JP2011015025A - Call reception control method and device for achieving communication quality guarantee and program therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a call reception control technique for achieving quality guarantee and capable of efficient resource allocation independent of the number of session flows by turning logical packet loss in a network indicating the characteristics of a communication session flow by a token bucket model stipulated by a peak rate, an average rate and a bucket size to zero.SOLUTION: Three kinds of resource allocation methods of average allocation, peak allocation and third resource allocation are made to function (S103-S104), and when resource amounts obtained by the three kinds of resource allocation methods are settled within an equipment amount (S105:Y), or when the resource amounts obtained by the three kinds of resource allocation methods are plotted on a graph for which a band and the resource amount of a buffer are a horizontal axis and a vertical axis and there is a point at which they are settled within the range of the equipment amount on a line segment connecting each of three points (S106:Y) even when all the resource amounts obtained by the three kinds of resource allocation methods exceed the equipment amount, the connection of a new session flow is permitted (S107).

Description

  The present invention realizes quality assurance in response to a connection request from a user by a network operator that provides a quality assurance service using an IP (Internet Protocol) network or an MPLS (MultiProtocol Label Switching) network. The present invention relates to a call admission control method and apparatus for accepting a connection request and providing a service by accepting a connection request if quality assurance is feasible and rejecting the connection request if quality assurance is not feasible, and a program therefor.

  In recent years, broadband lines have been provided to users at low prices, and the Internet has been rapidly spreading. With this spread, various services have been provided via the Internet, and the communication quality has become important.

  Especially for real-time video and audio services, the data transfer quality on the network has a large effect, but since there is no mechanism to guarantee the data transfer quality on the Internet, the network is now constructed as an IP closed network and data It is considered to provide a quality assurance type service in combination with a technology that guarantees transfer quality.

  In an IP network, service models for guaranteeing data transfer quality are mainly classified into Intfserv (Integrated Services) and Diffserv (Differentiated Services) studied by IETF (Internet Engineering Task Force).

  Intfserv is a network that uses RSVP (Resource Reservation Protocol; a protocol that reserves the bandwidth to the destination on the network and secures communication quality), which is a protocol for reserving the bandwidth on the network path end-to-end. It is a model that receives a service according to the secured quality after securing the bandwidth on the route, but it is necessary to manage all information related to RSVP in real time, so it is known that it has a problem of not scaling. (See Non-Patent Document 1 and Non-Patent Document 2).

  Diffserv is a model that classifies transfer packets by marking them in the network and differentiates transfer quality for each class by packet scheduling in a network transfer node. In the Diffserv model, transfer quality at each network transfer node is ensured by a specified PHB (Per-Hop Behavior).

  The PHB corresponding to the service class defines EF (Expedited Forwarding), which is completely prioritized, and AF (Assured Forwarding), which is the minimum bandwidth guarantee, and is also being implemented in commercially available network forwarding nodes.

  Since Diffserv only differentiates the transfer quality for each network forwarding node and only regulates the control at each forwarding node, the problem of not scaling in Intfserv is solved. However, in order to provide quality-assurance-type services, the service provider must determine the quality of service provision according to the resources in the network, and the issue of having to make admission judgments regarding the use bandwidth unified within the service provision network. The problem that must be specified still remains (see Non-Patent Document 3, Non-Patent Document 4, and Non-Patent Document 5).

  Regarding the provision of service quality, ITU-T recommends parameter definitions, transfer quality class specifications, and service models for defining service quality.

  However, the means to realize the transfer quality specified in ITU-T recommendation Y.1541 by the service model in Y.1221 has not been studied, and the service provider specifies the service provision quality according to the resources in the network. The problem of having to remain still remains (see Non-Patent Document 6, Non-Patent Document 7, and Non-Patent Document 8).

  As a reception determination model related to the use band, there is a virtual buffer / trunk model that performs reception determination on input traffic to the network according to token bucket parameters. In the virtual buffer / trunk model, the network bandwidth is managed by modeling the worst condition of traffic according to the token bucket model as ON / OFF traffic and virtually allocating a proportional buffer amount to the declared bandwidth. The management server accumulates and manages the effective bandwidth in consideration of the buffer capacity of each network forwarding node based on the following concept, and determines whether or not the logical packet loss rate is zero.

<Maximum amount of traffic sent from token bucket>
Regarding the traffic defined by the token bucket, the maximum amount of traffic A (t) generated from time 0 to t is
A (t) = min {Pt, σ + ρt} (1)
(See FIG. 7). Here, min {,} indicates a minimum value in {,}, and P, ρ, and σ are parameters of the token bucket, and represent a peak rate, an average rate, and a bucket size, respectively.

At this time, assuming that the maximum time length during which traffic can be transmitted at the peak rate is T, from Equation (1), PT = σ + ρT,
T = σ / (P-ρ) (2)
Get.

<Characteristics of resource amount (b, c) used by traffic transmitted from token bucket>
As shown in FIG. 8, let us consider that one session flow sent from a token bucket is serviced in a deterministic queuing system.

  v (t) is the amount of buffer used at time t, b is the maximum value of the amount of buffer used, c is the link bandwidth allocated to the session flow, where ρ ≦ c ≦ P.

When the flow sends data at the maximum amount A (t), the amount of buffer used increases from the time 0 to T at a rate of Pc ≧ 0, and thereafter decreases at the rate of c-ρ. To become
b = v (T) = (P−c) T = σ (P−c) / (P−ρ) (3)
Is obtained (see FIG. 9).

<Virtual buffer / trunk model>
Now, consider that a session flow k (k = 1, 2,..., K) having token bucket parameters (Pk, ρk, σk) is multiplexed on a node having a buffer amount B and a link bandwidth C.

At this time, as shown in FIG. 10, it is considered that a link bandwidth ck (ρk ≦ ck ≦ Pk) is virtually allocated to each session flow k. However, the total amount of virtual resources should not exceed the link bandwidth C. That is,
c1 + c2 + ... + cK≤C (4)
And

At this time, the maximum value bk of the virtual buffer resource used by the session flow k is similar to Expression (3),
bk = σk (Pk−ck) / (Pk−ρk) (5)
Can be expressed as

Since the maximum value b of the actual buffer usage by all flows does not exceed the sum of the virtual buffer usage,
Σbk = Σσk (Pk−ck) / (Pk−ρk) ≦ B (6)
If this holds, the logical packet loss becomes zero.

  In other words, if equations (4) and (6) hold for K sets of virtual resources (bk, ck) satisfying equation (5), the logical packet loss in the link becomes zero.

The first method for determining the virtual resources to be assigned to each session flow is to ck
ck = Pk / (1 + B (Pk−ρk) / (Cσk)) (7)
(See Non-Patent Document 9).

In this method, the resulting bk value is
bk / B = ck / C (8)
There is an advantage that the ratio of the virtual bandwidth and the virtual buffer amount to the original resource amount is equal and the resource management is simplified.

As a second method of virtual resource allocation, an allocation method that minimizes the total amount of virtual buffers represented by Expression (6) is considered. When formulating this method,
min {Σσk (Pk−ck) / (Pk−ρk)} (9)
However,
Σck ≦ C, ρ ≦ ck ≦ Pk (10)
It becomes. The virtual bandwidth allocation that is the solution of Equation (9) is obtained by the following procedure.

First, a virtual band equal to ρk is assigned to all flows.
Thereafter, for the remaining bandwidth C-Σρk, for each flow, the following equation:
Tk = σk / (Pk−ρk) (11)
The peak rate transmittable time Tk calculated in (1) is newly added and assigned in order from the flow with the largest Tk value. It is known that when the virtual band ck is assigned by this method, the equation (6) is minimized (see Non-Patent Document 10).

  The admission judgment method that satisfies the condition of Equation (6) can guarantee the flow quality because the logical packet loss is zero and the maximum delay time at the node can be guaranteed by B / C. It is a simple method.

  Comparing the first method and the second method, the first method has an advantage that the resource allocation calculation is simple.

  In the second method, the maximum buffer usage calculated in the second method is guaranteed to be less than or equal to the maximum buffer usage calculated in the first method, so more sessions can be accepted when used for acceptance determination. There is an advantage that the accommodation efficiency is excellent.

  As described above, by applying the virtual buffer / trunk model, it is possible to perform acceptance determination with guaranteed quality. However, in the first method, the accommodation efficiency is problematic, and in the second method, the computational complexity and the amount of information to be managed are problems.

S. Shenker and J. Wroclawski, “General Characterization Parameters for Integrated Service Network Elements,” Internet Engineering Task Force, RFC2215, Sep. 1997. R. Braden, L. Zhang, S. Berson, S. Herzog and S. Jamin, “Resource ReSerVation Protocol (RSVP)-Version 1 Functional Specification,” Internet Engineering Task Force, RFC2205, Sep. 1997. S. Blake, D. Black, M. Carlson, E. Davies, Z. Wang and W. Weiss, “An Architecture for Differentiated Service,” Internet Engineering Task Force, RFC2475, Dec. 1998. B. Davie, A. Charny, JC \ R. Bennett, K. Benson, JY Le Boudec, W. Courtney, S. Davari, V. Firoiu and D. Stiliadis, “An Expedited Forwarding PHB (Per-Hop Behavior), ”Internet Engineering Task Force, RFC3246, Mar. 2002. J. Heinanen, F. Baker, W. Weiss and J. Wroclawski, “Assured Forwarding PHB Group,” Internet Engineering Task Force, RFC2597, Jun. 1999. “IP packet transfer and availability performance parameters,” ITU-T Recommendation Y.1540 “Network Performance Objectives for IP-Based Services,” ITU-T Recommendation Y.1541 “Traffic control and congestion control in IPbased networks,” ITU-T Recommendation Y.1221 A. Elwalid, D. Mitra and RH \ Wentworth, “A New Approach for Allocating Buffers and Bandwidth to Heterogeneous, Regulated Traffic in an ATM node, '' IEEE Journal on Selected Areas in Communications, vol. 13, no. 6, pp 1115-1127, Aug. 1995. FL Preste, Z.-L.Zhang, J. Kurose and D. Towsley, “Source Time Scale and Optimal Buffer / Bandwidth Tradeoff for Heterogenous Regulated Traffic in a Network Node,” IEEE / ACM Transactions on Networking, vol. 7, no 4, pp. 490--501, Aug. 1999.

The technology related to virtual resource allocation can be expressed as follows.
Now, with respect to the traffic parameters (Pk, ρk, σk) of the session flow k, a direction vector on a two-dimensional plane that takes a band resource amount in the X-axis direction and a buffer resource amount in the Y-axis direction,
vp (k) = (Pk, 0), va (k) = (ρk, σk) (12)
far.

At this time, the resource amount (ck, bk) used by the session flow k is an internal dividing point of two direction vectors, that is,
v (k, α) = (ck, bk) = αk · vp (k) + (1−αk) · va (k), (0 ≦ αk ≦ 1)
... (13)
When αk = 1, it corresponds to peak rate allocation (Pk, 0), and when αk = 0, it corresponds to average rate allocation (ρk, σk) (FIG. 11).

  Next, consider the case where there are multiple session flows. When the number of session flows = 2, the traffic parameters of session flows k and l are (Pk, ρk, σk) and (Pl, ρl, σl), respectively. However, σk / (Pk−ρk) ≧ σl / (Pl−ρl).

At this time, the total amount of resources allocated to the session flows k and l is a combination of the internal dividing points of the direction vectors represented by Expression (13),
αk · vp (k) + (1−αk) · va (k) + αl · vp (l) + (1−αl) · va (l), (0 ≦ αk, αl ≦ 1)
・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ （14）
Indicated by However, vp (k) = (Pk, 0), va (k) = (ρk, σk), vp (l) = (Pl, 0), va (l) = (ρl, σl).

  The range that the total amount of resources according to Equation (14) can take is given by the inner region of the quadrilateral R1-R2-R4-R3 in FIG. Here, R1 is an average rate allocation for both session flows k and l, R2 is a peak allocation to session flow k, an average allocation to session flow l, R3 is an average allocation to session flow k, a peak allocation to session flow l, and R4 is a session This corresponds to the case where peak assignment is performed for both flows k and l, line segment R1-R2 is average assignment to session flow l, line segment R2-R4 is peak assignment to session flow k, and line segment R1-R3 is session flow k. The average allocation and the line segments R3-R4 correspond to the case where the session allocation is peak allocation.

  From the viewpoint of efficient resource allocation, the allocation on the line segment R1-R2 or R2-R4 is a smaller allocation for both the band resource and the buffer resource, and the allocation of the line segment R1-R3 or R3-R4 or the quadrilateral This is preferable to the allocation of the internal area.

In general, when the number of session flows is larger, as shown in FIG. 13, the total value of resource allocation can be represented by a two-dimensional polygonal area. Points when average allocation is made for all session flows,
RA = va (1) + va (2) +... Va (n)
And points when peak assignment is performed for all session flows RP = vp (1) + vp (2) +... + Vp (n)
The lower limit value of the curve connecting the lines can be created as follows.

(Procedure 1) The connection session flows are rearranged in descending order of Tk in equation (11).
(Procedure 2) For session flow 1 with the largest value of Tk, va (1) is replaced with vp (1). RA '= vp (1) + va (2) + ... + va (n)
Consider the line connecting RA and RA. This line segment corresponds to the total resource allocation amount when only the resource allocation for the session flow 1 is changed within the range of the equation (13).

(Procedure 3) Next, the same operation is performed for the session flow 2 having the largest Tk value. Thereafter, the process is repeated in order up to session flow n, and a curve reaching point RP is obtained.

  Since Tk in Expression (11) corresponds to the slope of the line segment created in (Procedure 2) and (Procedure 3), resource allocation below this cannot be realized and becomes the lower limit. In addition, the slope of the curve is non-decreasing monotonously and becomes a downwardly convex curve.

Similarly, the upper limit of the curve connecting RA and RP is
(Procedure 1 ′) The connection session flows are rearranged in descending order of Tk in equation (11).
(Procedure 2 ′) For session flow n with the smallest Tk value, va (n) is replaced with vp (n). RA ″ = va (1) + va (2) +... + Vp (n)
Consider the line connecting RA and RA. This line segment corresponds to the total resource allocation amount when only the resource allocation for session flow n is changed within the range of equation (13).
(Procedure 3 ') The same operation is performed for the session flow n-1 having the next smallest Tk value. Thereafter, the process is repeated until session flow 1 to obtain a curve reaching point RP.

  Similar to the lower limit curve, resource allocation exceeding (procedure 1 ') to (procedure 3') is not feasible and is an upper limit. In addition, the slope of the curve is monotonically non-increasing and becomes an upwardly convex curve. Therefore, the region surrounded by both curves is a convex polygon.

  A portion where this region intersects with the rectangular region OBCD defined by the equations (4) and (6) corresponds to an actual value of resource allocation that can reduce the logical packet loss to zero. In particular, a downwardly convex curve connecting RA and RP corresponds to efficient resource allocation.

In the first method in the prior art, the value of α is α = (Cσ−Bρ) / (Cσ + B (P−ρ) so that the gradient of the vector (ck, bk) indicating the resource amount matches C / B. )) ・ ・ ・ ・ (15)
(See FIG. 14).

  In this first method, α varies depending on the traffic parameter of the session flow. However, the calculation method is uniquely given by Equation (14), and is not affected by other session flows. The ratio of all the buffer resources and the ratio of the allocated band resources to the all band resources are equal, and there is an advantage that only one of the resource allocations needs to be managed. However, the first method has a problem that the resource allocation method is not always efficient.

  In FIG. 15, the total amount of resource allocation to two session flows according to the first method is indicated by S1, but the ratio of the allocated buffer resources to the total buffer resources and the allocated bandwidth resources There is a more efficient resource allocation to a two-session flow such that the total value of resource allocations with equal ratios to bandwidth resources is S2.

  Another problem of the first method is that, as shown in FIG. 16, resource allocation according to equation (7) cannot be executed for a session flow that satisfies (σk / ρk <B / C). is there. In this case, for example, as shown in FIG. 16, buffer resources are consumed more than necessary, such as (ρk, ρk × B / C).

  The second method in the prior art corresponds to obtaining a point RQ on the side CD by tracking a downward convex curve from the point RA in FIG. 13 to the point RP. Since the Y-axis coordinate of the point RR is equal to or less than B and the intersection of the polygonal area and the quadrangle OBDC is equivalent, the second method in the prior art is one of the most suitable virtual resource allocation methods. One can say.

  However, in the second method, it is necessary to always manage information on all points constituting the downwardly convex curve RA-RP. In this regard, the second method has the following problems.

(1) The curve RA-RP is composed of a number of points equal to the number of connected session flows, and increases with the number of session flows.

(2) To construct a downwardly convex curve, it is necessary to rearrange the values so that the Tk values in Equation (11) are in descending order, and the order must be changed each time a new session flow is connected. There is.

(3) Since the positions of all points change due to connection termination or addition of a new session flow, recalculation is required.
The above problems have led to the problem of increasing the processing time in acceptance determinations that require real-time performance.

  Therefore, the present invention is a network that represents the characteristics of a communication session flow with a token bucket model defined by a peak rate, an average rate, and a bucket size, and logical packet loss is made zero by an exclusive resource allocation method for each session flow. The call admission control method and apparatus capable of efficiently allocating resources without requiring a certain amount of management information and processing amount, without depending on the number of connection session flows, and the quality assurance Is to provide a program.

  In the present invention, one point in the downward convex polygonal region in FIG. 13 (hereinafter, this point is referred to as RR) is always managed together with the points RA and RP, and the curve RA-RR-RP intersects the quadrangle OBDC. Whether or not resource allocation is possible is determined based on whether or not the resource is allocated. Since the curve RA-RR-RP obtained by this method always exists in the convex polygon region due to the nature of the convex polygon, all points on the curve can be candidates for the total resource allocation.

  Since this curve is always above the downwardly convex curve obtained by the conventional second method, the second method is better in terms of resource allocation efficiency. Regardless of the number of flows, the number is constant at 3 points and can be updated using only information on a new connection session or an end session. Therefore, this method overcomes the problems of the second method.

  In the present invention, it is possible to cope with a session flow in which the resource allocation according to Expression (7) in the first method in the prior art cannot be performed without any problem, and the efficiency is superior to the first method. I can expect that.

Hereinafter, a more specific configuration of the present invention will be described.
a) A call admission control method according to the present invention is a call using a computer that allocates two types of resources, a link bandwidth and a buffer amount, individually to a connection session flow, and accepts a new connection session flow within a range that fits in the facility amount. In the admission control method, a procedure for simultaneously functioning three types of resource allocation methods including the third resource allocation method in addition to the two types of resource allocation methods of the average rate resource allocation method and the peak rate resource allocation method by computer control And the amount of resources obtained by computer control when the total amount of resources obtained by any one of the three types of resource allocation methods is within the amount of equipment, or by each of the three types of resource allocation methods Even if the sum of all exceeds the amount of equipment, 2 of the link bandwidth and buffer amount When the total of the resource amounts obtained by each of the three resource allocation methods is plotted on a graph with the horizontal and vertical axes representing the type of resource, on the line segment connecting each of the three plotted points And a procedure for permitting connection of a new session flow when there is a point that falls within the range of the facility amount.

b) In addition, in the above a), the third resource allocation method uses a session flow k (k = 1, 2,..., K) having token bucket parameters of a peak rate Pk, an average rate ρk, and a bucket size σk. ) Is multiplexed on a node having a buffer amount B and a link bandwidth C, the peak rate resource allocation method is used for a session flow in which σk / (Pk−ρk) exceeds a certain threshold X, and σk / (Pk− It is a resource allocation method that performs an average rate resource allocation method for a session flow in which ρk) is less than the certain threshold value X.

  In this way, quality assurance can be realized by eliminating logical packet loss by the exclusive resource allocation method, and it does not depend on the number of connected session flows, only a certain amount of management information and processing amount are required. Thus, efficient resource allocation becomes possible.

c) In the above a) or b), in the third resource allocation method, when there is a situation where the resource allocation by the third resource allocation method exceeds the equipment amount, or when the link bandwidth resource exceeds When the resource allocation method is changed to the resource allocation method that gives fewer link bandwidth resources and more buffer resources, and conversely, when there is a situation where the buffer resource allocation exceeds the facility amount, the link bandwidth resource Is changed to a resource allocation method that gives more buffer resources and less buffer resources.
In this case, the third resource allocation method can reduce the probability of resource allocation exceeding the amount of equipment, increase the number of receivable session flows, and increase accommodation efficiency.

d) In c) above, in the third resource allocation method, when there is a situation where the resource allocation by the third resource allocation method exceeds the facility amount for a certain threshold value X, the band resource is exceeded. The fixed threshold value X is greatly changed by a fixed value dx, and when the buffer resource exceeds, the fixed threshold value X is changed by a fixed value dx.
By adopting such a third resource allocation method, it is possible to increase the number of session flows that can be accepted by reducing the probability that the resource allocation exceeds the facility amount, and to increase the accommodation efficiency.

e) Also, in the above c) or d), the change is performed when the number of connections in the session flow becomes zero.
In this way, by changing the resource allocation method or changing a certain threshold value when the number of connections in the session flow becomes 0, resources allocated to individual session flows by the third resource allocation method The amount can be prevented from changing during connection of the session flow.

f) A program according to the present invention is a program for causing a computer to execute each procedure in the call admission control method of any one of the above-mentioned a) to e).

g) A call admission control device according to the present invention is a call using a computer that allocates two types of resources, a link bandwidth and a buffer amount, individually to a connection session flow, and accepts a new connection session flow within the scope of the facility amount. In the reception control apparatus, means for simultaneously functioning three types of resource allocation methods including the third resource allocation method in addition to the two types of resource allocation methods of the average rate resource allocation method and the peak rate resource allocation method by computer control And the amount of resources obtained by computer control when the total amount of resources obtained by any one of the three types of resource allocation methods is within the amount of equipment, or by each of the three types of resource allocation methods Even if the sum of all exceeds the amount of equipment, the link bandwidth and buffer amount When plotting the sum of the resource amounts obtained by each of the three resource allocation methods on a graph with two types of resource amounts on the horizontal and vertical axes, a line segment connecting each of the three plotted points And a means for permitting connection of a new session flow when there is a point within the range of the equipment amount.

h) A communication router or network control device according to the present invention is characterized in that it incorporates the call admission control device of g) above, and the network according to the present invention uses such a communication router or network control device. It is a network characterized by

  According to the present invention, in a network that represents the characteristics of a communication session flow with a token bucket model defined by a peak rate, an average rate, and a bucket size, logical packet loss is reduced to zero by an exclusive resource allocation method for each session flow. The call admission control method and apparatus capable of efficiently allocating resources without requiring a certain amount of management information and processing amount, without depending on the number of connection session flows, and the quality assurance Can be realized.

(Example 1) which is the flowchart for demonstrating the call admission control method which concerns on this invention. The sum of the resource amounts by the peak rate resource allocation method, the average rate resource allocation method, and the third resource allocation method are all outside the rectangle OBCD, but the line segment (P ′, 0) − (F ′, G ′) It is a figure for demonstrating the example in case where cross | intersects quadrangular OBCD. Although the sum of the resource amounts by the peak rate resource allocation method, the average rate resource allocation method, and the third resource allocation method are all outside the rectangle OBCD, the line segment (F ′, G ′) − (ρ ′, σ ′ ) Is a diagram for explaining an example in the case of intersecting with a rectangular OBCD. (Example 3) which is a flowchart for demonstrating the call admission control method which concerns on this invention. It is a figure for demonstrating Example 6 which concerns on this invention (example which incorporated the call admission control method concerning this invention in the communication router). It is a figure for demonstrating Example 7 which concerns on this invention (The example which incorporated the call admission control method which concerns on this invention in the network control apparatus). It is a figure which shows time passage of the maximum amount of the data transferred from the session flow prescribed | regulated with a token bucket. It is the figure which simulated a mode that the data transferred from one session flow prescribed | regulated with a token bucket added to a virtual buffer / trunk system. It is a figure which shows the time passage of the virtual buffer usage amount when the session flow prescribed | regulated with a token bucket is added to the virtual buffer / trunk system. It is a figure imitating a mode that a plurality of session flows prescribed by a token bucket are added to a virtual buffer / trunk system. It is a figure for demonstrating the resource amount which a session flow uses using the two-dimensional plane which takes a bandwidth resource amount in a X-axis direction and takes a buffer resource amount in a Y-axis direction (the 1). It is a figure for demonstrating the resource amount which a session flow uses using the two-dimensional plane which takes a bandwidth resource amount in a X-axis direction and a buffer resource amount in a Y-axis direction (the 2; range which the total of resource amount can take Showing). It is a figure for demonstrating the resource amount which a session flow uses using the two-dimensional plane which takes a bandwidth resource amount in a X-axis direction, and takes a buffer resource amount in a Y-axis direction (the case 3; there are many session flows) . It is a figure for demonstrating the resource amount which a session flow uses using the two-dimensional plane which takes a bandwidth resource amount in the X-axis direction and a buffer resource amount in the Y-axis direction (part 4; a vector indicating the resource amount is (link If it matches the bandwidth C / buffer amount B)). It is a figure for demonstrating the resource amount which a session flow uses using the two-dimensional plane which takes a bandwidth resource amount in a X-axis direction and takes a buffer resource amount in a Y-axis direction (the 5). It is a figure for demonstrating the resource amount which a session flow uses using the two-dimensional plane which takes a bandwidth resource amount in a X-axis direction, and takes a buffer resource amount in a Y-axis direction (the 6).

<Overview>
In addition to simple allocation methods such as a peak rate resource allocation method and an average rate resource allocation method, the present invention performs a reception determination only by newly calculating a third resource allocation method. By implementing a method that can be implemented even in a large-scale network by avoiding the situation where the amount of information and calculation required for acceptance determination increases with the number of session flows, such as is there.

  In addition, since the area where resource allocation can be performed is a convex polygon, the property that all points on the line segment connecting three points corresponding to the three types of resource allocation methods are within the area where execution is possible is used. Judgment that more session flows can be accepted by determining that there are cases where efficient resource allocation is possible and can be executed even if all three resource allocations exceed the amount of equipment. The method was made feasible.

  Furthermore, in the third resource allocation method, a method is proposed in which the accommodation efficiency can be further improved by periodically reviewing (updating) the allocation method.

<Explanation of symbols>
Symbols used in the description of the embodiments of the present invention are defined as follows.
Assume that the total value of the peak rates of the connection session flow is P, the total value of the average rates is ρ, the total value of the bucket size is σ, and the total number of connection session flows is N. Assume that the total resource amount when peak allocation is performed is RP = (P, 0), and the total resource amount when average allocation is performed is RA = (ρ, σ).

  A third resource allocation method other than the peak rate resource allocation method and the average rate resource allocation method is determined for the traffic parameter (Pk, ρk, σk) of the session flow k, and a bandwidth resource determination method based on the allocation method is defined as a function f. (Pk, ρk, σk) and the buffer resource determination method is a function g (Pk, ρk, σk).

  Assume that the total amount of resources when the third resource allocation method is performed is RR = (F, G). The value when the total resource amount RR when the third resource allocation method is performed exceeds the equipment amount is defined as (C *, B *).

  In order to perform the acceptance determination, R ′, ρ ′, σ ′, F ′, and G ′ are used as symbols representing the total value of the resource amounts of the already connected session flow and the session flow that requests new acceptance.

  As an installation amount of the network node, an upper limit value of the bandwidth resource amount is C and an upper limit value of the buffer resource amount is B.

<Example 1>
FIG. 1 is a flowchart for explaining an embodiment of a call admission control method according to the present invention.
(Step S101)
Assuming that there is no connection session flow as an initial condition, P, ρ, σ, F, and G are all set to zero.

(Step S102)
It is determined whether a session flow k requesting a new reception having the traffic parameters (Pk, ρk, σk) has arrived or the connection has ended. If a session flow k requesting a new reception has arrived, the process proceeds to step S103 to connect. If completed, the process moves to step S109.

(Step S103)
If a new session flow k arrives at step S102, the traffic parameters (Pk, ρk, σk) are extracted.

(Step S104)
The total amount of resources required when the session flow k is connected is calculated as follows.
P ′ = P + Pk, ρ ′ = ρ + ρk, σ ′ = σ + σk, F ′ = F + f (Pk, ρk, σk), G ′ = G + g (Pk, ρk, σk)

(Step S105)
Any of the sum (P ′, 0), (ρ ′, σ ′), and (F ′, G ′) of the resource amounts by the peak rate resource allocation method, average rate resource allocation method, and third resource allocation method is a quadrangle Inside the OBCD, ie
P ′ ≦ C
ρ ′ ≦ C and σ ′ ≦ B
F ′ ≦ C and G ′ ≦ B
If any of the above holds (step S105: Y), it is determined that the resource allocation including the session flow k within the equipment amount (C, B) is within the equipment amount range, and the process proceeds to step S107. If all of the above conditions are not satisfied (step S105: N), the process proceeds to step S106.

(Step S106)
The sum of the resource amounts (P ′, 0), (ρ ′, σ ′), (F ′, G ′) of the peak rate resource allocation method, average rate resource allocation method, and third resource allocation method are all rectangular OBCD. Even when the line segment (P ′, 0) − (F ′, G ′) intersects the quadrangle OBCD (see FIG. 2), or the line segment (F ′, G ′). -When (ρ ', σ') intersects with the quadrangle OBCD (see Fig. 3), the resource allocation including the session flow k within the range of the equipment amount (C, B) falls within the equipment amount range. Therefore, the process proceeds to step S107. If the above condition is not satisfied, the process proceeds to step S108.

(Step S107)
It is determined that it is possible to allocate resources that fall within the capacity range for the already connected session and session k, accept connection of session flow k, and the peak rate resource allocation method, average rate resource allocation method, third The sum of the resource amounts by the resource allocation method is updated so that P = P ′, ρ = ρ ′, σ = σ ′, F = F ′, and G = G ′.

(Step S108)
For the already-connected session flow and session flow k, it is determined that resource allocation that falls within the range of the facility amount is impossible, and connection of session flow k is rejected.

(Step S109)
When a termination request for an already connected session flow j having a traffic parameter (Pj, ρj, σj) arrives, the parameter Pj, ρj, σj is extracted.

(Step S110)
The session flow j connection is terminated. The sum of the resource amounts by the peak rate resource allocation method, the average rate resource allocation method, and the third resource allocation method is updated, and P = P−Pj, ρ = ρ−ρj, σ = σ−σj, F = F−f (Pj, ρj, σj), G = G-g (Pj, ρj, σj).

<Example 2>
In the present embodiment, in particular, the following equation is used as the third resource allocation method in the first embodiment.
f (Pk, ρk, σk) = ρk if σk / (Pk−ρk) ≦ X
= Pk if σk / (Pk−ρk)> X

g (Pk, ρk, σk) = σk if σk / (Pk−ρk) ≦ X
= 0 if σk / (Pk−ρk)> X

  This is equivalent to a method in which peak rate resource allocation is performed for session flows in which the value of σk / (Pk−ρk) exceeds a certain threshold X, and average rate resource allocation is performed for session flows less than X. This is an example corresponding to one point on the downwardly convex curve in FIG. Other implementation methods are the same as those in the first embodiment.

<Example 3>
In this embodiment, in order to improve the efficiency of the third resource allocation method in the first embodiment, the functions f and g shown in the second embodiment are periodically updated. FIG. 4 is a flowchart for explaining the present embodiment.

(Step S201)
As an initial condition, it is assumed that there is no connection session flow, and P, ρ, σ, F, G, N, C *, and B * are all 0.

(Step S202)
It is determined whether a session flow k requesting a new reception having the traffic parameters (Pk, ρk, σk) has arrived or the connection has ended. If a session flow k requesting a new reception has arrived, the process proceeds to step S203 to connect. If completed, the process moves to step S211.

(Step S203)
If a new session flow k arrives at step S202, the traffic parameters (Pk, ρk, σk) are extracted.

(Step S204)
The total amount of resources required when the session flow k is connected is calculated as follows.
P ′ = P + Pk, ρ ′ = ρ + ρk, σ ′ = σ + σk, F ′ = F + f (Pk, ρk, σk), G ′ = G + g (Pk, ρk, σk)

(Step S205)
Any of the sum (P ′, 0), (ρ ′, σ ′), and (F ′, G ′) of the resource amounts by the peak rate resource allocation method, average rate resource allocation method, and third resource allocation method is a quadrangle Inside the OBCD, ie
P ′ ≦ C
ρ ′ ≦ C and σ ′ ≦ B
F ′ ≦ C and G ′ ≦ B
If any of the above holds, it is determined that the resource allocation including the session flow k within the equipment amount (C, B) is within the equipment amount range, and the process proceeds to step S209. If all of the above conditions are not satisfied, the process proceeds to step S206.

(Step S206, Step S207)
When the total amount (F ′, G ′) of the resource amount by the third resource allocation method goes out of the quadrangle OBCD, the value of (C *, B *) is not updated from the initial value (step S206: Y) and updating by (F ′, G ′) (step SS207). If the value of (C *, B *) has already been updated (step S206: N), the process proceeds to step S208 without further updating.

(Step S208)
The sum of the resource amounts (P ′, 0), (ρ ′, σ ′), (F ′, G ′) of the peak rate resource allocation method, average rate resource allocation method, and third resource allocation method are all rectangular OBCD. Line segment (P ′, 0) − (F ′, G ′) intersects the quadrangle OBCD (see FIG. 2) or line segment (F ′, G ′) − If (ρ ′, σ ′) intersects the quadrangle OBCD (see FIG. 3), it can be determined that the resource allocation including session k within the range of the equipment amount (C, B) falls within the equipment amount range. Therefore, go to step 6. If the above condition is not satisfied, the process proceeds to step S209.

(Step S209)
For the already connected session flow and session flow k, it is determined that the resource allocation that falls within the range of the facility amount is possible, and the connection of session flow k is accepted. Update the sum of the resource amounts by the peak rate resource allocation method, average rate resource allocation method, and third resource allocation method, and P = P ′, ρ = ρ ′, σ = σ ′, F = F ′, G = G ', N = N + 1.

(Step S210)
For the already-connected session flow and session flow k, it is determined that resource allocation that falls within the range of the facility amount is impossible, and connection of session flow k is rejected.

(Step S211)
When a termination request for an already connected session flow j having a traffic parameter (Pj, ρj, σj) arrives, the parameter Pj, ρj, σj is extracted.

(Step S212)
The session flow j connection is terminated. The sum of the resource amounts by the peak rate resource allocation method, the average rate resource allocation method, and the third resource allocation method is updated, and P = P−Pj, ρ = ρ−ρj, σ = σ−σj, F = F−f (Pj, ρj, σj), G = G−g (Pj, ρj, σj), and N = N−1.

(Step S213)
If N = 0, the connection session flow does not exist as in the initial condition. Therefore, if the value of (C *, B *) is updated (step S213: Y), the third resource allocation method is executed. It is determined that there is room for improvement, and the functions f (.) And g (.) Are updated in step S214.

  As shown in FIG. 2, if (B * / C *> B / C), it is determined that the third resource allocation method allocates more buffer resources, the buffer resources are allocated less, and the bandwidth Consider updating the functions f (.) And g (.) So that the allocation of resources increases. On the other hand, as shown in FIG. 3, if (B * / C * <B / C), the third resource allocation method determines that the allocation of bandwidth resources is large, and the allocation of bandwidth resources is small. Also, consider updating the functions f (.) And g (.) So that the allocation of buffer resources increases.

  This update (change) is executed when the number of connections in the session flow becomes zero. Thereby, it is possible to prevent the amount of resources allocated to individual session flows by the third resource allocation method from changing during connection of session flows.

  If (B * / C * = B / C), the third resource allocation method determines that the allocation is balanced between the bandwidth resource and the buffer resource, and in particular, functions f (.), G We will not consider updating (.).

<Example 4>
In the present embodiment, in particular, the following formula is used as the third resource allocation method in the third embodiment.
f (Pk, ρk, σk) = Pk if σk / (Pk−ρk) ≦ X
= Ρk if σk / (Pk−ρk)> X
g (Pk, ρk, σk) = 0 if σk / (Pk−ρk) ≦ X
= Σk if σk / (Pk−ρk)> X

  This is equivalent to a method in which peak rate resource allocation is performed for session flows in which the value of σk / (Pk−ρk) exceeds a certain threshold X, and average rate resource allocation is performed for session flows less than X. This is an example corresponding to one point on the downwardly convex curve in FIG. Other implementation methods are the same as those in the third embodiment.

<Example 5>
In the present embodiment, in the method shown in the fourth embodiment, the following method is particularly used as a method of updating the functions f (.) And g (.).
If B * / C *> B / C, then X = X-dx
If B * / C * <B / C, then X = X + dx
That is, the threshold value X used in the fourth embodiment is changed by a constant value dx according to the magnitude relationship between B * / C * and B / C.
For example, if (B * / C *> B / C), the threshold value X is decreased by dx, and adjustment is made so that bandwidth resource allocation increases by increasing the session flow for peak allocation, If B * / C * <B / C), the threshold value X is increased by dx, and adjustment is made so that the allocation of buffer resources is increased by increasing the number of session flows to be averaged. Other implementation methods are the same as those in the fourth embodiment.

  This change (adjustment) of the certain threshold value is executed when the number of session flow connections becomes zero. Thereby, it is possible to prevent the amount of resources allocated to individual session flows by the third resource allocation method from changing during connection of session flows.

  Note that the call admission control method using the computer according to the present invention (especially, refer to the flowcharts of FIGS. 1 and 4), as will be described later (see Examples 6 and 7), network devices such as communication routers and network control devices are used. This is realized by reading and executing a program corresponding to the processing by hardware such as a CPU and a memory of a computer constituting the computer. Needless to say, a program corresponding to the processing can be distributed to the market via a network such as the Internet or a recording medium such as a CD-ROM, DVD, or FD.

<Example 6>
The present embodiment is an embodiment in which the call admission control method according to the present invention is realized on the network shown in FIG. 5 (when the communication router has the function of the call admission control device according to the present invention).

(Step (a))
A connection request for a new session flow is notified from the user terminal device (TE) 11 to the network control device 12 together with the traffic parameter information. The UNI 13 is a user network interface device.

(Step (b))
The network control device 12 notifies the traffic parameters of the new session flow to the communication routers (R) 14 and 15 on the communication path of the new session.

(Step (c))
Each communication router (R) 14 and 15 accepts a new session flow. Thereby, it is determined using any of the methods in the first to sixth embodiments whether or not resources can be allocated within the range of the equipment amount possessed by the communication router (R) 14, 15,. Notify the control device 12.

(Step (d))
When the network session control device 12 has received permission to accept a new session flow from all the communication routers (R) 14, 15... On the communication path of the new session flow, the network control device 12 grants connection permission for the new session flow to the user terminal device. (TE) 11 and communication router (R) 14, 15... Are notified (connection of new session flow).

  If any communication router (R) is notified of the inability to accept, a new session flow is not accepted, so the user terminal device (TE) 11 is notified of connection refusal, and the communication routers (R) 14 and 15 are notified. Notifies the end of communication of the new session flow (to step (g)).

(Step (e))
When ending the session flow during communication, the user terminal device (TE) 18 notifies the network control device 12 of the end of the session.

(Step (f))
The network control device 12 notifies the communication router (R) 16 on the communication path of the session of the end of communication of the session.

(Step (g))
The communication router (R) 16 deletes information regarding the session flow for which communication has ended.

<Example 7>
The present embodiment is an embodiment in which the call admission control method according to the present invention is realized on the network shown in FIG. 6 (when the network control device has the function of the call admission control device according to the present invention).

(Step (a))
A connection request for a new session flow is notified from the user terminal device (TE) 11 to the network control device 12 together with the traffic parameter information. The UNI 13 is a user network interface device.

(Step (b))
The network control device 12 receives the new session flow at the communication routers (R) 14, 15, and 16 on the communication path of the new session flow, and thereby the facility amount of each communication router (R) 14, 15, and 16 is increased. Whether or not resource allocation within the range is possible is determined by using any one of the methods in the first to sixth embodiments, and whether or not reception is possible is determined.

(Step (c))
When the network control device 12 determines that all the communication routers (R) on the communication path of the new session flow can allocate resources within the facility amount, the network control device 12 grants connection permission for the new session flow to the user terminal device ( (TE) (new session flow connection).

  If any communication router (R) determines that resource allocation within the range of the facility amount is impossible, a new session flow is not accepted, and a connection rejection is notified to the user terminal device (TE) 11.

(Step (d))
When ending the session flow during communication, the user terminal device (TE) 18 notifies the network control device 12 of the end of the session flow.

(Step (e))
The network control device 12 deletes information related to the session flow.

  According to the above-described embodiment, quality assurance can be realized by reducing logical packet loss to zero, and efficient resource allocation can be performed only by requiring a certain amount of management information and processing amount without depending on the number of connected session flows. It becomes possible.

11: TE (user terminal equipment)
12: Network control device 13: UNI (user network interface device)
A (t): Maximum amount of traffic P: Peak rate T: Maximum time length at which traffic can be transmitted at the peak rate σ: Bucket size v (t): Used buffer amount at time t b: Maximum used buffer amount c: Link bandwidth allocated for session flow B: Buffer amount C: Link bandwidth ρ: Average rate

Claims (10)

In a call admission control method using a computer that allocates two types of resources, a link bandwidth and a buffer amount, individually to a connection session flow, and accepts a new connection session flow within a range that fits in the facility amount,
A procedure for simultaneously operating three types of resource allocation methods including a third resource allocation method in addition to two types of resource allocation methods of average rate resource allocation method and peak rate resource allocation method by computer control;
When the total amount of resources obtained by any of the three types of resource allocation methods is within the amount of equipment under computer control, or the total amount of resources obtained by each of the three types of resource allocation methods Even if all of them exceed the installed capacity, the resources obtained by each of the three resource allocation methods on the graph with the horizontal axis and the vertical axis representing the two types of resource amounts of the link bandwidth and the buffer amount And a procedure for permitting connection of a new session flow when there is a point that is within the range of the equipment amount on a line segment connecting each of the three plotted points when the total amount is plotted. A featured call admission control method. In the call admission control method according to claim 1,
In the third resource allocation method, a session flow k (k = 1, 2,..., K) having a token bucket parameter with a peak rate Pk, an average rate ρk, and a bucket size σk is buffer amount B, link bandwidth C Is multiplexed to a node having σk / (Pk−ρk) exceeding a certain threshold value X, the peak rate resource allocation method is used for the session flow, and σk / (Pk−ρk) is less than the certain threshold value X. A call admission control method, which is a resource allocation method for performing an average rate resource allocation method for a session flow. In the call admission control method according to claim 1 or 2,
In the third resource allocation method, when there is a situation where the resource allocation by the third resource allocation method exceeds the equipment amount, or when the link bandwidth resource exceeds, the resource allocation method uses fewer link bandwidth resources. Change to the resource allocation method that gives more buffer resources, and conversely, if there is a situation where the buffer resource allocation exceeds the facility capacity, the resource allocation method is resource allocation that gives more link bandwidth resources and less buffer resources. A call admission control method characterized by changing to a law. In the call admission control method according to claim 3,
In the third resource allocation method, when there is a situation where the resource allocation by the third resource allocation method exceeds the facility amount for the certain threshold value X, when the bandwidth resource exceeds, the certain threshold value X is set. A call admission control method characterized in that the constant threshold value X is greatly changed by a constant value dx, and when the buffer resource is exceeded, the constant threshold value X is changed by a constant value dx. In the call admission control method according to claim 3 or 4,
The change is executed when the number of connections in the session flow becomes zero.   The program for making a computer perform each procedure in the call admission control method in any one of Claim 1 to 5. In a call admission control device using a computer that individually allocates two types of resources such as a link bandwidth and a buffer amount to a connection session flow, and accepts a new connection session flow within a range that fits in the facility amount,
Means for simultaneously controlling three types of resource allocation methods including a third resource allocation method in addition to the two types of resource allocation methods of average rate resource allocation method and peak rate resource allocation method by computer control;
When the total amount of resources obtained by any of the three types of resource allocation methods is within the amount of equipment under computer control, or the total amount of resources obtained by each of the three types of resource allocation methods Even if all of them exceed the installed capacity, the resources obtained by each of the three resource allocation methods on the graph with the horizontal axis and the vertical axis representing the two types of resource amounts of the link bandwidth and the buffer amount And a means for permitting connection of a new session flow when there is a point that falls within the facility amount range on a line segment connecting each of the three plotted points when the total amount is plotted. Call admission control device.   A communication router comprising the call admission control device according to claim 7.   A network control device, comprising the call admission control device according to claim 7.   A network using the communication router according to claim 8 or the network control device according to claim 9.

Images