JPH0358646A - Band assignment system in packet communication network - Google Patents

Abstract

PURPOSE:To surely realize quality assurance by relating transmission quality and priority to assign a band of a virtual line. CONSTITUTION:Calls handled by a high speed packet network are classified into 3 classes, and a band assignment device 110 assigns the capacity of a transmission line 100 with respect to the classes of calls. In this case, the class 1 is referred to as a call for a fixed band R1, the class 2 is a burst call having a large maximum band in which the band is secured with the maximum band R2 to meet the quality requirements and the class 3 is a call whose band is secured with a virtual band R3 because the maximum band is small and statistic multiplex effect is to be expected. The classes 1, 2 are defined as high priority classes and the class 3 is as a low priority class, a buffer 120 is exclusively used for the high priority classes and a buffer 121 is exclusively used for the low priority class and a priority controller 130 applies write readout control of the buffer. Thus, no deteriorated quality is affected onto the classes 1, 2 due to the probable hand fluctuation of calls of the class 3.

Description

[Detailed Description of the Invention] (Field of Industrial Application) The present invention relates to a high-speed packet switching network for collectively exchanging voice, data, images, etc. The present invention relates to a system for efficiently accommodating calls.

(Prior Art) In a high-speed Pagano switching network compatible with such a multiplexing method, the generation of packets has statistical properties. Therefore, when setting up a virtual circuit for a call, in order to effectively utilize the transmission path while maintaining the required transmission quality such as packet delay and discard, allocate a certain amount of bandwidth to the call within the network and need to be managed. In that case, in order to simplify the protocol, it is better to measure the actual traffic volume and perform flow control by estimating the human load (bandwidth) at the time of call setup and permitting call setup by allocating bandwidth.
A method of refusing is preferable.

Regarding such a band allocation method, one is known from Watanabe et al., ``Study of statistical multiplex control method in high-speed packet switching,'' described in IEICE technical report SE87-138, pages 43-48. , when configuring a virtual circuit, the average bandwidth, maximum bandwidth, and burst length (
The bandwidth of the transmission path is allocated using a "virtual bandwidth" that is greater than the average bandwidth and less than the maximum bandwidth, taking into account various parameters such as (a value indicating how many Paqueso I* occur continuously).

(Problem to be Solved by the Invention) According to the prior art, the bandwidth allocation for each call is equal to or greater than the average bandwidth.
Although calls are made using a "virtual bandwidth" that is less than the maximum bandwidth, from the perspective of bandwidth allocation, calls handled by high-speed packet networks can be classified into the following three types.

O class 1. ．． ．． Fixed-band call O class where average bandwidth and maximum bandwidth are equal 2. ．． ．． Since the maximum bandwidth is large (for example, more than 1/10 of the link capacity), burst calls must be allocated at the maximum bandwidth in order to guarantee quality. ．． ．． Since the maximum bandwidth is small, burst call class 1 and class 2 calls, for which quality can be guaranteed even if bandwidth is allocated using a virtual bandwidth smaller than the maximum bandwidth, can only obtain the same multiplicity as circuit switching (fixed bandwidth allocation). However, since the transmission line capacity is allocated using the maximum band, that is, the deterministic band,
This does not cause quality deterioration to other calls. On the other hand, for class 3 calls, a higher number of multiplexes than circuit switching can be expected. This is called statistical multiple effects. However, although class 3 calls can be expected to have a statistical multiplexing effect, there are stochastic fluctuations in the band, and there is a possibility that a load beyond the virtual band will be added. In other words, if Class 3 calls are allowed, 1
- There is a possibility that the traffic may flow into the network without causing quality deterioration such as delay or buffer overflow to other calls.

Generally, calls such as video or high-speed still image transfers are Class 1,
However, these calls have stricter quality requirements than class 3 calls such as data communications. However, the problem here is class 1, 2. If the intra-network packet transfer processing for calls 3 is treated uniformly, the quality of class 1 and 2 calls will deteriorate due to the stochastic band allocation due to class 3 calls, even though the bands are secured in a deterministic band. is a possibility that may occur. This is the problem with the first prior art.

The next problem is how to calculate the class 3 virtual bandwidth. The virtual bandwidth of a burst call is determined by the multiplexing efficiency under conditions that satisfy the required quality, but generally this depends largely on the size of the bandwidth of each call in line with the link capacity. Figure 3 is an example of the virtual bandwidth Rv for the maximum bandwidth of the terminal (normalized by link speed).
An example of this is described in ``Methods for Accommodating Burst Traffic in ATM Networks''. In FIG. 3, the maximum band and the average band are both shown. The virtual bandwidth is defined by a value below the maximum bandwidth and above the average bandwidth. From this figure, it can be seen that the larger the maximum band for a call, the larger the virtual band needs to be set.

It should be noted here that this virtual bandwidth depends on the maximum bandwidth normalized by the link capacity. Here, link capacity means the capacity of a link on which class 3 burst calls are multiplexed.

For example, as shown in Figure 4, class 1 and 2
When calls are multiplexed, it means the remaining capacity obtained by subtracting the bandwidth used by these calls from the bandwidth of the transmission path. In reality, since the bandwidth used by class 1 and 2 calls changes from moment to moment, the remaining bandwidth changes from moment to moment, and the virtual band for each burst call must be redefined each time. Therefore,
In the conventional technology that allocates bandwidth using virtual bandwidth, the amount of calculation for the virtual bandwidth for class 3 calls increases significantly. This is the second problem with the prior art.

The present invention provides a method for addressing Class 3 calls. The objective is to provide a bandwidth allocation method that does not affect the quality of calls of classes 1 and 2 even if the area allocation is changed, and that requires less calculation of bandwidth allocation for class 3 calls. I will do it.

(Means for Solving the Problems) According to the first aspect of the present invention, two types of priorities are provided in a packet communication network, and for a call requesting the first priority, packet transmission from an information source is performed. Bandwidth is allocated to the passing transmission path using a band equal to the maximum speed,
For calls that require the second priority, bandwidth is allocated to the transmission path passing through using a virtual band that is greater than or equal to the average value and less than the maximum value of the packet exit speed of the information source, and A bandwidth allocation method in a packet communication network is obtained in which a packet having a priority of 1 is transmitted preferentially over a packet having a second priority.

Further, according to the second invention, in the packet communication network,
Bandwidth is allocated to the transmission path through which each call passes, using a bandwidth equal to the maximum value of the packet exit speed of the information source.The first call type and the average value of the packet sending speed of the information source, the maximum value The total capacity of each transmission path is divided into a second call type, in which bandwidth is allocated to the transmission path passing through the following virtual bands, and the capacity is allocated to the first call type, and the capacity is allocated to the second call type. This provides a bandwidth allocation method that logically separates and operates in advance.

Further, in the second aspect of the present invention, the first call type and the second
It is also possible to reallocate logically separate transmission line capacities to the first call type and the second call type in response to long-term load fluctuations of the call types.

(Operation) In the first invention, a high priority is given to a class 1.2 call in which the maximum bandwidth is allocated to a transmission path, and a class 3 call is allocated in a probabilistically determined virtual band. Mutual influence between classes is eliminated by giving low priority to the class and performing intra-network transfer control. By introducing this priority, calls in the high priority class are not affected by bandwidth fluctuations in calls in the low priority class. Since the highest priority class 1 and 2 calls are allocated the maximum bandwidth, reliable transmission quality can be guaranteed. On the other hand, class 3 can obtain a statistical multiplexing effect by multiplexing based on virtual bandwidth allocation, but stochastic quality deterioration can be made to occur only between class 3 calls.

In the second invention, the link capacities for classes 1, 2 and 3 are virtually separated and allocated in advance. Therefore, the virtual link capacity for class 3 is always constant without being affected by the call load status of classes 1 and 2. The virtual bandwidth for a class 3 call can be calculated by normalizing the bandwidth with the amount of virtual links allocated in advance, and once the virtual bandwidth is determined, there is no need to recalculate it.

Further, in the second invention, multiplexing can be performed more efficiently by reallocating the virtual link capacity to both classes on a monthly basis depending on the long-term load situation. When reallocating, it is necessary to recalculate the virtual bandwidth, but since reallocation in response to long-term load fluctuations does not occur frequently, the increase in the amount of calculation is small.

(Embodiment) Next, a band allocation method in a packet communication network according to the present invention will be described with reference to the drawings.

FIG. 1 is a schematic diagram showing an embodiment of the invention according to claim 1. According to FIG. 1, the embodiment of the present invention
．． 3, a priority control device 130 that controls packet writing/reading of the bucket buffers 120, 121, a transmission path 100, and a bandwidth allocation device 110 that manages the transmission path 100. Consisted of.

Here, class 1 is a fixed band R1 call. Class 2 is a burst call, but it is a burst call with a large maximum bandwidth, and in order to not meet the required quality, the bandwidth must be secured in the maximum imperial area R2. Examples of these calls require high transmission quality (for example, packet discard probability of 10-9 or less) from the packet transmission network, such as video information or large-capacity still image transfer.

Class 3 is a call in which the virtual band R3 is reserved for the transmission line 100 because the maximum band is small and a statistical multiplexing effect can be expected. Examples include LAN-to-LAN communications and terminal-to-host conversational communications, which are burst communications that are relatively slow compared to the transmission path speed.

Here, the band allocation device 110 allocates the capacity of the transmission path 100 to each of these classes of calls. Particularly for class 3, the virtual bandwidth is determined from the average bandwidth and maximum bandwidth of the call. Figure 3 shows the relationship between the maximum bandwidth of each call, the link capacity, and the virtual bandwidth, and the bandwidth allocation device 110 obtains the virtual bandwidth using such a relational expression (Step 1). For new call setup requests, calculate the maximum bandwidth or the sum of virtual bandwidth for each call (
For example, Rl + R2 + R3), this is the transmission line 10
Call setup is allowed only when the capacity C is smaller than 0.

Here, calls of class], 2 will never receive packets beyond the allocated bandwidth, but calls of class 3 will have their bandwidth stochastically fluctuated, and packets will be input to the network beyond the allocated virtual bandwidth. It may be entered. For that reason, 1 for other calls...
There is a possibility of quality deterioration. As a measure for that purpose,
In the present invention, classes 1 and 2 are defined as high priority classes, and class 3 is defined as a low priority class, and intra-network transmission is performed. Buffer 12
0 is dedicated to the high priority class, and the buffer 121 is dedicated to the low priority class, and the priority control device 130 controls writing and reading of the buffer. When there are packets in the high priority buffer, low priority packets are not sent. In this way, the quality of classes 1 and 2 will not be degraded due to stochastic band fluctuations in class 3 calls.

Next, a second embodiment of the invention will be described using FIG. 2.

(12) The transmission line capacity C is logically separated into a virtual link capacity C1 for classes 1 and 2 and a virtual link capacity C3 for class 3. The allocation of 01 and 02 is determined by statistically predicting the long-term ratio of both l-rahic. Here class 1,
If the band of call 2 is set as Rl and R2, C1≧R1+R2
The total virtual bandwidth of class 3 calls is ΣR3 so that
If it deviates, band allocation is performed so that C2≧ΣR3. This bandwidth allocation is based on the bandwidth allocation shown in Figure 1 for device 1.
10.

As a result, the remaining capacity for class 3, which is obtained by subtracting the usage capacity of class 1 and 2 calls from the transmission path capacity JtC, is always 02 or more. Therefore, the virtual bandwidth described above may be calculated using the relationship shown in FIG. 3 from the maximum bandwidth and average bandwidth of each call normalized with C2 as the link capacity. At this time, C2 does not depend on the usage capacity of class 1 and 2 calls, so once the virtual bandwidth is calculated, there is no need to recalculate it. Furthermore, although it has been described that the allocation of C1 and 02 is performed based on long-term traffic prediction, reallocation may also be performed based on long-term traffic fluctuations (due to time, day, or human events). This is 01 and 0 in Figure 2.
It is to dynamically move the virtual boundary (logical boundary) of 2. This allows the transmission path to be used effectively.

At this time, it is necessary to recalculate the class 3 virtual bandwidth due to the change in C2, but since reallocation of Cl and C2 does not occur frequently, the amount of calculation does not increase much.

Note that the first invention and the second invention can be implemented independently or can be used in combination. Further, in this embodiment, an example of application to a single-stage transmission link has been shown, but the present invention can also be applied to bandwidth allocation in all transmission paths of multiplexing equipment and exchanges through which bucket calls pass.

Further, in the embodiment of the first invention, a structure is shown in which buffers are provided for each priority, but this example is not limited to this buffer structure.
The present invention can also be applied to priority control using other buffer structure methods.

(Effects of the Invention) As described above, according to the present invention, by allocating the bandwidth of a virtual circuit by associating transmission quality and priority,
Reliable quality assurance can be achieved.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an embodiment of the first invention, FIG. 2 is a diagram showing an example of a bandwidth allocation situation when link capacity is virtually divided according to an embodiment of the second invention, and FIG. is a diagram illustrating an example of a method for determining a virtual:
The figure is a diagram showing an example of a band allocation situation according to the prior art. In the figure, 100... Transmission line, 110... Bandwidth allocation device, 12
0, 1.21...Buffer, 130...Priority control device.

Claims (3)

[Claims] (1) In the packet communication network, two types of priorities are established,
For a call requesting the first priority, bandwidth is allocated to the transmission path passing by a band equal to the maximum value of the packet transmission rate of the information source, and for a call requesting the second priority. For this purpose, bandwidth is allocated to the transmission path passing through using a virtual bandwidth that is greater than or equal to the average value and less than the maximum value of the packet transmission speed of the information source, and packets with the first priority within the network are assigned the second priority. A bandwidth allocation method in a packet communication network that is characterized in that packets are transmitted preferentially over packets with higher priorities. (2) In a packet communication network, the first type of call and information source packet transmission in which each call is allocated a bandwidth to the transmission path through which the call is assigned a bandwidth equal to the maximum value of the packet transmission speed of the information source. a second call type that allocates bandwidth to transmission paths passing through a virtual band whose speed is above the average value and below the maximum value; and a capacity where the total capacity of each transmission path is allocated to the first call type; Bandwidth allocation method that operates by logically separating the capacity allocated to the second call type in advance. (3) In the band allocation method for a packet communication network according to claim 2, the first call type and the second call type are A bandwidth allocation method in a packet communication network characterized by reallocating logically separate transmission path capacities for two call types.