KR20100096599A - Packet scheduling apparatus and method - Google Patents

Abstract

PURPOSE: A packet scheduling device and a method of the same are provided to leave an imaginary lower frame in the inner side of the frame and arrange packets in an imaginary lower frame and transmit packets, thereby providing flexible delay performance without synchronizing frame between nodes of the same size frame. CONSTITUTION: A scheduler(120) schedules received packets. A multi queue device(130) includes a plurality of lower frame queues(131~13n). The multi queue device respectively stores packets in a plurality of lower frame queue. An output interface(140) transmits the stored packets of a first lower frame queue. If all the packets are transmitted within a lower frame period, the output interface services stored packets of best-effort queue during remaining time.

Description

Packet Scheduling Apparatus and Method

The present invention relates to a packet scheduling algorithm that has been studied and proposed from ATM to packet switched network.

In recent years, the rapidly increasing real-time distributed media application services (ex., Grid, e-science, remote collaboration), unlike the one-way multimedia application service provided around the server / client model in the past, real-time interaction and multi-dimensional interaction between users It is evolving into an interactive multimedia application service based on high quality media (5.1ch audio, HD video, etc.) sources. Since they usually utilize multiple high quality / real-time media streams simultaneously, it is essential to guarantee bandwidth (Mbps to Gbps) and to ensure stable performance against network delay and jitter.

Guaranteeing Quality of Service (QoS) (including bandwidth, delay and jitter) in a comprehensive sense for real-time distributed application services is one of the key issues of the future Internet. The existing Internet was designed with the highest priority in providing connectivity to distributed nodes, and application services have also been able to use as many networking resources as they need without understanding network infrastructure. Under this structure, several methods have been proposed at the application / network layer to accommodate the increasing demands on the quality of service of users. In particular, in networks, call admission control, traffic access control, packet scheduling, buffer management, flow and congestion control, QoS routing, etc. This has been studied as a way to guarantee the quality of service.

Among them, packet scheduling is one of the representative methods to guarantee the network performance (bandwidth, delay, jitter) required by each connection, and many studies and proposals have been made from ATM to packet switched network.

They have been proposed mainly to guarantee the performance (bandwidth, delay, jitter) for each connection. Recently, due to the activation of real-time distributed media application services, the guarantees in terms of time such as network delay and delay displacement are quantitative ( As well as guaranteeing bandwidth). Indeed, Fair Queuing (FQ) and its variants (eg, WFQ, WF2Q, SCFQ, SFQ) have been developed primarily to guarantee the bandwidth requested by each connection and to provide an equal level of service, but Delay-EDD (Early Due) Scheduling schemes such as Date, Jitter-EDD, Stop-and-Go (SG), and Hierarchical Round Robin (HRR) focus on ensuring the performance of network latency or jitter by using allocated bandwidth rather than providing fair services. Is fitting. However, of these methods, such as Delay-EDD and Jitter-EDD, operate on the basis of computed sorted priority-based, which provides relatively good performance against delay and jitter, whereas N connections It has a minimum O (logN) complexity for. This is not suitable in high-speed network environments where packet processing must be very fast. On the other hand, frame-based scheduling methods such as SG and HRR have the advantage of being very simple and high-speed packet processing because the packets are processed in the order of arrival for each time interval of a predetermined length.

The time frame-based packet processing scheme has two problems. First, since a packet is transmitted based on a fixed time frame, there is a causal relationship between the delay performance and the number of connections that can be allocated to the frame according to the size of the frame. Increasing the size of a frame increases the number of connections that can be allocated, while reducing latency. Second, since all network nodes must process packets based on the same sized synchronized time frame, it is difficult to apply to large networks. Multiple-framing Stop-and-Go and HRR have attempted to solve this problem by using a multi-stage frame structure. However, because the traffic specification is based on the frame structure and its size based on a fixed time origin, To ensure network end-to-end, you still need to know all the information about the frames configured on the I / O link. In addition, since the delay performance for each connection is fixed according to the size of the frame associated with the connection, the call acceptance step becomes complicated and inflexible, which is undesirable in terms of actual implementation and service provision.

In summary, since most sorted-priority based scheduling algorithms using indexing such as time-stamp have a minimum O (logN) complexity for N connections, It is not suitable for high speed network environment. In contrast, the frame-based packet scheduling scheme has a structural problem that is difficult to apply to a large network because it requires synchronization between frames instead of improving complexity.

In order to solve the above problems, an object of the present invention is to provide a packet scheduling method and apparatus having a low complexity that is suitable for a high-speed network environment and guarantees the delay performance required by a real-time distributed media application service.

According to an embodiment of the present invention, a method of scheduling a packet between an input link and an output link includes dividing a transmission time axis of an output link into a plurality of frames each including at least two subframes; When packets arrive, allocating the packets to subframes of the plurality of frames, and transmitting the packets in the respective allocated subframes.

Herein, the allocating step includes determining whether a packet is from a connection established in advance on the output link when the packet arrives, and when the packet is from the connection, an arrival frame index indicating a frame position on the output link. Converting the packet to a packet, calculating a virtual service frame of the packet, comparing the virtual service frame index of the packet with the arrival frame index, and determining a service frame index according to the result; And transmitting the calculated service frame.

The method of claim 2, wherein the determining of the service frame index comprises: if the packet is the first packet arrived for the connection, determining the next lower frame of the packet's arrival frame index as a virtual service frame index. Can be.

The determining of the service frame index may include determining the calculated virtual service frame index as the service frame index when the virtual service frame index of the packet lags behind the arrival service frame index in time.

Here, in the determining of the service frame, if the calculated virtual service frame index is temporally ahead of an arrival frame index, the next lower frame value of the lower frame value of the arrival frame index is a lower frame value of the service frame index. It may be a step of setting.

Here, when the lower frame value of the virtual service frame index exceeds the maximum lower frame value, the first lower frame value of the next frame of the virtual service frame may be a lower frame value of the service frame index.

Here, the lower frame of the virtual service frame may be calculated by the following equation.

는 기준 하위 프레임의 위치이고,

는 하위 프레임 내에서 전송 가능한 패킷 수이고,

은 하나의 프레임 내의 최대 하위 프레임의 수이고,

는 해당 프레임에 도착된 패킷의 수이다. From here,

Is the position of the base subframe,

Is the number of packets that can be transmitted within the lower frame,

Is the maximum number of subframes within one frame,

Is the number of packets arriving in the frame.

The method of scheduling a packet between the input link and the output link may further include increasing a q value indicating the number of packets arriving in the corresponding frame after the service step by one.

The method of scheduling a packet between the input link and the output link may include, if the packet is not from the connection, storing the received packet as a low best-effort traffic and storing it separately, and later if the output link is idle. The method may further include sequentially transmitting the stored packets.

Further, an apparatus for scheduling a packet connected between an input link and an output link according to an embodiment of the present invention includes an input interface connected to the input link to receive, buffer and output packets from the input link, and an output link. A scheduler for allocating the packets to subframes of the plurality of frames when packets arrive in a state in which a transmission time axis of the subframe is divided into a plurality of frames each including at least two subframes, and packets for each subframe. A multiple buffer device for queuing, and an output interface for transmitting the packets from the multiple buffer device.

Herein, when the packet arrives, the scheduler determines whether the packet is from a connection established in advance on the output link, and if the packet is from the connection, converts the arrival time of the packet into an arrival frame index indicating a frame position on the output link. Calculate a virtual service frame of the packet, compare the virtual service frame index of the packet with the arrival frame index, determine a service frame index according to the result, and match the packet to the calculated service frame. Can be passed to the buffer device.

Here, if the packet arrives for the first time for the connection, the scheduler may determine the next lower frame of the packet arrival frame index as the virtual service frame index.

Here, when the virtual service frame index of the packet lags behind the arrival service frame index in time, the scheduler may determine the calculated virtual service frame index as the service frame index.

Here, the scheduler may set the next lower frame value of the lower frame value of the arrival frame index as the lower frame value of the service frame index when the calculated virtual service frame index is temporally ahead of the arrival frame index.

Here, when the lower frame value of the virtual service frame index exceeds the maximum lower frame value, the first lower frame value of the next frame of the virtual service frame may be set to the lower frame value of the service frame index.

Here, the lower frame of the virtual service frame may be calculated by the following equation.

는 기준 하위 프레임의 위치이고, 는 하위 프레임 내에서 전송 가능한 패킷 수이고,

은 하나의 프레임 내의 최대 하위 프레임의 수이고,

는 해당 프레임에 도착된 패킷의 수이다. From here,

Is the position of the base subframe, Is the number of packets that can be transmitted within the lower frame,

Is the maximum number of subframes within one frame,

Is the number of packets arriving in the frame.

Here, the scheduler may increase the number of packets arriving in the corresponding frame by one after the service step.

Here, the apparatus for scheduling the packet further includes a separate queue for storing a low best-effort packet, the scheduler is a low-best best-effort packet arrived if the packet is not from the connection The traffic can be stored as a separate queue and stored in a separate queue. If the output link is idle, the stored packets can be sequentially transmitted.

Here, the multi-queue device may include a plurality of subframe queues for queuing packets respectively corresponding to the subframes.

The present invention provides a low complexity packet scheduling method and apparatus for guaranteeing the delay performance required by a real-time distributed media application service and suitable for a high speed network environment. Although the present invention is a frame-based scheduling scheme, virtual subframes are placed inside a frame, and packets to be transmitted in each frame are distributed and transmitted in virtual subframes without frame synchronization between nodes having the same frame size. It provides flexible delay performance and effectively solves the coupling problem of delay performance and the number of assignable connections. That is, since the number of lower frames in the frame can be changed variably, the delay performance can be independently adjusted by adjusting the number of subframes even if the size of the frame is increased. Therefore, the proposed method can provide predictable and differentiated delay performance in the call acceptance process of the service, and can guarantee this from end to end. This performance-based packet transmission technology can be actively utilized in the existing mission / time-critical application service fields (IPTV, telemedicine, remote collaboration, etc.) and is expected to greatly contribute to the development of new application services and related industries. .

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. Like reference numerals are used for like elements in describing each drawing.

The terms first, second, A, B, etc. may be used to describe various elements, but the elements should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component. And / or < / RTI > includes any combination of a plurality of related listed items or any of a plurality of related listed items.

When a component is referred to as being "connected" or "connected" to another component, it may be directly connected to or connected to that other component, but it may be understood that other components may be present in between. Should be. On the other hand, when a component is said to be "directly connected" or "directly connected" to another component, it should be understood that there is no other component in between.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in the commonly used dictionaries should be construed as having meanings consistent with the meanings in the context of the related art and shall not be construed in ideal or excessively formal meanings unless expressly defined in this application. Do not.

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

The present invention can guarantee the transmission quality of various application services in a high-speed packet switched network environment and effectively solve the problems of the existing scheduling scheme through a two-layer frame structure having virtual subframes in the frame structure and a packet scheduling algorithm using the same. Provide a way. Packet transmission is very simple like the existing frame-based algorithms, but by distributing the packets in the inner subframes, it can effectively solve the network synchronization problem and provide better delay performance.

1 is a diagram illustrating a general packet switched network.

Referring to FIG. 1, two devices 2, 4 transmit packets to node A 10. When Node A 10 receives packets from two devices 2 and 4, these packets are scheduled for transmission and sent to another node, such as Node B 20 or Node C 30. Node A 10 may be a router or a switch device. The present invention can be implemented in the nodes 10, 20, 30 relaying the packet in this way.

One node can theoretically guarantee local performance if each connection follows the traffic specification initially established. Therefore, traffic that needs to guarantee performance must declare their desired traffic specification during connection establishment. Using this specification, a node can compare the amount of resources available on its output link with the amount of resources requested to accept or reject new connections. Of course, this performance guarantee must be done end-to-end, so this connection establishment must succeed on all nodes that exist in the established path.

However, the present invention does not deal with the establishment of the path, but deals with the scheduling of packets arriving from the accepted connections through call admission control at the output link of each node on the established path.

In the present invention, it is assumed that the traffic specification of each connection follows the (r, T) -smooth proposed by Stop-and-go. This means that up to r? T (bits) of traffic occur within any time interval T (sec) (i.e., frame). Where r is the bandwidth allocated for the connection.

In the present invention, it is assumed that the packet size is fixed (e.g., L bits), and for any connection having an allocated bandwidth of r, the number of packets that can be transmitted in one frame may be calculated as (r? T) / L. We define this as Q.

Every switch (or router) in the path through which a connection passes stores information about each accepted connection. The information on each connection includes the number Q of packets that can be transmitted in one frame, the number q of packets serviced in the current frame, and the ID of the connection. Immediately after the connection is established, the q value is set to 0. The present invention places virtual subframes inside the frame structure to effectively solve the coupling problem between the transmission frame size and the delay performance of the conventional frame-based transmission scheme. In addition, the present invention provides a layer 2 frame-based packet scheduling algorithm for distributing and transmitting packets to be transmitted in a corresponding frame to a virtual lower frame and an apparatus therefor.

2 is a block diagram of an apparatus for executing a layer 2 frame-based packet scheduling method according to the present invention.

Referring to FIG. 2, the layer 2 frame-based packet scheduling apparatus includes an input interface 110, a scheduler 120, a multiple queue apparatus 130, and an output interface 140.

The input interface 110 may be connected to the input link to receive and buffer packets provided from the plurality of connections. In addition, the input interface 110 provides the scheduler 120 with packets provided from the plurality of connections.

The scheduler 120 schedules packets provided from the plurality of connections and provides them to the multiple queue device 130. Specifically, the scheduler 120 assigns the arriving packet on a frame on the output link when the packet arrives.

Here, the frame structure according to the present invention will be described with reference to FIG.

3, the frame according to the present invention has a two-layer frame structure. The two-tier frame structure consists of a frame F and n subframes f within each frame. In the present invention, in order to process the arrival time and service time of each packet under a two-layer frame structure, it is assumed that a unique number is assigned to a frame on the output link, and a lower frame of each frame has a number of 1 to n. This is expressed as {F, f}. That is, the position of the first lower frame of the first frame may be represented by {1,1}, and the position of the first lower frame of the fourth frame may be represented by {4,1}.

The number of lower frames included in one frame is variable. Accordingly, even if the frame increases, the delay performance may be adjusted by controlling the number of lower frames. The scheduler 120 thus allocates the received packet on the two-layer frame.

The operation of the scheduler will be described in detail with reference to FIG. 4.

4 is a flowchart illustrating the operation of a scheduler according to an exemplary embodiment of the present invention. As described above, it is assumed that the transmission time axis of the output link is divided into a plurality of frames each including at least two subframes.

Referring to FIG. 4, the scheduler 120 determines whether a packet has arrived in step 210. When a new packet arrives, the scheduler 120 first checks the ID of the packet in step 220 to determine whether it is from connections established on the output link. The scheduler 120 proceeds to step 240 if the arriving packet is from a connection established on the output link. In addition, the scheduler 120 considers the packet arriving in step 230 as a low best-effort traffic if it has not come from a connection established on the output link, and stores it separately before the server is idle. If there is, send it sequentially.

The scheduler 120 then converts the arrival time of the packet to the frame position on the output link in step 240. That is, the scheduler 120 converts the arrival time of the packet into an arrival frame index indicating the frame and lower frame positions on the output link.

For example, if the size of the frame is 1 sec and the number of subframes is 10, if the arrival time of the packet is 4.3 sec with respect to the transmission time axis of the output link, then the destination frame index {F = 5th (frame), f = 3rd (sub-frame)}, i.e., {5,3}.

Next, the scheduler 120 calculates the virtual service frame index of the packet in step 250 and determines the service frame index through comparison with the arrival frame index. The scheduler 120 determines whether the service frame determined in step 260 is a new frame through comparison with a reference frame index.

Here, the virtual service frame index and the service frame index indicate a position on the frame where the packet is to be serviced and a position {F, f}} on the frame being served, respectively. Immediately after the connection is established, the {F ^R , f ^R } value, which is the reference frame index of the connection, is set to 0. Here, the reference frame index is a reference value when determining a service frame index to which a packet is allocated in each connection.

If the determined service frame is a new frame, the scheduler 120 proceeds to step 270 and updates the service frame index with the reference frame index.

Hereinafter, the steps 250 to 270 will be described in detail with reference to FIG. 5.

5 is a diagram illustrating a relationship between a virtual service frame index and a service frame index with respect to an arrival frame index.

Referring to FIG. 4, when the value of Q (the number of packets that can be transmitted in one frame) of the connection i is 2, it is assumed that four consecutive packets having the arrival frame index arrive sequentially. And, when a packet first arrives at the switch or router from each connection, the reference frame index is set to {F ^R = 0, f ^R = 0}.

First, the scheduler 120 converts the arrival time of the packet to the frame position on the output link when the packet arrives from the connection i. If the arrival frame index of the packet is {F = 5, f = 1}, the arrival frame index of the packet F (= 5) of Since the value is larger than the F ^R (= 0) value of the reference frame index of the connection, the scheduler 120 may know that a packet has arrived in a new frame.

Then, the scheduler 120 allocates and transmits the next lower frame of the arrival frame index on the received output link as the service frame index of the packet in the case of the packet first arriving from the connection. . For example, if the arrival frame index of the packet is {5,1}, the scheduler 120 may have the service frame index of the packet to be {5,2}. Thus, for the first packet arriving for the connection, the scheduler 120 determines the next lower frame of the packet's arrival frame index as the virtual service frame index, which becomes the service frame index. The scheduler 120 increments q by 1, which indicates the number of packets serviced in the current frame.

The service frame position of this packet is a reference frame index in the frame. Therefore, the reference frame index is {F ^R = 5, f ^R = 2}. This reference frame index can be updated each time a packet is serviced in a new frame.

On the other hand, when a packet arrives, a case where the calculated virtual service frame index of the packet is ahead of the arrival frame index, and the same case, falls behind.

For example, if the virtual service frame index of the packet lags behind the arrival service frame index in time, the calculated virtual service frame index becomes the actual service frame index, which will be described in detail below.

After a packet with a destination frame index of {F = 5, f = 1}, if a packet with a destination frame index of {F = 5, f = 2} arrives, the F (= 5) value of the packet's arrival frame index is Equivalent to the reference frame index F ^R (= 5) of the concatenation. As a result, the scheduler knows that a packet has arrived in a frame that has already started.

As described above, since one packet is already served in the fifth frame, the q value is 1 and the reference frame index is {F ^R = 5, f ^R = 2}. The scheduler 120 checks the q value of the connection and calculates the virtual service frame index for which this packet will be serviced.

에 기초하여 패킷이 전송될 하위 프레임의 위치를 계산한다. 다시 말해, 스케줄러(120)는 패킷이 서비스될 가상 서비스 프레임 인덱스 즉, 가상 서비스 프레임 인덱스중 하위 프레임값을 다음과 같은 수학식 1에 따라 계산한다. Specifically, the scheduler 120 distributes the packets arriving from the connection to the lower frames in the frame and transmits them to the reference frame index of the first packet arriving in the frame.

Calculate the position of the lower frame to which the packet is to be transmitted. In other words, the scheduler 120 calculates a lower frame value among the virtual service frame indexes to which the packet is to be serviced, that is, the virtual service frame index, according to Equation 1 below.

는 기준 하위 프레임 값이고,

는 하위 프레임 내에서 전송 가능한 패킷 수이고,

은 하나의 프레임 내의 하위 프레임의 수이고,

는 해당 프레임에 도착된 패킷의 수이다.From here,

Is the base subframe value,

Is the number of packets that can be transmitted within the lower frame,

Is the number of subframes within one frame,

Is the number of packets arriving in the frame.

(=4) 이 되어 해당 패킷이 서비스될 가상 서비스 프레임 인덱스는 {F =5, f=4}가 된다. 이 패킷은 도착 프레임 인덱스가 {F =5, f=2}이고, 가상 서비스 프레임 인덱스가 {F =5, f=4}이므로, 결과적으로 서비스될 시점보다 해당 패킷이 먼저 도착한 것이 되므로, 서비스 프레임 인덱스는 {F =5, f=4}가 된다. 그리고, 스케줄러(120)는 현재 프레임 내에 서비스된 패킷의 수를 나타내는 q를 1만큼 증가시킨다.In this case, the lower frame value

(= 4), and the virtual service frame index to which the packet is to be served is {F = 5, f = 4}. Since the packet has a destination frame index of {F = 5, f = 2} and a virtual service frame index of {F = 5, f = 4}, the packet arrives earlier than the point of service. The index is {F = 5, f = 4}. The scheduler 120 increments q by 1, which indicates the number of packets serviced in the current frame.

In another example, as described above, when the calculated virtual service frame index becomes the service frame index, if the lower frame value of the service frame index exceeds the maximum lower frame value, the scheduler 120 may lower the first lower frame of the new frame. The frame value is set to a lower frame value of the service frame index, which will be described in detail below.

When a packet with an arrival frame index of {F = 5, f = 4} arrives, the F (= 5) value of the packet's arrival frame index is equal to the reference frame index F ^R (= 5) of the corresponding connection. This allows the scheduler to know that a packet has arrived in a frame that has already started.

(=6)이 된다. 따라서, 해당 패킷의 서비스 프레임 인덱스는 {F=5, f=6}이 된다. 그러나, 6번째 하위 프레임은 존재하지 않으므로, 이 패킷은 6번째 프레임의 첫번째 하위 프레임에서 서비스되어야 한다. 따라서, 서비스 프레임 인덱스는 {F=6, f=1}이 된다. 이 경우, 패킷이 새로운 프레임에 서비스되기 때문에, 기준 프레임 인덱스는 {F^R =6, f^R=1}로 업데이트 된다. 또한, 스케줄러(120)는 6번째 프레임에서 하나의 패킷이 서비스되었으므로, q값을 1로 초기화한다. 또 다른 예에서, 계산된 가상 서비스 프레임 인덱스가 도착 프레임 인덱스 보다 시간적으로 앞서면, 스케줄러(120)는 도착 프레임 인덱스의 하위 프레임 값의 바로 다음 하위 프레임 값을 서비스 프레임 인덱스의 하위 프레임 값으로 설정하는데, 이하 상세히 설명한다.The scheduler 120 checks the q value of the connection. In this example, the value of q is 2 since two packets are already served in the fifth frame. The lower frame value of the virtual service frame of that packet is

(= 6). Therefore, the service frame index of the packet is {F = 5, f = 6}. However, since the sixth subframe does not exist, this packet must be serviced in the first subframe of the sixth frame. Therefore, the service frame index is {F = 6, f = 1}. In this case, because the packet is serviced in a new frame, the reference frame index is updated to {F ^R = 6, f ^R = 1}. In addition, the scheduler 120 initializes q to 1 since one packet was serviced in the sixth frame. In another example, if the calculated virtual service frame index is temporally ahead of the arrival frame index, the scheduler 120 sets the next lower frame value of the lower frame value of the arrival frame index to the lower frame value of the service frame index. It will be described in detail below.

When a packet with an arrival frame index of {F = 6, f = 4} arrives, the F (= 6) value of the packet's arrival frame index is equal to the reference frame index F ^R (= 6) of the corresponding connection. As a result, the scheduler knows that a packet has arrived in a frame that has already started.

In this case, the scheduler 120 checks the q value of the connection and calculates the virtual service frame index of the packet. As described above, since one packet has already been serviced in the sixth frame, the q value is one.

(=3)이 된다. 따라서, 해당 패킷의 가상 서비스 프레임 인덱스는 {F=6, f=3}이 된다. 그런데, 해당 패킷의 도착 프레임 인덱스가 {F=6, f=4}이므로, 해당 패킷은 서비스될 시점보다 늦게 도착한 것이다. 이 경우 스케줄러(120)는 해당 패킷의 하위 서비스 프레임값을 도착 프레임 인덱스의 하위 프레임 값의 바로 다음 값으로 설정한다. 그에 따라, 이 패킷의 서비스 프레임 인덱스는 {F=7,f=1}이 된다. 또한, 패킷이 새로운 프레임에 서비스되기 때문에, 기준 프레임 인덱스는 {F^R =7, f^R=1}로 업데이트 된다. 또한, 스케줄러(120)는 7번째 프레임에서 하나의 패킷이 서비스되었으므로, q값을 1로 초기화한다.The lower frame value of the virtual service frame of that packet is

(= 3). Therefore, the virtual service frame index of the packet becomes {F = 6, f = 3}. However, since the arrival frame index of the packet is {F = 6, f = 4}, the packet arrives later than the time of service. In this case, the scheduler 120 sets the lower service frame value of the packet to the next value of the lower frame value of the arrival frame index. Accordingly, the service frame index of this packet is {F = 7, f = 1}. Also, because the packet is serviced in a new frame, the reference frame index is updated to {F ^R = 7, f ^R = 1}. In addition, the scheduler 120 initializes q to 1 since one packet was serviced in the seventh frame.

Accordingly, when the packet arrives, the scheduler 120 determines whether the virtual service frame of the packet is ahead of, equal to, or behind the arrival frame index, and determines the actual service frame according to the result. The operation of can be programmed as shown in Table 1 below.

The scheduler 120 then services the packet in the determined service frame at step 280.

The scheduler 120 allocates the received packet on the two-layer frame according to the calculated service frame and outputs the packet to the multi-queue device 130. The multiple queue device 130 stores the packets to be serviced in the lower frame of the frame in the same queue.

If you put a single queue on the output link and want to send packets that have been determined by the service frame sequentially, every time a new packet comes in, all packets in the queue need to be sorted in order of service frame again. It is not suitable for environments that require packet processing.

The present invention provides a structure in which 2n (n is the number of subframes) of multiple queues correspond to service frame indexes, and packets corresponding to the service frame indexes are stored in each queue and sequentially serviced.

For example, initially, from the first subframe queue to the second nth subframe queue, {1,1}, {1,2},... , {1, n}, {2,1}, {2,2},... , The service frame index of {2, n} corresponds. Accordingly, packets having a service frame index of {1,1} are stored in the first subframe queue, and packets having a service frame index of {2, n} are stored in a second subframe queue, respectively.

Specifically, the multi-queue device 130 includes a plurality of subframe queues 131 to 13n. The multiple queue apparatus 130 stores packets to be serviced in the future in the plurality of subframe queues 131 to 13n for each subframe, and thus has a complexity of O (1).

In this embodiment, the multi-queue device 130 has a multi-queue structure for storing packets allocated to up to 2n consecutive subframes. Since the traffic specification of each connection follows the (r, T) -smooth characteristic, no packets more than 2 · r · T arrive in one frame period. Therefore, when using 2n multiple queues while traversing, there is no problem handling input traffic. The service is sequentially performed for each queue, and the service continues until all the packets in the queue are serviced. Here, since the scheduled packet is stored without considering the number of packets already in the queue of the subframe, the number of packets smaller or larger than the amount of packets that can be processed within the T / n time is included in each subframe queue. This can go in.

In the present invention, in order to prevent each packet from being serviced earlier than the position of the subframe to be serviced, when the transmission is started, the output interface 140 transmits the packets stored in the first subframe queue and the packet is within the corresponding subframe period. If all are transmitted, the packets stored in the best-effort queue 150 are serviced for the remaining time. If there is no time left, the output interface 140 immediately starts serving the packet stored in the next lower frame queue.

That is, the output interface 140 is configured to service all the packets stored in the queue while switching each queue at a minimum T / n interval.

The lower frame queue in which transmission is completed is updated with the corresponding service frame index. For example, when all the packets stored in the first lower frame queue are serviced, the corresponding service frame index is updated to {3,1} and the corresponding packets start to be stored again.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the present invention as defined by the following claims It can be understood that

1 is a diagram illustrating a general packet switched network.

2 is a block diagram of an apparatus for executing a layer 2 frame-based packet scheduling method according to the present invention.

3 is a view showing a frame structure according to the present invention.

4 is a flowchart illustrating the operation of a scheduler according to an exemplary embodiment of the present invention.

5 is a diagram illustrating a relationship between a virtual service frame index and a service frame index with respect to an arrival frame index.

<Explanation of symbols for the main parts of the drawings>

110: input interface

120: scheduler

130: multiple queue device

140: output interface

Claims (19)

A method of scheduling a packet between an input link and an output link, Dividing the transmission time axis of the output link into a plurality of frames, each comprising at least two subframes, When packets arrive, allocating the packets to subframes of the plurality of frames; And transmitting the packets in each of the allocated lower frames. The method of claim 1, The allocation step, Determining if the packet is from a connection previously established on the output link; Converting the arrival time of the packet to an arrival frame index indicating a frame position on an output link if the packet is from the connection; Calculating a virtual service frame of the packet; Comparing the virtual service frame index of the packet with the arrival frame index and determining a service frame index according to the result; Transmitting the packet in the calculated service frame. The method of claim 2, Determining the service frame index, If the packet is the first packet arrived for the connection, determining the next lower frame of the packet arrival frame index as a virtual service frame index. The method of claim 2, Determining the service frame index, If the virtual service frame index of the packet lags behind the arrival service frame index in time, determining the calculated virtual service frame index as a service frame index. The method of claim 2, Determining the service frame, If the calculated virtual service frame index is earlier in time than the arrival frame index, setting the next lower frame value of the lower frame value of the arrival frame index to the lower frame value of the service frame index. . The method according to claim 4 or 5, If the lower frame value of the virtual service frame index exceeds the maximum lower frame value, the first lower frame value of the next frame of the virtual service frame is set to the lower frame value of the service frame index. The method of claim 2, The lower frame of the virtual service frame is calculated by the following equation.

From here,

Is the position of the base subframe,

Is the number of packets that can be transmitted within the lower frame,

Is the maximum number of subframes within one frame,

Is the number of packets arriving in the frame. The method of claim 7, wherein And increasing the q value representing the number of packets arriving in the corresponding frame by one after the service step. The method of claim 1, If it is not a packet from the connection, storing the received packet as the best best-effort traffic and storing it separately; And subsequently transmitting the stored packets sequentially if the output link is idle. An apparatus for scheduling a packet connected between an input link and an output link, the apparatus comprising: An input interface coupled to the input link to receive, buffer, and output packets from the input link; A scheduler for allocating packets to subframes of the plurality of frames when packets arrive in a state in which a transmission time axis of an output link is divided into a plurality of frames each including at least two subframes; A multiple buffer device for queuing packets for each of the lower frames; And an output interface for transmitting the packets from the multiple buffer device. The method of claim 10, When the packet arrives, the scheduler determines whether the packet is from a connection established in advance on the output link. If the packet is from the connection, the scheduler converts the arrival time of the packet into an arrival frame index indicating a frame position on the output link. Calculates a virtual service frame of the packet, compares the packet's virtual service frame index with the arrival frame index, and determines a service frame index according to the result, and transmits the packet to the multi-buffer device corresponding to the calculated service frame. Packet scheduling apparatus characterized in that the transfer. The method of claim 11, And the scheduler determines the next lower frame of the packet arrival frame index as the virtual service frame index if the packet is the first packet arrived for the connection. The method of claim 11, And if the virtual service frame index of the packet lags behind the arrival service frame index in time, the scheduler determines the calculated virtual service frame index as a service frame index. The method of claim 11, The scheduler sets the next lower frame value of the lower frame value of the arrival frame index as the lower frame value of the service frame index when the calculated virtual service frame index is ahead of the arrival frame index. Device. The method according to claim 13 or 14, And when the lower frame value of the virtual service frame index exceeds the maximum lower frame value, the first lower frame value of the next frame of the virtual service frame is set to the lower frame value of the service frame index. The method of claim 11, And a lower frame of the virtual service frame is calculated by the following equation.

From here,

Is the position of the base subframe,

Is the number of packets that can be transmitted within the lower frame,

Is the maximum number of subframes within one frame,

Is the number of packets arriving in the frame. The method of claim 16, And the scheduler increases the number of packets arriving in the corresponding frame by one after the service step. The method of claim 10, Further includes a separate queue to store low best-effort packets, If the scheduler is not a packet from the connection, the scheduler considers the received packet as a low best-effort traffic and stores it in the separate queue. Packet scheduling apparatus characterized in that. The method of claim 10, The multi-queue device comprises a plurality of sub-frame queues for queuing packets respectively corresponding to the sub-frames.

Images