KR102551887B1 - Apparatus and method for designing network capacity - Google Patents

Abstract

The present application relates to a communication network capacity designing device and a communication network capacity designing method. The communication network capacity designing method according to an embodiment of the present invention includes the steps of analyzing a topology in a communication network and classifying links within the communication network; setting a traffic distribution model for each link by reflecting the traffic distribution characteristics of the service flowing through the links; and calculating Service Level Agreement (SLA) link capacities for the links by applying the traffic distribution model.

Description

Apparatus and method for designing network capacity {Apparatus and method for designing network capacity}

The present application relates to a communication network capacity design device and a communication network capacity design method, and more particularly, to a communication network capacity design device and a communication network capacity design method capable of applying a traffic distribution model reflecting service traffic characteristics.

It is common to design an Internet communication network by analyzing traffic circulating in the communication network to create a model for the traffic, and designing the capacity of the communication network based on the generated modeling. Conventionally, when modeling traffic for a communication network, capacity design was performed on the assumption that the network utilization probability distribution of traffic followed a normal distribution. That is, after assuming the network utilization probability distribution of traffic, based on the assumption, SLA link capacity that satisfies an appropriate service level agreement (SLA) for traffic distribution is calculated, and link capacity design for the communication network is performed. can do.

However, in order to assume that the probability distribution of network utilization is a normal distribution function, the sample size is determined by the assumption that individual traffic characteristics of individual services circulating in the network must be multiplexed enough to be ignored, and by the Central Limit Theorem. The premise must be established that the population standard deviation must be known or can be replaced with the sample standard deviation while exceeding the reference value.

In the case of links constituting a communication network, links belonging to a core network generally satisfy the above preconditions, but most links belonging to an access network do not satisfy the preconditions. Therefore, it is difficult to provide a high level of reliability for the design result of traffic modeling and link capacity design based on the assumption that all link sections constituting the communication network are collectively normal distribution.

The present application is intended to provide a communication network capacity designing device and a communication network capacity designing method capable of performing reliable link capacity design by using a traffic distribution model reflecting service traffic characteristics.

A communication network capacity design method according to an embodiment of the present invention relates to a traffic modeling-based communication network capacity design method, comprising the steps of analyzing a topology in a communication network and classifying links within the communication network; setting a traffic distribution model for each link by reflecting the traffic distribution characteristics of the service flowing through the links; and calculating Service Level Agreement (SLA) link capacities for the links by applying the traffic distribution model.

Here, the communication network includes an aggregation node connected to customer terminals, an intermediate node to which a plurality of aggregation nodes are connected, a static connection node connected to the intermediate node, and a TCP (Transmission Control Protocol) protocol service connected to the static connection node It may include a first server node connected to a server, and a second server node connected to the static connection node and connected to a service server of a User Datagram Protocol (UDP) protocol.

Here, the service server of the TCP protocol may provide a best effort (BE) service or a unicast download and play (D&P) service, and the service server of the UDP protocol may provide a multicast streaming service or a unicast streaming service. .

Here, the classifying step may include a first link between the aggregation node and the intermediate node, a second link between the intermediate node and the static connection node, a third link between the static connection node and the first server node, and a static connection node and the second link. Each can be classified as a fourth link between two server nodes.

Here, in the modeling step, in the case of the BE service, a traffic distribution model according to a Pareto distribution for the first link and a normal distribution for the second link and the third link may be applied.

In the modeling step, in the case of a multicast streaming service, a traffic distribution model according to a binomial distribution for the first link, a Poisson distribution for the second link, and a uniform distribution for the third link may be applied.

Here, in the modeling step, in the case of a unicast streaming service, a traffic distribution model according to a binomial distribution for the first link and a Poisson distribution for the second link and the third link may be applied.

Here, in the modeling step, in the case of a unicast D&P service, a traffic distribution model according to a binomial distribution for the first link and a Poisson distribution for the second link and the fourth link may be applied.

An apparatus for designing communication network capacity according to an embodiment of the present invention includes a classification unit that analyzes a topology within a communication network and classifies links within the communication network; a model setting unit that sets a traffic distribution model for each link by reflecting the traffic distribution characteristics of the service flowing through the links; and a calculation unit that calculates a service level agreement (SLA) link capacity for each link by applying the traffic distribution model.

In addition, the solution to the above problem does not enumerate all the features of the present invention. Various features of the present invention and the advantages and effects thereof will be understood in more detail with reference to specific embodiments below.

According to the communication network capacity design apparatus and communication network capacity design method according to an embodiment of the present invention, it is possible to perform traffic modeling and link capacity design in which individual service characteristics are reflected for a link for which a normal distribution function cannot be assumed. Therefore, it is possible to increase the reliability of link capacity design compared to the conventional normal distribution function, and to increase the efficiency of Internet service provision by the established communication network.

1 is a schematic diagram showing the topology of a communication network according to an embodiment of the present invention.
2 is a block diagram showing an apparatus for designing communication network capacity according to an embodiment of the present invention.
3 is a schematic diagram illustrating traffic modeling using a Pareto model according to an embodiment of the present invention.
4 is a flowchart illustrating a design traffic calculation method based on a Pareto model according to an embodiment of the present invention.
5 is a schematic diagram showing traffic modeling of a normal distribution model according to an embodiment of the present invention.
6 is a flowchart illustrating a design traffic calculation method according to a Poisson distribution model according to an embodiment of the present invention.
7 is a flowchart illustrating a communication network capacity design method according to an embodiment of the present invention.

Hereinafter, preferred embodiments will be described in detail so that those skilled in the art can easily practice the present invention with reference to the accompanying drawings. However, in describing a preferred embodiment of the present invention in detail, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. In addition, the same reference numerals are used throughout the drawings for parts having similar functions and actions.

In addition, throughout the specification, when a part is said to be 'connected' to another part, this is not only the case where it is 'directly connected', but also the case where it is 'indirectly connected' with another element in between. include In addition, 'including' a certain component means that other components may be further included, rather than excluding other components unless otherwise stated. In addition, terms such as “~unit” and “module” described in the specification refer to a unit that processes at least one function or operation, and may be implemented as hardware or software or a combination of hardware and software.

1 is a schematic diagram showing the topology of a communication network according to an embodiment of the present invention. The topology of the actual communication network may be more complex, but is simplified and displayed here for explanation.

Referring to FIG. 1, the communication network includes an aggregation node n1 connected to customer terminals 1, an intermediate node n2 to which a plurality of aggregation nodes n1 are connected, and a static connection node connected to the intermediate node n2. (n3), the first server node (n4) connected to the static connection node and connected to the service server (S1) of the TCP protocol, connected to the static connection node (n3) and connected to the service servers (S2, S3) of the UDP protocol It may include second server nodes (n5, n6) to be. Here, the service server S1 of the TCP protocol may provide a BE service or a unicast D&P (Download and Play) service, and the service servers S2 and S3 of the UDP protocol may provide a multicast streaming service or a unicast streaming service. can provide

In addition, a first link L1 is provided between the aggregation node n1 and the intermediate node n2, a second link L2 is provided between the intermediate node n2 and the static connection node n3, and the static connection node n3 and A third link L3 is provided between the first server nodes n4, and a fourth link L3 is provided between the static connection node n3 and the second server nodes n5 and n6.

Here, according to an embodiment of the present invention, the capacity of links included in the communication network may be designed by applying the communication network capacity design apparatus 100 . That is, using the communication network capacity design device 100, a traffic distribution model can be set by reflecting individual service characteristics of each link that cannot be assumed to be normally distributed, and through this, the link capacity of the entire communication network can be designed. can

Here, in order to perform modeling according to traffic distribution characteristics of services flowing into each link, it is necessary to analyze service traffic characteristics and classify links according to service traffic types. Link classification may exist in various analysis and classification methods depending on the viewpoint, but the communication network capacity design apparatus 100 according to an embodiment of the present invention classifies each link into a best effort (BE) service, a multicast type streaming service, It can be classified into unicast type streaming service and unicast type D&P service.

Specifically, the BE service is applied to traditional Internet access services such as WWW, FTP, and e-mail services, and corresponds to a service that guarantees only access to the Internet independently of individual applications within the limit of idle resources of the network. It is usually called Best Effort Service in the sense of providing the best resources for Internet network access within the limit possible.

The BE service has a strong self-similarity in traffic-causing characteristics and a burst characteristic, that is, a very large peak value compared to an average value of traffic. In addition, it has a probability distribution property in the form of a heavy-tail. Therefore, traffic modeling that can express strong self-similarity and clustering while satisfying the long-tail distribution function requirement in terms of probability distribution is required for a link to which the BE service is applied.

In addition to traditional Internet access services, there are various types of video services. Even if a video service is provided by an SLA of the same level as the BE service, the corresponding video service can be included in the traffic model for the BE service. However, in the case of a video service that requires a higher level of quality assurance than the SLA of the BE service, it is necessary to know the front end path of the traffic, and new modeling according to the traffic distribution model is also required.

Quality-guaranteed video services can be divided into multicasting streaming services, unicasting streaming services, and unicasting download and play services.

The multicast type streaming service includes a real-time channel video providing service such as an IP-TV service, and the unicast type streaming service includes a real-time Video on Demand (VoD) providing service. Here, the multicast-type streaming service and the unicast-type streaming service may use a UDP (User Datagram Protocol) protocol as a traffic transfer method due to the nature of the streaming service. In the case of a unicast type D&P service, providing a real-time VoD service is the same as a unicast type streaming service, but there is a difference in that a Transmission Control Protocol (TCP) protocol is used as a traffic transfer method. That is, in the unicast type D&P service, a slight difference is made between a download start time and a play start time in order to prevent video breakage during playback. This is a buffering technique that implements perfect video playback by waiting until enough traffic for video playback is downloaded and then starting playback. If the communication quality level required by the communication network is not satisfied, playback delay may occur, but of playback quality can be maintained.

In addition, an internet video phone service may exist, and in the case of the internet video phone service, the service may be provided in a manner similar to the above-described video providing method. In the case of a video service, the service provision direction is unidirectional, but the video phone service is bidirectional. That is, the one-way unicast streaming service corresponds to a video service, and the two-way unicast streaming service corresponds to a video phone service. If the two-way unicast streaming service is a two-way video phone service, the two-way multicast streaming service corresponds to a multi-party video phone service with three or more people. In addition, the voice-only Internet phone service corresponds to an audio-only service excluding video from the video phone service.

Therefore, if traffic analysis and modeling are performed by classifying into four types of BE service, multicast-type streaming service, unicast-type streaming service, and unicast D&P service, it is possible to perform analysis and modeling covering all service types. .

Specifically, referring to FIG. 2 , the communication network capacity design apparatus 100 may include a classification unit 110 , a model setting unit 120 and a calculation unit 130 .

The classification unit 110 may analyze the topology within the communication network and classify links within the communication network. That is, each of the first link L1, the second link L2, the third link L3, and the fourth link L4 can be distinguished from the communication network topology of FIG.

The model setting unit 120 may set a traffic distribution model for each link by reflecting traffic distribution characteristics of services flowing into each link L1 , L2 , L3 , and L4 . Specifically, as shown in Table 1, the model setting unit 120 may set a traffic distribution model for each link classification.

division 1st link 2nd link 3rd link 4th link BE service Pareto Distribution normal distribution - normal distribution multicast
streaming service binomial distribution Poisson distribution uniform distribution - Unicast
streaming service Poisson distribution - Unicast D&P service - Poisson distribution

The calculation unit 130 may calculate the SLA link capacity for each link by applying the traffic distribution model. That is, by applying the SLA link capacity to each link, it is possible to perform capacity setting for the entire communication network.

Specifically, referring to Table 1, in the case of BE service, the model setting unit 120 may set a traffic distribution model using a Pareto model in the first link L1. As shown in Figure 3, the Pareto model corresponds to an on/off model that can express general Internet traffic such as WWW, FTP, and E-mail services, and strong self-similarity and clustering are achieved by assuming a Pareto distribution for burst sessions. it is possible to express In addition, the long-tail distribution function requirement for service provision time can be satisfied. That is, the Pareto model may have characteristics suitable for traffic design of the first link L1 having a small scale of connected customer terminals 1 compared to the case of assuming a normal distribution.

Accordingly, the model setting unit 120 may apply the Pareto model to the first link L1 of the BE service, and at this time, a traffic distribution model using the Pareto model may be set using the design parameters in Table 2.

division meaning note average transmission volume The average of the bands used by one customer on that link Estimated average transmission amount at design time Max Transmission Amount Maximum amount of bandwidth available to one customer on that link Maximum transmission amount of customer subscription products Average ON time Average traffic generation time per unit time Calculated using Equation 1 and Equation 2 Average OFF time Average time without traffic per unit time Calculated using Equation 1 and Equation 2 TailIndex How large or long the tail is in the Pareto distribution A median value of 1.5 is usually used.

Then, as shown in FIG. 4 , the calculation unit 130 may calculate the SLA link capacity for the first link L1 using the Pareto model. That is, simulation can be performed using each of the design parameters set in the model setting unit 120, and 'design traffic of a unit link compared to the number of accepted customers' can be calculated through the simulation. For example, if up to 24 customer terminals 1 can be accommodated in the first link L1, and as a result of simulation for the 24 customer terminals 1, the effective bandwidth calculation result is 60 Mbps, '60 Mbps' + αbps' may be set as the SLA link capacity for the BE service of the first link (L1). Here, α is a correction coefficient for stable operation of the link, and corresponds to surplus bandwidth to ensure an appropriate reliability level (SLA).

In addition, the calculation unit 130 may set SLA link capacities for the second link L2 and the fourth link L4 in the case of the BE service, based on a normal distribution. In this case, a traffic distribution model in the form of a normal distribution may appear as shown in FIG. 5, and the SLA link capacity is (average traffic forecast per customer at the time of design) x (number of customers) + K*(busy hour traffic) standard deviation). Here, K corresponds to the loss rate and may be set to 3.09 depending on the embodiment.

Meanwhile, the model setting unit 120 may set a traffic distribution model of the first link L1 for the video service as a binomial distribution model instead of a Poisson distribution. This is due to the nature of the first link L1 connected to the aggregation node n1 directly connected to the customer terminal 1, the number of customers receiving services through the first link L1 is not large, and accordingly, Erlang- This is because the Poisson distribution assumption applying Formula B is inappropriate. Therefore, instead of the Poisson distribution, the traffic distribution model can be set as a binomial distribution model based on concurrent usage (simultaneous viewership).

In this case, the calculation unit 130 calculates (number of customers) x (simultaneous viewing rate) x (required band per issue) + α, where 0 < simultaneous viewing rate < 1, and α is a correction coefficient for stable operation of the link, α = k(=6, loss rate 0%) x √ (number of customers * concurrent utilization rate * (1-concurrent utilization rate) * required bandwidth per call) SLA link capacity can be set.

In addition, the model setting unit 120 may set a Poisson distribution model as a traffic distribution model for the second link L2, the third link L3, and the fourth link L4, in which case the calculation unit 130 The SLA link capacity can be set using the Erlang-B formula.

In the case of a traffic distribution model using the Erlang-B formula, it is possible to satisfy the requirement to guarantee a certain bandwidth during service duration, and it is possible to easily extract the required bandwidth for each service.

Specifically, the Erlang-B formula is as follows, and design parameters for calculating the SLA link capacity using the Erlang-B formula are shown in Table 3.

However, when the blocking probability is φ, E(C,ρ) < φ,

ρ = (Mean call arrival rate * Mean holding time of a call) / 3600

= (M × a × λ ÷ μ) / 3600

parameter Notation note number of registered customers M According to access node size Busy hour call attempt rate a Depends on individual service Average arc duration 1/μ same average call arrival rate λ load ρ ρ = λ/μ Blocking Probability φ band required ω Depends on individual service

Here, E(C, ρ) corresponds to the blocking probability when C lines are loaded with ρ, and it is possible to set C and ρ to satisfy the desired blocking probability, respectively. That is, when the load ρ applied to the link is determined, the number of circuits to satisfy the blocking probability can be derived.

Therefore, referring to FIG. 6, the calculation unit 130 first multiplies the busy hour call attempt rate per customer by the average call duration to obtain the design traffic per customer in Erlang units (S210), and then the number of customers receiving the unit link. By multiplying by the design traffic per customer, unit link design traffic can be obtained in Erlang units (S220). Here, since the unit link design traffic corresponds to the load applied to the unit link, the number of logically required lines satisfying a specific call blocking probability can be calculated by applying the Eralng-B formula (S230). Then, by multiplying the required number of logical lines by the required bandwidth per line, unit link design traffic can be obtained in units of bandwidth (S240). That is, the calculation unit 130 may set respective SLA link capacities for the second link L2, the third link L3, and the fourth link L4 in the video service.

On the other hand, in the case of the third link (L3) of the multicast streaming service, the model setting unit 120 may set a traffic distribution model with a uniform distribution. That is, since the third link L3 of the multicast streaming service distributes all live channel traffic in service for the live channel service, constant traffic can be maintained. Accordingly, the calculation unit 130 may set the SLA link capacity for the third link L3 as (required bandwidth per channel) x (number of lines).

7 is a flowchart illustrating a communication network capacity design method according to an embodiment of the present invention. Here, each step of the communication network capacity design method may be performed by the communication network capacity design device described above.

Specifically, the communication network capacity design apparatus may analyze the topology of the communication network and classify links within the communication network (S10). Here, the communication network includes an aggregation node connected to customer terminals, an intermediate node to which a plurality of aggregation nodes are connected, a static connection node connected to the intermediate node, and a first server connected to the static connection node and connected to the service server of the TCP protocol. node, and a second server node connected to the static connection node and connected to a UDP protocol service server. At this time, the communication network capacity design device includes a first link between the aggregation node and the intermediate node, a second link between the intermediate node and the static connection node, a third link between the static connection node and the first server node, and a static connection node and the second link. Each can be classified as a fourth link between server nodes.

Thereafter, a traffic distribution model for each link may be set by reflecting the traffic distribution characteristics of the service flowing into the links (S20). Specifically, in the case of the BE service, the communication network capacity design device applies a traffic distribution model according to a Pareto distribution for the first link and a normal distribution for the second link and the third link, and in the case of a multicast streaming service, the first link A traffic distribution model according to a binomial distribution, a Poisson distribution for the second link, and a uniform distribution for the third link may be applied.

In addition, in the case of the unicast streaming service, the first link applies a binomial distribution, the second link and the third link apply a traffic distribution model according to the Poisson distribution, and in the case of the unicast D&P service, the first link has a binomial distribution, Link 2 and link 4 may apply a traffic distribution model according to Poisson distribution.

When the traffic distribution model for each link is set, the communication network capacity design device may calculate SLA (Service Level Agreement) link capacity for the links by applying the traffic distribution model (S30). However, since the SLA link capacity calculation method for each traffic distribution model has been described above, a detailed description thereof will be omitted.

The present invention is not limited by the foregoing embodiments and accompanying drawings. It will be clear to those skilled in the art that the components according to the present invention can be substituted, modified, and changed without departing from the technical spirit of the present invention.

100: communication network capacity design device 110: classification unit
120: model setting unit 130: calculation unit

Claims (9)

In the communication network capacity design method of the communication network capacity design device based on traffic modeling,
By analyzing the topology in the communication network including at least one of a first server node connected to the service server of the TCP (Transmission Control Protocol) protocol and a second server node connected to the service server of the UDP (User Datagram Protocol) protocol, the communication network classifying a plurality of links included therein;
setting a traffic distribution model for each link by reflecting the traffic distribution characteristics of the service flowing through the links; and
Calculating Service Level Agreement (SLA) link capacity for the links by applying the traffic distribution model,
The service server of the TCP protocol provides a best effort (BE) service or a unicast download and play (D&P) service, and the service server of the UDP protocol provides a multicast streaming service or a unicast streaming service,
Setting the traffic distribution model
Reflecting the traffic distribution characteristics of the BE service, unicast D&P service, multicast streaming service, or unicast streaming service, individually setting the traffic distribution model for the link for each service,
A communication network capacity design method characterized in that for at least one link among the plurality of classified links, a traffic distribution model according to any one of a Pareto distribution, a Poisson distribution, a binomial distribution, and a uniform distribution is set.
The method of claim 1, wherein the communication network
An aggregation node connected to customer terminals, an intermediate node to which a plurality of aggregation nodes are connected, a static connection node connected to the intermediate node, a first server node connected to the static connection node and connected to a service server of the TCP protocol, the A communication network capacity design method comprising a second server node connected to a static connection node and connected to a service server of a UDP protocol.
delete The method of claim 2, wherein the classifying step
A first link between the aggregation node and the intermediate node, a second link between the intermediate node and the static connection node, a third link between the static connection node and the first server node, and a fourth link between the static connection node and the second server node. A communication network capacity design method characterized in that each is classified into links.
The method of claim 4, wherein setting the traffic distribution model
In the case of a BE service, a communication network capacity design method characterized in that a traffic distribution model according to a Pareto distribution is applied to the first link and a normal distribution is applied to the second and third links.
The method of claim 4, wherein setting the traffic distribution model
In the case of a multicast streaming service, a traffic distribution model according to a binomial distribution for a first link, a Poisson distribution for a second link, and a uniform distribution for a third link is applied.
The method of claim 4, wherein setting the traffic distribution model
In the case of a unicast streaming service, a traffic distribution model according to a binomial distribution for the first link and a Poisson distribution for the second link and the third link is applied.
The method of claim 4, wherein setting the traffic distribution model
In the case of unicast D&P service, a traffic distribution model according to a binomial distribution for the first link and a Poisson distribution for the second link and the fourth link is applied.
In the traffic modeling-based communication network capacity design device,
By analyzing the topology in the communication network including at least one of a first server node connected to the service server of the TCP (Transmission Control Protocol) protocol and a second server node connected to the service server of the UDP (User Datagram Protocol) protocol, the communication network a classification unit for classifying a plurality of links included therein;
a model setting unit that sets a traffic distribution model for each link by reflecting the traffic distribution characteristics of the service flowing through the links; and
A calculation unit for calculating Service Level Agreement (SLA) link capacity for each link by applying the traffic distribution model,
The service server of the TCP protocol provides a best effort (BE) service or a unicast download and play (D&P) service, and the service server of the UDP protocol provides a multicast streaming service or a unicast streaming service,
The model setting unit
Reflecting the traffic distribution characteristics of the BE service, unicast D&P service, multicast streaming service, or unicast streaming service, individually setting the traffic distribution model for the link for each service,
A communication network capacity design device, characterized in that for at least one link among the plurality of classified links, a traffic distribution model according to any one of a Pareto distribution, a Poisson distribution, a binomial distribution, and a uniform distribution is set.

Images