KR19980040849A - Improvement of Schedule Characteristics in Transfer of Jitter Information by Counter Interworking in Asynchronous Transmission Mode Networks - Google Patents

Abstract

According to the present invention, the jitter is controlled at each node without exchanging information with the previous node by reading the counter value at the time of arrival of the cell at each node of the asynchronous transmission mode communication network with constant transmission delay and calculating information related to jitter by itself. In the high-speed communication network, the delay limit at each node is set to be larger than the time period corresponding to the bandwidth, thereby improving the schedule characteristics and reducing the burden on the synchronization of multimedia services to the terminal. The present invention relates to a method for improving schedule characteristics in jitter information transmission by counter interworking in an asynchronous transmission mode communication network that can reduce the complexity. A counter (C _{i, j} ) is generated per per session of each node. do the counter (C _{i, j)} is counted in units of slots as separate installation of the connection units by Frame according to an asynchronous transfer mode communication network transmission delay is constant which is defined, the bandwidth requirements of the frame _(j = r Delay limit compared to ) Is considered to be a packet that can be divided into several cells in each node, the delay limit ( To increase the time interval of one cycle ( = ) Up to q _j cells.

Description

Improvement of Schedule Characteristics in Transfer of Jitter Information by Counter Interworking in Asynchronous Transmission Mode Networks

The present invention relates to a method for improving schedule characteristics in jitter information transmission by counter interworking in an asynchronous transmission mode communication network. In particular, a counter value at the time when a cell arrives at each node of an asynchronous transmission mode communication network with constant transmission delay is read. As the information related to jitter is calculated by itself, the delay limit at each node is set to be larger than the time period corresponding to the bandwidth in the high-speed communication network in which jitter is adjusted at each node without exchanging information with the previous node. Improvement of the schedule characteristics in the transfer of jitter information by the counter interworking in the asynchronous transmission mode communication network, which can reduce the burden on the synchronization of multimedia services to the terminal and the terminal, thereby reducing the complexity of the terminal and the efficient use of network resources. It is about a method.

In general, a broadband integrated service digital network (B-ISDN) connects subscribers and service providers that are concentrated or discrete based on broadband transmission and switching technology to provide continuous real-time services with a wide bandwidth distribution and clustered data services. It is a digital communication network that comprehensively provides various services.

As such, in order to effectively process various services, ITU-T adopts an asynchronous transfer mode (ATM) as a communication method of BISDN. The asynchronous transmission mode communication method is a packet transfer method using an asynchronous time division multiplexing method, and various services can be uniformly processed by the advantages of the existing circuit switching method and the packet switching method.

In the future, asynchronous transmission mode networks will have many services that are unpredictable at present, and will have very different traffic characteristics, both qualitatively and quantitatively. In general, the service should be able to be guaranteed with various quality of service for yield, delay, jitter, loss rate, etc., especially delay and jitter as real-time services such as video and voice become the main services of broadband network. The requirements for are becoming very strict.

In real-time services, when information is not delivered within the time limit, it has the same effect as loss, and it is becoming very important to effectively satisfy the requirements for delay and jitter. In the asynchronous transmission mode communication network, as cells from different connections interact with each other in the switch, the cells must be properly controlled to guarantee the quality of service to the user.

1 is a diagram illustrating a call connection state through a node in a general asynchronous transmission mode communication network. In the drawings, reference numerals i-1 to i + n denote asynchronous transmission mode communication network nodes, and A to F denote subscribers.

The asynchronous transmission mode communication network is composed of a group of interconnected nodes, and data is transmitted from a source to a destination through a network of nodes. The figure shows a general example of the concept of data transmission, wherein the nodes are connected by a transmission path. On the other hand, when data is input from a predetermined subscriber to a network, the data is transferred from a predetermined node to a predetermined node while going to a predetermined destination.

For example, when data is transmitted from subscriber A to subscriber D, it is first sent to node i-1 and then through node i and node i + 1 or node i + 2. To D).

On the other hand, the queue service method is a method of determining the transmission order of cells in consideration of the relationship between the cells stored in the queue. The queue service method manages three resources of a network, that is, bandwidth, delay limit, and buffer space, and these three resources are directly related to yield and delay loss rate, which are performance parameters of a service required by a user. Therefore, when the queue service method is effectively used and the three resources are used flexibly, the quality of service of the user can be guaranteed.

In addition, the queue service method is largely divided into work-conserving and nonwork-conserving. In the work preservation method, when a cell exists in a queue, the server never rests. In non-work preservation, even if a cell exists in the queue, it can not be serviced.

Moreover, in the past, average delay and average yield are the main parameters of performance. In the future, the study of the work preservation method is a major part, but in the future integrated network, guaranteeing the performance of continuity real-time service is an important issue. This makes the thresholds for delay and jitter quite important.

In using the work preservation scheme, a big problem is that clustering gradually occurs as traffic passes through the network, which wastes network resources, thereby significantly reducing the jitter characteristic of the traffic. On the contrary, in the non-work preservation method, the server utilization rate is reduced because it does not properly utilize the server that is at rest, but it passes through the network with almost the traffic characteristics maintained and is sufficiently supported by a small amount of resources. The jitter characteristic is improved. Thus, the non-work preservation scheme is more suitable for supporting continuity real-time services.

In recent years, many non-working scheduling algorithms have been proposed to effectively support real-time services. However, these algorithms have become quite inefficient by employing a frame structure or using a packet header that is not actually supported in an asynchronous transmission mode network. do. The algorithm may include a hierarchical round robin (HRR) algorithm, a stop-and-go algorithm, an jitter earlylies-due-date algorithm, rate-controlled static priority (RCSP), and the like. ) Algorithms.

The hierarchical turn algorithm and the stop-and-go algorithm both use a frame technique. In this hierarchical turn algorithm, the number of cells that can be serviced in one frame period for each connection is limited. Cell rate jitter is guaranteed. However, the frame used in the algorithm is defined independently without interaction with neighboring nodes so that the delay occurring when the cell passes through the node varies from cell to cell, thereby preventing end-to-end delay jitter.

FIG. 2a shows a switching node i having an output link l connected to input links l ', l' 'in a typical stop-and-go queue service scheme, and FIG. 2b is shown in FIG. 2a. It is a figure which shows the structure of the frame in each link.

In the stop-and-go algorithm, a cell is transmitted in a one-to-one correspondence between frames of an input link and frames of an output link at an exchange node. Almost the same delay occurs in all cells belonging to such a connection, and end-to-end delay jitter is generated only by position translation in the frame, which is ensured by a fairly small amount.

Both algorithms have their own problems due to the adoption of the frame technique, that is, the problem of coupling between delay limits and band allocation units. As a solution, both algorithms adopt a layered frame structure. It's a bit different. In the hierarchical ordering algorithm, a frame of a lower level is defined by passing a portion of time slots of a higher level, and a process of selecting a cell to be serviced because one time slot belongs only to a frame of a specific level is simplified. do.

However, although the number of time slots that can be allocated at each level is limited, it may not be possible to accept new connections even if there is excess bandwidth. Frame will be constructed.

Therefore, the complexity of the implementation is increased by increasing the number of levels by using the static priority of the level unit in the process of selecting a cell to be serviced because one time slot is simultaneously included in the frames of several levels. In each level, the extra bandwidth is fully utilized, thereby preventing the degradation of bandwidth usage efficiency caused by the sequential algorithm. In both algorithms, a layered frame structure is introduced to solve the coupling problem to some extent, but the coupling problem at each level still exists.

3 is a conceptual diagram illustrating a delay jitter control method in a general data communication network. Meanwhile, the EDD algorithm and the RCSP algorithm output scheduling information in the header of the packet to guarantee delay jitter, and the next node transmits information about how fast it is transmitted compared to the deadline. do. In contrast, virtually all cells are output with maximum delay at every node, ensuring end-to-end delay jitter.

As described above, in the conventional non-working preservation algorithm, the jitter is adjusted by scheduling using a frame structure or the jitter information is transmitted by using the packet header so that the jitter is adjusted at each node, and the hierarchical order algorithm guarantees the cell rate jitter. In contrast, the remaining three algorithms guarantee delay jitter.

In the hierarchical order algorithm and the stop-and-go algorithm, the frame scheme is adopted to simplify implementation, but the coupling problem occurs, resulting in poor bandwidth allocation and delay limit allocation characteristics, and in the jitter EDD algorithm and the RCSP algorithm. Although there is an advantage that the allocation of the delay limit and the delay limit can be separated, scheduling related information is sent for every packet, which causes overhead. This is a big problem for networks with small packet sizes.

Accordingly, the present invention is to solve the above problems, the counter value of the time when the cell arrives in each node of the asynchronous transmission mode communication network with a constant transmission delay is read and the information related to jitter is calculated by itself to separate from the previous node In a high-speed communication network in which jitter is adjusted at each node without exchanging information, the delay limit at each node is set to be larger than the time period corresponding to the bandwidth, thereby improving the schedule characteristics and reducing the burden of synchronizing multimedia services to the terminal. The present invention relates to a method for improving schedule characteristics in jitter information transmission by counter interworking in an asynchronous transmission mode communication network that can reduce network complexity and reduce terminal complexity.

In order to achieve the above object, the present invention provides an asynchronous transmission mode communication network in which a transmission delay in which an individual frame is defined in a connection unit is provided by performing a counter in a slot unit while a counter is installed for each node. In the case where the delay limit is larger than the bandwidth required for the frame, each node regards several cells as one packet that can be divided, and the delay limit is increased to serve up to qj cells during one period of time interval. It is characterized by.

According to the present invention configured as described above, the counter value at the time of arrival of the cell is read from each node of the asynchronous transmission mode communication network with a constant transmission delay, and the information related to jitter is calculated by itself, so that each node is not exchanged with the previous node. In a high-speed communication network in which jitter is controlled at a node, a delay limit at each node is set to be larger than a time period corresponding to a bandwidth, thereby improving schedule characteristics and reducing burden on synchronization of multimedia services to a terminal. Efficient use and complexity of the terminal can be reduced.

1 is a view showing a connection state of a subscriber through a node in a general asynchronous transmission mode communication network,

FIG. 2a shows a switching node i with an output link l connected to an input link l ', l' 'in a typical stop-and-go queue service scheme; FIG.

FIG. 2B is a diagram showing the structure of a frame in each link shown in FIG. 2A;

3 is a conceptual diagram illustrating a delay jitter control method in a general data communication network;

FIG. 4 is a diagram illustrating an embodiment of a method for improving schedule characteristics in jitter information transmission by counter linkage in an asynchronous transmission mode communication network according to the present invention.

Explanation of symbols on the main parts of the drawings

i-1 to i + n: node, A to F: subscriber.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described in detail.

On the other hand, when the end-to-end delay and jitter can be effectively ensured, the general jitter control service method briefly described, and the proposed algorithm in detail.

In the jitter control service method, a server is conceptually divided into two components, a jitter controller and a scheduler. The jitter controller compensates for the distortion of traffic generated in the network, and the scheduler determines a service order between the cells that are distortion-compensated. Done. This functional separation allows delay limits to be flexibly allocated regardless of bandwidth, and greatly improves end-to-end jitter.

When the traffic is partially or totally reconfigured at each node so that the distortion of the traffic can be compensated for, the original traffic characteristic passes through the network without being greatly distorted. Therefore, the required buffer space can be reduced to save network resources, and the jitter characteristic of the transmitted traffic becomes good, thereby reducing the burden of jitter compensation processing on the receiving side.

In addition, when reconfiguring the characteristics of the traffic, the eligible time (ET) of the arriving cell is calculated, and the destination cell is temporarily stored in the jitter controller until the time is reached, and then at the moment when the eligibility time is reached. Will print to the scheduler. In other words, the cell is not scheduled as soon as it arrives, but is considered as if the cell has not arrived by the cell's qualifying time.

The jitter regulator is divided into a rate jitter regulator and a delay jitter regulator by a method of calculating the eligibility time, wherein the rate jitter regulator defines the eligibility time based on the distance from the previous cell, but the delay jitter regulator at the previous node. The eligibility time is calculated based on the deadline of and the traffic is completely reconstructed.

On the other hand, in the scheduler, the service order is determined by the delay requirements so that each cell performs the service within the delay limit. The scheduling method is a first-in-first-out (FCFS) method and a static priority method. Will be classified as.

The first-in, first-out method is the simplest to implement, but only one delay limit is provided regardless of the type of traffic, making it difficult to use in various network characteristics, and the fixed-priority method is fixed after dividing the traffic into different classes. Serving as a priority provides delay limits for each class.

However, there is a limit to the efficiency that can be obtained because fixed priorities are always used regardless of network conditions. In the dynamic priority scheme, priority is given in units of cells, and then services are provided from the highest priority. This method is the most efficient, but it causes the hassle of sorting all the cells in the queue in order for the highest priority cell to be selected.

The non-working preservation algorithm can be interpreted by the jitter adjustment service method, and the jitter adjusters and schedulers adopted by the existing non-work preservation algorithm are listed in Table 1 below.

TABLE 1

In general, the conventional jitter control service method is divided into two types according to the jitter control method, one of which is to control jitter by transmitting information related to scheduling to one node by using one section of the packet header. The other is a method in which jitter is controlled by introducing a frame technique to transmit and exchange cells in units of frames.

On the other hand, since there is no extra section for carrying scheduling related information in the ATM cell header, the first method cannot be applied to an asynchronous transmission mode communication network, and the second method can be used in an asynchronous transmission mode communication network, but there are various problems. Moreover, the efficiency is degraded by the coupling nature between delay limits and bandwidth allocation units, which are inherent in the adoption of frame schemes, and the use of layered frame schemes can alleviate the problem to some extent, but is still difficult to solve. It is a problem.

After that, the proposed algorithm overcomes the disadvantages inherent in the frame technique by defining and using individual frames as counters are used for each connection.In addition, the boundary of each frame is matched between neighboring nodes when accepting the connection. Can be recognized by itself.

In addition, in the network (i) in the network so that jitter is adjusted, the qualifying time is defined through interaction with the previous node (i-1) irrespective of the relationship between neighboring cells. That is, it defines the eligible time as if the cell has experienced the maximum allowable delay, e.g., the size of the frame, at the previous node (i-1). And the qualified time of cell (k) in connection j for the node after the source node ( ) Is defined in the following equation (where i = 1, 2, ..., N).

[Equation 1]

Τ _i denotes a propagation delay between the previous node i-1 and node i, Denotes the frame size of connection j. Equation 1 shows that the qualified time of the k-th cell arriving at node i is determined by the previous node i-1 of the cell. This means that the previous node i-1 should output the qualified time information of the k-th cell to the node i. In the conventional method, the information is loaded on the header of the k-th cell.

However, in the present algorithm, it is possible to know itself as soon as the cell arrives at the node without having to send the information every time. That is, for connection j By operating the per-session counter (C _{i, j} ) to establish the relationship between ), The value of the counter is read out and immediately known. Thus, only one counter per connection can be used to know the virtual arrival times for all cells of the connection so that delay jitter can be effectively adjusted.

Therefore, each node has one counter per-session, and each counter is counted in slot units to define individual frames. The counter decrements the count value by 1 for each slot, and is initialized again by the size value of the frame as soon as the counter value becomes zero. The counter is interlocked by accurately reflecting the transmission delay between neighboring nodes. That is, if the cell starting when the counter value is c in the previous node (i-1) arrives at the node (i), it interlocks so that the value of the counter of the node (i) is also c. This can be accomplished by synchronizing the operation of the counters in the neighboring nodes.

The value of the counter has two meanings, which means the time remaining until the deadline for the cells in the scheduler and the time remaining until the qualified time for the cells in the jitter adjuster. Thereafter, when the value of the counter becomes 0, the service deadline of the cell in the scheduler becomes a service, and the cell in the scheduler must be serviced before the counter becomes 0.

For example, if the service is performed when the value of the counter is 3, the cell is serviced before the 3 slots of the deadline, and when the cell arrives at the next node, the value of the counter at that time becomes 3 Will be stored in the jitter adjuster for three slots. If it is stored for 3 slots in the jitter controller and then output to the scheduler, it is the same as having received the deadline service from the previous node and reaching the scheduler of this node.

FIG. 4 is a diagram illustrating an embodiment of a method for improving schedule characteristics in jitter information transmission by counter linkage in an asynchronous transmission mode communication network according to the present invention.

In this case, regardless of when the service is received from the previous node, the time input to the scheduler of the next node is always constant, and the delay jitter is completely removed by performing the service after all cells have experienced the maximum delay at the node.

In the method of the present invention, the packet size is small, such as the asynchronous transmission mode network, and can be effectively applied to control jitter in a high speed communication network that does not support the transmission of scheduling-related information. It is entirely dependent on the operation of the counter, making it possible to use it only in synchronization networks with constant transmission delay.

On the other hand, the required bandwidth rj, for example Delay limit compared to In the case of large), each node is regarded as one packet that can be divided, and the delay limit is increased to effectively support multiple cells during one period of time, for example, a frame.

In this way, the bandwidth is used equally and only the end-to-end delay increases, and the time interval of one cycle For example, in case of serving up to qj, that is, n cells during nTj, it should be adjusted to input qj cells per frame at the entrance of the network. This is defined in Equation 2 below for the eligibility time at the node at the network entrance.

[Equation 2]

),

, [ ]}, ( <K <∞)

Therefore, the nodes in the network define the qualifying time as in the previous algorithm.

4 illustrates a process in which a plurality of cells pass through a network through aligned frames between nodes in one frame. Is 20 and the number of cells qj is two, for example. There are various effects obtained by adjusting the frame size by the method of the present invention.

First, as the schedule characteristics increase, that is, as the frame size increases, so does the delay limit at each node, thereby providing a smaller delay limit for subsequent connection establishment requests. As it becomes larger, the basic unit of bandwidth allocation gradually decreases, thereby reducing the loss due to bandwidth quantization.

In addition, when various delay limits are provided by adjusting the size of a frame, synchronization of multimedia services can be efficiently performed. For example, if the real-time information of each medium in the source arrives at the same destination through different connections, synchronization occurs at the destination terminal based on the connection that causes the largest end-to-end delay.

Thus, a connection with a small bandwidth usually suffers from a large end-to-end delay such that many cells belonging to a connection with a large bandwidth are stored for a long time at the destination terminal for inter-media synchronization, which requires a significant amount of buffers. In addition, when the network provides various delay limits as necessary, the terminal can sufficiently synchronize with a small amount of buffer, thereby greatly reducing the burden on the terminal.

In addition, in case of providing a large delay limit for a large bandwidth, the amount of buffers required in the network increases, but unlike terminals, resources in the network can be shared by all the connections in the network. Will be.

On the other hand, assuming that all connections pass through the same hop number N, the amount of buffers required by the terminal and the network for inter-media synchronization is respectively , to be. Here, rj represents the bandwidth of the connection j belonging to the set 연결 of the connection requiring synchronization, r 'represents the minimum value among them. Therefore, if the bandwidth of each connection is significantly larger than the minimum bandwidth, it can be seen that there is almost no difference in the buffer amount.

As described above, according to the present invention, a counter value at the point of arrival of a cell is read from each node of an asynchronous transmission mode communication network having a constant transmission delay, and information related to jitter is calculated by itself so that no information is exchanged with the previous node. In a high-speed communication network in which jitter is controlled at each node, the delay limit at each node is set to be larger than the time period corresponding to the bandwidth, thereby improving the schedule characteristics and reducing the burden on the synchronization of multimedia services to the terminal. Efficient use of the terminal and complexity of the terminal will be reduced.

Claims (2)

Counter (C _{i, j} ) is installed as a per-session of each node, and this counter (C _{i, j} ) performs counting by slot unit, so that transmission delay in which individual frames are defined in connection unit is constant. In the asynchronous transmission mode communication network, The required bandwidth of the frame (rj = Delay limit compared to ) Is considered to be a packet that can be divided into several cells in each node, the delay limit ( To increase the time interval of one cycle ( = A method of improving schedule characteristics in jitter information transfer by counter interworking in an asynchronous transmission mode communication network, characterized in that a maximum number of qj cells are served. The qualified time of the cell q _j for a predetermined connection j, )silver = 0 (0 <k < ), = max { + , [ ]} ( A method for improving schedule characteristics in jitter information transmission by counter linkage in an asynchronous transmission mode communication network, characterized in that &lt; k &lt;