KR20030027496A - Quality of Service control system and method for video service using dynamic resource allocation - Google Patents

Abstract

The present invention improves utilization of network resources by dynamically allocating network resources in order to provide reliable video services in the Internet and asynchronous transmission mode (ATM) networks, and provides the user with the desired (maximum packet delay, packet loss rate) perspective. The present invention relates to a new traffic control apparatus and method for providing QoS.

According to the present invention, a traffic predictor for measuring the incoming video traffic for each connection to predict the traffic to arrive in the next time interval; QoS observer discarding when the connection time of the video traffic stays in the buffer exceeds the maximum packet delay; A dynamic resource manager that determines a band to allocate to the connection-specific video traffic at a next frame time according to the predicted future input traffic of the connection-specific video traffics, the number of delayed packets waiting in the buffer, and the degree of QoS provided so far; And a connection control manager that accepts the connection of the new video source if the QoS provided to all existing connections is above a preset level and the sum of the average input data rates of the existing connections is below the output link capacity tolerance level. An apparatus for controlling QoS of a video service using dynamic resource allocation is provided.

Description

QOS control apparatus and method for video service using dynamic resource allocation {Quality of Service control system and method for video service using dynamic resource allocation}

The present invention relates to an apparatus and method for controlling a QOS of a video service using dynamic resource allocation, and more particularly, to dynamically provide resources of a communication network in order to provide a reliable video service in an internet and asynchronous transmission mode (ATM) communication network. The present invention relates to an apparatus and method for controlling a QOS of a video service using dynamic resource allocation capable of improving the utilization of communication network resources and providing QoS of a desired (maximum packet delay, packet loss rate) perspective.

In a high-speed communication network such as ATM, a management structure for guaranteeing QoS for variable bit rate (VBR) video traffic requires the user to declare the traffic characteristics of the video source to the network in advance when establishing a connection. On the basis of the request, the communication network performs call admission control and resource allocation.

However, in real time video applications such as Internet broadcasting and interactive video, there is a problem that it is impossible or difficult for the user to inform the network in advance of the characteristics of the video traffic in advance. By allocating, a problem arises in that utilization of a network is lowered with over-allocation of resources.

The Dynamic Resource Allocation structure dynamically allocates bandwidth and buffers by observing input traffic, and can improve network utilization without requiring traffic information to the user in advance. Much research is being done. Existing dynamic resource allocation typically uses linear or nonlinear filters to predict future traffic and to determine the band to be allocated in the next time interval based on the predicted traffic. Several representative methods have been proposed so far for real-time on-line prediction of VBR video traffic.

First, there is a method for predicting JPEG and MPEG traffic using Recursive Least Square (RLS) and Time Delay Neural Network (TDNN) algorithms.

Second, there is an adaptive linear prediction method using Least Mean Square (LMS).

Third, there is an adaptive wavelet prediction method for improving the late convergence speed of the LMS method.

The existing dynamic band allocation based on this prediction uses Equation 1 below.

here, Is an estimate of the traffic to be input next time based on the input traffic of video connection i so far, Q _i (n) is the queue length of connection i at the current time n, and K is the current video source The number of That is, the existing band allocation method divides the band for each connection within the limit of the link capacity C of the communication device based on the sum of the next predicted traffic and the amount of packets currently waiting in the buffer (A _i (n + 1)). Done.

Until now, the dynamic resource allocation method has been mainly focused on improving the accuracy of traffic prediction. For dynamic bandwidth allocation, a simple structure based on the prediction traffic and the current queue length is used. However, such an existing structure cannot estimate the QoS that can be provided to the user. Therefore, there is a need for a dynamic and flexible QoS control mechanism suitable for multimedia traffic with efficient use of resources through prediction.

An object of the present invention for solving the problems of the prior art as described above is to dynamically allocate the resources of the network in order to provide a reliable video service in the Internet and asynchronous transmission mode (ATM) communication network to improve the utilization of the network resources, It is an object of the present invention to provide a new traffic control apparatus and method capable of providing QoS in terms of a user desired (maximum packet delay, packet loss rate).

1 is a block diagram of an apparatus for controlling QoS of a video service using dynamic resource allocation according to an embodiment of the present invention;

FIG. 2 is a conceptual diagram illustrating a method of operating a delayed packet counter in the QoS observer shown in FIG. 1;

3 is a configuration diagram of the dynamic resource manager 140 shown in FIG. 1,

4 is a flowchart illustrating a process band allocation policy procedure of the process band allocator shown in FIG.

5 is a diagram illustrating a change in PQD according to an embodiment of the present invention.

※ Explanation of code about main part of drawing ※

110: connection control manager 120: traffic predictor

130: QoS Observer 140: Dynamic Resource Manager

310: required band estimator 320: process band allocator

In order to achieve the above object, according to the present invention, there is provided a QoS control apparatus for a video service using dynamic resource allocation, comprising: a traffic predictor for predicting traffic arriving at a next time interval by measuring incoming video traffic for each connection; QoS observer discarding when the connection time of the video traffic stays in the buffer exceeds the maximum packet delay; Dynamic to determine the bandwidth to allocate to the connection-specific video traffic at the next frame time according to the future input traffic of the connection-specific video traffics predicted by the traffic predictor, the number of delayed packets waiting in the buffer and the degree of QoS provided so far. Resource manager; And a connection control manager that accepts the connection of the new video source if the QoS provided for all existing connections is above a predetermined level and the sum of the average input data rates of the existing connection is below the output link capacity tolerance level. An apparatus for controlling QoS of a video service using a dynamic resource allocation is provided.

In addition, a QoS control method of a video service using dynamic resource allocation, the method comprising: a traffic prediction step of predicting traffic arriving at a next time interval by measuring video traffic for each connection input; A QoS observation step of monitoring a QoS degree of currently present traffic and discarding when the connection time of the video traffic stays in a buffer exceeds a maximum packet delay; Dynamic to determine the bandwidth to allocate to the connection-specific video traffic at the next frame time according to the future input traffic of the connection-specific video traffic predicted in the traffic prediction step, the number of delay packets waiting in the buffer and the QoS degree provided so far. Resource management step; And a connection control management step of accepting a connection of a new video source if the QoS provided to all existing connections is equal to or greater than a preset level and the sum of the average input data rates of the existing connection is equal to or less than the output link capacity tolerance level. A QoS control method of a video service using dynamic resource allocation is provided.

In addition, the traffic prediction step of predicting the traffic to arrive in the next time interval by measuring the video traffic for each connection input to the computer; A QoS observation step of discarding when the connection time of each video traffic stays in a buffer exceeds a maximum packet delay; Dynamic to determine the bandwidth to allocate to the connection-specific video traffic at the next frame time according to the future input traffic of the connection-specific video traffic predicted in the traffic prediction step, the number of delay packets waiting in the buffer and the QoS degree provided so far. Resource management step; And a connection control management step of accepting a connection of a new video source if the QoS provided to all existing connections is above a preset level and the sum of the average input data rates of the existing connections is below the output link capacity tolerance level. In the dynamic resource management step, the required bandwidth is determined by using the QoS level currently provided to the sum of the predicted traffic for the next frame time and the number of packets waiting in the current buffer, and the required bandwidth for each video traffic for each connection. A computer readable recording medium having a program recorded thereon capable of receiving an estimated value and performing band allocation in comparison with a link capacity is provided.

Hereinafter, an apparatus and a method for controlling a QOS of a video service using dynamic resource allocation according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

Traffic prediction in the present invention uses a conventional video prediction method. Based on the predicted traffic and the observed QoS, the band to be allocated in the next frame time is determined. In order to perform QoS observation in real time, a packet delay counter is used to update the packet loss rate using the counter value every frame time. While each video source maintains a video service, a simple connection admission control method is used to prevent the quality of the existing video connection from dropping below the desired level due to the connection acceptance of the new video source in the network.

1. System structure

1 is a block diagram of a QoS control apparatus for a video service using dynamic resource allocation according to an embodiment of the present invention, wherein the QoS control apparatus includes a connection control manager 110, a traffic predictor 120, a QoS observer 130, and A dynamic resource manager 140.

Traffic prediction, QoS measurement, resource allocation, etc. are made at every same time interval, and the present invention uses one frame time of video (1/30 seconds for 30 frames per second) in this time interval. In the QoS control apparatus proposed in the present invention, the following functional blocks are defined.

First, the traffic predictor 120 estimates traffic to arrive at the next time interval by measuring input traffic for each connection.

In addition, the QoS observer 130 monitors the degree of QoS being serviced for each connection with respect to the defined QoS parameter, and determines how it meets the user QoS requirements.

In addition, the dynamic resource manager 140 calculates the required bandwidth and allocates resources for the next time interval in order to provide the QoS requested by the user for each video connection.

In addition, the connection control manager 110 determines whether to accept the connection of the new video source in consideration of the simple traffic characteristics of the current video sources and the QoS being provided.

In the present invention, the following two parameters are defined as QoS parameters provided by each node of a communication network for each connection for real-time video application. Here, each parameter value separates QoS requirements from actual users in units of nodes.

The packet loss ratio is defined as lossR _i , and the maximum packet delay delay is defined as d max (i).

The maximum packet delay in the present invention is defined in units of frame time of the video in consideration of the characteristics of the video. That is, 'dmax (i) = 2' means that the maximum packet delay of connection i is 2 frame times (2/30 seconds for 30 frames per second).

2. QoS observer and observation device

In the QoS control of the present invention, packets whose latency in the buffer exceeds dmax (i) are discarded, which leads to packet loss. The measurement of QoS currently offered in real-time video transmission should be made with a simple implementation. QoS observations are made every frame time, and the provided QoS value is updated.

The measurement of the packet loss rate, which is the first QoS parameter, is calculated as the ratio of the number of lost (discarded) packets among the input packets of the connection i up to the current n frame time as shown in Equation 2 below.

Where X _i (m) is the number of input packets of connection i at frame time m, and L _i (m) is the number of double lost packets.

A simple counter in delay frames is used to know if any packet has exceeded the delay limit.

FIG. 2 is a conceptual diagram illustrating a method of operating a delayed packet counter in the QoS observer illustrated in FIG. 1.

Each video connection i has dmax (i) delay packet counters QC _i ^d (n). QC _i ^d (n) represents the number of packets in the current buffer that have experienced (d-1, d] frame delay at the end of n frames, where (d-1, d] is greater than d-1, d It means less than or equal to the range.

QC _i ^{dmax (i)} Packets that have not been serviced during the (n + 1) frame time of the (n) packets exceed the maximum delay limit dmax (i), and in this method these packets are discarded in the buffer. Each connection buffer in this system architecture is large enough to queue all the input packets within the maximum allowable limit, and the packet loss is not serviced within the maximum delay limit at the front of the buffer as shown in FIG. It happens when you can't. The number of packets L _i (n + 1) discarded during the (n + 1) frame time is obtained using Equation 3 below.

Here, A _i (n + 1) is the bandwidth (unit: number of packets) allocated to connection i at (n + 1) frame time.

At the end of each frame time, the packet delay counter for each connection is updated. The proposed delay packet observation method does not calculate the delay time for each input packet, but uses only the number of input packets of the corresponding frame time, the dynamically allocated bandwidth (the number of output packets), and the value of each delay packet counter. The number of packets waiting for each delay time can be calculated using Equation 4 below.

here, This is the total number of packets waiting in the buffer of connection i at the end of n frames. The update of the packet delay counter and the provided QoS measurement at the generalized (n + 1) end point can be obtained using Equation 5 below.

3. Dynamic resource manager and its method

The dynamic resource manager 140 determines a bandwidth to be allocated to each connection at the next frame time according to future input traffic of each connection, the number of delayed packets waiting in the buffer, and the degree of QoS provided so far.

3 is a configuration diagram of the dynamic resource manager 140 shown in FIG. 1, which includes a required band estimator 310 and a process band allocator 320.

The dynamic resource manager 140 is composed of two functional blocks as shown in FIG.

The required bandwidth estimator 310 estimates a bandwidth required to support future stable QoS for each connection, and the process bandwidth allocator 320 connects each connection within a currently available resource limit. Implement fair resource distribution.

The required band estimator 310 determines whether the QoS currently provided for the corresponding video connection is provided over or under the required level, and dynamically calculates the required band R _i (n + 1) required for the next frame time. It is obtained using Equation 6.

Where S _f is the scaling factor, PQD _i (n) is the Provided QoS Degree provided to connection i until the end of n frames, lossR _i is the packet loss rate required by the i th video connection, lossM _i is defined as the packet loss rate observed for the i th connection up to n frame times.

The required bandwidth estimation method of Equation (6) basically increases the required bandwidth according to the QoS level PQD _i (n) currently provided to the sum of the predicted traffic for the next frame time and the number of packets waiting in the current buffer. Or reduce it. If a lower level of quality is provided than the required QoS (i.e., PQD _i (n) <0), then the bandwidth required for the next frame time compensates for the prediction error and can be allocated more bands in the process band allocation in the next section. Calculate it. Conversely, if a higher level of quality is provided than the required QoS (i.e. PQD _i (n)> 0), then the required band is It is calculated smaller than the value. The S _f value is set differently according to the range of PQD _i (n) value to prevent sudden QoS deterioration due to a large reduction in required bandwidth when QoS is provided higher than the requested value.

The process band allocator 320 performs fair band allocation in terms of providing QoS for each connection in consideration of the available link capacity based on the required bandwidth estimate requested for each connection. Two class sets are defined according to the QoS level provided for each connection. The LOW class is a set of connections whose PQD _i (n) value is less than zero, and the HIGH class is a set of connections where PQD _i (n) is greater than or equal to zero.

FIG. 4 is a flowchart illustrating a process band allocation policy procedure of the process band allocator shown in FIG. 3.

First, in step S410, it is determined whether the sum of required bands estimated for each connection is larger than the link capacity.

As a result of the determination in step S410, if the sum of the estimated required bands is smaller than the link capacity, in step S420, the requested required bands are the allocated bands of each connection as they are.

As a result of the determination in step S410, if the sum of the estimated required bands is larger than the link capacity, in step S430, it is determined whether the packets in the buffers of the connections belonging to LOW can be serviced at the next frame time and can be serviced. If so, in step S440, the band is first distributed for packets that would violate the delay requirement, and the remaining available band is distributed to all connections belonging to HIGH and LOW based on the required band estimate. By using this policy, connections provided with low QoS can receive the requested QoS after the next frame time, and connections already provided above the request value use less resources for other LOW class connections. .

As a result of the determination in step S430, if the sum of the bands allocated for the respective connections in the next frame time is small compared to the total link capacity, in step S450, the remaining surplus band may be used for other applications except real time video. That is, redundant band is used for the existing Internet Best-Effort service or UBR service class of ATM. This greatly improves the utilization of the network in a real network environment.

4. Connection Control Manager

A simple connection admission control method is used to ensure that existing video does not experience QoS degradation below the requested level due to the addition of new video sources. The method used in the present invention is simple and scalable. Each node of the network accepts the connection of the new video source if the two conditions shown in Equation 7 below are satisfied.

Here, f (K) is a marginal factor for a new connection, and may be expressed as a function of K by a network operator.

The first condition checks whether the QoS provided for all existing K connections is above the desired level. The second condition checks whether the sum of the average input rates of existing connections is below the output link capacity tolerance level.

If there is a connection request of a new video source in the packet mode network, the network uses the connection control manager 110 to determine whether to accept the new connection. Since the network does not know the traffic characteristics of the new connection, the service for the new video is started or rejected based on the quality of the service provided to the existing video connection currently being serviced by Equation 7 above. . For service-initiated video, the network allocates an appropriate band to each video source every frame time to satisfy the QoS of the desired {packet loss rate, maximum packet delay}.

The band required for the next frame time is determined by the following procedure and method. The traffic predictor 120 is used to predict the traffic of the next frame time based on the input traffic so far. In addition, the QoS provided through a simple delay packet counter update process is observed, and the number of delayed packets in units of frames for each connection is measured.

The dynamic resource manager 140 fairly allocates a band necessary to provide the QoS requested by each service user at the next frame time based on the predicted traffic and the provided QoS level.

This operation is performed in network equipment such as all switches or routers that perform video service of the network every frame time.

FIG. 5 is a diagram illustrating that a QoS control apparatus of a video service using dynamic resource allocation and a method thereof according to an embodiment of the present invention can provide a desired quality of a video connection more accurately than a conventional method. .

The results of FIG. 5 show that frame time 32000 (approximately 30 minutes) at frame time 0 (start of video) when the films 'silence of sheep' and 'Talk Show' MPEG video traffic are respectively input to the router input buffer shown in FIG. Shows the 'provided QoS level (PQD)' observed.

Figure 5 (i) is the most ideal case where the PQD is 0, (ii) is the result of applying the present invention to the film 'silence of sheep', (iii) is the result of applying the prior art to the film 'silence of sheep' (Iv) is the result of applying the present invention to the movie 'Talk Show', and (v) is the result of applying the prior art to the movie 'Talk Show'.

As can be seen in FIG. 5, the present invention suggests that the service can be very closely approximated to the desired level of the user (PQD = 0, the user QoS requirement and the actual provided quality), but when using the conventional method. As a result, the QoS provided is too high or too low to meet the desired QoS requirements.

While the invention has been described above based on the preferred embodiments thereof, these embodiments are intended to illustrate rather than limit the invention. It will be apparent to those skilled in the art that various changes, modifications, or adjustments to the above embodiments can be made without departing from the spirit of the invention. Therefore, the protection scope of the present invention will be limited by the appended claims, and should be construed as including all such changes, modifications or adjustments.

As described above, according to the present invention, it is possible to provide QoS for each video service without describing MPEG video service traffic in a user's traffic parameter in a communication network structure such as ATM and Internet communication network. By using this, it is possible not only to provide reliable service quality to users when allocating resources in the network, but also to achieve efficient use of network resources.

Claims (15)

An apparatus for controlling QoS of a video service using dynamic resource allocation, A traffic predictor for measuring the incoming video traffic for each connection and predicting the traffic to arrive at the next time interval; QoS observer discarding when the connection time of the video traffic stays in the buffer exceeds the maximum packet delay; Dynamic to determine the bandwidth to allocate to the connection-specific video traffic at the next frame time according to the future input traffic of the connection-specific video traffics predicted by the traffic predictor, the number of delayed packets waiting in the buffer and the degree of QoS provided so far. Resource manager; And And a connection control manager that accepts the connection of the new video source if the QoS provided to all existing connections is above a preset level and the sum of the average input data rates of the existing connections is below the output link capacity tolerance level. QoS control device for video service using dynamic resource allocation. The method of claim 1, The QoS observer, QoS control of video service using dynamic resource allocation, which discards at the front of the buffer because the packet which is not serviced during (n + 1) frame time among the video packets currently buffered at the end of n frame exceeds the delay limit. Device. The method according to claim 1 or 2, The QoS observer, Video using dynamic resource allocation, characterized in that the number of packets waiting for each delay time is determined using the number of input packets, dynamically allocated bandwidth, and delay packets of the corresponding frame time without calculating the delay time for each input packet. QoS control device of the service. The method of claim 1, The dynamic resource manager, A required bandwidth estimating unit for determining a required bandwidth by using the QoS level currently provided to the sum of the predicted traffic for the next frame time and the number of packets waiting in the current buffer; And And a process band divider configured to receive the requested bandwidth estimation value for each video traffic per connection and compare the link capacity with the link capacity to perform bandwidth distribution. The method of claim 4, wherein The required band estimation unit, In the video service using the dynamic resource allocation, it is determined whether the QoS provided to the associated video traffic is provided more than the required level, and the required bandwidth for the next frame time is dynamically obtained by the following Equation 1. QoS control device. [Equation 1] R _i (n + 1) is the required band to be obtained, S _f is the scaling factor, and PQD _i (n) is the Provided QoS Degree provided to connection i until the end of n frames. The method of claim 4, wherein The process band distribution unit, If the sum of estimated bandwidths for connection-specific video traffic is less than the link capacity, the requested bandwidth is still the allocated bandwidth of each connection; if greater than the link capacity, the packets in the buffer are not serviced at the next frame time. And a first bandwidth is allocated to packets that violate the delay requirement, and the remaining available band is distributed to all the connections based on the required band estimate. The method of claim 1, The connection control manager, QoS control device for a video service using dynamic resource allocation, characterized in that it starts or rejects a service for a new video based on the quality of service provided to each existing video connection. [Equation 2] lossR _i is the packet loss rate required by the i th video connection, lossM _i is the packet loss rate observed on the i th connection up to n frame times, X _i (m) is the number of input packets of connection i at frame time m, L _i (m) is the number of double lost packets, and f (K) is the marginal factor for new connections, defined as a function of K by the network operator. A QoS control method of a video service using dynamic resource allocation, A traffic prediction step of predicting traffic arriving at a next time interval by measuring video traffic for each connection input; A QoS observation step of monitoring a QoS degree of currently present traffic and discarding when the connection time of the video traffic stays in a buffer exceeds a maximum packet delay; Dynamic to determine the bandwidth to allocate to the connection-specific video traffic at the next frame time according to the future input traffic of the connection-specific video traffic predicted in the traffic prediction step, the number of delay packets waiting in the buffer and the QoS degree provided so far. Resource management step; And If the QoS provided for all existing connections is above a predetermined level and the sum of the average input data rates of the existing connections is below the output link capacity tolerance level, then including a connection control management step of accepting the connection of the new video source. QoS control method of video service using dynamic resource allocation. The method of claim 8, The QoS observation step, QoS control of video service using dynamic resource allocation, which discards at the front of the buffer because the unserviced packet during (n + 1) frame time among the video packets currently buffered at the end of n frame exceeds the delay limit. Way. The method according to claim 8 or 9, The QoS observation step, Video using dynamic resource allocation, characterized in that the number of packets waiting for each delay time is determined using the number of input packets, dynamically allocated bandwidth, and delay packets of the corresponding frame time without calculating the delay time for each input packet. QoS control method of service. The method of claim 8, The dynamic resource management step, Determining a required bandwidth by using the QoS level currently provided to the sum of the predicted traffic for the next frame time and the number of packets waiting in the current buffer; And And a sub-step of receiving the required bandwidth estimation value for each video traffic for each connection and comparing the link capacity with the link capacity to perform bandwidth allocation. The method of claim 11, The sub-step of determining the required bandwidth, In the video service using the dynamic resource allocation, it is determined whether the QoS provided to the associated video traffic is provided more than the required level, and dynamically obtains the required bandwidth for the next frame time by the following Equation 3. QoS control method. [Equation 3] R _i (n + 1) is the required band to be obtained, S _f is the scaling factor, and PQD _i (n) is the Provided QoS Degree provided to connection i until the end of n frames. The method of claim 11, The sub-step of performing the band distribution, A sub-sub step of designating the requested required band as the allocated band of each connection as it is if the sum of the estimated required bands for the video traffic per connection is smaller than the link capacity; If it is larger than the link capacity, it allocates priority bands to packets in the buffer that would otherwise violate delay requirements if not serviced at the next frame time, and distributes the remaining available band to all connections based on required bandwidth estimates. QoS control method of video service using dynamic resource allocation, characterized in that it comprises a sub-sub step. The method of claim 8, The connection control management step, QoS control method of a video service using dynamic resource allocation, characterized in that it starts or rejects a service for a new video based on the quality of service provided to each existing video connection. [Equation 4] lossR _i is the packet loss rate required by the i th video connection, lossM _i is the packet loss rate observed on the i th connection up to n frame times, X _i (m) is the number of input packets of connection i at frame time m, L _i (m) is the number of double lost packets, and f (K) is the marginal factor for new connections, defined as a function of K by the network operator. On your computer, A traffic prediction step of predicting traffic arriving at a next time interval by measuring video traffic for each connection input; A QoS observation step of discarding when the connection time of each video traffic stays in a buffer exceeds a maximum packet delay; Dynamic to determine the bandwidth to allocate to the connection-specific video traffic at the next frame time according to the future input traffic of the connection-specific video traffic predicted in the traffic prediction step, the number of delay packets waiting in the buffer, and the QoS degree provided so far. Resource management step; And If the QoS provided for all existing connections is above a predetermined level and the sum of the average input data rates of the existing connections is below the output link capacity tolerance level, a connection control management step of accepting connections of the new video source, The dynamic resource management step, The required bandwidth is determined by using the QoS level currently provided to the sum of the predicted traffic for the next frame time and the number of packets waiting in the current buffer, and the required bandwidth estimation value for each video traffic for each connection is input and compared with the link capacity. And a computer-readable recording medium having recorded thereon a program capable of executing band allocation.