KR100818303B1 - Method for predicting traffic characteristics and traffic estimation devices for the wired and wireless networks - Google Patents

Abstract

A method and an apparatus for predicting wired/wireless network traffics through probabilistic traffic characteristic estimation are provided to model characteristics of dynamically-changed traffics, and extract probabilistic characteristics of the traffic from limited sampling data, thereby probabilistically predicting effects of traffics, applied to a specific flow or a specific link, on service quality of the other flow or link for providing services within a network. A frame of a flow flowed into a variation band traffic within a preset time in wired/wireless networks is sampled(S601,S602). A probability distribution function for a continuous function corresponding to a histogram of continuous frame duration extracted through the sampling, continuous frame duration configured from frame pause period data and continuous frame appearance period is calculated(S609). Through the flow frame sampling increase, the probability distribution function of the continuous frame duration and the continuous frame appearance period is modified. Based on the modified probability distribution function, the traffic is predicted.

Description

{Method for predicting traffic characteristics and traffic estimation devices for the wired and wireless networks}

1 is a view for explaining an example in which a congestion link occurs in a transmission network to which the present invention can be applied;

FIG. 2 is a diagram illustrating a model applied to define an end-to-end traffic or a traffic characteristic mixed with same class traffics in the transport network of FIG. 1; FIG.

3 is a diagram illustrating a resource allocation method when a frame stream introduced from three links in a bridge 136 of the transport network of FIG. 1 is output to a specific link;

4 is a diagram for explaining a method of estimating traffic characteristics from a continuous function approximating a histogram constructed from restricted traffic stream data;

5 is a view for explaining a process of finding a probability density function in a histogram constructed from limited traffic stream data by a numerical method;

6 is a flowchart illustrating a method of estimating characteristics of Ton and Toff, which are characteristic values of the traffic shown in FIG. 2, according to an embodiment of the present invention;

FIG. 7 is a functional block diagram illustrating an apparatus for estimating traffic of an Ethernet switch for estimating characteristics of Ton and Toff, which are characteristic values of traffic shown in FIG. 2, according to an embodiment of the present invention.

The present invention relates to a method for estimating traffic characteristics of a link or a flow of a wired network and a wireless network, and more particularly, to an apparatus for estimating and estimating a traffic model of a wired or wireless network based on probabilistic characteristics and estimation of variable band traffic. And to a method thereof.

A technique for estimating traffic characteristics of a conventional network consists of an apparatus and an algorithm for collecting data in real time for every frame and processing the same. As a result, a large database and a high performance traffic analysis processor were required for traffic analysis of a high-speed link or a large network.

In addition, the technique of estimating the traffic characteristics of a large network has limitations in traffic characterization due to the limitation of the size of the database and the processor.

In particular, in the case of traffic having variable bandwidth characteristics in a network, it is difficult to develop models and algorithms that can predict the characteristics of traffic.

A first object of the present invention for solving the above problems is to provide an apparatus and method for traffic model and prediction of a wired / wireless network based on stochastic characteristics and estimation of variable band traffic.

A second object of the present invention is to provide an apparatus and method for a traffic model and prediction that can be applied to QoS control of wired / wireless converged networks based on stochastic estimation of traffic characteristics having fluctuating bands with limited data only. To provide.

A third object of the present invention is an apparatus for traffic model and prediction that can be applied to network capacity planning, resource allocation, and connection admission control through probabilistic characteristics and estimation of variable band traffic with limited data only; To provide that method.

In order to achieve the above object, a traffic prediction method of a wired or wireless network according to an embodiment of the present invention includes sampling a frame of a flow within a time limit, collecting a continuous frame duration and a continuous frame appearance period data, and collecting the collected data. Construct a histogram of successive frame durations and successive frame emergence cycles, calculate a continuous function approximating the histogram, calculate a probability distribution function from the continuous function, and periodically extend the sampling database, And a second step of modifying the probability distribution function of the successive frame duration and the successive frame appearance period by the extension database, and estimating traffic based on the modified probability distribution function.

According to an embodiment of the present invention, the first step includes: extracting frame arrival time data through frame sampling; Distinguishing whether a newly arrived frame from the frame arrival time database is a continuous frame or a frame first arrived after a frame rest period; According to the determination, modifying the frame arrival time interval database and determining whether to continue frame sampling; Constructing a histogram of successive frame durations and a histogram of frame appearance periods from successive frame duration and frame idle duration data when the flow characteristic extraction period is reached; And approximating a probability distribution function of successive frame durations, and approximating a probability distribution function of frame appearance periods.

The second step also includes determining traffic estimation links and flows, and controlling frame sampling; Managing a database consisting of frame arrival time data through frame sampling and traffic characteristic data periodically calculated until the traffic characteristic estimation and prediction is completed; And predicting traffic of the link or flow from the traffic estimation data.

On the other hand, the apparatus for the construction of the transmission network according to an embodiment of the present invention for achieving the above object, the traffic characteristic basic data by extracting the arrival time information of the corresponding frame received as variable band traffic through the network interface in the wired or wireless network A frame sampler for generating in frame duration Ton and frame pause time Toff data; A flow characteristic extraction unit which starts and ends traffic characteristic estimation of a link or a flow by a network management module or a connection signaling protocol, and periodically calculates characteristics of the traffic; And a flow estimator for predicting a traffic characteristic of a link or flow to be multiplexed by the frame switch from the traffic characteristic information.

In the present embodiment, the frame sampler extracts frame arrival time information received for a predetermined period of time through the network interface, and determines a start time of the Toff, the traffic characteristic basic data, based on a previous frame arrival time. Store information in the flow statistics database.

The flow characteristic extraction unit performs calculations representing a dispersion function of the tons and tcycles as a continuous function based on the histogram data of tons and frame appearance cycles configured from the tons and toff data and stores them in a flow statistics database. Reset the frame sampling period.

The flow estimator estimates the characteristics of the integrated flow on the output link based on the frame switching information and the Ton and Toff characteristic data of the flow statistics database when requesting for multiplexing of flows or resource allocation of a new flow.

The apparatus for configuring a transmission network according to an embodiment of the present invention further includes a signal quality (QoS) control unit for controlling resource allocation by controlling a connection admission control or an output queue scheduler based on the flow characteristic information estimated by the flow estimating unit. Include.

According to the present invention, it is possible to model the characteristics of dynamically changing traffic, and to provide a method of extracting the stochastic characteristics of traffic from limited sampling data, so that the traffic flowing into a specific flow or a specific link is in service in the network. Capacity planning, resource allocation, connection admission control, and signal quality of wired or wireless networks that accommodate variable bit rate traffic can be predicted with a probabilistic estimate of how it will affect the quality of service of the flow or link. Applicable to (QoS) control, etc., it is possible to guarantee the quality of service that could not be provided in the existing transport network.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, whereby those skilled in the art may easily implement the technical idea of the present invention. There will be. In addition, in describing the present invention, when it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. The following merely illustrates the principles of the invention. Therefore, those skilled in the art, although not explicitly described or illustrated herein, can embody the principles of the present invention and invent various devices that fall within the spirit and scope of the present invention. Furthermore, all conditional terms and embodiments listed herein are in principle clearly intended for the purpose of understanding the concept of the invention and are not to be limited to the specifically listed embodiments and states. Should be.

In addition, it is to be understood that all detailed descriptions, including the principles, aspects, and embodiments of the present invention, as well as listing specific embodiments, are intended to include structural and functional equivalents of these matters. In addition, these equivalents should be understood to include not only the equivalents currently known but also the equivalents developed in the future, that is, all devices invented to perform the same function regardless of the structure.

1 is a view for explaining an example in which a congestion link occurs in a transmission network to which the present invention can be applied. 1 illustrates a configuration of a bridged local area network (Bridged LAN).

As shown, the bridged local area network may be comprised of Ethernet switch bridges 114, 116, 132, 134, 136, and extended local area communication networks 122, 124, 126, 142 in which a transport network that accommodates end stations 102, 104, 106 is provided.

As such, the bridge local area network connects a single local area network 122 that accommodates an end station 1 102 to each other by a media access control bridge 132. By interconnecting or relaying or filtering data frames between the separated bridges, the design of an extended local area communication network is possible.

Here, the bridges 114, 116, 132, 134, and 136 separate the traffic by distinguishing the user from the constituent user of the bridge local area network and queuing the frame according to the traffic class, or selecting and transmitting the queued frame.

In the figure, Termination 1 102 and Termination 3 106 communicate via the first path 101 while Termination 2 104 and Termination 3 106 communicate with the second path 103. When trying to communicate, the bridges 116 and 136, which are Ethernet switches located in a path where the first path 101 and the second path 103 overlap, are used simultaneously.

Bridge 1 116 separates the traffic according to the end classification and queues frames according to the class, or in the same class, mixes the same queue and transmits the same to the next bridge. Bridge 2 136 receives the traffic generated at the end 102 and the traffic generated at the end 104, queues according to the class, and transmits the traffic to the link 126.

On the other hand, when a large number of frames are to be delivered to the bridges 116 and 136 simultaneously, congestion occurs and the overflowed frames may be deleted. This phenomenon is likely to occur even if the average network utilization is low.

In order to control this, the bridges 116 and 136 allocate resources by controlling the time for using the links 101 and 103 by dividing the traffic by each end or by the same class. In this way, the quality of service requirements can be maintained by a scheduling method for determining the start time and service period for servicing the queue for each class or accepting the inflow of new traffic into the network.

 FIG. 2 is a diagram for explaining a model applied to define a traffic characteristic mixed with end-to-end traffic or same class traffics shown in the transport network of FIG. 1.

Referring to the drawings, a period of generating data to be delivered periodically (Tcycle) 200 is expressed by expressing data generated from an application service or aggregated traffic according to a behavior characteristic of a user or a characteristic of a service. The frame duration (Ton) 201 and a period in which the user absorbs data or the data does not occur for a certain time according to a service characteristic, that is, the frame pause period (Toff) 202 is repeated. have.

The generated data is divided into frames of a certain length according to the capacity of the network, and a random unit time is defined to represent the time occupied by the link to transmit the data as discrete data and is defined as a slot (203).

The size of the frame duration 201 for generating data to be delivered and the frame idle period 202 for which no data is generated varies depending on the application service or the absorbed traffic. In general, the traffic characteristics are the frame duration 201 where the data is generated. And a probability distribution of the frame idle period 202 that is not generated. This model can be applied in the same way to represent the characteristics of traffic input to the bridge or traffic queued and output on the link.

3 is a diagram illustrating a resource allocation method when a frame stream introduced from three links in a bridge 136 of the transport network of FIG. 1 is output to a specific link.

Referring to the drawings, the frame streams received through the three input links are stored in the input queues 301, 302, and 303, respectively, and then forwarded to the output queue 304 to be transmitted to a specific link under the control of the queue scheduler 300. do. The input frame stream may be represented by a stochastic distribution of Ton and a stochastic distribution of Toff according to the frame stream characteristic definition model of FIG. 2.

The histogram 311 according to the ton length of the stream input to the input queue 301 represents the characteristics of the input stream and maintains the same characteristics when occupying the output link by the queue scheduler 300.

The frame stream characteristics of the output link are changed according to the Toff characteristic and the scheduling control scheme as well as the Ton of the input stream.

In the three input links, a stream having the histogram characteristics 311, 312 and 313 of Ton and the histogram characteristic of Toff is forwarded through the input queue scheduler 300, and then the integrated frame stream transmitted through the output queue 304 is Ton. The characteristics change with the histogram 314 of and histogram 315 of Toff. This property is occupied by link resources.

Therefore, the control scheme for optimal resource allocation should be based on the prediction of the stream characteristics of the output link formed by the stream and queue scheduling of the input link.

4 is a diagram for explaining a method of estimating traffic characteristics from a continuous function approximating a histogram constructed from restricted traffic stream data.

Referring to the figure, it is required to predict the characteristics of the traffic for the resource allocation of FIG. The characteristics of the traffic are the length of the frame or the arrival time. This value has a random probability distribution. If the probability distribution over the frame length is known, it is possible to estimate the characteristics of the traffic to be input in the future.

The probability density function for traffic estimation can be expressed as an approximation from the actually observed traffic data. After a histogram is obtained from the frame length or inter-arrival time data of the traffic stream collected for a limited time (401), a continuous function f (x) approximating the histogram is obtained (402). F (x) can be obtained through the equation 'f (x) / f (x) dx'.

Here, the traffic characteristics can be estimated stochastically from the probability density function F (x). The continuous function f (x) approximated from the histogram constructed from the traffic stream data is divided into n constant intervals 405 between the minimum value 403 and the maximum value 404 to obtain functions approximated to the histogram of each interval and expressed as the sum. Can be.

FIG. 5 is a diagram for explaining a process of finding a probability density function by a numerical method in a histogram constructed from limited traffic stream data.

In the present embodiment, a method of selecting a continuous function approximating a histogram may apply a general numerical analysis method.

As shown in FIG. 5, in order to select a continuous function approximating the displayed histogram 500, adjacent functions may be selected from four positions 501, 502, 503, and 504, which are main inflection points, and expressed as a combination thereof. Equation 1 below is applied as a function approximating each interval as a method of finding such a combination of functions.

f _i (x) = a _i (xx _i ) ³ + b _i (xx _i ) ² + c _i (xx _i ) + d _i

The condition that such a function is continuous is shown in Equation 2 below.

f _i (x _{i + 1} ) = f _{i + 1} (x _{i + 1} ), f ′ _i (x _{i + 1} ) = f ′ _{i + 1} (x _{i + 1} ), f ″ _i (x _{i + 1} ) = f ″ _{i + 1} (x _{i + 1} )

If we find a _i , b _i , c _i , and d _i , we can determine the continuous function for each interval. Accordingly, a combination of the functions 506, 507, 508, and 509 determined in each section may provide a continuous function approximating the histogram.

6 is a flowchart illustrating a method of estimating characteristics of Ton and Toff, which are characteristic values of the traffic shown in FIG. 2, according to an embodiment of the present invention.

As shown, the bridge 136, which is an Ethernet switch, sets a data extraction period after selecting a traffic characteristic estimation target flow or a target link (S601). The bridge 136 extracts a frame sampling Ton (i) time when the corresponding frame is found from the input link (S602).

The bridge 136 subtracts the sampling Ton (i) time of the previous arrival frame from the sampling Ton (i-1) time of the current arrival frame, which is longer than the transmission time of the maximum transmittable frame length, and predicts probabilistic traffic characteristics. It is compared whether the database is larger than the 3rd quartile value of Toff (S603).

The bridge 136 stores Ton (0) in the Ton database at the beginning of the data stream if the interval between Ton (i-1) and Ton (i) is longer than the maximum frame length transmission time and is greater than the 3rd quartile value of Toff. And store Ton (i) in the Toff database at the end of the data stream (604), and at the same time store Ton (i) in Ton (0) as the starting point of new data generation and initialize it with the value of i (1). S605). The bridge 136 stores Ton (i) information without storing the Toff value if the interval between Ton (i-1) and Ton (i) is shorter than the transmission time of the maximum frame length or smaller than the 3rd quartile value of Toff. And increase i (S606).

The bridge 136 determines whether the data extraction period of the flow has timed out (S607). If the flow data extraction cycle has not timed out, the bridge 136 continuously monitors the input link to perform steps S602 to S607.

If it is determined that the flow data extraction cycle has ended, the bridge 136 constructs a Ton histogram and a Tcycle histogram from the stored Ton database and the Toff database (S608). The Ton and Tcycle histogram data are divided into arbitrary intervals based on the maximum value and the minimum value, a continuous function is calculated according to the method of FIG. 5, and converted into a probability density function (S609).

Accordingly, the bridge 136 adds the flow extraction data to the existing database (DB), expands the characteristic prediction database by recalculating the probabilistic characteristics of traffic by flow or by link (S610), and flows in the next cycle. Characteristic estimation (steps S601 to S610) is repeated.

FIG. 7 is a functional block diagram illustrating an apparatus for estimating traffic of an Ethernet switch for estimating characteristics of Ton and Toff, which are characteristic values of traffic shown in FIG. 2, according to an embodiment of the present invention.

As shown, in the traffic estimating apparatus, the frame sampler 702 is configured to estimate the characteristics of the flow through a frame received through the network port interface 701 on the Ethernet network side. Extract frame arrival time information while The frame sampler 702 determines the start time of Toff based on the previous frame arrival time, and stores this information in the Flow Statistics DB 703.

The flow infor extractor 704 manages the control of the frame sampler 702 and the flow statistics database 703. In addition, the flow characteristic extractor 704 constructs a histogram of Ton and Tcycle based on Ton and Toff data, performs a calculation that expresses a dispersion function of Ton and Tcycle as a continuous function, and stores it in the flow statistics database 703. Reset the frame sampling period.

The flow estimator 705 determines characteristics of the integrated flow on the output link based on the frame switching information and the Ton and Toff characteristic data of the flow statistics database 703 when multiplexing the flow or requesting resource allocation of a new flow. Estimate. The flow estimating unit 705 provides the estimated information to the signal quality (QoS) controller 706.

The signal quality control unit 706 performs resource allocation control by controlling the connection admission control or the output queue scheduler based on the information provided from the flow estimating unit 705.

In the above, specific preferred embodiments of the present invention have been illustrated and described. However, the present invention is not limited to the above-described embodiments, and any person having ordinary skill in the art to which the present invention pertains may make various modifications and other equivalents without departing from the gist of the present invention attached to the claims. Implementation will be possible. Therefore, the true technical protection scope of the present invention should be defined only by the appended claims.

According to the present invention, it is possible to model the characteristics of dynamically changing traffic, and to provide a method of extracting the stochastic characteristics of traffic from limited sampling data, so that the traffic flowing into a specific flow or a specific link is in service in the network. Capacity planning, resource allocation, connection admission control, and signal quality of wired or wireless networks that accommodate variable bit rate traffic can be predicted with a probabilistic estimate of how it will affect the quality of service of the flow or link. Applicable to (QoS) control, etc., it is possible to guarantee the quality of service that could not be provided in the existing transport network.

Claims (8)

The frame of the flow flowing into the fluctuation band traffic within the set time in the wired / wireless network is sampled, and the histogram of the continuous frame duration and the continuous frame appearance period configured from the continuous frame duration and the frame idle data extracted through the sampling. Calculating a probability distribution function for the corresponding continuous function; And And modifying the probability distribution function of the continuous frame duration and the continuous frame appearance period by increasing the flow frame sampling, and estimating the traffic based on the modified probability distribution function. A traffic prediction method for wired and wireless networks. The method of claim 1, The first step, Extracting frame arrival time data through frame sampling; Distinguishing whether a newly arrived frame from the frame arrival time database is a continuous frame or a frame first arrived after a frame rest period; According to the determination, modifying the frame arrival time interval database and determining whether to continue frame sampling; Constructing a histogram of successive frame durations and a histogram of frame appearance periods from successive frame duration and frame idle duration data when the flow characteristic extraction period is reached; And Approximating a probability distribution function of successive frame durations, and approximating a probability distribution function of frame appearance periods. The method according to claim 1 or 2, The second step, Determining traffic estimation links and flows, and controlling frame sampling; Managing a database consisting of frame arrival time data through frame sampling and traffic characteristic data periodically calculated until the traffic characteristic estimation and prediction is completed; And Predicting traffic of a link or a flow from the traffic estimation data. A frame sampler for extracting arrival time information of a corresponding frame received through variable bandwidth traffic through a network interface in the wired / wireless network to generate frame duration Ton and frame pause Toff data which are basic data of traffic characteristics; Estimation and start of traffic characteristics of a link or flow are performed by a network management module or a connection signaling protocol. The distribution function of Ton and Tcycle is derived based on the histogram data of Ton and frame appearance cycles configured from Ton and Toff data. A flow characteristic extracting unit configured to calculate a characteristic for the traffic periodically by performing a calculation expressed as a continuous function, storing the result in a flow statistics database, and resetting a frame sampling period; And And a flow estimator for predicting a traffic characteristic of a link or a flow to be multiplexed by the frame switch from the traffic characteristic information. The method of claim 4, wherein The frame sampler is, Extracting frame arrival time information received for a predetermined period through the network interface, determining the start time of the traffic characteristic basic data Toff based on a previous frame arrival time, and storing the information in a flow statistics database. Traffic prediction device of a wired / wireless network frame switch. delete The method of claim 4, wherein The flow estimating unit, Traffic prediction apparatus for wired / wireless network frame switches, when the multiplexing of flows or resource allocation request of a new flow is performed, the characteristics of the integrated flow on the output link are estimated based on the frame switching information and the Ton and Toff characteristic data of the flow statistics database. . The method of claim 7, wherein Based on the flow characteristic information estimated by the flow estimating unit, the traffic estimation apparatus of the wired / wireless network frame switch further comprising a signal quality (QoS) control unit for controlling resource allocation by controlling a connection admission control or an output queue scheduler. .

Images