WO2009145011A1 - 最適化評価システム、最適化評価装置、最適化評価方法、及び最適化評価用プログラム - Google Patents
最適化評価システム、最適化評価装置、最適化評価方法、及び最適化評価用プログラム Download PDFInfo
- Publication number
- WO2009145011A1 WO2009145011A1 PCT/JP2009/057268 JP2009057268W WO2009145011A1 WO 2009145011 A1 WO2009145011 A1 WO 2009145011A1 JP 2009057268 W JP2009057268 W JP 2009057268W WO 2009145011 A1 WO2009145011 A1 WO 2009145011A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- communication traffic
- traffic
- optimization
- communication
- probability density
- Prior art date
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
- H04L43/0823—Errors, e.g. transmission errors
- H04L43/0829—Packet loss
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/123—Evaluation of link metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/125—Shortest path evaluation based on throughput or bandwidth
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/70—Routing based on monitoring results
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/04—Processing captured monitoring data, e.g. for logfile generation
- H04L43/045—Processing captured monitoring data, e.g. for logfile generation for graphical visualisation of monitoring data
Definitions
- the present invention provides an optimization evaluation system, an optimization evaluation apparatus, an optimization evaluation method, and an optimization evaluation that evaluates the effect of optimization by the optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics. Related to the program.
- Optimization generally means making the best decision in a given situation, or choosing the best of several options.
- Such an optimization problem is formulated as a mathematical model as follows.
- the objective function f is a real-valued function defined on a suitable set including S.
- S represents the set of values that the variable x can take in this optimization problem.
- the Internet which is a communication network.
- the TCP protocol has a mechanism called slow start for optimally controlling the transmission amount of communication traffic so as to eliminate the spread of congestion on the network.
- backpressure congestion control in Ethernet registered trademark
- MAC control protocol to perform flow control
- an optimization method or the like for transferring only traffic for a specific application on a communication network as quickly as possible is also performed in a P2P (Peer-to-Peer) network or the like.
- Non-Patent Document 4 J. C. Doyle, S. H. Low et al.'S group and Fukuda et al. Also reported that the cause of such traffic behavior is the feedback control in TCP and Ethernet and the buffer function (or delay) mechanism in the feedback itself (non- See Patent Documents 1 to 3). In addition, it has been confirmed that traffic efficiency is maximized at a critical point in phase transition in a system having such a mechanism (see Non-Patent Document 4).
- the optimal control method for communication traffic is basically only that whose effectiveness has been confirmed by a simple network model on simulation, including those mentioned above as examples. Therefore, it is not possible to know whether those optimization methods are really optimal (that is, whether the objective function is minimized or maximized) on the actual Internet or other communication networks.
- an object of the present invention is to provide an optimization evaluation system, an optimization evaluation device, an optimization evaluation method, and a program for optimization evaluation that can quantitatively evaluate the effects of optimization of traffic characteristics in a communication network. .
- the optimization evaluation system is an optimization evaluation system for evaluating the effect of optimization by an optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics, and uses measured communication traffic data.
- Communication traffic analysis means for obtaining fluctuation distribution of communication traffic, and communication traffic characteristics that are executed by the optimization function based on whether the fluctuation distribution calculated by the communication traffic analysis means is a power function or not
- communication traffic evaluation means for performing processing for quantitatively evaluating the effect of the optimization.
- the optimization evaluation apparatus is an optimization evaluation apparatus for evaluating the effect of optimization by the optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics, and uses measured communication traffic data.
- Communication traffic analysis means for obtaining fluctuation distribution of communication traffic, and communication traffic characteristics that are executed by the optimization function based on whether the fluctuation distribution calculated by the communication traffic analysis means is a power function or not
- communication traffic evaluation means for performing processing for quantitatively evaluating the effect of the optimization.
- the optimization evaluation method is an optimization evaluation method for evaluating the effect of optimization by the optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics, and is used to measure communication traffic data. Based on the fluctuation distribution of communication traffic, and quantitatively evaluating the effect of the optimization of the communication traffic characteristics executed by the optimization function based on whether the calculated fluctuation distribution is a power function or not It is characterized by
- the optimization evaluation program is an optimization evaluation program for evaluating the effect of optimization by the optimization function in a communication network provided with an optimization function for optimizing communication traffic characteristics, The function of obtaining the fluctuation distribution of communication traffic based on the measured communication traffic data, and the optimum of the communication traffic characteristic executed by the optimization function based on whether the calculated fluctuation distribution is a power function or not And the function of quantitatively evaluating the effect of
- the effect of optimization of traffic characteristics in a communication network can be quantitatively evaluated.
- Embodiment 1 Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.
- Embodiments of the present invention relate to an optimization method for optimizing traffic characteristics in a communication network, in particular a communication network having a feedback mechanism and a buffer (or delay) function, and to a method of evaluating the optimization method.
- the structure and various conditions of the modeled network in the simulation used to verify the effects of each optimization method are different, and as shown in the first problem, it is optimal.
- a general algorithm for optimizing communication traffic basically, communication traffic is observed and feedback is applied with a delay of a predetermined time (such as use of statistics).
- a delay of a predetermined time such as use of statistics.
- the fluctuation distribution of communication traffic exhibits a phase transition, and the fluctuation distribution of communication traffic exhibits a power law at its critical point, and the efficiency of communication traffic is maximized. Therefore, as a second problem, the effects of adopting an optimization method for traffic characteristics of a large-scale communication network can not generally be compared with the effects of using other optimization methods. .
- the traffic efficiency of the communication network is optimized at the critical point of the phase transition, and the communication traffic fluctuation distribution at that time is free. It is possible to evaluate how much the variables set in the optimization method or the optimization method operate optimally with regard to the optimization method that optimizes the communication traffic efficiency by paying attention to showing the rule Provide a comprehensive evaluation system.
- the traffic efficiency of the communication network is optimized at the critical point of the phase transition, and the fluctuation distribution of the communication traffic at that time has power law.
- an optimization method for optimizing some communication traffic efficiency focusing on the indication, a system capable of quantitatively evaluating the effectiveness of the optimization method and the effectiveness of the optimization variable is provided.
- FIG. 1 is a block diagram showing an example of the configuration of an optimization evaluation system according to an embodiment of the present invention.
- FIG. 1 shows, as an embodiment of the present invention, an optimization method or its optimization method for optimizing traffic characteristics in a communication network, particularly a communication network having a feedback mechanism and a buffer (or delay) function.
- An evaluation method for evaluating how optimally a variable set in the optimization method is operating and a configuration of a system capable of quantitatively grasping the effectiveness are shown.
- the optimization evaluation system communicates on the Internet, which is an existing communication network, a network managed by an Internet service provider, a local area network (LAN), and almost all other small or large communication networks. It can be applied to applications that evaluate the effects of traffic characteristics optimization. In particular, it can be applied to the application of quantitatively evaluating the effectiveness of an optimization method used for traffic characteristics in a communication network having a feedback mechanism and a buffer (or delay) function, and the effectiveness of an optimization variable. In addition, the present invention can be applied to an application for evaluating how optimally these optimization methods or variables set in these optimization methods operate.
- the optimization evaluation system includes a plurality of routers 1 (1a-1, 1a-2, ..., 1b-1, 1b-2, ...), and band control devices 2a, 2b. And traffic monitors 3a and 3b.
- the number of band control devices is not limited to two.
- the optimization evaluation system may include three or more band control devices.
- two traffic monitors 3a and 3b are shown in FIG. 1, the traffic monitoring is not limited to two, and for example, the optimization evaluation system may include three or more traffic monitors.
- the band control device 2 when the band control devices 2a and 2b are comprehensively expressed, or when any one of the band control devices 2a and 2b is referred to, the band control device 2 is also simply expressed. Also, hereinafter, when expressing the traffic monitors 3a and 3b comprehensively, or when referring to either of the traffic monitors 3a and 3b, they are also expressed simply as the traffic monitor 3.
- the router 1 has a function of forwarding the arrived communication packet to the next router in accordance with the routing control setting described therein. Specifically, the router 1 stores in advance a route table for performing route control in a storage unit such as a memory. Then, upon receiving the communication packet, the router 1 performs path control in accordance with the setting contents of the route table, and transmits the received communication packet to the next router.
- a route table for performing route control in a storage unit such as a memory.
- the band control device 2 is realized by, for example, a shaping / filtering device.
- the band control device 2 has a function of determining whether or not to transfer the arrived communication packet to the next router according to the control setting described therein.
- the band control device 2 has a function of transferring only the communication packet set to be transferable to the next router.
- the band control device 2 stores in advance a route table for performing transferability determination and path control in a storage unit such as a memory. Then, when receiving the communication packet, the band control device 2 determines transferability of the received communication packet according to the setting contents of the route table. Also, when determining that transfer is possible, the band control device 2 transmits a communication packet to the next router according to the setting contents of the route table. On the other hand, when determining that transfer is not possible, the band control device 2 controls not to transmit the communication packet to the next router.
- the traffic monitor 3 is realized by an information processing apparatus such as a personal computer which operates according to a program.
- the traffic monitor 3 is basically disposed at the hub regardless of whether the network topology is known or unknown.
- the traffic monitor 3 has a function of observing the state of traffic on the outgoing line side of the router 1 according to the monitoring function setting described therein. Specifically, the traffic monitor 3 stores in advance a monitoring table for monitoring traffic in a storage unit such as a memory. Then, the traffic monitor 3 monitors the state of traffic on the departure side of the router 1 according to the setting contents of the monitoring table.
- the traffic monitor 3 is installed on the outgoing side of the router at least one hop before the router where the band control device 2 is installed on the outgoing side in order to be able to observe as many types of communication traffic as possible. Ru.
- the traffic monitor 3a is installed on the outgoing line side of the router 1a-2 one hop before the router 1a-1 on which the band control device 2a is installed. It is shown.
- the traffic monitor 3a may be disposed in the router 1a-3 two hops before the band control device 2a.
- the traffic monitor 3b is installed on the outgoing line side of the router 1b-2 one hop before the router 1b-1 on which the band control device 2b is installed is shown.
- the traffic monitor 3b may be disposed in the router 1b-3 two hops before the band control device 2b.
- Areas 40a and 40b shown in FIG. 1 indicate communication networks, respectively. Areas 40a and 40b relate particularly to an optimization method for optimizing traffic characteristics in a communication network having a feedback mechanism and a buffer (or delay) function, and which variable is set in the optimization method or the optimization method
- the scope of observable networks is an evaluation method for evaluating whether the system is operating to an extent optimum, and an agent that constructs a system capable of quantitatively grasping the effectiveness.
- the traffic monitor 3a can monitor communication traffic of the communication network indicated by the area 40a.
- the traffic monitor 3b can monitor the communication traffic of the communication network indicated by the area 40b.
- the regions 40a and 40b are comprehensively expressed or when any of the regions 40a and 40b is pointed out, the regions 40a and 40b are also simply expressed as the region 40.
- At least one traffic monitor 3 is disposed in the observable area 40.
- the traffic monitor 3 determines the cumulative probability density distribution of the fluctuation of the communication traffic characteristic such that X ⁇ x when considering the variable X and the traffic characteristic to be optimized that can be expressed by the traffic fluctuation distribution directly or indirectly and the variable X Calculate the complementary probability distribution (CDF) of the cumulative probability density distribution (CDF) of the variation of its communication traffic characteristics such that CDF) or X ⁇ x.
- the variable X is, for example, the congestion duration time or the size of the packet round trip time (RTT).
- the value x is, for example, a numerical value of the congestion duration time or the packet round trip time observed by the traffic monitor 3.
- the traffic monitor 3 further holds (stores) the observation start time and the observation end time of the congestion duration time and the packet round trip time in the memory means 5 described later.
- FIG. 2 is a block diagram showing an example of the configuration of the traffic monitor 3 shown in FIG.
- the traffic monitor 3 includes a communication traffic monitor unit 4, a communication traffic analysis unit 5, a memory unit 6, a communication traffic evaluation unit 7, and a feedback unit 8.
- the communication traffic monitor unit 4 is realized by the CPU and the network interface unit of the information processing apparatus that operates according to a program.
- the communication traffic monitoring unit 4 has a function of observing communication traffic and monitoring its packet size, packet arrival interval, and the like.
- the communication traffic monitor unit 4 receives a communication packet from the router 1 and measures the packet size and packet arrival interval of the received communication packet.
- the communication traffic analysis unit 5 is specifically realized by the CPU of the information processing apparatus that operates according to a program.
- the communication traffic analysis unit 5 has a function of reconstructing observation data by (1) traffic characteristics (packet length, packet arrival interval, etc.) based on observation results of communication traffic by the communication traffic monitoring unit 4. Also, (2) the traffic characteristics to be optimized that can be represented by the traffic fluctuation distribution directly or indirectly, and the communication traffic characteristics such that X ⁇ x when the variable X is considered. It has a function of calculating a cumulative probability density distribution (CDF) of variation, or a complementary distribution (CCDF) of a cumulative probability density distribution (CDF) of variation of the communication traffic characteristic where X ⁇ x.
- CDF cumulative probability density distribution
- CCDF complementary distribution
- the memory means 6 is realized by a storage device such as a memory or a hard disk device provided in the information processing apparatus.
- the memory means 6 holds (1) threshold value (preset value) for each traffic characteristic, (2) traffic characteristic (packet length, packet arrival interval, etc.), and (3) observation start / end time of each packet (storage ).
- the communication traffic evaluation unit 7 is specifically realized by the CPU of the information processing device that operates according to a program.
- the communication traffic evaluation unit 7 uses the analysis result of the communication traffic analysis unit 5 and the information stored in the memory unit 6 to (1) observe data for each traffic characteristic (packet length, packet arrival interval, etc.) with an origin of 10 It has a function to compare with a power function of exponent -1 to -1.3 when taken to the 0-th power of (ie, "1"). Also, (2) the communication traffic evaluation unit 7 is in a state where the observed communication traffic can sufficiently afford the limitations of the buffer of the router 1 and the communication bandwidth, etc., or congestion that exceeds these limitations. It has a function to evaluate (determine) whether or not it is in a state of occurrence. Further, the communication traffic evaluation unit 7 has a function of evaluating (calculating) the degree of the margin and the degree of congestion.
- the feedback unit 8 is realized by a CPU and a network interface unit of an information processing apparatus that operates according to a program.
- the feedback unit 8 has a function of feeding back the evaluation result obtained by the communication traffic evaluation unit 7 to an apparatus / system having an optimization function of optimizing communication traffic characteristics according to the optimization method.
- the feedback unit 8 transmits information indicating the evaluation result obtained by the communication traffic evaluation unit 7 to the band control device 2 realized by the shaping / filtering device. Also, for example, the feedback unit 8 may transmit information indicating the evaluation result obtained by the communication traffic evaluation unit 7 to an apparatus / system that implements a slow-start algorithm in TCP retransmission control. Also, for example, the feedback unit 8 may transmit the evaluation result obtained by the communication traffic evaluation unit 7 to an apparatus / system for realizing backoff control in Ethernet. Furthermore, for example, even if the feedback unit 8 transmits the evaluation result obtained by the communication traffic evaluation unit 7 to an apparatus / system that realizes a network filter (NW filter: for example, only data by a certain application is allowed to pass). Good.
- NW filter for example, only data by a certain application is allowed to pass.
- the storage device (not shown) of the traffic monitor 3 is various for evaluating the effect of optimization by the optimization function in the communication network provided with the optimization function for optimizing communication traffic characteristics.
- the storage device of the traffic monitor 3 is based on communication traffic analysis processing for obtaining a fluctuation distribution of communication traffic based on the measured communication traffic data in a computer, and whether or not the calculated fluctuation distribution is a power function.
- an optimization evaluation program for executing communication traffic evaluation processing which executes processing of quantitatively evaluating the effect of optimization of communication traffic characteristics executed by the optimization function.
- FIG. 3 is a flow chart showing an example of processing in which the optimization evaluation system evaluates the effect of optimization of communication traffic characteristics.
- Preparation 1 Traffic function to calculate the cumulative probability density distribution (CDF) such that X ⁇ x when considering the variable X and the traffic characteristics to be optimized that can be represented by traffic fluctuation distribution directly or indirectly It is mounted on the monitor 3 in advance.
- the variable X is, for example, the congestion duration time or the size of the packet round trip time (RTT).
- the value x is, for example, a numerical value of the congestion duration time or the packet round trip time observed by the traffic monitor 3.
- X ⁇ x indicates a case where the value X of the variable is greater than or equal to a certain value x. That is, XXx, where the horizontal axis represents the variable variable X in the coordinate space, and the vertical axis represents the cumulative probability density distribution, all values of the cumulative probability density distribution when the variable is X are x or more It means that it is the sum of the cumulative probability density distribution of the case.
- the case of obtaining the cumulative probability density distribution (CDF) in which X ⁇ x is obtained is shown, but the cumulative probability density distribution (CDF) in which X ⁇ x may be obtained.
- the cumulative probability density distribution (CDF) satisfying X ⁇ x is calculated, if the complementary probability density distribution (CCDF), which is the complementary distribution, is recalculated, the original accumulation Xxx. It becomes probability density distribution (CDF). Therefore, as the cumulative probability density distribution (CDF), any one of X ⁇ x or X ⁇ x may be calculated after all.
- Preparation 2 The traffic monitor 3 is arranged in advance in the router 1 two or more hops ahead of the bandwidth control device 2, and the total communication traffic on the outgoing side is observed. In addition, it is desirable to arrange at least two traffic monitors 3 in the observation area 40 in consideration of redundancy.
- the traffic monitor 3 needs to be able to obtain all or almost (for example, 90% or more) of the variable X shown in Preparation 1 for communication traffic. As described above, this means that "the total traffic and application-specific traffic may not show the power law", and "the current dominant traffic P2P and the Web are a set of them ( It is because that it is known that aggregated traffic shows power law.
- the traffic monitor 3 observes the communication traffic communicated within the observation area 40 of the communication network as needed by performing the above-mentioned preparation (step S101). Specifically, the traffic monitor 3 receives a communication packet from the router 1 at the previous stage, and measures the packet size and packet arrival interval of the received communication packet.
- the traffic monitor 3 accumulates data indicating the observed communication traffic in the memory means 6 as needed (step S102). Specifically, the traffic monitor 3 stores data of the measured packet size and packet arrival interval in the memory means 6 as needed.
- the traffic monitor 3 calculates the cumulative probability density distribution (CDF) of the fluctuation of the communication traffic characteristics such that X ⁇ x at a predetermined timing (for example, every predetermined time) (step S103).
- the traffic monitor 3 may calculate the complementary distribution (CCDF) of the cumulative probability density distribution (CDF) of the fluctuation of the communication traffic characteristic in which X ⁇ x.
- the traffic monitor 3 uses the calculation function loaded in advance in preparation 1, and the cumulative probability density distribution (CDF) or complementary cumulative probability density distribution (CCDF) of the fluctuation of the communication traffic characteristic observed is: It is determined whether the power law is indicated (step S104). Hereinafter, with reference to FIG. 3, the process of determining whether or not the power law is indicated will be described.
- CDF cumulative probability density distribution
- CCDF complementary cumulative probability density distribution
- the traffic monitor 3 uses the calculation function loaded in preparation 1 in advance, and the cumulative probability density distribution of the fluctuation of the communication traffic characteristics and the index when the origin is taken to the 10th power of 10 (that is, “1”) A process of fitting a power function which is 1 to -1.3 using a curve or linear approximation method such as the least square method is performed.
- the region where the value of the variable observed in the cumulative probability density distribution of the fluctuation of the communication traffic characteristics is large decreases exponentially due to the upper limit of the buffer size of the router 1 and the upper limit of the communication capacity.
- the region where the variable to be observed is small is saturated due to the roughness of its accuracy because normal observation is performed by sampling. For these reasons, it is necessary for the region to be fitted with a power function of exponent -1 to -1.3 when taking the origin to 10 to the power of 10 (that is, "1") be between those regions. is there.
- the traffic monitor 3 accumulates the fluctuations of the communication traffic characteristics obtained by observation and the power functions of the exponent -1 to -1.3 when the origin is set to 10 to the power of 10 (that is, "1"). It is determined whether the probability density distribution fits within one decade (one digit) or more. Also, it is assumed that the traffic monitor 3 conforms within 1 decade (one digit) or more range as the condition that the cumulative probability density distribution of the fluctuation of the communication traffic characteristic obtained by the observation shows the power law. It is determined whether the power law is indicated. Then, when determining that the power law is indicated (Y in step S104), the traffic monitor 3 defines an area indicating the power law as a scaling area, and stores the scaling area in the memory means 6 (step S105).
- the traffic monitor 3 uses the calculation function loaded in advance in preparation 1 to calculate the cumulative probability density distribution of the fluctuation of the communication traffic characteristics and the origin 0 to the power of 10 (that is, “1”
- the scaling region is (Ie, it is set in the range of 0.1 to 0.001 of the power function of exponent-1 when taken to (1).
- the traffic monitor 3 stores the set scaling area in the memory means 6 (step S106).
- the traffic monitor 3 Based on the cumulative probability density distribution (or its complementary distribution) and the scaling area obtained by performing the above processing, the traffic monitor 3 sets an optimization method related to communication traffic characteristics and a setting on the optimization method. Evaluate the optimality of variables.
- the traffic monitor 3 determines whether or not the observed communication traffic can sufficiently afford the limitations of the buffer of the router 1 and the bandwidth of the communication band, or congestion occurs beyond these limitations. It is evaluated (judged) whether or not it is in the state of doing. In addition, when congestion occurs, the traffic monitor 3 evaluates (calculates) the degree of the margin and the degree of congestion.
- the cumulative probability density distribution has a point of 10 at 0 In the scaling region defined by ( ⁇ ) above in the power function of exponent -1 to -1.3 when taken to the power (ie, "1"), the variation of the observed communication traffic characteristics with respect to the power function The cumulative probability density distribution is on the lower side (negative side). Conversely, when congestion occurs, the cumulative probability density distribution of the fluctuation of the observed communication traffic characteristic is on the upper side (positive side) with respect to the power function in the scaling area.
- the traffic monitor 3 determines the index -1 to -1 when the cumulative probability density distribution takes the origin at 10 to the power of 0 (that is, "1") in the scaling region defined in (A) above ( ⁇ ). .3 It is assumed that the case where the cumulative probability density distribution is on the upper side is positive, and the case where the cumulative probability density distribution is on the lower side is negative, from the optimal state. Define the direction of deviation. By doing so, the traffic monitor 3 determines whether or not there is sufficient communication traffic, or congestion occurs, depending on the optimization method applied and the setting variables on the optimization method. It can be evaluated (judged) whether it is or not.
- the traffic monitor 3 performs scaling based on the cumulative probability density distribution (or its complement) obtained in step S103 and the scaling area stored in the memory means 6 in steps S105 and S106. In the area, with respect to the power function, it is determined whether the cumulative probability density distribution of the fluctuation of the communication traffic characteristic observed is the positive side or the negative side (step S107).
- step S107 If the cumulative probability density distribution of fluctuations in communication traffic characteristics observed with respect to the power function is on the positive side (above step S107), the traffic monitor 3 is in a state where communication traffic exceeds the optimum state and there is no margin. It determines (step S108). On the other hand, if the cumulative probability density distribution of the fluctuation of the communication traffic characteristic observed with respect to the power function is on the negative side (below step S107), the traffic monitor 3 has a state where the communication traffic has room until the optimum state. It is determined that (step S109).
- the traffic monitor 3 measures the degree of margin or congestion as a power function of exponent -1 to -1.3 when the origin is taken to the 10th power of 10 (that is, "1"), It is determined by the difference with the cumulative probability density distribution of the fluctuation of the communication traffic characteristic obtained by the observation in the scaling region defined in the above ( ⁇ ). That is, the traffic monitor 3 performs processing to evaluate the amount of deviation of the communication traffic characteristics from the optimum state (step S110). In this case, the traffic monitor 3 obtains the difference using the average value on the horizontal axis in the scaling area defined in the above ( ⁇ ) as the difference to be obtained.
- the traffic monitor 3 can perform quantitative evaluation of the optimization method applied and the degree of optimumness of the setting variables on the optimization method. It becomes possible. Therefore, it is possible to quantitatively grasp the effectiveness of the optimization method and the effectiveness of the optimization variable.
- the cumulative probability density distribution of communication traffic characteristic variation and the origin were taken to the 10th power of 10 (that is, “1” using the calculation function previously installed in Preparation 1 shown in the above ( ⁇ )
- the method of fitting the power function with the time index of -1 to -1.3 using the curve or linear approximation method such as the least squares method is the same as the value obtained from the observed traffic data by Traffic Monitor 3 May be.
- the data result calculated to the double logarithm used in order to show a normal power law may be used, and the data result which calculated either of the vertical axis / horizontal axis to the single logarithm may be used.
- the traffic monitor 3 uses the calculation function loaded in advance in preparation 1 as described in ( ⁇ ) above, and uses the cumulative probability density distribution of fluctuations in communication traffic characteristics and the origin of 10 If it is not determined that the power law state is obtained as a result of comparison with a power function of exponent -1 to -1.3 taken on the 0th power of 0 (ie, "1"), the scaling region Is set in the range of 0.1 to 0.001 of the power function of exponent-1 when the origin is taken to the tenth power of 0 (ie, “1”). Then, the traffic monitor 3 stores the set scaling area in the memory means 6. In the case of “1”, the range of the power function of exponent-1 when the origin to be the scaling area is 10 powers of 0 (that is, “1”) is any one decade (1 digit) from 0.1 to 0.001. Or it may be a range beyond it.
- the traffic monitor 3 may obtain the difference with the cumulative probability density distribution of the change of the communication traffic characteristic using any value on the horizontal axis of the scaling region defined in the above ( ⁇ ). Also, the traffic monitor 3 may be obtained using the difference value in the largest value of the scaling area defined in the above ( ⁇ ).
- the traffic monitor 3 may obtain the difference according to the method shown in FIG. 4 as “the difference between the variation of the characteristic and the cumulative probability density distribution”. That is, the traffic monitor 3 accumulates the data obtained by observation with the traffic monitor 3 and the power function of the exponent -1 to -1.3 when the origin is set to 10 to the power of 0 (that is, "1"). A value obtained by integrating the difference between the probability density distribution and its scaling region may be obtained.
- FIG. 4 shows, by way of example, the integral of the difference between the power function and the cumulative probability density distribution in the cumulative probability density distribution of fluctuations in communication traffic characteristics shown in Non-Patent Document 4.
- the traffic monitor 3 may obtain the difference according to the method shown in FIG. 5 as “the difference between the characteristic variation and the cumulative probability density distribution”. That is, traffic monitor 3 accumulates the data obtained by observation at traffic monitor 3 with a power function of exponent -1 to -1.3 when the origin is set to 10 to the power of 10 (that is, "1").
- FIG. 5 shows the integral of the difference between the power function and the cumulative probability density distribution in the cumulative probability density distribution of fluctuations in communication traffic characteristics shown in Non-Patent Document 4.
- the traffic monitor 3 feeds back the evaluation results obtained in the steps S107 to S110 to an apparatus / system (for example, the band control apparatus 2) provided with an optimization function for optimizing communication traffic characteristics according to the optimization method. Specifically, the traffic monitor 3 transmits information indicating whether or not there is enough communication traffic or a table indicating whether congestion is occurring or not to the device / system having the optimization function. . Also, for example, the traffic monitor 3 transmits information indicating the degree of margin of communication traffic and the degree of congestion to a device / system having an optimization function.
- a system administrator who manages an apparatus / system (for example, the band control apparatus 2) having the optimization function operates a terminal for system management, for example, based on evaluation information transmitted from the traffic monitor 3 Confirm the evaluation result of the optimization effect by the optimization function.
- a table indicating whether information communication terminal has room for communication traffic or a table indicating whether congestion occurs or not Output (for example, display) information.
- the terminal for system management outputs (for example, displays) information indicating the degree of margin of communication traffic and the degree of congestion.
- the terminal for system management uses the cumulative probability distribution of variation in communication traffic characteristics and the power function on the coordinate axis. Or the integral value obtained as the degree of margin or the degree of congestion.
- whether communication traffic has a margin or whether congestion occurs may be divided into a plurality of levels and output.
- the traffic monitor 3 determines whether or not the values of the calculated margin and the degree of congestion are equal to or greater than the first threshold, and if equal to or greater than the first threshold, the margin level of communication traffic or It determines that the congestion level is "high". Also, the traffic monitor 3 determines that the margin level and congestion level of the communication traffic are "medium” if the values of the determined margin and congestion degree are smaller than the first threshold but are greater than or equal to the second threshold. judge. Furthermore, the traffic monitor 3 determines that the communication traffic margin level or the congestion level is “small” if the obtained margin or the value of the degree of congestion is smaller than the second threshold. Then, the terminal for system management outputs (displays) the communication traffic margin level or congestion level in three stages of "large”, “medium”, and “small” based on the evaluation result from the traffic monitor 3, for example. Do.
- the system administrator determines whether it is necessary to change various setting variables or the like for optimizing communication traffic characteristics based on the output evaluation result, and also determines the value of the setting variables to what extent Decide if you should. Then, the system administrator operates the terminal for system management, for example, to change the setting of the device / system (for example, the band control device 2) having the optimization function.
- the device / system for example, the band control device 2
- the device / system (for example, the band control device 2) provided with the optimization function, based on the evaluation result from the traffic monitor 3 according to the operation of the system administrator, performs setting variables etc. for optimization of communication traffic characteristics. Execute the process to change.
- the device / system having the optimization function is a device / system that implements a network filter
- the cumulative probability density distribution (CDF) of communication traffic is obtained based on the observed (measured) communication traffic data, and the calculated cumulative probability density distribution (CDF) is free.
- a process of quantitatively evaluating the effect of optimization of the communication traffic characteristics performed by the optimization function is performed. That is, based on observation data of a certain existing communication network, the effects of optimization are quantitatively evaluated. Therefore, the optimization method applied to the traffic characteristics of the communication network and the benchmark of the optimization method can be performed at least with relative quantitative nature.
- the effects of optimization of traffic characteristics in the communication network can be quantitatively evaluated.
- the effects of implementing optimization techniques for traffic characteristics of large scale communication networks can be compared with the effects of using other optimization techniques.
- the above-mentioned effect is a property that the fluctuation of the traffic characteristic shows power law at the optimum efficiency if the total communication traffic is observed even if observation data by local measurement is used. That is, the phase transition and the occurrence of the power law at its critical point are obtained by measuring the total communication traffic, and furthermore, the power law indicates self-similarity (that is, it does not depend on the scale to be measured).
- the power law shows self-similarity (that is, it does not depend on the scale to be measured, and does not depend on the system size), so the amount of calculation when evaluating the effect of optimization and the calculation therefor Resources can be reduced significantly.
- the traffic monitor 3 is basically disposed at the hub regardless of whether the network topology is known or unknown. Therefore, in the cumulative probability density distribution (CDF) or the complementary probability density distribution (CCDF) of the acquired traffic, it is possible to reduce fluctuation errors at both ends (high or low portion of the probability distribution).
- CDF cumulative probability density distribution
- CCDF complementary probability density distribution
- the band control device 2 is arranged in advance two or more hops in the router 1 (however, observation is performed in consideration of redundancy) It is desirable to place at least two traffic monitors 3 in the area 40.
- the arrangement method of the traffic monitors 3 is not limited to that shown in this embodiment, for example, the topology of the target network.
- the traffic monitor 3 may be arranged as follows.
- the output order (number of output links) of the router 1, which is a node is randomly selected from the maximum Do. Then, the traffic monitor 3 is selectively arranged on the server connected to the selected router.
- the traffic monitor 3 is placed on a server connected to a router (hub) having a large Betweenness (*).
- Betweenness of router i is calculated by the number of shortest paths between any two routers passing through node i (ie, router i) Ru.
- the small world nature makes the distance between any two nodes short by passing through the hub, so that the hub's Betweenness tends to be high on average.
- bridge router bridge router
- Betwennness (B (v)) passing through the router v is expressed by the following equation (2).
- ⁇ w, w ' is the number of shortest paths from the node w to the node w'.
- ⁇ w, w ′ (v) is the number of shortest paths from node w through node v to node w ′.
- the traffic monitor 3 may be arranged in the network according to the rule shown in (1) or (2).
- FIG. 6 is a block diagram showing an example of the configuration of the optimization evaluation system in the second embodiment. As shown in FIG. 6, the present embodiment differs from the first embodiment in that the optimization evaluation system includes a center node 9 in addition to the components shown in FIG.
- the information on the cumulative probability density distribution of changes in communication traffic characteristics observed by a plurality of traffic monitors 3 is periodically or irregularly transmitted to the center node 9 in an autonomous distributed manner or centralized control. You may do it. Then, the center node 9 may perform the error detection by comparing the information of the cumulative probability density distribution of the communication traffic characteristics received from each of the traffic monitors 3.
- each traffic monitor 3 has the information of the cumulative probability density distribution of the fluctuation of the observed communication traffic characteristics at a predetermined timing (for example, every predetermined period) ) Has a function to transmit to the center node 9.
- the other functions of each traffic monitor 3 are the same as the functions shown in the first embodiment.
- the center node 9 is specifically realized by an information processing apparatus such as a personal computer operating according to a program.
- the center node 9 has a function to perform error detection between the traffic monitors 3 by comparing the information of the cumulative probability density distribution of the communication traffic characteristics received from each traffic monitor 3.
- the center node 9 is an accumulated probability density distribution (CDF) distribution (or an accumulated probability thereof satisfying X ⁇ x) such that X ⁇ x of the observed traffic fluctuation held (stored) in the memory means 6 of the traffic monitor 3
- CDF accumulated probability density distribution
- a complementary distribution (CDF) of a density distribution (CDF) may be received from each traffic monitor 3 and an observation start time and an observation end time.
- CDF complementary distribution
- the center node 9 analyzes the time-series traffic fluctuation for each traffic monitor using the cumulative probability density distribution (or its complement) and the observation start time and the observation end time. The center node 9 then calculates its Lyapunov index (or Hurst index).
- the functions of the router 1 and the band control device 2 are the same as those shown in the first embodiment.
- the center node 9 performs comparison processing of the information of the cumulative probability density distribution received from each traffic monitor 3 Therefore, error detection between the traffic monitors 3 can be performed.
- the deviation of the trajectory with a very long (or infinite) time difference based on the cumulative probability density distribution (or its complementary distribution) and the observation start time and the observation end time can be evaluated whether the optimization method applied and the setting variables on the optimization method are still valid or not .
- FIG. 7 is a block diagram showing a minimum configuration example of the optimization evaluation system.
- the optimization evaluation system is a system for evaluating the effect of optimization by the optimization function in a communication network provided with an optimization function for optimizing communication traffic characteristics. As shown in FIG. 1, the optimization evaluation system includes at least a communication traffic analysis unit 5 and a communication traffic evaluation unit 7 as minimum components.
- the communication traffic analysis unit 5 has a function of obtaining a fluctuation distribution of communication traffic based on the measured communication traffic data.
- the communication traffic evaluation unit 7 quantifies the effect of the optimization of the communication traffic characteristic executed by the optimization function based on whether the fluctuation distribution calculated by the communication traffic analysis unit 5 is a power function or not. It has a function to execute the process to evaluate.
- the effect of optimization of traffic characteristics in the communication network can be quantitatively evaluated.
- the optimization evaluation system is an optimization evaluation system that evaluates the effect of optimization by the optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics, and uses measured communication traffic data
- Communication traffic analysis means for example, realized by the communication traffic analysis unit 5
- a fluctuation distribution of communication traffic for example, cumulative probability density distribution (CDF), complementary cumulative probability density distribution (CCDF)
- Communication traffic evaluation that executes processing to quantitatively evaluate the effect of optimization of communication traffic characteristics performed by the optimization function based on whether the fluctuation distribution calculated by the communication traffic analysis means is a power function or not Means (for example, realized by the communication traffic evaluation unit 7) And butterflies.
- the communication traffic evaluation means determines whether the fluctuation distribution calculated by the communication traffic analysis means is positive or negative with respect to the power function, and the fluctuation distribution is a power function On the other hand, if it is on the positive side, it is judged that congestion of communication traffic is occurring, and if the fluctuation distribution is on the negative side with respect to the power function, it is judged that communication traffic has a margin. It may be done.
- the communication traffic evaluation means obtains the difference between the fluctuation distribution calculated by the communication traffic analysis means and the power function, and obtains the difference between the obtained fluctuation distribution and power function as the communication traffic margin. Alternatively, it may be configured to calculate as the degree of congestion.
- the optimization evaluation system is an optimization evaluation system that evaluates the effect of optimization by the optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics, and is two hops of a band control device
- the traffic monitor (for example, traffic monitor 3), which is disposed in a router (for example, router 1) located in the previous stage and observes the total communication traffic of the outgoing line of the router, the traffic monitor observes the communication traffic, Based on communication traffic measurement means (for example, realized by the communication traffic monitor unit 4) that measures the packet size or packet arrival interval of communication packets, and based on the measurement results by the communication traffic measurement means, the observation data is reconstructed according to traffic characteristics.
- Traffic fluctuation distribution Communication traffic analysis means for giving the traffic characteristic to be converted and the variable X and calculating the complementary probability distribution of the cumulative probability density distribution of the fluctuation of the traffic characteristic with X ⁇ x or the fluctuation of the traffic characteristic with X ⁇ x (For example, realized by the communication traffic analysis unit 5) and a memory means (for example, realized by the memory means 6) for storing a threshold for each traffic characteristic, traffic characteristic, observation start time and observation end time of each packet
- the index is -1 to -1.3 when the origin is 0 to the power of 10 using the analysis result by the communication traffic analysis means and the information stored in the memory means using the analysis result by the communication traffic analysis means and the information stored in the memory means Compared with the power function, the observed communication traffic has room for the buffer or communication bandwidth limitation of the router Communication traffic evaluation means (for example, for determining the degree of margin and the degree of congestion as an evaluation result) while determining whether or not congestion is occurring beyond the buffer or communication band limit , And a feedback unit (for example, the feedback unit
- the traffic evaluation means is a power function whose index is -1 to -1.3 when the origin is 0 to the power of 10, and communication traffic characteristics measured by the communication traffic measurement means It is judged whether the cumulative probability density distribution of the variation of the distribution fits within the range of 1 decade, and the cumulative of the variation of the communication traffic characteristic measured that the power function and the cumulative probability density distribution fit within the range of 1 decade It is judged that the probability density distribution shows the optimal traffic efficiency as the condition that the power law is shown, and the fitting range of the power function and the cumulative probability density distribution is the scaling area, and the communication traffic characteristics obtained by other measurement It may be configured to be determined as a comparison area when performing comparison processing with the cumulative probability density distribution of variation.
- the traffic evaluation means uses the average value on the horizontal axis of the scaling area and the cumulative probability density distribution when taking the origin to the zero power of 10 is the index -1 to -1. It is configured to calculate the degree of communication traffic margin or congestion by calculating the difference between the cumulative probability density distribution of variation of communication traffic characteristics obtained by measurement in the scaling domain of power function which is 3 and the power function. It may be done.
- the traffic evaluation means is an accumulation of data measured by the communication traffic measurement means and a power function whose index is -1 to -1.3 when the origin is set to 10 to the 0th power Scaling of the power function whose cumulative probability density distribution is exponential -1 to -1.3 when the origin is taken to the 0th power of 10 using any value on the horizontal axis of the scaling region in the probability density distribution
- the degree of communication traffic margin or the degree of congestion may be calculated by obtaining the difference between the cumulative probability density distribution of fluctuations in communication traffic characteristics measured by the communication traffic measurement means in the area and the power function. .
- the traffic evaluation means is an accumulation of data measured by the communication traffic measurement means and a power function whose index is -1 to -1.3 when the origin is set to 10 to the 0th power
- the degree of communication traffic margin or the degree of congestion may be calculated by obtaining the difference between the cumulative probability density distribution of fluctuations and the power function.
- the traffic evaluation means accumulates the data measured by the communication traffic measurement means, and the power function whose index is -1 to -1.3 when the origin is set to 10 to the 0th power
- the variation of the communication traffic characteristic measured by the communication traffic measuring means in the scaling area of the power function whose cumulative probability density distribution is the index-1 to -1.3 As a value obtained by integrating the difference in the scaling area with the probability density distribution, the variation of the communication traffic characteristic measured by the communication traffic measuring means in the scaling area of the power function whose cumulative probability density distribution is the index-1 to -1.3.
- the degree of communication traffic margin or the degree of congestion may be calculated by obtaining the difference between the cumulative probability density distribution and the power function.
- the traffic evaluation means is a value obtained on the basis of the traffic data measured by the communication traffic measurement means, a data result calculated on double logarithms used to show a normal power law, or The cumulative probability density distribution of the fluctuation of the communication traffic characteristics measured by the communication traffic measurement means using the data result of calculating either the vertical axis or the horizontal axis in a single logarithm, and when the origin is taken to be 10 to the power of 10 It may be configured to fit a power function whose index is -1 to -1.3 using a least squares curve or linear approximation method.
- the optimization evaluation system uses the estimation result of the configuration of the communication network searched using the traceroot command, and the outdegree of the router that is a node is at most
- the router may be randomly selected from among the routers, and the traffic monitor may be arranged in advance in a server connected to the selected router.
- the optimization evaluation system When the network topology to be measured is known, and the outdegree distribution of the nodes shows a power law, the optimization evaluation system previously sends traffic on a server connected to a router with a large value of Betweenness function It may be configured to deploy a monitor.
- the optimization evaluation system includes a center node (for example, the center node 9), and the traffic monitor uses information indicating the cumulative probability density distribution of fluctuations in communication traffic characteristics measured by the communication traffic measurement means to the center node.
- the transmitting, center node may be configured to perform error detection between traffic monitors by comparing information indicating accumulated probability density distributions received from each traffic monitor.
- the traffic evaluation means is a power function in the scaling domain of the power function whose index is -1 to -1.3 when the cumulative probability density distribution takes the origin at the 0th power of 10
- the communication traffic can be spared by the applied optimization method and the setting variable in the optimization method. It is determined that the communication traffic is in a congestion state if it is determined that there is a certain state, and if the similarity probability density distribution is on the positive side with respect to the power function in the scaling region, It may be configured.
- the center node is a cumulative probability density distribution in which Xxx of the traffic fluctuation stored in the memory means or a complementary distribution of the cumulative probability density distribution in which X ⁇ x, an observation start time and
- the setting variable on the optimization method may be configured to determine whether it is valid in the future.
- the optimization evaluation device is an optimization evaluation device (for example, traffic monitor 3) for evaluating the effect of optimization by the optimization function in a communication network provided with an optimization function that optimizes communication traffic characteristics.
- Communication traffic analysis means for example, communication traffic analysis unit 5) for obtaining fluctuation distribution of communication traffic (for example, cumulative probability density distribution (CDF), complementary cumulative probability density distribution (CCDF)) based on the measured communication traffic data
- CDF cumulative probability density distribution
- CCDF complementary cumulative probability density distribution
- Communication traffic evaluation means for example, the communication traffic Characterized by comprising a to
- the present invention is the effect of optimizing communication traffic characteristics in the existing communication network, the Internet, a network managed by an Internet service provider, a local area network (LAN), and almost any other small or large communication network. It can be applied to the purpose of evaluating In particular, it can be applied to the application of quantitatively evaluating the effectiveness of an optimization method used for traffic characteristics in a communication network having a feedback mechanism and a buffer (or delay) function, and the effectiveness of an optimization variable. In addition, the present invention can be applied to an application for evaluating how optimally these optimization methods or variables set in these optimization methods operate.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Environmental & Geological Engineering (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Description
制約条件:x∈S 式(1)
以下、本発明の第1の実施形態について図面を参照して説明する。本発明の実施形態は、通信ネットワーク、特にフィードバック機構とバッファ(あるいは遅延)機能とを有する通信ネットワークにおけるトラフィック特性について最適化を行う最適化手法に関し、その最適化手法の評価方法に関する。
w≠w’≠v 式(2)
次に、本発明の第2の実施形態を図面を参照して説明する。図6は、第2の実施形態における最適化評価システムの構成の一例を示すブロック図である。図6に示すように、本実施形態では、最適化評価システムが、図1で示した構成要素に加えて、センターノード9を含む点で、第1の実施形態と異なる。
2 帯域制御装置
3 トラフィックモニタ
4 通信トラフィックモニタ部
5 通信トラフィック解析部
6 メモリ手段
7 通信トラフィック評価部
8 フィードバック部
9 センターノード
Claims (29)
- 通信トラフィック特性を最適化する最適化機能を備えた通信ネットワークにおける前記最適化機能による最適化の効果を評価する最適化評価システムであって、
計測した通信トラフィックデータに基づいて、通信トラフィックの変動分布を求める通信トラフィック解析手段と、
前記通信トラフィック解析手段が算出した前記変動分布がベキ関数となっているか否かに基づいて、前記最適化機能により実行された通信トラフィック特性の最適化の効果を定量的に評価する処理を実行する通信トラフィック評価手段とを有することを特徴とする最適化評価システム。 - 前記通信トラフィック評価手段は、
前記通信トラフィック解析手段が算出した変動分布がベキ関数に対して正側であるか負側であるかを判定し、
前記変動分布が前記ベキ関数に対して正側であれば、通信トラフィックの輻輳が発生していると判定し、
前記変動分布が前記ベキ関数に対して負側であれば、通信トラフィックに余裕がある状態であると判定する請求項1記載の最適化評価システム。 - 前記通信トラフィック評価手段は、
前記通信トラフィック解析手段が算出した変動分布とベキ関数との差分を求め、
求めた前記変動分布と前記ベキ関数との差分を、通信トラフィックの余裕度又は輻輳の度合いとして算出する請求項1又は請求項2記載の最適化評価システム。 - 帯域制御装置の2ホップ以上前段に位置するルータに配置され、当該ルータの出線の総通信トラフィックを観測するトラフィックモニタを有し、
前記トラフィックモニタは、
通信トラフィックを観測し、通信パケットのパケットサイズ又はパケット到着間隔を計測する通信トラフィック計測手段と、
前記通信トラフィック計測手段による計測結果に基づいて、トラフィック特性別に観測データを再構築し、トラフィック変動分布で表現可能な最適化すべきトラフィック特性と変数Xとを与え、X≦xとなるトラフィック特性の変動の累積確率密度分布又はX≧xとなるトラフィック特性の変動の累積確率密度分布の補分布を計算する通信トラフィック解析手段と、
トラフィック特性毎の閾値、トラフィック特性、各パケットの観測開始時間及び観測終了時間を記憶するメモリ手段と、
前記通信トラフィック解析手段による解析結果及び前記メモリ手段が記憶する情報を用いて、トラフィック特性別の観測データを、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と比較し、観測した通信トラフィックがルータのバッファ又は通信帯域の制限に対して余裕がある状態であるか否か、又はバッファ又は通信帯域の制限を超えて輻輳が発生しているか否かを評価結果として判定するとともに、余裕の程度及び輻輳の程度を評価結果として求める通信トラフィック評価手段と、
前記通信トラフィック評価部が判定又は求めた評価結果を、最適化機能を備えた装置又はシステムにフィードバックする処理を行うフィードバック手段とを含む請求項1記載の最適化評価システム。 - 前記トラフィック評価手段は、
原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とが、1decadeの範囲で適合するか否かを判定し、
前記ベキ関数と前記累積確率密度分布とが1decadeの範囲で適合することを、計測した通信トラフィック特性の変動の累積確率密度分布がベキ則を示す条件として、最適なトラフィック効率を示す状態と判定し、
前記ベキ関数と前記累積確率密度分布とのフィッティング範囲を、スケーリング領域として、他の計測で得られた通信トラフィック特性の変動の累積確率密度分布との比較処理を行う際の比較領域として求める請求項1から請求項4のうちのいずれか1項に記載の最適化評価システム。 - 前記トラフィック評価手段は、スケーリング領域の横軸上の平均値を用いて、原点を10の0乗にとったときの累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における計測で得られた通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項5記載の最適化評価システム。
- 前記トラフィック評価手段は、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測したデータの累積確率密度分布とにおけるスケーリング領域の横軸上のいずれかの値を用いて、原点を10の0乗にとったときの累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項6記載の最適化評価システム。
- 前記トラフィック評価手段は、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測したデータの累積確率密度分布とにおけるスケーリング領域の最も大きい値における差分値を用いて、累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項6記載の最適化評価システム。
- 前記トラフィック評価手段は、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測したデータの累積確率密度分布とにおけるスケーリング領域内の差を積分した値として、累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項6記載の最適化評価システム。
- 前記トラフィック評価手段は、通信トラフィック計測手段が計測したトラフィックデータに基づいて得られた値、通常のベキ則を示すために用いられる両対数に計算したデータ結果、又は縦軸若しくは横軸のいずれかを片対数に計算したデータ結果を用いて、前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布と、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数とを、最小2乗法による曲線又は直線近似手法を用いてフィッティングする請求項5から請求項9のうちのいずれか1項に記載の最適化評価システム。
- 計測対象とするネットワークのトポロジが不明な場合に、tracerootコマンドを用いて探索した通信ネットワークの構成の推定結果を用いて、ノードであるルータの出次数が最大であるルータの中からランダムにルータを選択し、選択したルータに接続されたサーバに予め選択的にトラフィックモニタを有する請求項4から請求項10のうちのいずれか1項に記載の最適化評価システム。
- 計測対象とするネットワークトポロジが既知であり、且つノードの出次数分布がベキ則を示す場合に、Betweenness関数の値が大きいルータに接続されたサーバ上に予めトラフィックモニタを有する請求項4から請求項10のうちのいずれか1項に記載の最適化評価システム。
- センターノードを備え、
前記トラフィックモニタは、通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布を示す情報を、前記センターノードに送信し、
前記センターノードは、各トラフィックモニタから受信した累積確率密度分布を示す情報の比較を行うことによって、トラフィックモニタ間の誤差検出を行う請求項4から請求項12のうちのいずれか1項に記載の最適化評価システム。 - 通信トラフィック特性を最適化する最適化機能を備えた通信ネットワークにおける前記最適化機能による最適化の効果を評価する最適化評価装置であって、
計測した通信トラフィックデータに基づいて、通信トラフィックの変動分布を求める通信トラフィック解析手段と、
前記通信トラフィック解析手段が算出した前記変動分布がベキ関数となっているか否かに基づいて、前記最適化機能により実行された通信トラフィック特性の最適化の効果を定量的に評価する処理を実行する通信トラフィック評価手段とを
備えたことを特徴とする最適化評価装置。 - 前記通信トラフィック評価手段は、
通信トラフィック解析手段が算出した変動分布がベキ関数に対して正側であるか負側であるかを判定し、
前記変動分布が前記ベキ関数に対して正側であれば、通信トラフィックの輻輳が発生していると判定し、
前記変動分布が前記ベキ関数に対して負側であれば、通信トラフィックに余裕がある状態であると判定する請求項14記載の最適化評価装置。 - 前記通信トラフィック評価手段は、
前記通信トラフィック解析手段が算出した変動分布とベキ関数との差分を求め、
求めた前記変動分布と前記ベキ関数との差分を、通信トラフィックの余裕度又は輻輳の度合いとして算出する請求項14又は請求項15記載の最適化評価装置。 - 帯域制御装置の2ホップ以上前段に位置するルータに配置され、
前記ルータの出線の総通信トラフィックを観測し、通信パケットのパケットサイズ又はパケット到着間隔を計測する通信トラフィック計測手段と、
前記通信トラフィック計測手段による計測結果に基づいて、トラフィック特性別に観測データを再構築し、トラフィック変動分布で表現可能な最適化すべきトラフィック特性と変数Xとを与え、X≦xとなるトラフィック特性の変動の累積確率密度分布又はX≧xとなるトラフィック特性の変動の累積確率密度分布の補分布を計算する通信トラフィック解析手段と、
トラフィック特性毎の閾値、トラフィック特性、各パケットの観測開始時間及び観測終了時間を記憶するメモリ手段と、
前記通信トラフィック解析手段による解析結果及び前記メモリ手段が記憶する情報を用いて、トラフィック特性別の観測データを、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と比較し、観測した通信トラフィックがルータのバッファ又は通信帯域の制限に対して余裕がある状態であるか否か、又はバッファ又は通信帯域の制限を超えて輻輳が発生しているか否かを評価結果として判定するとともに、余裕の程度及び輻輳の程度を評価結果として求める通信トラフィック評価手段と、
前記通信トラフィック評価部が判定又は求めた評価結果を、最適化機能を備えた装置又はシステムにフィードバックする処理を行うフィードバック手段とを有することを特徴とする請求項14記載の最適化評価装置。 - 前記トラフィック評価手段は、
原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とが、1decadeの範囲で適合するか否かを判定し、
前記ベキ関数と前記累積確率密度分布とが1decadeの範囲で適合することを、計測した通信トラフィック特性の変動の累積確率密度分布がベキ則を示す条件として、最適なトラフィック効率を示す状態と判定し、
前記ベキ関数と前記累積確率密度分布とのフィッティング範囲を、スケーリング領域として、他の計測で得られた通信トラフィック特性の変動の累積確率密度分布との比較処理を行う際の比較領域として求める請求項14から請求項17のうちのいずれか1項に記載の最適化評価装置。 - 前記トラフィック評価手段は、スケーリング領域の横軸上の平均値を用いて、原点を10の0乗にとったときの累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における計測で得られた通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項18記載の最適化評価装置。
- 前記トラフィック評価手段は、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測したデータの累積確率密度分布とにおけるスケーリング領域の横軸上のいずれかの値を用いて、原点を10の0乗にとったときの累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項19記載の最適化評価装置。
- 前記トラフィック評価手段は、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測したデータの累積確率密度分布とにおけるスケーリング領域の最も大きい値における差分値を用いて、累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項19記載の最適化評価装置。
- 前記トラフィック評価手段は、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数と、通信トラフィック計測手段が計測したデータの累積確率密度分布とにおけるスケーリング領域内の差を積分した値として、累積確率密度分布が指数-1~-1.3であるベキ関数のスケーリング領域における前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布とベキ関数との差分を求めることによって、通信トラフィックの余裕度又は輻輳の度合いを算出する請求項19記載の最適化評価装置。
- 前記トラフィック評価手段は、通信トラフィック計測手段が計測したトラフィックデータに基づいて得られた値、通常のベキ則を示すために用いられる両対数に計算したデータ結果、又は縦軸若しくは横軸のいずれかを片対数に計算したデータ結果を用いて、前記通信トラフィック計測手段が計測した通信トラフィック特性の変動の累積確率密度分布と、原点を10の0乗にとったときの指数が-1~-1.3であるベキ関数とを、最小2乗法による曲線又は直線近似手法を用いてフィッティングする請求項18から請求項22のうちのいずれか1項に記載の最適化評価装置。
- 計測対象とするネットワークのトポロジが不明な場合に、tracerootコマンドを用いて探索した通信ネットワークの構成の推定結果を用いて、ノードであるルータの出次数が最大であるルータの中からランダムにルータを選択し、選択したルータに接続されたサーバに予め選択的に配置されている請求項17から請求項23のうちのいずれか1項に記載の最適化評価装置。
- 計測対象とするネットワークトポロジが既知であり、且つノードの出次数分布がベキ則を示す場合に、Betweenness関数の値が大きいルータに接続されたサーバ上に予め配置されている請求項17から請求項23のうちのいずれか1項に記載の最適化評価装置。
- 通信トラフィック特性を最適化する最適化機能を備えた通信ネットワークにおける前記最適化機能による最適化の効果を評価する最適化評価方法であって、
計測した通信トラフィックデータに基づいて、通信トラフィックの変動分布を求め、
算出した前記変動分布がベキ関数となっているか否かに基づいて、前記最適化機能により実行された通信トラフィック特性の最適化の効果を定量的に評価する処理を実行することを特徴とする最適化評価方法。 - 前記最適化の効果を評価する際に、
算出した変動分布がベキ関数に対して正側であるか負側であるかを判定し、
前記変動分布が前記ベキ関数に対して正側であれば、通信トラフィックの輻輳が発生していると判定し、
前記変動分布が前記ベキ関数に対して負側であれば、通信トラフィックに余裕がある状態であると判定する請求項26記載の最適化評価方法。 - 通信トラフィック特性を最適化する最適化機能を備えた通信ネットワークにおける前記最適化機能による最適化の効果を評価するための最適化評価用プログラムであって、
コンピュータに、
計測した通信トラフィックデータに基づいて、通信トラフィックの変動分布を求める処理と、
算出した前記変動分布がベキ関数となっているか否かに基づいて、前記最適化機能により実行された通信トラフィック特性の最適化の効果を定量的に評価する処理を実行する処理とを実行させることを特徴とする最適化評価用プログラム。 - 前記コンピュータに、
前記最適化の効果を評価する際に、
算出した変動分布がベキ関数に対して正側であるか負側であるかを判定する処理と、
前記変動分布が前記ベキ関数に対して正側であれば、通信トラフィックの輻輳が発生していると判定する処理と、
前記変動分布が前記ベキ関数に対して負側であれば、通信トラフィックに余裕がある状態であると判定する処理とを実行させる請求項28記載の最適化評価用プログラム。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2010514413A JP5287853B2 (ja) | 2008-05-29 | 2009-04-09 | 最適化評価システム、最適化評価装置、最適化評価方法、及び最適化評価用プログラム |
US12/991,779 US8493882B2 (en) | 2008-05-29 | 2009-04-09 | Optimization evaluation system, optimization evaluation device, optimization evaluation method, and optimization evaluation program |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2008141283 | 2008-05-29 | ||
JP2008-141283 | 2008-05-29 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2009145011A1 true WO2009145011A1 (ja) | 2009-12-03 |
Family
ID=41376897
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2009/057268 WO2009145011A1 (ja) | 2008-05-29 | 2009-04-09 | 最適化評価システム、最適化評価装置、最適化評価方法、及び最適化評価用プログラム |
Country Status (3)
Country | Link |
---|---|
US (1) | US8493882B2 (ja) |
JP (1) | JP5287853B2 (ja) |
WO (1) | WO2009145011A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2015132734A (ja) * | 2014-01-14 | 2015-07-23 | キヤノン株式会社 | 通信装置、通信装置の制御方法 |
CN117983668A (zh) * | 2024-04-07 | 2024-05-07 | 东北大学 | 一种基于性能评估的热轧过程厚度活套张力优化控制方法 |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AR055163A1 (es) | 2005-09-13 | 2007-08-08 | Iwics Inc | Determinacion de la posicion de estaciones moviles en una red inalambrica |
KR101261182B1 (ko) * | 2011-12-14 | 2013-05-09 | 한국과학기술원 | 네트워크 트래픽 측정 주기 및 알람을 조정하는 시스템 및 방법 |
US10652318B2 (en) * | 2012-08-13 | 2020-05-12 | Verisign, Inc. | Systems and methods for load balancing using predictive routing |
US9525617B2 (en) * | 2014-05-02 | 2016-12-20 | Cisco Technology, Inc. | Distributed predictive routing using delay predictability measurements |
CN109729538B (zh) * | 2017-10-31 | 2021-02-09 | 华为技术有限公司 | 移动通信系统扩容方法、设备及存储介质、程序产品 |
US11444860B2 (en) * | 2020-09-24 | 2022-09-13 | Cisco Technology, Inc. | Automating and extending path tracing through wireless links |
CN113052455B (zh) * | 2021-03-22 | 2024-04-19 | 广西电网有限责任公司 | 一种流域径流水文条件精细化评价的方法及装置 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6144639A (en) * | 1996-09-03 | 2000-11-07 | Sbc Technology Resources, Inc. | Apparatus and method for congestion control in high speed networks |
US7724660B2 (en) * | 2005-12-13 | 2010-05-25 | Alcatel Lucent | Communication traffic congestion management systems and methods |
-
2009
- 2009-04-09 US US12/991,779 patent/US8493882B2/en not_active Expired - Fee Related
- 2009-04-09 WO PCT/JP2009/057268 patent/WO2009145011A1/ja active Application Filing
- 2009-04-09 JP JP2010514413A patent/JP5287853B2/ja not_active Expired - Fee Related
Non-Patent Citations (4)
Title |
---|
HIDENORI HASHIMOTO ET AL.: "Algorithms for Estimating Flow Rate Distribution by Packet Sampling", IEICE TECHNICAL REPORT, vol. 106, no. 578, 1 March 2007 (2007-03-01), pages 507 - 512 * |
KENSUKE FUKUDA ET AL.: "Phase Transition Phenomena Originating in Network Protocols", IEICE TECHNICAL REPORT, vol. 101, no. 8, 12 April 2001 (2001-04-12), pages 1 - 6 * |
SHUNJI ABE ET AL.: "Performance Evaluation Method by Using Observed Traffic Data for an Academic Information Network", IEICE TECHNICAL REPORT, vol. 103, no. 492, 4 December 2003 (2003-12-04), pages 31 - 36 * |
TOSHIHISA OZAWA: "New Trend in the Theories for Queueing Systems and Performance Evaluation", SYSTEMS CONTROL AND INFORMATION, vol. 43, no. 3, 15 March 1999 (1999-03-15), pages 1 - 6 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2015132734A (ja) * | 2014-01-14 | 2015-07-23 | キヤノン株式会社 | 通信装置、通信装置の制御方法 |
CN117983668A (zh) * | 2024-04-07 | 2024-05-07 | 东北大学 | 一种基于性能评估的热轧过程厚度活套张力优化控制方法 |
Also Published As
Publication number | Publication date |
---|---|
US8493882B2 (en) | 2013-07-23 |
US20110058498A1 (en) | 2011-03-10 |
JP5287853B2 (ja) | 2013-09-11 |
JPWO2009145011A1 (ja) | 2011-10-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2009145011A1 (ja) | 最適化評価システム、最適化評価装置、最適化評価方法、及び最適化評価用プログラム | |
US10171332B2 (en) | Probing technique for predictive routing in computer networks | |
EP3138247B1 (en) | Centralized predictive routing using delay predictability measurements | |
US20190363951A1 (en) | Distributed and learning machine-based approach to gathering localized network dynamics | |
EP2954643B1 (en) | Traffic-based inference of influence domains in a network by using learning machines | |
JP5812282B2 (ja) | トラヒック監視装置 | |
WO2012117549A1 (ja) | 障害解析装置、そのシステム、およびその方法 | |
WO2012029422A1 (ja) | 通信品質監視システム、通信品質監視方法、及び記憶媒体 | |
US9992081B2 (en) | Scalable generation of inter-autonomous system traffic relations | |
EP2589183B1 (en) | A method and an apparatus for evaluating network performance | |
Wei et al. | A congestion control scheme based on fuzzy logic for wireless sensor networks | |
WO2015011220A1 (en) | Monitoring network use of resources | |
Wang et al. | Efficient measurement of round-trip link delays in software-defined networks | |
Abushagur et al. | Hybrid software-defined network monitoring | |
Kawahara et al. | Mobile QoS tomography using compressed sensing | |
CN107592269B (zh) | 传输路径的负载信息的方法和网络节点 | |
JP5155284B2 (ja) | P2pトラヒック量推定方法と装置およびプログラム | |
Gao et al. | The diagnosis of wired network malfunctions based on big data and traffic prediction: An overview | |
JP5537692B1 (ja) | 品質劣化原因推定装置、品質劣化原因推定方法、品質劣化原因推定プログラム | |
Dong et al. | Measure correlation analysis of network flow based on symmetric uncertainty | |
JP5659393B2 (ja) | ネットワーク装置、及び、パケット処理方法 | |
Markovich et al. | Modeling of loss processes arising from packet flows at bottleneck links | |
Bazmi et al. | A NEURAL NETWORK BASED TRAFFIC-AWARE FORWARDING STRATEGY IN NAMED DATA NETWORKING | |
Indurkar et al. | Performance failure detection and path computation | |
Zhang et al. | Performance analysis of available bandwidth estimation algorithm based on ewma and kalman filter |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 09754519 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 12991779 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2010514413 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 09754519 Country of ref document: EP Kind code of ref document: A1 |