when the load is applied. The network state estimation apparatus according to 1. The network state estimation means is
For the measurement information at the current time, search the past measurement information closest to the measurement information at the current time from the measurement information storage means,
In the past time when the searched past measurement information is measured, a new service is added, and the measurement information after the addition of the new service is searched from the measurement information storage unit,
3. The network according to claim 2, wherein the measurement information after adding a new service retrieved from the measurement accumulation information is output as measurement information of each virtual network when a new service is added at the current time. State estimation device. The network state estimation apparatus is
Obtain a plurality of past measurement information from the measurement information storage means, obtain the relationship between the plurality of past measurement information and the current time measurement information,
A new service at the current time using the past measurement information after adding the new service and the relationship between the past measurement information and the past measurement information from the past measurement time of the plurality of past measurement information. The network state estimation apparatus according to claim 2 or 3, wherein measurement information of each virtual network when the is added is output. In a network state estimation program for estimating the state of one or a plurality of virtual networks constructed by dividing a bandwidth of a physical network into a plurality of pieces,
Computer
Bearer characteristic measuring means for measuring the bearer characteristic of a packet transmitted intermittently based on any one of a provided service, a communication protocol, an IP address, a port number, or a packet length, or a combination of all or a part thereof,
Service quality measurement means for measuring the quality of services provided for each service,
Measurement information acquisition means for acquiring measurement information obtained by measuring traffic information of a transmission packet by one or a plurality of node devices existing on the physical network;
From the measurement information measured by the bearer characteristic measurement unit and the service quality measurement unit and the measurement information acquired by the measurement information acquisition unit, intermittently by the bearer characteristic measurement unit and the service quality measurement unit. Measurements that tend to be similar based on the slight difference between the measurement result measured at the time when the measurement packet was sent and the measurement result measured at the time when the measurement packet was not sent by the measurement information acquisition means Group creation means for obtaining information by packet and performing grouping,
Characterized in that packets with the same group by the group creation means determines that communicate using the same virtual network, to function as the network state estimation means for outputting an estimated value for each of the network state virtual network Network status estimation program.

Images