JP2015011419A - Content charging ratio setting device and method by isp (internet service provider) - Google Patents

Abstract

PROBLEM TO BE SOLVED: To set a content charging ratio so that one ISP makes its revenue maximum when any one of ISPs combines a content charging for collecting a certain ratio of a content distribution fee with a conventional transit charging in an environment that there are two ISPs.SOLUTION: The invention obtains ISP information (parameter) and determines CP strategy type aggregation using a conventional transit charging, and then calculates an optimal value α* of a certain ratio α that the content distribution fee which is charged to the CP strategy type aggregation using the conventional transit charging makes its revenue maximum when the ISP-1 collects the certain ratio of the content distribution fee to the CP in an environment that the ISP-1 and ISP-2, namely two ISPs exist.

Description

  The present invention relates to a content billing ratio setting apparatus and method by an ISP (Internet Services Provider), and in particular, a content distribution of a fee distribution type in which a content provider (hereinafter referred to as “CP”) collects a content distribution fee from a user. In an environment where there are two ISPs, ISP-1 and ISP-2, in the service, ISP-1 uses content billing that collects a certain percentage of the content distribution fee for CP in combination with conventional transit billing. In this case, the present invention relates to a content billing ratio setting apparatus and method by an ISP for obtaining a content billing ratio that maximizes its own profit.

  Distribution of large-capacity rich content using moving images and images on a network (NW) is becoming widespread. The business model of content providers (CPs) has traditionally been the business model of obtaining advertising fees from advertisers, but recently, paid distribution that obtains viewing fees from users has been widely used, The spread is expected in the future. When rich content is distributed, a large amount of traffic is transferred on the NW. For example, in the case of moving picture content of HDTV (High Definition TV) quality, it has a bit rate of about 25 Mbps, but if it is 100 minutes of moving picture content, about 19 Gbytes of data are transferred. ISPs need to maintain stable quality, and the increase in equipment costs accompanying the increase in rich content distribution is a major burden. ISPs need to cover facility costs with fees obtained from users and CP, but it is difficult for ISPs to ask users for additional fees for rich content distribution in NW access services where price competition is intense. In addition, in the case of paid distribution, in addition to the content viewing fee paid to the CP, it seems that the user is also reluctant to pay the additional service fee to the ISP. However, on the other hand, CP can earn a large amount of revenue by paid distribution, and considers that there is room to return a part of the revenue as compensation for the contribution of the ISP supporting the business (for example, Non-Patent Document 1). reference).

J. Crowcroft, "Net Nuetrality: The Techinical Side of the Debate: A White Paper," ACM SIGCOMM CCR, Vol. 37, No.1, pp. 49-55, 2007.

  At present, CP is generally connected to ISP as a user of transit service and pays transit fee. However, pay-per-use system according to 95% of average data transfer rate every 5 minutes is common. It is. If the transit fee is proportional to the total amount of transferred data, even if the total amount of transferred data increases due to rich content distribution, the fee that the CP pays to the ISP also increases proportionally. In principle it is possible. In many cases, however, the rate of increase in transit charges decreases as the amount of transferred data increases, and ISPs cannot sufficiently fund the capital investment costs that increase rapidly as rich content distribution increases.

  Therefore, as a method of providing a fair share among NW's capital investment players and ISPs covering the capital investment costs, content charges that ISPs obtain from CP for each content distribution are combined with transit charges. It seems to be effective. That is, ISP obtains a part of the distribution fee that CP obtains from the user from CP, so that ISP can secure the source of equipment cost.

  In recent years, business relations between ISPs and CPs have been modeled, and appropriate pricing methods and methods for ISPs to secure profits in content distribution services have been widely studied, and content billing for ISPs' CPs is assumed. However, at present, content charging is not realized, and the above-described transit charging is used.

  If there is only one ISP, the CP has no choice but to contract with this ISP, and the ISP can always increase revenue by introducing content billing. However, when there are multiple competing ISPs, the CP may switch the transit contract to another ISP because it dislikes the burden of introducing content charging, and the introduction of content charging is not always effective for the ISP. Therefore, it is important to analyze the feasibility and effectiveness of content billing for ISP CPs. However, there are no reports on such analysis. In addition, as a billing model for the CP of the conventional ISP, only a billing model in which the CP pays the ISP for each content distribution or a billing model that is simply proportional to the total amount of transferred data is assumed. No comparison has been made with an increase rate reduction type pay-as-you-go billing generally used in transit billing, and the effect of content billing on ISP revenue has not been analyzed. There are also studies analyzing the feasibility of introducing content billing for ISPs in an environment where there are multiple ISPs, but how to determine the billing rate for content billing has not been studied.

  The present invention has been made in view of the above points. In an environment in which two ISPs exist, content charging in which any ISP collects a certain percentage of content distribution charges with respect to CP is referred to as conventional transit charging. It is an object of the present invention to provide a content billing ratio setting apparatus and method by an ISP that can set a content billing ratio so that one ISP maximizes its own revenue when used together.

According to one aspect, in a paid distribution type content distribution service in which a content provider (CP) collects a content distribution fee from a user, there are two ISPs, ISP (Internet Service Provider) -1 and ISP-2 In the content billing rate setting device by the ISP, the ISP-1 for setting the optimum content billing rate that maximizes its own revenue,
ISP information input means for acquiring information related to the ISP service;
CP transit strategy estimation means for obtaining a set by CP strategy using transit billing,
An optimal charging rate deriving unit that calculates a fixed rate α at which the content distribution fee charged for the CP strategy set using the transit charging is maximized in its own revenue;
There is provided a content billing ratio setting apparatus by an ISP characterized by comprising:

  According to one aspect, in a paid distribution type content distribution service in which a content provider (CP) collects a content distribution fee from a user, in an environment where two ISPs, ISP-1 and ISP-2, exist, ISP-1 When using content billing that collects a certain percentage α of content delivery charges for CP in combination with conventional transit billing, ISP-1 has an optimal content billing rate that maximizes its revenue. There is an effect that can be set.

It is a block diagram of the content billing ratio setting device in an embodiment of the present invention. It is a flowchart of operation | movement of the content charge ratio setting apparatus in one embodiment of this invention. It is an example of the relationship of each player in one embodiment of this invention. It is an example of the delivery route in one embodiment of this invention. It is an example of the range of the charging parameter α in each CP strategy in one embodiment of the present invention. It is a figure which shows the profit of CP (content provider) in one embodiment of this invention. It is a figure which shows the optimal value ((alpha) *) of CP ratio (alpha) using each strategy in one embodiment of this invention. It is a diagram showing a profit of each player _{(R 1, R 2, R} c) in one embodiment of the present invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 shows the configuration of a content billing ratio setting apparatus according to an embodiment of the present invention.

  The apparatus shown in the figure includes an ISP information input unit 101, a CP transit strategy estimation unit 102, an optimal charging rate derivation unit 103, and an optimal charging rate output unit 104.

ISP information input unit 101, a parameter representing the transit whether differences in communication quality due to the agreement with CP of ISP gamma, the number of users accommodated u _y of the ISP, content billing amount P, access ratio epsilon _x of the CP, each user The average number of viewing times d per month and the transit cost F of CP are entered.

  The CP transit strategy estimation unit 102 estimates a set of CPs using each transit strategy.

  The optimal charging rate deriving unit 103 calculates an optimal charging rate α that ISP-1 maximizes its own profit.

  The optimal charging ratio output unit 104 outputs the optimal value (α *) of α calculated by the optimal charging ratio deriving unit 103.

  Hereinafter, a method for optimally setting the content billing ratio in the content billing ratio setting apparatus having the above configuration will be described.

  FIG. 2 is a flowchart of the operation of the content billing ratio setting device according to the embodiment of the present invention.

Step 101) The ISP information input unit 101 obtains the results (1) to [4] shown below in advance before the processing or values (β, n _t , ξ, γ, L, u _y , P, ε _x , d, F) are acquired as parameters and stored in a memory (not shown).

  The parameters input here are as follows.

γ: Parameter indicating the difference in communication quality depending on the presence of transit contract with ISP CP;
u _y : Number of users accommodated by each ISP;
P: Content billing fee;
ε _x : Access ratio of each CP;
d: Average number of viewing times per month for each user;
F: CP transit line cost Step 102) Next, the CP transit strategy estimation unit 102 reads the above parameters from a memory (not shown). Assume that ISP-1 and ISP-2 exist, ISP-1 uses both content charging and transit charging, and ISP-2 uses only conventional transit charging.

κ0: No contract with either ISP-1 or ISP-2;
κ1: Contract only with ISP-1;
κ2: Contract only with ISP-2;
κ3: Contract with both ISP-1 and ISP-2;
As described above, a set according to strategy of each CP is derived by the processing shown in [6] and [7] described later using the above parameters, and the set according to strategy of CP is stored in a memory (not shown).

  Step 103) The optimum billing ratio deriving unit 103 reads the CP strategy-by-strategy set and parameters from a memory (not shown), and estimates the earnings that the ISP obtains monthly by the method [8] described later.

Step 104) The optimum charging ratio deriving unit 103 defines elements (φ _{x, k} , ρ _{x, k} , σ _{x, k} ) under which CP-x selects a strategy, and sets CP (A ₁ , A2) ₂ ) A ₁ ≡ {x∈κ: ρ _{x, b} ≦ ρ _{x, a} , ≦ ρ _{x, c} }, A ₂ ≡ {x∈κ: ρ _{x, c} ≦ ρ _{x, a} , ≦ ρ _{x , b} }, ISP-1's monthly revenue (R ₁ ) changes discontinuously only in the value of the charging parameter α for which the CP strategy-specific set κ _k changes, and will be described later [8]. The monthly revenue (R ₁ ) of ISP-1 when the charging parameter α is set at each element point of the boundary point set M defined in で is calculated, and α when the value is the maximum is calculated as the optimal value α Select as *.

  Step 105) The optimum billing ratio output unit 104 outputs the optimum value α *.

[1] Structure between players:
Consider the three players that constitute the content distribution service: the CP that provides the content, the user u that requests and views the content, and the ISP that distributes the content from the CP to the user.

u ₁ and u two ISP (ISP-1 and ISP-2) for accommodating the user ₂ is present, u ₁ and u ₂ are fixed. Thereafter, the total number of users u = u ₁ + u ₂ is constant, and u ₁ = ζu, u ₂ = (1−ζ) u using a parameter ζ that takes a value in the range of 0 <ζ <1. In addition, there are N CPs, each of which is an independent entity, and each independently selects a contract ISP. A set of CPs contracted with only ISP-1, only ISP-2, and both ISPs is denoted as κ ₁ , κ ₂ , and κ ₃ , respectively. A CP may choose not to contract with both ISPs, and such a set of CPs is denoted as κ ₀ .

FIG. 3 shows the relationship between each player. In the figure, CP-1 is contracted with ISP-1, CP-2 is contracted with both ISPs, and CP-N is contracted with ISP-2. Each CP-x pays the transit fee T _xy to the contract ISP-y and distributes the content to the user via the contract ISP. Both ISPs are connected to each other by a peering contract, and CP can distribute content to all users of two ISPs by signing a transit contract with any one or more ISPs. For example, as shown in FIG. 4A, CP-1 connected only to ISP-1 delivers content to ISP-1 users via ISP-1 only. -2 users will receive content via both ISP-1 and ISP-2. On the other hand, as shown in Fig. 4 (b), CP-2 connected to both ISPs is only ISP-1 for ISP-1 users and ISP-2 for ISP-2 users. Content is distributed via -2 only. It is assumed that ISP-1 and ISP-2 are connected by free peering.

[2] Amount of generated traffic Let d be the average value of the number of content distributions received by each user during one month, and for the sake of simplicity in this specification, the content preferences of all users are the same. _{Let the} viewing ratio be ε _x . However,

となる。CPがトランジット契約を締結し、配信サーバを接続しているISPの収容ユーザと、そうでないISPの収容ユーザとでは、コンテンツ配信時のスループットや安定性等の通信品質に差異が生じることが予想される。ユーザがコンテンツ配信サービスによって得られる満足度は通信品質に依存することから、ISP-yに収容された各ユーザが月間にCP-xからコンテンツ配信サービスを受ける回数の平均値ｄ_xyは通信品質に依存する。ｄ_xyは、各NWの容量や負荷状態、ISP間の接続リンクの容量や負荷状態等、様々なものに依存するが、本明細書では、マクロな傾向を分析するため、d_xyは収容先のISPのCPとのトランジット契約の有無にのみ依存するとし、接続ISPに対してはｄ_xy＝ε_xdであり、非接続ISPに対してはd_xy=γε_xdとする。但し、γは０＜γ＜１の値をとる実数パラメタである。すなわち、両方のISPを経由することによる品質劣化の度合いをγで表す。

It becomes. It is expected that there will be a difference in communication quality such as throughput and stability at the time of content distribution between the accommodated user of the ISP with which the CP has signed a transit contract and the distribution server is connected, and the accommodated user of the ISP that is not. The Since the degree of satisfaction obtained by the content distribution service depends on the communication quality, the average value d _{xy of the} number of times each user accommodated in ISP-y receives the content distribution service from CP-x per month is determined by the communication quality. Dependent. d _xy depends on various factors such as the capacity and load status of each NW, the capacity and load status of the connection link between ISPs, etc. In this specification, d _xy of the only dependent on the presence or absence of transit contract with CP of ISP, for the connection ISP is d _xy = ε _x d, for non-connection ISP and d _xy = γε _x d. However, γ is a real parameter that takes a value of 0 <γ <1. That is, γ represents the degree of quality degradation caused by going through both ISPs.

For content users of non-transit contract ISPs of CP, content is delivered from CP via the other ISP, so transit traffic occurs at the other ISP. Therefore, d _xy is obtained by the following equation.

[3]課金モデル
CPのユーザに対する課金モデルとしては、毎月の定額料金で任意の回数のコンテンツ視聴が可能な定額制と、各々の配信毎に課金する従量制が一般的である。本明細書では、従量制を想定し、CPはユーザから各々のコンテンツ配信に対して、コンテンツ種別とは無関係に同一の料金Pをコンテンツ料として得る場合を想定する。

[3] Billing model
As a billing model for CP users, a flat-rate system that allows a user to view content any number of times with a flat-rate monthly charge and a pay-per-use system that charges for each distribution are generally used. In this specification, a pay-as-you-go system is assumed, and the CP assumes a case where the same fee P is obtained as a content fee regardless of the content type for each content distribution from the user.

Now, consider a case where ISP-1 uses content billing collected from each CP-x as a content delivery cost for each delivery in addition to the transit cost _Tx1 . Specifically, the content distribution fee of αP is collected from the CP for each distribution using a charging parameter α that takes a value in the range of 0 <α <1. That is, when the content is distributed via the transit link with ISP-1, the CP obtains only the charge of (1-α) P, not P, from the user. On the other hand, it is assumed that ISP-2 does not introduce content billing and uses only transit charges as before.

  FIG. 3 also shows money flow between players. When there are three or more ISPs, an analysis in the case where a plurality of ISPs introduce content billing is a future problem and is outside the scope of the present invention.

Next, transit billing is modeled. Currently, many ISPs collect transit costs from other ISPs with a CP that has a transit contract, but generally charge a fee according to the total amount of data that flows on the transit link during the month. According to the analysis of transit costs used by 20 regional ISPs in the US in 2004, the monthly transit cost T is proportional to the data transfer rate V (Mbps) to the 0.75th power and can be approximated by T = 100V ^0.75 . For example, T = 560 USD when V = 10 Mbps, and T = 100,000 USD when V = 10 Gbps. In this specification, the power of V is generalized and expressed as β,

と表記する。

Is written.

Many ISPs use 95% of the data transfer rate every 5 minutes as V, but assume that the 95% value is three times the average transfer rate. If the average content size is L (Mbyte) and the number of days in a month is 30 days, the V value Vxy that each ISP-y applies to each CP-x is
V _xy = (3 × 8Ld _xy ) / (30 × 24 × 3600) = 1.08 × 10 ^-5 Ld _xy
It becomes. Therefore, CP-x pays ISP-y, and the monthly transit cost T _xy is

となる。ただし、n_tを

It becomes. Where n _t

と定義する。

It is defined as

[4] Cost model Next, we will model the cost of each CP and ISP. In addition to the transit costs and content distribution costs described in [3], the CP must bear the line costs to the connection point with the contracted ISP. Here, for the sake of simplicity, let's consider the line cost per month when an arbitrary CP-x connects to an arbitrary ISP-y with a fixed amount of F. In this specification, considering the cost of using a dedicated line, the NTT East (registered trademark) ATM Mega Link Service distance of 30 km, Dual class, and 100 Mbps are assumed, and the line cost F is F = 40,000 USD.

On the other hand, the ISP incurs a cost for transferring the content distributed from the contract CP. ISP-1 distributes content from the connection point with CP-x to the user for each distribution directed to its own user distributed from CP-x with x∈κ ₁ or x∈κ ₃ , also, for each delivery towards CP-x ISP-2 users distributed from the X∈kappa _1, it is necessary to distribute the content from the connection point of the CP-x to peering points between ISP-2 is there. Additionally, for each delivered towards its own user distributed from CP-x of X∈kappa _2, must be delivered from the peering point between ISP-2 to the user. In this specification, it is assumed that a certain distribution cost G is generated by each of these contents distribution.

It is difficult to accurately estimate the distribution cost G. Therefore, for example, Valancius et al. Modeled the traffic delivery cost of a relay ISP with a linear function of distance, a function that is convex upward with respect to distance, a function of rough distance classification such as metro, domestic, and international, etc. The effect of differentiating the charges according to the etc. is analyzed. In this specification, monthly transit costs between ISPs are applied to G. In 2013, the average transit price per Mbps in the United States was 1.57 USD, and peak demand for commercial VoD services was reported to be about 1.8 times the average.
G = (1.57 × 1.8 × 8Lq) / (30 × 24 × 3600) = n _s q
Give in. However, L is the average size of the content, q is the number of content distributions per month that flows on the NW of each ISP,
n _s = 8.72 × 10 ^-6 L
It is defined as

[5] Analysis of the impact of introducing content billing
Analyze the strategy for determining the transit contract between CP and each ISP, and analyze the feasibility of content billing and the impact on ISP revenue. The transit charging setting parameter β is fixed, and only the ISP-1 that uses content charging together can freely set the charging parameter α. Then, analyze the strategies for determining the transit contract between each CP and each ISP, and analyze the feasibility of content billing and the impact on ISP revenue. The transit charging setting parameter β is fixed, and only the ISP-1 which uses content charging together can freely set the charging parameter α. Then, each CP knows the set value of α and decides the ISP to contract. Therefore, CP and ISP actions consist of a first stage in which ISP-1 determines α and a second stage in which each CP determines a contracted ISP. Since it is a complete information stackelberg game with CP and ISP-1 as players, the equivalent solution is a subgame perfect hash equivalence point (SPE) and is equivalent using backward induction (backward induction) Can be derived. In other words, a perfect hash equalization solution can be obtained by first obtaining an optimal CP strategy and then obtaining an ISP-1 optimal strategy based on it.

[6] CP transit strategy Each CP's possible strategy k is not contracted with both ISPs (k = 0), contracted only with ISP-1 (k = 1), contracted only with ISP-2 (k = 2), defined as contracting with both ISPs (k = 3). If the monthly profit obtained when CP-x uses strategy k is φ _{x, k} ,

となる。但し、φ_x,0＝０であり、z_x=ε_xdと定義する。CPの行動を決める条件に関して、以下の命題が成立する。

It becomes. However, it is defined as φ _{x, 0} = 0 and z _x = ε _x d. The following propositions hold for the conditions that determine CP behavior.

Proposition 1: Conditions C ₁ , C ₂ , and C ₃ for selecting k = 1, k = 2, and k = 3 by CP-x are expressed by the following equations, respectively.

C ₁ : α <min (ρ _{x, a} , ρ _{x, b} , σ _{x, a} ) (11)
C ₂ : α> max (ρ _{x, a} , ρ _{x, c} ) (12)
C ₃ : ρ _{x, b} <α <min (ρ _{x, c} , σ _{x, b} ) (13)
However, boundary points ρ _{x, a} , ρ _{x, b} , ρ _{x, a} , σ _{x, a} , σ _{x, b} with respect to α are defined by the following equations.

［7］CPの戦略別集合κ_kの導出
当該処理は、CPトランジット戦略推定部１０２で行われる処理である。

[7] Derivation of CP strategy-specific set κ _k This process is performed by the CP transit strategy estimation unit 102.

The following proposition holds for the magnitude relationship of ρ _{x, a} , ρ _{x, b} , ρ _{x, c} .

Proposition 2. There is a pattern of 3 × 2 = 6 in the magnitude relationship between ρ _{x, a} and ρ _{x, b} and ρ _{x, c} , but ρ _{x, b} ≦ ρ _{x, a} ≦ ρ _{x, c} or ρ _{x , c} ≦ ρ _{x, a} ≦ ρ _{x, b} .

Proof: Assuming ρ _{x, a} <ρ _{x, b} <ρ _{x, c} , when ρ _{x, a} <α <ρ _{x, b} , φ _{x, 2} > φ _{x, 1} , φ _{x, 1} > φ _{Since x, 3} , φ _{x, 3} > φ _{x, 2} , a contradiction arises. Similarly, ρ _{x, a} <ρ _{x, c} ≦ <ρ _{x, b} , or ρ _{x, b} <ρ _{x, c} <ρ _{x, a} , or ρ _{x, c} <ρ _{x, b} <ρ _x, assuming _a, respectively _{ρ x, a, ρ x,} b, ρ x, contradiction for the magnitude relationship of _c occurs. Therefore, either ρ _{x, b} ≦ ρ _{x, a} ≦ ρ _{x, c} or ρ _{x, c} ≦ ρ _{x, a} ≦ ρ _{x, b} holds.

So CP sets A ₁ and A ₂ respectively
A ₁ ≡ {x∈κ: ρ _{x, b} ≦ ρ _{x, a} ≦ ρ _{x, c} }
A ₂ ≡ {x∈κ: ρ _{x, c} ≦ ρ _{x, a} ≦ ρ _{x, b} }
When defined as all CP is included in either of A ₁ or A _2.

FIG. 5 shows CP-1 monthly revenue φ ₁ obtained from equations (8) to (10) when N = 1 (ε ₁ ), d = 10, P = 1, γ = 0.7, and ξ = 0.8. , Φ ₂ , φ ₃ are plotted against φ _k , ρ _k , σ _k (N = 1, so φ _{x, k} is expressed as φ _k ), and ISP-1 charging parameter α. However, u was set to 2 × 10 ⁷ or 2 × 10 ⁶ . In the figure, α = ρ _a at the intersection of φ ₁ and φ ₂ , and similarly α = ρ _b at the intersection of φ ₁ and φ ₃ and α = ρ _c at the intersection of φ ₂ and φ _3. . FIG. 5A corresponds to the case where x∈A ₁ and FIG. 5B corresponds to the case where x∈A ₂ .

CP-x, when x∈A ₁ , if α <min (ρ _b , σ _a ), then strategy k = 1, if σ _a <α <ρ _b , then strategy k = 0, ρ _b Select strategy k = 3 if <α <min (ρ _c , σ _b ), select strategy k = 0 if σ _b <α <ρ _c , select strategy k = 2 if ρ _c <α Doing so maximizes your earnings. Therefore, when an area of the alpha CP-x selects strategy k and S _k, S _k is the region as shown in Figure 6 (a). Therefore, when x∈A ₁ , ρ _a is included in the region S ₃ and thus does not affect the behavior of CP-x.

Similarly, CP-x has a strategy k = 1 when α <min (ρ _{x, a} , σ _a ) and a strategy k = 0 when σ _a <α <ρ _a when x∈A2. If ρ _a <α, the strategy k = 2 is selected, and its own profit is maximized. In this case, S _k is an area as shown in FIG. _6B, and ρ _b and ρ _c do not affect the behavior of CP-x. Therefore, when x∈A ₂ , the strategy k = 3 is never selected. Therefore, when x∈A ₂ , the strategy k = 3 is never selected.

When x∈A ₁ and ρ _b > σ _a (ρ _b <σ _a ) and α = σ _a (α = ρ _b ), the strategy k = 0 (k = 3) and k = 1 for CP-x It is unclear which of these strategies to choose because the resulting revenues match. However, for the sake of convenience in this specification, it is assumed that k = 1 is selected. Further, it is assumed that the strategy k = 3 is selected when α = ρ _c and σ _b . Similarly, when x∈A ₂ and ρ _a > σ _a (ρ _a <σ _a ) and α = σ _a (α = ρ _b ), the strategy k = 0 (k = 2) and the strategy for CP-x Assume that the revenues obtained from k = 1 match, but for convenience, s = 1 is selected. From the above consideration, the following theorem is obtained.

Theorem 1: CP sets by strategy κ ₀ , κ ₁ , κ ₂ , κ ₃ are obtained by the following equations.

CPトランジット戦略推定部１０２は、上記のような方法により求められたCPの戦略別集合κ_kを最適課金比率導出部１０３に出力する。

The CP transit strategy estimation unit 102 outputs the CP-by-strategy set κ _k obtained by the above-described method to the optimum charging ratio deriving unit 103.

[8] ISP-1 Charging Parameter Setting Strategy The processing shown below is performed by the optimum charging ratio deriving unit 103.

Based on the prediction of the CP strategy set κ _k derived by the CP transit strategy estimation unit 102, the optimum charging ratio deriving unit 103 obtains an optimal setting strategy for the charging parameter α of ISP-1.

First, the following proposition holds regarding the revenue R ₁ that ISP-1 earns during the month.

  Proposition 3: The monthly profit R1 that ISP-1 earns is given by the following equation.

証明：ISP-1は、[3]で述べたように契約CPからトランジット費とコンテンツ費を得て、一方で[4]で述べたように配信コストGを負担する。即ちx∈κ₁のCP-xのコンテンツ配信によって、

Proof: ISP-1 obtains transit costs and content costs from the contract CP as described in [3], while it bears the distribution cost G as described in [4]. In other words, by delivering CP-x content with x∈κ ₁

の収益をISP-1は月間に得ることができる。同様に、x∈κ₃のCP-xのコンテンツ配信によって、

ISP-1 can earn a month of revenue. Similarly, by delivering CP-x content with x∈κ ₃

の収益をISPは月間で得ることができる。x∈κ₂のCP-xのコンテンツ配信からはISP-1は収入を得ることができないが、一方で、ISP-1のユーザへの配信に対しては、ISP-2とのピアリング点よりユーザまでのコンテンツ配信コストGが発生するため、−n_sz_xγu₁の収益（損失）が発生する。以上のことからISP-1が月間に得る収益R₁は、上記の式(23)で得られる。

ISPs can earn monthly revenue. ISP-1 cannot earn revenue from the content distribution of CP-x with x∈κ ₂ , but on the other hand, for the distribution to ISP-1 users, the user from the peering point with ISP-2 Content distribution cost G up to -n _s z _x γu ₁ is generated (loss). From the above, the profit R ₁ that ISP-1 obtains in a month is obtained by the above equation (23).

In ISP-1, the optimal strategy is to set α so that profit R ₁ is maximized, but R ₁ changes discontinuously only at the value of α at which the strategy-specific set κ _k of CP changes. Therefore, the following proposition holds for ISP-1's alpha setting candidates.

Proposition 4: When the set M is defined by
M≡M ₁ ∪M ₂ ∪M ₃ ∪M ₄ ∪M ₅ (24)
ISP-1 only needs to consider α∈M as a setting candidate for α. However, α boundary point sets M ₁ , M ₂ , M ₃ , M ₄ , and M ₅ are defined by the following equations.

M ₁ ≡ {ρ _{x, b} , ρ _{x, c} : x∈A ₁ }, (25)
M ₂ ≡ {σ _{x, a} : x∈A ₁ , σ _{x, a} <ρ _{x, b} }, (26)
M ₃ ≡ {σ _{x, b} ： x∈A ₁ , σ _{x, b} <ρ _{x, c} }, (27)
M ₄ ≡ {ρ _{x, a} ： x∈A ₂ }, (28)
M ₅ ≡ {σ _{x, a} ： x∈A ₂ , σ _{x, a} <ρ _{x, a} } (29)
Proof: As described in [7], for CP-x with x∈A ₁ , κ _k always changes when α = ρ _{x, b} and α = ρ _{x, c} . Moreover σ _{x, a} <ρ _x, in the case of _{b α} = _{σ x,} in _{a, also, σ x, b <ρ x} , in the case of _c α = σ _{x, κ k} is changed in _b. On the other hand, for CP-x with x∈A ₂ , κ _k always changes at α = ρ _{x, a} , and when σ _{x, a} <ρ _{x, a} κ at α = σ _{x, a k} changes. Therefore, κ _k changes only when α takes a point of each element of the set M defined by (24).

Assuming the CP behavior described in [7], since R ₁ increases monotonically with increasing α at the α boundary point where κ _k changes, any two consecutive elements of M R ₁ is constant even if α is changed in a range of m ₁ <α <m ₂ with respect to m ₁ and m ₂ . Thus R ₁ by setting the alpha = m ₂ so maximizes, the setting candidate of alpha in the range m ₁ <α ≦ m _2, it is sufficient to consider only the alpha = m _2. Therefore, ISP-1 only needs to consider M elements defined in (24) as α setting candidates.

  Therefore, the following theorem holds for the optimal behavior of ISP-1.

Theorem 2: By calculating the monthly revenue R ₁ when α is set for each element point of set M from Equation (23) and selecting α as the optimal value α * when the value is the maximum, ISP -1 can maximize R _1.

ISP-2's monthly revenue R ₂ is given by the following equation, regardless of ISP-1's billing strategy.

また、全CPの月間収益の総和をRcとすると、式(8)〜(10)から得られるφ_x,1,φ_x,2,φ_x,3を用いて、

Also, letting Rc be the sum of the monthly revenues of all CPs, using φ _{x, 1} , φ _{x, 2} , φ _{x, 3} obtained from equations (8) to (10),

より得られる。よって月間の社会的余剰（全プレイヤの利得の総和）Φは、式(23),(30),(31)より次式で得られる。

More obtained. Therefore, the monthly social surplus (sum of the gains of all players) Φ can be obtained from the equations (23), (30), (31) by the following equation.

コンテンツ課金に伴うお金の流れはISP-1とCP間で相殺されるため、Φはαとは無関係になる。

Since the flow of money associated with content billing is offset between ISP-1 and CP, Φ is independent of α.

[9] Performance evaluation conditions The performance evaluation conditions shown below are used to evaluate the effects of content billing on ISPs.

  The evaluation of the effect of content billing on ISP will be explained using numerical examples.

CP sets the fee obtained from the user for each content distribution to P = 1USD. In the case of Hikari TV using the flat-rate unlimited plan, the average number of viewing times per month is 10, and this number is considered the upper limit average value that allows users to view rich content during the month due to time constraints, etc. The monthly average content viewing count of each user is set to d = 10. In addition, the degree of decrease in the number of viewing times due to quality degradation when going through a plurality of ISPs is set to γ = 0.3. The number of CPs is N = 10, and the ratio ε _x for viewing the contents of each CP-x is the Zipf distribution of parameter 1, that is,

に設定する。ISP-1とISP-2のユーザ数の合計をu=1.0×10⁷とし、各ISPのユーザ数を与える配分パラメタをξ＝0.6に設定する。また、コンテンツの符号化レートとしてDVB-S2の40Mbps〜60Mbps程度を想定し、100分程度の映画コンテンツを考え、コンテンツの平均サイズをL=30Gbyteに設定する。

Set to. The total number of users of ISP-1 and ISP-2 is set to u = 1.0 × 10 ^7, and an allocation parameter that gives the number of users of each ISP is set to ξ = 0.6. Further, assuming that the content encoding rate is about 40 Mbps to 60 Mbps of DVB-S2, and considering movie content of about 100 minutes, the average size of content is set to L = 30 Gbyte.

[10] Contract CP number ratio and optimal value of α Figure 7 shows the optimal value α * of the content charging parameter α obtained by the method described in [8] above, γ (the presence or absence of transit contract between ISP and CP) Difference in communication quality), ξ (a real number parameter that takes a value in the range of 0 <ξ <1 with the number of users u being constant), u (number of users accommodated by each ISP), L (average size of k content) Plot against parameters.

In the same figure, the ratio Π _j of the CP contracted with each ISP-j is plotted in an equal state realized when ISP-1 sets α = α *. For the number of CPs s _k using each strategy k,
Π ₁ = (s ₁ + s ₃ ) / N,
Π ₂ = (s ₂ + s ₃ ) / N
It becomes. The results when both ISP-1 and ISP-2 use only transit billing (denoted as TR) and when only ISP-1 is used together with content billing (denoted as CC) are shown respectively.

[Pi ₁ of CC is equal to [pi ₁ of TR in the range of all parameters, ISP-1 is by setting the alpha = alpha *, it is possible to introduce the content charging while under contract number of CP. Also, [pi ₂ of the CC, the difference between [pi ₂ of TR in the entire region is small, ISP-2 can be confirmed to be unaffected upon content billing introduction of ISP-1.

As γ increases, CP benefits from contracting with only one ISP, but because u ₁ > u _{2 is} assumed, Π ₁ increases and Π ₂ decreases. In the region where γ is small, the advantage of ISP-1 over ISP-2 increases as γ increases, so α * increases. On the other hand, in the region where γ is large, α * decreases because the difference in the number of distributions obtained with either ISP-1 or ISP-2 contracts as γ increases.

Since u ₁ <u ₂ when ξ <0.5, many CPs contract with ISP-2, but when ξ> 0.5, u ₁ > u ₂ and many CPs are ISP-2 Contract. Also, as ξ increases, α-1 increases because ISP-1's competitiveness against ISP-2 increases. From the above, it can be confirmed that the feasibility of introducing content billing largely depends on the magnitude relationship of the number of users with competitors.

Since u ₁ > u ₂ is assumed, as u increases, ISP-1's competitiveness against ISP-2 increases and α * increases. Since the transit cost increases as the average content size L increases, it is necessary to reduce α in order to introduce content billing. In a region where L is large, it becomes difficult for each CP to secure profits, and the number of CPs using the strategy k = 0 (does not contract with either ISP-1 or ISP-2) increases.

[11] Revenue and social surplus of each player ISP-1 monthly revenue R ₁ , ISP-2 monthly revenue R ₂ , total CP total revenue Rc, and society Plot the target surplus Φ. As γ and ξ increase, the benefits of contracting only with a large ISP-1 for CP increase, so ISP-1's revenue increases and ISP-2's revenue decreases. In addition, as the number of ISPs decreases, the fixed cost F is halved. As a result, R _c increases as γ increases. As ξ increases, the competitiveness of ISP-1 increases, and because ISP-1 can set α to a large value, R _c decreases. The effect of γ and ξ on social surplus is small, and only the distribution of revenue among players is affected.

Since the total number of distributions increases as u increases, all of Φ, R ₁ , R ₂ , and Rc increase. The content distribution cost that ISP-1 can collect from CP cannot exceed the revenue obtained from the newly increased distribution by CP contracting with ISP-1, so the degree of increase in CP revenue accompanying u increase Is larger than ISP-1. ISP-2's revenue increase using only transit billing decreases with increasing u, so ISP-2's revenue increase is the smallest.

  As the average content size L increases, the traffic volume increases while the number of distributions is constant, so the CP revenue exceeds the increase in transit costs, so the ISP-1 revenue also decreases. On the other hand, ISP-2's revenue, which only receives transit costs, will increase. However, in a region where L is large, it is difficult for the CP to obtain profits, and as a result of the rapid increase in the CP that selects the strategy k = 0 that does not contract with both ISPs, the profits of all players rapidly decrease. Both the increase in the number of distributions and the increase in L are factors in increasing the traffic transfer cost, but introduction of content charging is effective for the increase in the number of distributions. It is necessary to cover facility and operating costs by methods other than content billing.

  It is possible to construct the operation of each component of the content billing ratio setting device as a program and install and execute it on a computer used as the content billing ratio setting device, or distribute it via a network.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made within the scope of the claims.

101 ISP Information Input Unit 102 CP Transit Strategy Estimation Unit 103 Optimal Charging Rate Deriving Unit 104 Optimal Charging Rate Output Unit

Claims (8)

In a paid distribution type content distribution service in which a content provider (CP) collects a content distribution fee from a user, in an environment where two ISPs (ISP (Internet Service Provider) -1 and ISP-2) exist, the ISP-1 Is a content billing ratio setting device by an ISP for setting an optimal content billing ratio that maximizes its own revenue,
ISP information input means to obtain information about ISP services;
CP transit strategy estimation means for obtaining a set by CP strategy using transit billing,
Content distribution fee charged for the CP strategy-specific set κ _k using the transit charging is an optimal charging ratio deriving unit that calculates a certain ratio α that maximizes its own revenue;
A content billing ratio setting apparatus by an ISP characterized by comprising: The CP transit strategy estimation means is:
When the ISP-2 uses only transit billing, the CP strategy set (κ) that only contracts with the ISP-1 under the assumption that the CP selects a transit contract ISP to maximize its revenue. ₁ ), a CP strategy set (κ ₂ ) that only contracts with ISP-2, a CP set (κ ₃ ) that contracts with both ISPs, and a CP strategy set (κ0) that does not contract with both ISPs. As information of the parameter γ indicating the difference in communication quality depending on whether or not there is a transit contract with the CP of the ISP, the number of users accommodated by each ISP u _y , the content charge P, the transit line cost F of the CP, and a power multiplier of the data transfer rate Using β
The parameter ρ _{x, a} is

And
The parameter ρ _{x, b} is

And
The parameter ρ _{x, c} is

And
The parameter σ _{x, a} is

And
The parameter σ _{x, b} is

When
The CP strategy-specific set κ _k ,

Including means derived by
The content billing ratio setting device by the ISP according to claim 1. The optimum billing ratio derivation means includes:
The ISP information includes a parameter γ indicating a difference in communication quality depending on whether or not there is a transit contract with the ISP CP, the number of users accommodated u _y of each ISP, a content charging fee P, a CP transit line cost F, a data transfer rate Using the power multiplier β,
Monthly revenue R ₁ that ISP-1 earns during the month,

2. The content billing ratio setting apparatus according to claim 1, further comprising means for estimating by the ISP. The optimum billing ratio derivation means includes:
The CP sets A ₁ and A ₂ are respectively represented as A ₁ ≡ {x∈κ: ρ _{x, b} ≦ ρ _{x, a} ≦ ρ _{x, c} }, A ₂ ≡ {x∈κ: ρ _{x, c} ≦ ρ When defined as _{x, a} ≦ ρ _{x, b} }, the monthly revenue of the ISP-1 is the revenue R ₁ obtained by the ISP-1 during the month in the value of α at which the CP strategy-specific set κ _k varies. Because it only changes discontinuously,
M ₁ ≡ {ρ _{x, b} , ρ _{x, c} : x∈A ₁ },
M ₂ ≡ {σ _{x, a} : x∈A ₁ , σ _{x, a} <ρ _{x, b} },
M ₃ ≡ {σ _{x, b} : x∈A ₁ , σ _{x, b} <ρ _{x, c} },
M ₄ ≡ {ρ _{x, a} : x∈A ₂ },
M ₅ ≡ {σ _{x, a} ： x∈A ₂ , σ _{x, a} <ρ _{x, a} }
The means for calculating the monthly profit R ₁ when α is set at each element point of the set M defined in (1), and setting α as the optimum value α * when the value is maximum is included. Content billing ratio setting device by ISP as described in 3. In a paid distribution type content distribution service in which a content provider (CP) collects content distribution charges from users, in an environment where there are two ISPs, ISP-1 and ISP-2, the ISP-1 generates its own revenue. A content billing ratio setting method by an ISP for setting an optimum content billing ratio to be maximized,
In an apparatus having ISP information input means, CP transit strategy estimation means, and optimum charging ratio derivation means,
The ISP information input means, an ISP information input step of acquiring information related to ISP services;
The CP transit strategy estimation means includes a CP transit strategy estimation step for obtaining a set by CP strategy using transit charging;
The optimum charging ratio deriving means calculates the constant charging ratio deriving step in which the content distribution fee charged for the CP strategy-based set using the transit charging is a fixed rate α that maximizes its own revenue;
Content billing ratio setting method by ISP characterized by performing In the CP transit strategy estimation step,
If the ISP-2 uses only transit billing, the CP set (κ ₁ ) that contracts only with the ISP-1 on the assumption that the CP selects a transit contract ISP so as to maximize its revenue. CP set (κ ₂ ) that only contracts with the ISP-2, CP set (κ ₃ ) that contracts with both ISPs, CP set that does not contract with both ISPs (κ0), and the ISP information Using parameter γ representing the difference in communication quality depending on whether or not there is a transit contract with CP, the number of users accommodated by each ISP u _y , content charge P, CP transit line cost F, data transfer rate power multiplier β,
The parameter ρ _{x, a} is

And
The parameter ρ _{x, b} is

And
The parameter ρ _{x, c} is

And
The parameter σ _{x, a} is

And
The parameter σ _{x, b} is

When
The set according to the CP strategy,

Derived by
The content charging ratio setting method by the ISP according to claim 5. In the optimum charging rate derivation step,
The ISP information includes a parameter γ indicating a difference in communication quality depending on whether or not there is a transit contract with the ISP CP, the number of users accommodated u _y of each ISP, a content charging fee P, a CP transit line cost F, a data transfer rate Using the power multiplier β,
Revenue R ₁ that ISP-1 earns in a month,

The content billing ratio setting method by the ISP according to claim 5, wherein In the optimum charging rate derivation step,
The CP sets A ₁ and A ₂ are respectively represented as A ₁ ≡ {x∈κ: ρ _{x, b} ≦ ρ _{x, a} ≦ ρ _{x, c} }, A ₂ ≡ {x∈κ: ρ _{x, c} ≦ ρ When defined as _{x, a} ≦ ρ _{x, b} }, the monthly revenue of the ISP-1 is obtained only when the revenue R1 obtained by the ISP-1 in the month is the value of α at which the CP strategy-specific set κ _k changes. Because it changes discontinuously,
M ₁ ≡ {ρ _{x, b} , ρ _{x, c} : x∈A ₁ },
M ₂ ≡ {σ _{x, a} : x∈A ₁ , σ _{x, a} <ρ _{x, b} },
M ₃ ≡ {σ _{x, b} : x∈A ₁ , σ _{x, b} <ρ _{x, c} },
M ₄ ≡ {ρ _{x, a} : x∈A ₂ },
M ₅ ≡ {σ _{x, a} ： x∈A ₂ , σ _{x, a} <ρ _{x, a} }
8. The monthly profit R ₁ when the α is set at each element point of the set M defined in (5) is calculated, and α when the value is maximum is set as an optimal value α *. Content billing ratio setting method by ISP.

Images