WO2014007166A1 - 流量予測装置、流量予測方法及び流量予測プログラム - Google Patents
流量予測装置、流量予測方法及び流量予測プログラム Download PDFInfo
- Publication number
- WO2014007166A1 WO2014007166A1 PCT/JP2013/067848 JP2013067848W WO2014007166A1 WO 2014007166 A1 WO2014007166 A1 WO 2014007166A1 JP 2013067848 W JP2013067848 W JP 2013067848W WO 2014007166 A1 WO2014007166 A1 WO 2014007166A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- flow rate
- prediction
- unit
- stochastic
- model
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 52
- 238000005309 stochastic process Methods 0.000 claims abstract description 68
- 238000005315 distribution function Methods 0.000 claims abstract description 14
- 238000009792 diffusion process Methods 0.000 claims description 71
- 230000006870 function Effects 0.000 claims description 32
- 238000001744 unit root test Methods 0.000 claims description 25
- 238000004364 calculation method Methods 0.000 claims description 23
- 238000005259 measurement Methods 0.000 claims description 9
- 230000010455 autoregulation Effects 0.000 claims description 2
- 238000004891 communication Methods 0.000 description 163
- 230000005540 biological transmission Effects 0.000 description 46
- 230000008569 process Effects 0.000 description 24
- 238000012360 testing method Methods 0.000 description 22
- 238000010586 diagram Methods 0.000 description 19
- 230000005654 stationary process Effects 0.000 description 15
- 238000012546 transfer Methods 0.000 description 12
- 230000005653 Brownian motion process Effects 0.000 description 11
- 238000005537 brownian motion Methods 0.000 description 11
- 238000012790 confirmation Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 7
- 238000013500 data storage Methods 0.000 description 3
- 230000007480 spreading Effects 0.000 description 3
- 230000003190 augmentative effect Effects 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 238000001278 Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test Methods 0.000 description 1
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000033001 locomotion Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000000551 statistical hypothesis test Methods 0.000 description 1
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
- H04L43/04—Processing captured monitoring data, e.g. for logfile generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- H04L41/142—Network analysis or design using statistical or mathematical methods
-
- 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
- H04L43/0876—Network utilisation, e.g. volume of load or congestion level
- H04L43/0894—Packet rate
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/83—Admission control; Resource allocation based on usage prediction
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/08—Testing, supervising or monitoring using real traffic
Definitions
- the present invention relates to a flow rate prediction device, a flow rate prediction method, and a flow rate prediction program for predicting a flow rate.
- a communication throughput prediction apparatus that predicts communication throughput that is a data size (data amount) distributed (transmitted) per unit time via a communication network is known.
- An example of a communication throughput prediction apparatus is the communication apparatus described in Patent Document 1.
- the communication device described in Patent Literature 1 is realized as a server device that distributes video stream data, for example.
- the server apparatus predicts the communication throughput without requiring feedback from the receiving terminal, and distributes the video stream data at a rate corresponding to the predicted communication throughput.
- the prediction of communication throughput can be used for various purposes.
- Non-Patent Document 1 describes an example of a communication throughput prediction method used in this type of communication throughput prediction apparatus.
- Non-Patent Document 1 fluctuations in communication throughput are modeled using a Brownian motion model with drift. Then, the drift and dispersion, which are parameters of the modeled model, are estimated from time-series data of a past predetermined communication throughput.
- the estimation is expressed by the word “identification”, and the word “identification” is used in the same meaning in the following description.
- Non-Patent Document 1 a probability distribution (probability density function) of future communication throughput is calculated based on the identified model, and future communication is calculated from the calculated probability density function. Calculate the stochastic spread of the throughput (stochastic diffusion).
- the communication throughput varies depending on various factors. For example, in the case of communication conforming to TCP / IP (Transmission / Control Protocol / Internet Protocol), the presence of end-to-end delay, the occurrence of packet loss, the influence of cross traffic, the fluctuation of radio field intensity in wireless communication, etc. Due to the complex action of various factors, the communication throughput varies from moment to moment.
- TCP / IP Transmission / Control Protocol / Internet Protocol
- Non-Patent Document 1 Even when this fluctuation is unstable and randomness is high, as described in Non-Patent Document 1, it is possible to predict the stochastic diffusion of communication throughput with high accuracy by using the Brownian motion model. .
- the actual communication throughput does not fluctuate completely randomly like the Brownian motion.
- the factors that influence the fluctuations in communication throughput as described above are in a stable state, the communication throughput does not fluctuate completely at random and stabilizes.
- Non-Patent Document 1 there is a problem that the prediction accuracy is lowered depending on whether the communication throughput is stable or unstable.
- An exemplary object of the present invention is to provide a flow prediction device, a flow prediction method, and a flow prediction program capable of predicting the stochastic diffusion of communication throughput with high accuracy regardless of the state of communication throughput. There is.
- the continuity determination for determining whether the state of the flow rate is a steady state or an unsteady state based on the time-series data of the measured flow rate.
- a stochastic process model having the flow rate as a random variable based on the discrimination result by the continuity discrimination unit, and then calculating parameters used in the identified stochastic process model based on the time series data
- a stochastic diffusion calculating unit for calculating a function.
- the continuity determination for determining whether the flow state is in a steady state or an unsteady state based on the time-series data of the measured flow rate. And identifying a stochastic process model using the flow rate as a random variable based on the determination result in the step and the stationaryity determining step, and then calculating parameters used in the identified stochastic process model based on the time series data A probability distribution function or probability density using the flow rate to be predicted as a random variable based on the prediction model identification step to be performed, the stochastic process model identified in the prediction model identification step, and the parameter calculated in the prediction model identification step And a stochastic diffusion calculating step for calculating a function.
- a flow rate prediction program for causing a computer to function as a flow rate prediction device for predicting a flow rate, Based on time series data of the measured flow rate, the computer determines whether the flow rate state is a steady state or an unsteady state, and based on the determination result by the continuity determination unit Identifying a stochastic process model having the flow rate as a random variable, and then calculating a parameter used in the identified stochastic process model based on the time series data, and a predictive model identifying unit A stochastic diffusion calculation unit that calculates a probability distribution function or a probability density function using the flow rate to be predicted as a random variable based on the identified stochastic process model and the parameter calculated by the prediction model identification unit; There is provided a flow rate prediction program characterized by functioning as a flow rate prediction device.
- pass ((sigma) epsilon 1) of the unsteady process for demonstrating embodiment of this invention. It is a figure showing the sample path
- HSDPA prediction result
- LTE prediction result
- WiFi prediction result
- the present invention is suitably applied to prediction of an arbitrary flow rate (for example, traffic volume, liquid flow rate, gas flow rate) as well as a flow rate such as communication throughput.
- a communication throughput prediction device will be described as an embodiment.
- FIG. 1 is a diagram showing the entire system including the throughput prediction apparatus 100 according to the present embodiment.
- a throughput prediction apparatus 100 according to the present embodiment is connected to an Internet protocol (IP) network 200.
- IP Internet protocol
- the throughput prediction device 100 is a device that predicts communication throughput that is a data size (amount of data) distributed (transmitted) per unit time via a communication network.
- the IP network 200 is a network that performs communication conforming to the Internet Protocol (IP), and connects various sub-networks around the world with the IP of the third layer of the OSI (Open Systems Interconnection) reference model. An expanding network.
- the IP network 200 includes a terminal that becomes a communication partner of the throughput prediction apparatus 100, a relay device such as a router, and the like.
- the IP network 200 may be realized by wired communication, but a part or all of the IP network 200 may be realized by wireless communication.
- the throughput prediction apparatus 100 and the IP network 200 are connected by a solid line, but this is not intended to limit the communication method to the wired communication method, and the throughput prediction apparatus 100 and the IP network 200 are HSDPA (The connection may be made by an arbitrary wireless connection conforming to an arbitrary communication method such as High (Speed) Downlink (Packet Access) or WiFi (Wireless Fidelity).
- HSDPA High (Speed) Downlink (Packet Access) or WiFi (Wireless Fidelity).
- the communication throughput prediction apparatus 100 includes a communication throughput measurement unit 101, a stationarity determination unit 102, a prediction model identification unit 103, and a stochastic diffusion calculation unit 104.
- the communication throughput measuring unit 101 measures the current (current) communication throughput in data transmission. Therefore, the communication throughput measuring unit 101 is connected to the IP network 200. Further, the measured communication throughput is sequentially accumulated and held as time series data. The held time series data is output to the continuity determination unit 102.
- the throughput prediction apparatus 100 may be a communication transmission side, a reception side, or a repeater (router or the like). Therefore, the communication throughput measuring unit 101 can measure the communication throughput for both reception data on the reception side and transmission data on the transmission side. A specific method for measuring the communication throughput will be described later.
- the stationarity determination unit 102 determines stationarity of communication throughput using a part of time series data of communication throughput measured by the communication throughput measurement unit 101. Regarding the stationarity and the determination of stationarity will be described later.
- the continuity determination unit 102 outputs the continuity determination result to the prediction model identification unit 103.
- the prediction model identification unit 103 identifies a prediction model for predicting the probabilistic spread (probabilistic diffusion) of the communication throughput according to the determination result given by the continuity determination unit 102. A specific identification method will be described later.
- the identified prediction model is notified to the probabilistic diffusion calculation unit 104.
- the stochastic diffusion calculation unit 104 calculates the stochastic diffusion based on the prediction model identified by the prediction model identification unit 103. A specific calculation method will be described later. Then, the stochastic diffusion calculation unit 104 outputs the calculation result to the outside or the inside of the throughput prediction apparatus 100.
- the output destination may be a storage medium either outside or inside the throughput prediction apparatus 100, or may be another function unit not shown in FIG. It may be transmitted to a device other than the throughput prediction device 100 via a network such as the network 200.
- the throughput prediction apparatus 100 is the transmission side. Further, it is assumed that data transmission is started from the transmission side to the reception side at time 0.
- xt represented by the following equation (1) is defined as the communication throughput of the time interval ⁇ at time t.
- the definition of communication throughput described this time is a definition based on St that can be measured on the transmission side. However, by replacing this St with the total data size Rt received on the reception side, the same definition of communication throughput on the reception side is also provided. Is possible.
- the communication throughput at an arbitrary time t ⁇ 0 can be measured.
- the time series data is a time interval for each ⁇ , but the above-described definition of communication throughput can be treated as a continuous time function, so any time interval may be used. That is, the time interval ⁇ may be a predetermined and arbitrary length time interval.
- time series data Although the measurement results are held as time series data, it is not necessary to hold all of them. For example, only the latest n (n is an integer of 1 or more) time series data is held. Good.
- time series data ⁇ x0, x ⁇ , x2 ⁇ , x3 ⁇ ,... ⁇ Is expressed as ⁇ x0, x1, x2, x3,.
- the stationarity determining unit 102 determines the stationarity of the communication throughput using a part of the time series data of the communication throughput measured by the communication throughput measuring unit 101.
- the stationarity of communication throughput is a property of whether this stochastic process is a stationary process or a non-stationary process when the communication throughput is considered as a stochastic process.
- the communication throughput is said to be steady, and when the stochastic process is a non-stationary process, the communication throughput is said to be non-stationary.
- the fact that the stochastic process is a steady process means that the expected value of the stochastic process is constant without depending on time, and the covariance between different time points depends only on the time difference.
- this stochastic process is said to be a non-stationary process.
- the stationarity of the communication throughput is determined based on the suitability of the model. Specifically, one model of each of the steady process and the unsteady process is selected, and it is verified to which model the time series data of the communication throughput to be discriminated this time is more suitable.
- the selection of the stationary process model and the non-stationary process model can be performed by an arbitrary method.
- the stationary process model and the non-stationary process model are selected from the AR (Autoregression) process will be described.
- the AR process uses the time series data ⁇ xt ⁇ 1, xt ⁇ 2,..., Xt ⁇ p ⁇ for the time series data xt at time t before t and the error term ⁇ t ⁇ .
- ⁇ 1, ⁇ 2,..., ⁇ p included in the equation (2) are coefficients.
- Equation (2) The stochastic process expressed as equation (2) is often written as an AR (p) model with particular emphasis on the use of the past p. Therefore, also in the description of the present embodiment, the stochastic process expressed as Equation (2) is expressed as “AR (p) model”.
- AR (p) model the condition for AR (p) to be a stationary process is that the absolute value of the root (solution) of the algebraic equation for ⁇ shown in the following equation (3) called the characteristic equation is All are known to be greater than one.
- the stationarity determining unit 102 calculates the absolute values of all the roots of the equation (3) of the characteristic equation. It will be tested whether or not is greater than 1.
- ⁇ t ⁇ is an error term of N (0, ⁇ 2).
- ⁇ is a constant term, that is, a model in which the deviation from the constant term follows AR (1).
- ⁇ is a coefficient of AR (1) and corresponds to ⁇ 1 in equation (2) (0 after ⁇ 2).
- Expression (4) can be expanded and written as the following Expression (5).
- the condition for AR (1) in Equation (4) to be a stationary process is that the absolute value of the root 1 / ⁇ of the characteristic equation is greater than 1, that is, ⁇ 1 ⁇ ⁇ 1 ( ⁇ ⁇ 0). is there. On the other hand, if ⁇ ⁇ 1 or ⁇ ⁇ ⁇ 1, the process is unsteady.
- equation (5) can be written as equation (6) below, which is called a unit root model. In particular, if this is considered as a continuous time function, it becomes a Brownian motion model.
- the null hypothesis (H0) and the alternative hypothesis (H1) must first be established.
- the null hypothesis of “with unit root” is changed to the alternative hypothesis of “without unit root”.
- the stochastic process model under the null hypothesis H0 is M0
- the stochastic process model under the alternative hypothesis H1 is M1, and is summarized in the following equation (8).
- FIG. 5 shows the characteristic value of the null distribution calculated by the Monte Carlo simulation.
- the above-described unit root test was performed on the two sample paths shown in FIGS.
- t statistic t ⁇ ⁇ 5.77 and p value is less than 0.01. In other words, the null hypothesis is rejected at the significance level of 1%, and the alternative hypothesis having no unit root is adopted.
- the unit root test described above is a test method generally referred to as a DF (Dickey-Fuller) test, and the deviation from the constant term ⁇ is AR ( It was a model according to 1). Similarly, in the DF test, not the deviation from the constant term but the deviation from the primary expression ⁇ + ⁇ t with respect to time t can be considered. In this case, the two models M0 and M1 to be verified are replaced as shown in the following equation (12).
- the parameter for calculating the t statistic t ⁇ increases by one, but the procedure is the same as when considering the deviation from the constant term as described above. However, it should be noted that the null distribution is different.
- the parameters ⁇ 1,..., ⁇ p are only increased, and the procedure is the same as that of the DF test, and the null distribution of the t statistic t ⁇ is the same as that of the DF test. Further, not the deviation from the constant term as in the DF test, but also the deviation from the primary expression ⁇ + ⁇ t with respect to the time t can be considered.
- the PP Phillips-Perron
- KPSS Kwiatowski, Phillips, Schmidthinand Shin
- LBI U Locally best invariant and unbiased
- ADF-GLS Algmented Dickey-Fully-Generalized Least Squares Method
- the stationarity determination unit 102 determines that “it is an unsteady process (with unit roots) based on given time-series data. Or “adopting that it is a steady process (adopting no unit root)”. However, if the null hypothesis and the alternative hypothesis are reversed, as in the KPSS test, “Accept a steady process (accept no unit root)” or “Unsteady process is adopted (with unit root). Adopt) ”.
- the stationarity can be determined by using any one of the unit root tests as exemplified above. However, in the environment where this embodiment is actually operated, if there is a time and resource allowance for further calculation processing, two or more unit root tests should be performed in order to strengthen the accuracy of stationary determination.
- the stationarity may be determined in combination.
- the first unit root test and the second unit root test of a different type from the first unit root test are performed, and both reject the null hypothesis (with unit root) and the alternative hypothesis (unit The alternative hypothesis (without unit root) may be adopted as a conclusion of the stationarity determination unit 102 only when “no root” is adopted. Further, three or more unit root tests may be performed, and the determination unit 102 may conclude based on a combination of these test results.
- the prediction model identification unit 103 identifies a prediction model for predicting the stochastic diffusion of communication throughput according to the determination made by the continuity determination unit 102. A specific prediction model identification method will be described below.
- the stationarity discriminating unit 102 performs the DF test on the two models shown in Expression (8) and identifies the prediction model based on the determination result.
- the prediction model identification unit 103 uses the non-stationary process model (unit root model) indicated by M0 in Expression (8) as the prediction model.
- M0 in Expression (8) the same formula as M0 in formula (8) is expressed as formula (14) below.
- the unknown parameter of this model is only the variance ⁇ 2 of the error term ⁇ t. That is, ⁇ 2 may be identified.
- the least square estimation amount of ⁇ 2 when time series data ⁇ x1, x2, x3,..., XT ⁇ is measured is obtained by the following equation (15).
- the prediction model can be identified.
- the prediction model identification unit 103 uses the steady process model indicated by M1 in Expression (8) as the prediction model.
- M1 in formula (8) the same formula as M1 in formula (8) is expressed as formula (16) below.
- the stochastic diffusion calculation unit 104 calculates the stochastic diffusion based on the prediction model identified by the prediction model identification unit 103.
- FIG. 7 is a schematic diagram showing a relationship between a change in communication throughput with time and stochastic diffusion.
- the past time series data is represented by the solid line on the left.
- a part of the possibility of future time-series data infinite in the dotted line on the right side is shown.
- this stochastic process is a Gaussian process
- the probability of communication throughput x after time t may be a Gaussian distribution (normal distribution). I know it. That is, the probability density function f (x, t) of the time t and the communication throughput x is expressed as the following equation (18).
- ⁇ t and ⁇ t are functions of t and depend on the prediction model (details will be described later).
- the stochastic diffusion is depicted in two, the upper side and the lower side.
- ⁇ is a constant.
- the processing in the stochastic diffusion calculation unit 104 is equivalent to calculating xt ⁇ . Furthermore, it is only necessary to calculate ⁇ t and ⁇ t for each prediction model.
- ⁇ t and ⁇ t in the unsteady model (unit root) M0 shown in the above equation (14) are calculated.
- ⁇ t and ⁇ t can be calculated by the following equation (20) and the following equation (21), respectively.
- FIG. 8 shows the difference in stochastic diffusion between the unsteady model and the steady model.
- Xt + in the case of. It can be seen that the stochastic diffusion of the non-stationary model is the largest. The steady model becomes closer to the non-stationary model as ⁇ is closer to 0, and the stochastic diffusion becomes smaller as ⁇ becomes smaller (if the absolute outside becomes larger).
- the communication throughput measuring unit 101 measures the communication throughput of the transmission data of the IP network 200 at every time interval ⁇ . Then, the communication throughput measuring unit 101 holds the measured communication throughput as time series data (step S11).
- the stationarity determination unit 102 determines stationarity by performing a unit root test based on a part of the time-series data held in step S11 (step S12). Then, the continuity determination unit 102 notifies the prediction model identification unit 103 of the determination result.
- the prediction model identification unit 103 selects a non-stationary model as the prediction model and identifies parameters of the prediction model ( Step S14-1). After identification, the process proceeds to step S15.
- the prediction model identification unit 103 adopts a steady model as the prediction model and identifies parameters of the prediction model (step S14). -2). After identification, the process proceeds to step S15.
- the stochastic diffusion calculation unit 104 calculates the stochastic diffusion of communication throughput based on the identification result in step S14-1 or step S14-2 (step S15).
- the embodiment described above can improve the prediction accuracy of the stochastic diffusion of communication throughput.
- the reason is that prior to predicting stochastic diffusion, it is determined whether the state of communication throughput is steady or non-stationary, and the parameters of the stochastic process model when the communication throughput is made a random variable according to the determined state It is because it identifies.
- FIG. 10 shows communication throughput when a file placed on a server on the Internet is downloaded via HSDPA (High Speed Packet Access), communication throughput when downloaded via LTE (Long Term Evolution), and WiFi (Wireless Fidelity). It is a figure showing the prediction precision when the stochastic spreading
- HSDPA High Speed Packet Access
- LTE Long Term Evolution
- WiFi Wireless Fidelity
- the curve expressed as Ideal value is the stochastic diffusion in the ideal Brownian motion (the same as the unit root model expressed by Equation (20) and Equation (21)).
- a curve expressed as HSDPA is the stochastic diffusion of communication throughput measured via HSDPA.
- the curve expressed as LTE is the stochastic diffusion of communication throughput measured via LTE.
- the curve represented as WiFi is the stochastic diffusion of communication throughput measured via WiFi.
- HSDPA can predict the probabilistic spread of communication throughput with relatively high accuracy, whereas the prediction accuracy of the probabilistic spread of communication throughput via other networks such as LTE and WiFi. Will fall. The reason for this will be described with reference to FIGS.
- LTE is the same as HSDPA in terms of a mobile packet network.
- LTE is not as unsteady as HSDPA because the number of users is still small in the current LTE, and because it is designed to shift to HSDPA when the radio wave environment deteriorates. For this reason, the prediction accuracy of LTE is worse than that of HSDPA for LTE.
- FIG. 13 is a diagram showing the prediction accuracy of the stochastic diffusion of communication throughput in HSDPA.
- FIG. 14 is a diagram showing the prediction accuracy of the stochastic diffusion of communication throughput in LTE.
- FIG. 15 is a diagram showing the prediction accuracy of the stochastic diffusion of communication throughput in WiFi.
- the prediction accuracy of stochastic diffusion in a general technique was high in HSDPA, but the accuracy was lowered in LTE and WiFi.
- the prediction accuracy of stochastic diffusion using the throughput prediction apparatus 100 of the present embodiment can realize high prediction accuracy not only in HSDPA but also in LTE and WiFi.
- continuity determination unit 102 of the present embodiment employs an optimal prediction model for prediction.
- the above throughput prediction device is configured with hardware such as an electronic circuit.
- the function of the above throughput prediction device can be realized by software or a combination of software and hardware.
- the throughput prediction method is also realized by hardware such as an electronic circuit, but can be realized by software or a combination of software and hardware.
- “realized by software” means realized by a computer reading and executing a program. For example, this is realized by incorporating software into a general-purpose server device or the like.
- An arithmetic processing device such as a CPU (Central Processing Unit) reads software and performs arithmetic processing, and various hardware is controlled according to the arithmetic processing result, thereby realizing the function of the throughput prediction device.
- CPU Central Processing Unit
- Non-transitory computer readable media include various types of tangible storage media.
- Examples of non-transitory computer readable media include magnetic recording media (eg, flexible disk, magnetic tape, hard disk drive), magneto-optical recording media (eg, magneto-optical disc), CD-ROM (Read Only Memory), CD- R, CD-R / W, semiconductor memory (for example, mask ROM, PROM (Programmable ROM), EPROM (Erasable ROM), flash ROM, RAM (random access memory)).
- the program may also be supplied to the computer by various types of temporary computer readable media. Examples of transitory computer readable media include electrical signals, optical signals, and electromagnetic waves.
- the temporary computer-readable medium can supply the program to the computer via a wired communication path such as an electric wire and an optical fiber, or a wireless communication path.
- FIG. 1 shows an example in which a throughput prediction device is connected to a network alone
- the throughput prediction device may be mounted on a communication device such as a transmission device, a reception device, or a relay device, a server device, a terminal device, and a transfer device. Good.
- FIG. 16 is a block diagram illustrating an example of a communication device that is a transmission device including a throughput prediction device.
- the communication device 300 includes a transmission unit 301, a rate controller 302, a data storage unit 303, a reception unit 304, and a throughput prediction device 305.
- the throughput prediction apparatus 305 has the same configuration as that shown in FIG.
- the transmission unit 301 of the communication device 300 transmits a packet to the reception device via the network, and the reception device transmits a transmission confirmation packet to the reception unit 304 of the communication device 300 via the network.
- the receiving unit 304 receives the transmission confirmation packet and sends it to the throughput device 305.
- the throughput predicting device 305 obtains the communication throughput by obtaining the number of packets transmitted to the receiving device per unit time via the network from the number of packets transmitted from the transmitting unit 301 and the transmission confirmation packet. Then, the throughput prediction apparatus 305 outputs the calculated stochastic diffusion to the rate controller 302.
- the rate controller 302 packetizes the data stored in the data storage unit 302, controls the transmission rate of the transmission packet based on the stochastic spreading, and outputs it to the transmission unit 301. What is stored in the data storage unit 303 is stream data such as video and audio.
- the throughput device 305 can predict the probabilistic spread of the communication throughput with high accuracy, by using the predicted communication throughput, a situation in which the reproduction of the video data is stopped in the receiving device is prevented, The video data can be steadily distributed and the video quality can be improved as much as possible.
- FIG. 17 is a block diagram showing an example of a server device provided with a throughput prediction device.
- the server device 400 includes a transmission unit 401, a content selection unit 402, a content storage unit 403, a reception unit 404, and a throughput prediction device 405.
- the throughput prediction apparatus 405 has the same configuration as that shown in FIG.
- the transmission unit 401 of the server device 400 transmits the content as a packet to the terminal device via the network, and the terminal device receives the packet and transmits a transmission confirmation packet to the reception unit 404 of the server device 400 via the network.
- the receiving unit 404 receives the transmission confirmation packet and sends it to the throughput prediction device 405.
- the throughput prediction device 405 obtains the communication throughput by obtaining the number of packets transmitted to the terminal device per unit time via the network from the number of packets transmitted from the transmission unit 401 and the transmission confirmation packet. Then, the throughput prediction device 405 outputs the calculated stochastic diffusion to the content selection unit 402.
- the content storage unit 403 stores content data such as text, images, and videos. Within the content data stored in the content storage unit 403, the content selection unit 402 within a predetermined time (for example, within 2 seconds) with a predetermined probability (for example, 95% probability) based on the stochastic diffusion of communication throughput The content that can be delivered to the terminal device is selected and the selected content is output to the transmission unit.
- the throughput prediction device 405 can predict the probabilistic spread of communication throughput with high accuracy, it is possible to steadily deliver content to the terminal device within a specified time by using the predicted communication throughput.
- FIG. 18 is a block diagram showing an example of a terminal device provided with a throughput prediction device.
- the terminal device 500 includes a content request unit 501, a content size acquisition unit 502, a reception unit 503, and a throughput prediction device 504.
- the throughput prediction apparatus 504 has the same configuration as that shown in FIG.
- the receiving unit 503 of the terminal device 500 receives the content requested by the content requesting unit 501 (to be described later) via the network, and notifies the throughput prediction device 504 of the communication throughput that is the number of packets received per unit time.
- the communication throughput prediction apparatus 504 outputs the calculated probabilistic spread of communication throughput to the content request unit 501.
- the content size acquisition unit 502 acquires the data size of the content requested by the terminal device 500.
- the content request unit 501 requests only content that can be received within a predetermined time with a predetermined probability based on the stochastic spread of communication throughput.
- the terminal device 504 can steadily acquire content within a specified time by using the predicted communication throughput. .
- FIG. 19 is a block diagram illustrating an example of a transfer apparatus including a throughput prediction apparatus.
- the transfer device 600 includes a transmission unit 601, a transmission schedule determination unit 602, a transmission buffer 603, a reception unit 604, and a throughput prediction device 605.
- the throughput prediction apparatus 605 has a configuration similar to that shown in FIG.
- Data to be transferred from the server device to the terminal device is stored as a queue in the transmission buffer 603 of the transfer device 600.
- the queue may be created for each destination terminal device, or may be created for each flow (data communication in which a combination of destination address, destination port number, source address, and source port number matches). .
- the transmission unit 601 extracts packets from the queue in the order determined by the transmission schedule determination unit 602 described later, and transmits the packets to the corresponding terminal device.
- the receiving unit 604 receives a transmission confirmation packet from the terminal device and sends it to the throughput prediction device 605.
- Throughput predicting device 605 obtains the number of packets transmitted to the terminal device per unit time via the network from the number of packets transmitted from transmitting section 601 and the transmission confirmation packet, and obtains the communication throughput for each terminal device. Then, the throughput prediction apparatus 605 outputs the calculated probabilistic spread of communication throughput for each terminal apparatus to the transmission schedule determination unit 602.
- the transmission schedule determination unit 602 performs transmission so as to maximize the number of packets that can be delivered within a predetermined time based on the stochastic spread of communication throughput for each terminal device and the queue size for each terminal device in the transmission buffer 603. Determine the order.
- the transfer device maximizes the number of packets that can be delivered to the terminal device within a specified time by using the predicted communication throughput. Can be realized.
- a continuity determination unit for determining whether the flow rate state is a steady state or an unsteady state;
- a prediction model that identifies a stochastic process model using the flow rate as a random variable based on the discrimination result by the stationarity discriminating unit, and then calculates parameters used in the identified stochastic process model based on the time series data An identification unit; Stochastic diffusion that calculates a probability distribution function or probability density function using the flow rate to be predicted as a random variable based on the stochastic process model identified by the prediction model identification unit and the parameter calculated by the prediction model identification unit A calculation unit;
- a flow rate predicting device comprising:
- the stochastic diffusion calculation unit further calculates a stochastic diffusion that is a stochastic spread of the flow to be predicted based on the probability distribution function or the probability density function.
- a continuity determination step for determining whether the state of the flow rate is a steady state or an unsteady state based on time-series data of the measured flow rate, A prediction model that identifies a stochastic process model using the flow rate as a random variable based on the discrimination result in the stationaryity discrimination step, and then calculates parameters used in the identified stochastic process model based on the time series data
- An identification step Stochastic diffusion for calculating a probability distribution function or probability density function using the flow rate to be predicted as a random variable based on the stochastic process model identified in the prediction model identification step and the parameter calculated in the prediction model identification step A calculation step;
- a flow rate prediction method comprising:
- a flow rate prediction program for causing a computer to function as a flow rate prediction device for predicting a flow rate, The computer, A continuity determination unit that determines whether the state of the flow rate is a steady state or an unsteady state based on time-series data of the measured flow rate; A prediction model that identifies a stochastic process model using the flow rate as a random variable based on the discrimination result by the stationarity discriminating unit, and then calculates parameters used in the identified stochastic process model based on the time series data An identification unit; Stochastic diffusion that calculates a probability distribution function or probability density function using the flow rate to be predicted as a random variable based on the stochastic process model identified by the prediction model identification unit and the parameter calculated by the prediction model identification unit A calculation unit; A flow rate prediction program that functions as a flow rate prediction device.
- the flow rate prediction apparatus is a fluid prediction device that is a communication throughput prediction device.
- a communication apparatus having a rate controller for controlling a transmission rate based on the stochastic spread output from the flow rate prediction apparatus.
- a server device including the flow rate prediction device including the flow rate prediction device according to supplementary note 9, A server device comprising: a content storage unit that stores a plurality of content data; and a content selection unit that selects content from the plurality of content data based on the stochastic diffusion output from the flow rate prediction device.
- a terminal device including the flow rate prediction device according to Supplementary Note 9 A terminal device having a content requesting unit that requests content that can be received under a certain condition based on the stochastic diffusion output from the flow rate prediction device.
- a transfer device including the flow rate prediction device according to supplementary note 9, A transfer apparatus comprising: a transmission buffer that accumulates transfer data; and a transmission schedule determination unit that determines a transmission order of the transfer data accumulated in the transmission buffer based on the stochastic diffusion output from the flow rate prediction apparatus.
- the present invention is suitable for prediction of an arbitrary flow rate (for example, traffic volume, liquid flow rate, gas flow rate) as well as a flow rate such as communication throughput.
- an arbitrary flow rate for example, traffic volume, liquid flow rate, gas flow rate
- a flow rate such as communication throughput.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Environmental & Geological Engineering (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Operations Research (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Description
前記コンピュータを、測定した流量の時系列データに基づき、前記流量の状態が定常状態であるのか又は非定常状態であるのかを判別する定常性判別部と、前記定常性判別部による判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定部と、前記予測モデル同定部で同定した確率過程モデルと前記予測モデル同定部で算出した前記パラメータとを基に、予測するべき前記流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出部と、を備える流量予測装置として機能させることを特徴とする流量予測プログラムが提供される。
前記定常性判別部による判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定部と、
前記予測モデル同定部で同定した確率過程モデルと前記予測モデル同定部で算出した前記パラメータとを基に、予測するべき前記流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出部と、
を備えることを特徴とする流量予測装置。
前記流量の前記時系列データを測定する測定部を更に備えることを特徴とする流量予測装置。
前記確率的拡散算出部は、前記確率分布関数又は前記確率密度関数に基づいて、予測するべき前記流量の確率的な広がりである確率的拡散を更に算出することを特徴とする流量予測装置。
前記定常性判別部は、測定した前記流量の状態が定常状態であるか又は非定常状態であるかの前記判別のために単位根検定を行なうことを特徴とする流量予測装置。
前記定常性判別部にて測定した前記流量の状態が非定常状態である判別された場合には、前記予測モデル同定部で同定する確率過程モデルは、非定常の確率過程モデルであり、
前記定常性判別部にて測定した前記流量の状態が定常と判別された場合には、前記予測モデル同定部で同定する確率過程モデルは、前記非定常の確率過程モデルとは異なる確率過程モデルである定常の確率過程モデルであることを特徴とする流量予測装置。
前記予測モデル同定部で同定する確率過程モデルがAR(Autoregression)モデルであることを特徴とする流量予測装置。
前記定常性判別ステップにおける判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定ステップと、
前記予測モデル同定ステップで同定した確率過程モデルと前記予測モデル同定ステップで算出した前記パラメータとを基に、予測するべき前記流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出ステップと、
を含むことを特徴とする流量予測方法。
前記コンピュータを、
測定した流量の時系列データに基づき、前記流量の状態が定常状態であるのか又は非定常状態であるのかを判別する定常性判別部と、
前記定常性判別部による判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定部と、
前記予測モデル同定部で同定した確率過程モデルと前記予測モデル同定部で算出した前記パラメータとを基に、予測するべき前記流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出部と、
を備える流量予測装置として機能させることを特徴とする流量予測プログラム。
前記流量予測装置は、通信スループット予測装置である流体予測装置。
前記流量予測装置から出力される確率的拡散に基づいて送信レートを制御するレートコントローラを有する通信装置。
複数のコンテンツデータを蓄積するコンテンツ蓄積部と、前記流量予測装置から出力される確率的拡散に基づいて、前記複数のコンテンツデータからコンテンツを選択するコンテンツ選択部とを有するサーバ装置。
前記流量予測装置から出力される確率的拡散に基づいて、一定条件下に受信を完了できるコンテンツを要求するコンテンツ要求部を有する端末装置。
転送データを蓄積する送信バッファと、前記流量予測装置から出力される確率的拡散に基づいて、前記送信バッファに蓄積された転送データの送信順序を決定する送信スケジュール決定部とを有する転送装置。
101 通信スループット測定部
102 定常性判別部
103 予測モデル同定部
104 確率的拡散算出部
200 IPネットワーク
300 通信装置
400 サーバ装置
500 端末装置
600 転送装置
Claims (8)
- 測定した流量の時系列データに基づき、前記流量の状態が定常状態であるのか又は非定常状態であるのかを判別する定常性判別部と、
前記定常性判別部による判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定部と、
前記予測モデル同定部で同定した確率過程モデルと前記予測モデル同定部で算出した前記パラメータとを基に、予測するべき流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出部と、
を備えることを特徴とする流量予測装置。 - 請求項1に記載の流量予測装置であって、
前記流量の前記時系列データを測定する測定部を更に備えることを特徴とする流量予測装置。 - 請求項1又は2に記載の流量予測装置であって、
前記確率的拡散算出部は、前記確率分布関数又は前記確率密度関数に基づいて、予測するべき前記流量の確率的な広がりである確率的拡散を更に算出することを特徴とする流量予測装置。 - 請求項1乃至3の何れか1項に記載の流量予測装置であって、
前記定常性判別部は、測定した前記流量の状態が定常状態であるか又は非定常状態であるかの前記判別のために単位根検定を行なうことを特徴とする流量予測装置。 - 請求項1乃至4の何れか1項に記載の流量予測装置であって、
前記定常性判別部にて測定した前記流量の状態が非定常状態である判別された場合には、前記予測モデル同定部で同定する確率過程モデルは、非定常の確率過程モデルであり、
前記定常性判別部にて測定した前記流量の状態が定常と判別された場合には、前記予測モデル同定部で同定する確率過程モデルは、前記非定常の確率過程モデルとは異なる確率過程モデルである定常の確率過程モデルであることを特徴とする流量予測装置。 - 請求項1乃至5の何れか1項に記載の流量予測装置であって、
前記予測モデル同定部で同定する確率過程モデルがAR(Autoregression)モデルであることを特徴とする流量予測装置。 - 測定した流量の時系列データに基づき、前記流量の状態が定常状態であるのか又は非定常状態であるのかを判別する定常性判別ステップと、
前記定常性判別ステップにおける判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定ステップと、
前記予測モデル同定ステップで同定した確率過程モデルと前記予測モデル同定ステップで算出した前記パラメータとを基に、予測するべき前記流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出ステップと、
を含むことを特徴とする流量予測方法。 - 流量を予測するための流量予測装置としてコンピュータを機能させるための流量予測プログラムであって、
前記コンピュータを、
測定した流量の時系列データに基づき、前記流量の状態が定常状態であるのか又は非定常状態であるのかを判別する定常性判別部と、
前記定常性判別部による判別結果に基づき、前記流量を確率変数とする確率過程モデルを同定し、その上で、前記時系列データに基づき、同定した前記確率過程モデルで用いるパラメータを算出する予測モデル同定部と、
前記予測モデル同定部で同定した確率過程モデルと前記予測モデル同定部で算出した前記パラメータとを基に、予測するべき前記流量を確率変数とした確率分布関数又は確率密度関数を算出する確率的拡散算出部と、
を備える流量予測装置として機能させることを特徴とする流量予測プログラム。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14/411,979 US9722892B2 (en) | 2012-07-02 | 2013-06-28 | Flow rate prediction device, flow rate prediction method, and flow rate prediction program |
JP2014523714A JP6256336B2 (ja) | 2012-07-02 | 2013-06-28 | 流量予測装置、流量予測方法及び流量予測プログラム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2012148284 | 2012-07-02 | ||
JP2012-148284 | 2012-07-02 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2014007166A1 true WO2014007166A1 (ja) | 2014-01-09 |
Family
ID=49881922
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2013/067848 WO2014007166A1 (ja) | 2012-07-02 | 2013-06-28 | 流量予測装置、流量予測方法及び流量予測プログラム |
Country Status (3)
Country | Link |
---|---|
US (1) | US9722892B2 (ja) |
JP (1) | JP6256336B2 (ja) |
WO (1) | WO2014007166A1 (ja) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2015222882A (ja) * | 2014-05-23 | 2015-12-10 | 日本電気株式会社 | 通信スループット予測装置、通信スループット予測方法、及び、プログラム |
WO2015184611A1 (zh) * | 2014-06-05 | 2015-12-10 | 华为技术有限公司 | 网络设备、网络业务预测装置和方法 |
WO2018034141A1 (ja) * | 2016-08-17 | 2018-02-22 | 日本電気株式会社 | ビットレート指示装置、ビットレート指示方法、及び、ビットレート指示プログラムが記録された記録媒体 |
US10445444B2 (en) | 2014-08-01 | 2019-10-15 | Nec Corporation | Flow rate prediction device, mixing ratio estimation device, method, and computer-readable recording medium |
US10749781B2 (en) | 2014-11-18 | 2020-08-18 | Nec Corporation | Setting device, setting method, recording medium to which setting program is recorded, communication system, client device, and server device |
US11509917B2 (en) | 2017-07-19 | 2022-11-22 | Nec Corporation | Data delivery device, system, method, and recording medium |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10271112B2 (en) * | 2015-03-26 | 2019-04-23 | Carnegie Mellon University | System and method for dynamic adaptive video streaming using model predictive control |
WO2019125247A1 (en) * | 2017-12-20 | 2019-06-27 | Telefonaktiebolaget Lm Ericsson (Publ) | Network node and method for deciding removal of a radio resource allocated to a ue |
TWI708197B (zh) * | 2019-04-26 | 2020-10-21 | 國立成功大學 | 生產機台組件的預測保養方法與其電腦程式產品 |
CN111200531B (zh) * | 2020-01-02 | 2022-01-04 | 国网冀北电力有限公司信息通信分公司 | 一种通信网流量预测方法、装置与电子设备 |
CN112989555B (zh) * | 2020-12-31 | 2024-04-16 | 京东科技控股股份有限公司 | 一种时间序列数据处理方法、装置、设备及计算机介质 |
TWI817519B (zh) * | 2022-05-24 | 2023-10-01 | 桓達科技股份有限公司 | 轉動式流量計於低流速狀態的流量推估方法 |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06175996A (ja) * | 1992-12-07 | 1994-06-24 | Nippon Telegr & Teleph Corp <Ntt> | 時系列予測方法 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI248508B (en) * | 2001-04-27 | 2006-02-01 | Matsushita Electric Ind Co Ltd | Method for extracting significant signal, recording medium and program |
JP2006129277A (ja) | 2004-10-29 | 2006-05-18 | Sharp Corp | 通信装置、通信方法、通信プログラム、通信プログラムを記録した記録媒体、および通信システム |
US8005707B1 (en) * | 2005-05-09 | 2011-08-23 | Sas Institute Inc. | Computer-implemented systems and methods for defining events |
WO2008119934A2 (en) * | 2007-04-02 | 2008-10-09 | British Telecommunications Public Limited Company | Handover technique for wireless communications enabled devices |
JP5483506B2 (ja) * | 2010-06-25 | 2014-05-07 | 日本電気通信システム株式会社 | 情報分類システム |
-
2013
- 2013-06-28 JP JP2014523714A patent/JP6256336B2/ja active Active
- 2013-06-28 US US14/411,979 patent/US9722892B2/en active Active
- 2013-06-28 WO PCT/JP2013/067848 patent/WO2014007166A1/ja active Application Filing
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06175996A (ja) * | 1992-12-07 | 1994-06-24 | Nippon Telegr & Teleph Corp <Ntt> | 時系列予測方法 |
Non-Patent Citations (2)
Title |
---|
HIROSHI YOSHIDA ET AL.: "Video Streaming Control by Predicting Stochastic Diffusion of TCP Throughput", INTERNET CONFERENCE 2011, 2011 * |
KUMAR, K. ET AL.: "Autoregressive integrated moving averages (ARIMA) modelling of a traffic noise time series", APPLIED ACOUSTICS, vol. 58, no. 3, 1999, pages 283 - 294 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2015222882A (ja) * | 2014-05-23 | 2015-12-10 | 日本電気株式会社 | 通信スループット予測装置、通信スループット予測方法、及び、プログラム |
WO2015184611A1 (zh) * | 2014-06-05 | 2015-12-10 | 华为技术有限公司 | 网络设备、网络业务预测装置和方法 |
CN105340311A (zh) * | 2014-06-05 | 2016-02-17 | 华为技术有限公司 | 网络设备、网络业务预测装置和方法 |
US10445444B2 (en) | 2014-08-01 | 2019-10-15 | Nec Corporation | Flow rate prediction device, mixing ratio estimation device, method, and computer-readable recording medium |
US10749781B2 (en) | 2014-11-18 | 2020-08-18 | Nec Corporation | Setting device, setting method, recording medium to which setting program is recorded, communication system, client device, and server device |
WO2018034141A1 (ja) * | 2016-08-17 | 2018-02-22 | 日本電気株式会社 | ビットレート指示装置、ビットレート指示方法、及び、ビットレート指示プログラムが記録された記録媒体 |
US10911804B2 (en) | 2016-08-17 | 2021-02-02 | Nec Corporation | Bitrate instruction device, bitrate instruction method, and non-transitory recording medium |
US11509917B2 (en) | 2017-07-19 | 2022-11-22 | Nec Corporation | Data delivery device, system, method, and recording medium |
Also Published As
Publication number | Publication date |
---|---|
JP6256336B2 (ja) | 2018-01-10 |
JPWO2014007166A1 (ja) | 2016-06-02 |
US20150180740A1 (en) | 2015-06-25 |
US9722892B2 (en) | 2017-08-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6256336B2 (ja) | 流量予測装置、流量予測方法及び流量予測プログラム | |
US10904162B2 (en) | System and method for selecting optimal path in multi-media multi-path network | |
US7693077B2 (en) | Method for estimating available bandwidth of network | |
Yoshida et al. | Constructing stochastic model of TCP throughput on basis of stationarity analysis | |
JP6337881B2 (ja) | データ予測装置 | |
CN107623641B (zh) | 限速方法及装置 | |
Van Rossem et al. | Vnf performance modelling: From stand-alone to chained topologies | |
US10361940B2 (en) | Monitoring quality of service | |
Wei et al. | HOAH: A hybrid TCP throughput prediction with autoregressive model and hidden Markov model for mobile networks | |
JP6390167B2 (ja) | 通信スループット予測装置、通信スループット予測方法、及び、プログラム | |
Wei et al. | A history-based tcp throughput prediction incorporating communication quality features by support vector regression for mobile network | |
US20150124636A1 (en) | System and Method for Delay Management for Traffic Engineering | |
WO2017169948A1 (ja) | 通信システム、可用帯域推定装置、可用帯域推定方法および可用帯域推定プログラムが記憶された記録媒体 | |
Yanev et al. | Retracted on December 7, 2023: Does TCP new congestion window validation improve HTTP adaptive streaming performance? | |
US9736078B2 (en) | Rendezvous flow control apparatus, method, and non-transitory tangible computer readable medium | |
Tunali et al. | Adaptive available bandwidth estimation for internet video streaming | |
JP2014112779A (ja) | データ送信制御装置、データ送信制御方法、および、コンピュータ・プログラム | |
Sequeira et al. | Describing the access network by means of router buffer modelling: a new methodology | |
JP2004172706A (ja) | パケット損失率測定方法及びシステム | |
Chen et al. | A probabilistic approach to estimate the mean waiting times in the earliest deadline first polling | |
Włodarski | Quality of Service for Aggregated VoIP Streams | |
Ninagawa | Edge Gateway Communication Performance Analysis | |
Ellens et al. | The effectiveness of throughput sampling for capacity management: a queueing approach | |
Antonova et al. | Evaluation of Information Transmission Resource While Processing Heterogeneous Traffic in Data Networks | |
Rizo‐Dominguez et al. | Internet delay forecasting for correlated and uncorrelated scenarios |
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: 13813599 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2014523714 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14411979 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 13813599 Country of ref document: EP Kind code of ref document: A1 |