KR20010009264A - Adaptive Scheduling Method for variable bit rate Service - Google Patents

Abstract

PURPOSE: A method of performing an adaptive scheduling process for a VBR(Variable Bit Rate) service is provided to find an optimum cell transmission rate relating to data sustained in a buffer by using a buffer size, a delay time, and a PC value, and to continuously deliver the data at regular amounts in the buffer, so as to stably transmit the data. CONSTITUTION: An ATM(Asynchronous Transfer Mode) network finds a maximum index value of time when data is temporarily stored and sustained in a packet memory. The ATM network finds a maximum index value of a capacity in which the stored data is not overabundant. The ATM network finds an index value of a capacity in which the stored data is not exhausted to be transmitted. The ATM network finds an optimum transmission rate relating to the stored data. The ATM network compares a minimum transmission rate which the stored data is not overabundant with a maximum transmission rate which the data is not exhausted. The ATM network finds regular transmission rate during a time section generated afterward, according to a compared result.

Description

Adaptive Scheduling Method for Variable Bit Rate Service

When designing a terminal equipment (TE) or a terminal adapter (TA) in an asynchronous transfer mode network, one should consider that when the sender generates traffic, the receiver It is necessary to generate the traffic that can satisfy the quality required by the network, and to make the best use of the functions of the transmission network. Variable Bit Rate (VBR) traffic in asynchronous transmission mode networks is characterized by Peak Cell Rate (PCR), Maximum Burst Size (MBS), and Sustainable Cell Rate (SCR). A traffic parameter descriptor (Traffic Descriptor) is defined, and the terminal device (TE) or the terminal device adapter (TA) exchanges traffic information with the network manager. Therefore, a traffic transmission method should be developed to effectively use network resources within a predefined traffic capacity limit.

In practice, when designing a terminal device adapter, the size of the buffer (packet memory) used and the time of staying data within the buffer are finite. In addition, since the variable bit rate that is allowed to be accessed through the terminal device adapter shows a characteristic that the traffic size is not uniform and the instantaneous increase / decrease width is increased, that is, a very bursty characteristic, a scheduling method should be effectively applied. .

Burstness for measuring burst characteristics is defined as burstiness = peak_rate / mean_rate. That is, it is defined as the ratio of the peak rate to the average rate, which is used to measure the degree of burst for generated traffic. In general, the larger this value becomes, the more difficult it is to control the traffic inside the network. Accordingly, resource allocation within the network must be increased.

On the other hand, scheduling means that a plurality of types of data are simultaneously transmitted in a section in which a single Asynchronous Cell (ATM) cell is to be transmitted, or data of a single service is intensively transmitted at an arbitrary time interval during a specific time interval. Control the flow of data to comply with previously negotiated traffic parameters. In other words, the scheduling of the transmission time for the cell unit is called scheduling.

Conventional scheduling methods have a problem in that the amount of traffic to be transmitted and received is not averaged over time and cannot be transmitted. There are two conventional techniques.

First, schedule cells at peak cell rate (PCR) within the maximum burst size (MBS) allowed by Usage Parameter Control (UPC), and sustainable cell rate (SCR) if the maximum burst size is exceeded. This is a method of scheduling with). Since this method exports individual traffic within the maximum burst size range without artificial traffic shaping, the result is that the burst characteristics are introduced into the network as it is, and the maximum burst size is exceeded. Since the cell rate is adjusted by the sustainable cell rate (SCR), unwanted data delay or delay variation may be caused.

Second, the second method is to buffer cells in the packet memory and to schedule the average traffic size of the packets in the buffer to overcome the first method to some extent. The problem with this method is that the delay is not uniform.

The present invention is to solve such a conventional problem, and to improve the efficiency of the network utilization and scheduling.

1 is a diagram of cell slot allocation for performing adaptive scheduling;

2 is a diagram illustrating a transmission wait structure for a cell of a specific connection stored in a packet memory.

An adaptive scheduling method for a variable bit rate service according to the present invention for solving the above technical problem,

In the asynchronous transfer mode (ATM) network for transmitting data,

Obtaining a maximum index value of a time at which data can be temporarily stored in a packet memory;

Obtaining a maximum index value of a capacity in which data may be temporarily stored in the packet memory and not overflowed;

Obtaining an index value of a capacity at which data can be transmitted without being exhausted in the packet memory;

Obtaining an optimal transmission rate for data stored in the packet memory;

Comparing a minimum transfer rate for not overflowing the data in the packet memory and a maximum transfer rate for avoiding running out of data; And

And calculating a constant transmission rate during a time interval that will occur in the future according to the comparison result.

The size of the data to be transmitted and received is preferably determined according to the transmission delay value.

It is preferable that the data during the future time interval does not affect the finite capacity of the packet memory.

Hereinafter, with reference to the drawings will be described an embodiment of the present invention. However, the following examples are only examples of the present invention and are not limited to the following examples.

In case of transmitting data in ATM network, 53 byte ATM cell is used as transmission unit. Therefore, dividing the maximum transmission rate of the channel by 53 bytes calculates the number of transmissions per unit time of the ATM cell. In this case, a time interval occupied by one ATM cell is called a cell slot. For example, in the case of a 155 Mbps SONET / SDH channel, 2.735usec becomes a time interval occupied by a single cell slot according to 53 bytes / 155 Mbps = 2.735 usec. When generating data having a mean rate of 77.5 Mbps, data is transmitted using only one cell slot among two cell slots generated in 2.735 usec units.

1 shows a diagram of cell slot allocation for performing adaptive scheduling.

Data is generated according to the variable bit rate (VBR). Therefore, it is necessary to adaptively determine the data transmission rate according to the variable amount of data and the size and delay of the effective buffer. When the circumference in FIG. 1 is called an arbitrary time interval, a cell slot corresponding to the cell rate of a link connected to the network is generated during this period. In Fig. 1, each of the small circles represents the starting point of the cell slot which occurs according to the line speed, i.e., the cell rate determined by the transmission speed of the channel, of which white circles represent unallocated cell slots. , And the gray circles represent the assigned cell slots, respectively.

The cell slot generation of the present invention is independent of the cell generation method since it does not depend on the physical medium but on the algorithm. Thus, as mentioned above, the 53 byte / line rate is the spacing of cell slots conceptually.

Through the process of finding the upper and lower limits of the adaptive rate and finding the optimal values between the upper and lower limits, we find the adaptive cell rate for a particular access and the reciprocal of the adaptive cell rate. The interval can be obtained and the cell transmission authority can be assigned to the connection of the interval corresponding to the inverse of the cell rate. In the present invention, a method of determining a cell rate schedule, a method of determining an optimized rate, or a method of allocating a cell's transmission right are used in the same sense.

2 shows a transmission wait structure for a cell of a particular connection stored in a packet memory. That is, it shows a data standby structure for a variable bit rate (VBR) connection that occurs during an individual cell slot interval and is stored in a packet memory. ΔT is the time for the inverse of the transmission link corresponding to the unit cell slot period, and if the index is j, the size of data generated during the time corresponding to the ((j-1) ΔT, jΔT) period is s. (j), the size of data to be transmitted may be represented by r (j). If the maximum delay value stored in the access memory is D and the size of the packet memory is B, the index values for the slots generated in each cell are as follows.

At this time, s (j-J) represents the size of data waiting to be transmitted, and cs (j-J) represents the size of data remaining after being transmitted for ΔT time among the data of s (j-J). The transmission rate of the lower limit is determined according to the transmission delay value (D) of the size r (j) of the transmitted data.

At this time, 1≤p≤J-1.

Equation 2 is the lower limit values of the transfer rate of the data stored in the packet memory determined according to the condition for smoothing the resultant value of Equation 1 to the maximum, i. Represents a lower limit for the transfer rate.

Through Equations 1 and 2, the lower limit transfer rate for data temporarily stored in the packet memory can be obtained.

In general, the capacity of a buffer is finite. If the data flows in excess of the capacity of the buffer, the overflow of the buffer may occur, resulting in loss of data, resulting in a deterioration of service quality. Therefore, in order to prevent the overflow from occurring in the buffer of the finite size, Equation 3 indicates that the data inputted to the buffer after jΔT does not overflow.

At this time, 1≤h.

In this case, PCR represents a negotiated value for a given variable bit rate (VBR) connection, and B represents the size of the buffer.

Unlike Equation 3, if data in the buffer is depleted, this means that data is bursted into the network. Therefore, the inflow of bursted data becomes inefficient in the use of network resources and makes it difficult to control the network resources.

Through Equation 4, a condition in which data to arrive after jΔT time in the packet memory can be smoothly transmitted without causing a data exhaustion of the buffer can be obtained.

At this time, 1≤h.

Equations 1, 2, 3, and 4 are values for optimizing cell occupancy, and are used to obtain an index value that does not overflow the amount of data stored in the buffer and does not delay the data in the buffer for too long. . Equation 1 is the initial value, Equation 2 is the maximum value that can be delayed, that is, a value that cannot be delayed more than this value, Equation 3 is the value just before data overflow in the buffer, and Equation 4 Represents an optimization value for not affecting the transmission process when another user transmits data. That is, the above equations can be used to obtain an approximated optimal value r (j) that keeps the data in the buffer in an appropriate amount and only stays in the buffer for an appropriate time.

That is, until now, the conditions for obtaining data under optimal conditions have been obtained.

In the following, a method of obtaining an optimal transmission rate will be described.

When sending data to the network, it is important to send the data continuously without causing a transmission conflict with the data of other networks. In other words, it is necessary to always have a constant data rate.

Thus, the procedure for obtaining the approximate optimal value r (j) is as follows.

The maximum value of the lower limit values of the transfer rates for the data buffered in the packet memory becomes the common value of the buffered data, and the optimized transfer rate for these data.

First, the rr value of Equation 5 is obtained. This rr can be used to find the optimal data rate for data values buffered in packet memory.

Equation 5 is to calculate the transmission rate for the data already generated and stored in the packet memory, while the future data should not affect the finite capacity of the packet memory. Therefore, the optimal transmission rate is obtained only when information about data to be generated in the future is reflected. The minimum transfer rate for avoiding data overflow in the buffer and the maximum transfer rate for avoiding data depletion in the buffer should be determined for as long as possible.

Therefore, second, the two values as shown in Equation 6 are compared using the rr value obtained above.

Third, the maximum value of h that satisfies Equation 6 is obtained, and the value H is determined, and the following Equation 7 is obtained.

The inverse of r (j) obtained in this way is allocated to the next transmission cell slot for the variable bit rate (VBR) connection.

The value determined in Equation 5 is an optimal data rate obtained using the buffered values, whereas Equation 7 obtained according to the value of h satisfying Equation 6 is a section in which the same data rate can be obtained during a future time interval hΔT. Therefore, this period becomes an optimal transmission rate at which the same transmission rate is obtained.

According to the present invention, an optimal cell transfer rate is obtained by using a buffer size, a delay time, and a PCR value for data remaining in the buffer, and thus data is continuously transmitted in a constant amount in the buffer, thereby enabling stable data transmission. Do.

Claims (3)

In the asynchronous transfer mode (ATM) network for transmitting data, Obtaining a maximum index value of a time at which data can be temporarily stored in a packet memory; Obtaining a maximum index value of a capacity in which data may be temporarily stored in the packet memory and not overflowed; Obtaining an index value of a capacity at which data can be transmitted without being exhausted in the packet memory; Obtaining an optimal transmission rate for data stored in the packet memory; Comparing a minimum transfer rate for not overflowing the data in the packet memory and a maximum transfer rate for avoiding running out of data; And And obtaining a constant rate during a future time interval according to the comparison result. In claim 1, The size of the data to be transmitted and received is adaptive scheduling method for a variable bit rate service, characterized in that determined according to the transmission delay value. In claim 2, Adaptive scheduling method for a variable bit rate service, characterized in that the data during the future time interval does not affect the finite capacity of the packet memory.