KR102391804B1 - Optimization method of FQ-CoDel parameter for network congestion control - Google Patents

Abstract

The present invention is to provide a method for optimizing a parameter of an FQ-CoDel algorithm capable of adaptively optimizing an FQ-CoDel parameter to ensure QoS of a service and control a network congestion in a military tactical environment wherein various application services exist. In the method for optimizing the parameter of the FQ-CoDel algorithm, the present invention comprises: a first step of designing, by the FQ-CoDel algorithm, a level priority of x number, a service class of y number, and a dynamic flow of z number, calculating a target delay, in units of service class, to determine an in-queue deletion policy of the FQ-CoDel algorithm and to optimize three parameters of a target delay for which is an allowable waiting delay time, an interval rate that determines a frequency for which the policy is updated, and a quantum that determines a weight per flow, and calculating an interval rate in flow unit; and a second step of calculating the quantum in flow unit with different purposes for each priority.

Description

FQ-CoDel algorithm parameter optimization method {Optimization method of FQ-CoDel parameter for network congestion control}

The present invention relates to a control method for improving network performance while ensuring quality for each service in a military-operated communication network, and more particularly, to ensure Quality of Service (QoS) of individual application services and to manage network congestion control It relates to a parameter optimization method of the FQ-CoDel algorithm capable of adaptively optimizing FQ-CoDel (Flow Queue-Controlled Delay) parameters that combine queue management and scheduler techniques.

The content described in this section merely provides background information for the present embodiment and does not constitute the prior art.

The tactical communication network used in the military operation environment needs to classify the flow for each application service for the purpose of achieving military priority according to the importance of information in operational operation. In the flow-based service, network congestion varies according to traffic properties and causes different queuing delays and packet loss.

Real-time multimedia data is sensitive to loss, and service quality is determined according to network conditions. In the current network system, congestion occurs frequently as user traffic increases. Network congestion causes queuing delay and packet loss, and as a service cannot be transmitted within a limited time, it affects the user's QoS (Quality of Service) guarantee.

Currently, packets entering the router are traditionally handled in a tail drop manner. The drop-off method stores as many packets as possible until there is storage space left in the queue, and when there is no storage space in the queue, packets arriving later are dropped unconditionally. If the queue fills up frequently, network congestion increases and all TCP connections are slowed down at the same time. In addition, the buffer space is unfairly allocated in the traffic flow, and performance is degraded in terms of network utilization efficiency.

Korea Patent Publication No. 10-2011184 relates to a cascade structure-based queue management apparatus and method. The queue management apparatus recalculates the current queue length based on the inner loop control unit that adjusts the current window size in consideration of the packet discard probability, the window size adjusted by the inner loop control unit, and the capacity of the input traffic, and then the recalculation It consists of an outer loop control unit that recalculates the packet discard probability based on the determined queue length and the target queue length and provides it to the inner loop control unit.

In addition, Korean Patent Publication No. 10-1831082 relates to the configuration of a system for managing real-time traffic video flow. In the patent, a node comprises a processor configured to receive a first real-time video traffic flow, wherein a state variable may be associated with a first real-time video traffic flow at the node, and wherein the state variable is associated with a second real-time video traffic flow at the node. configured to be The first real-time video traffic flow may include a plurality of packets, and each packet may include a lost packet indicator. A node is configured to update a state variable associated with the first real-time video traffic flow in the node to indicate the dropped packet to drop a first packet in the first real-time video traffic flow, based on the state variable. and update the lost packet indicator for a second packet in the real-time video traffic flow.

However, the above patents use a queue management technique that performs queue management based on a queue size independent of actual service quality or guarantees service quality based on separate feedback information.

Korea Patent Publication No. 10-2011184 (Title of the invention: cascade structure-based queue management apparatus and method) Korean Patent Publication No. 10-1831082 (Title of the invention: Quality of experience-based queue management for routers for real-time video applications)

Accordingly, it is an object of the present invention to provide a parameter optimization method of the FQ-CoDel algorithm that can adaptively optimize the FQ-CoDel parameters to ensure QoS of services and control network congestion in a military tactical environment in which various application services exist. is in

In order to solve the above technical problem, the parameter optimization method of the FQ-CoDel algorithm according to an embodiment of the present invention is a parameter optimization method of the FQ-CoDel algorithm for queue management,

The FQ-CoDel algorithm is designed with x priorities (level priority), y service classes (service class), and z flows (dynamic flow),

To determine the dequeue policy of the FQ-CoDel algorithm, the target delay, which is the allowable waiting delay, the interval rate, which determines the frequency at which the policy is updated, and the quantum that determines the weight for each flow To optimize the three parameters of

Step 1 of calculating the target delay in units of service classes and calculating the interval rate in units of flows; A second step of calculating the quantum in units of flows by different purposes for each priority is included.

Preferably, the FQ-CoDel algorithm is designed with two priorities (level priority), five service classes (service class), and 64 flows (dynamic flow).

Preferably, for the differential processing of each tactical priority, the first priority processes the guaranteed application service (Assured Voice, Assured Multimedia Conferencing) as the top priority.

Preferably, in the second priority, three application services (Short Message, Non-Assured Voice, Broadcast Video) are processed later.

Preferably, the FQ-CoDel algorithm creates a logical queue for each service flow and operates in a Deficit Weighted Round Robin (DWRR) method to individually process the queues.

Preferably, the flow corresponding to each application service operates in a Deficit Weighted Round Robin (DWRR) method, and is controlled by a PQ (Priority Queuing) method in the final stage.

Preferably, for step-by-step performance verification of the parameter optimization method of the FQ-CoDel algorithm, the total amount of packet discards, the time of queue stabilization, and the amount of outdated packets are analyzed in the first step.

Preferably, round times are compared in the second step to verify the performance of each step of the parameter optimization method of the FQ-CoDel algorithm.

In order to solve the above technical problem, the parameter optimization method of the FQ-CoDel algorithm according to an embodiment of the present invention is a parameter optimization method of the FQ-CoDel algorithm for queue management, the FQ-CoDel algorithm is designed with x level priorities, y service classes, and z dynamic flows, and the target delay, which is the allowable waiting delay time for determining the dequeue policy of the FQ-CoDel algorithm. To optimize (target delay),

Step 1 of recording a time stamp when a packet is entered into a queue, tracking the delay time at which the packet stays, and comparing it with a target delay value, which is a target delay time; a second step of calculating an allowable queue length not to exceed the target delay time using the size C of the bandwidth of the bottleneck and the target delay time; Equation (2) is applied to determine the optimal target delay value according to the network environment in the FQ-CoDel protocol in which multiple flows enter one queue, but all flows belonging to the same service class receive the same target delay value. used, and the target delay value is calculated in units of service classes.

Equation (2)

(Bottleneck bandwidth C, number of flows N _s , propagation delay time T _p , first packet arrival rate λ ₀ exceeding the minimum allowable queue length β _min )

Preferably, the optimal target delay is calculated by Equation (4).

Equation (4)

Here, t _s is the sum of the target delays of all flows in the same service class, t _min is the minimum target delay, and N _s is the number of flows in the service class s.

Preferably, the optimal target delay represents the optimal target delay of an individual flow.

Preferably, the sum of target delays of all flows within the same service class is calculated by Equation (3).

Equation (3)

Here, t _stable,s is the maximum value of the stable range of the target delay according to the network environment, t _min is the minimum target delay value, and t _max,s is the maximum target delay value in each service class.

Preferably, the parameters variable according to the service class s are N _s and t _max,s .

Preferably, in order to optimize an interval rate that determines the frequency at which a policy is updated in the FQ-CoDel algorithm, the optimal interval rate is calculated by Equation (8).

Equation (8)

Here, α _j denotes an optimal interval rate index obtained by mapping the round (r _j ) predicted in the flow j at an appropriate ratio.

Preferably, the optimal interval rate exponent (α _j ) is calculated by Equation (7).

Equation (7)

Preferably, the round r _j predicted in the flow j is calculated by Equation (6).

Equation (6)

Here, p _100,j denotes the number of packets discarded at 100 ms, which is the first packet discard time in flow j. β _opt is the allowable queue length using the optimal target delay.

Preferably, the predicted round (r _j ) in the flow j represents a relative packet loss ratio compared to β _opt in p _100,j .

Preferably, the total delay time required to process all packets in the flow is defined by Equation (9).

Equation (9)

Here, q _j is the quantum value, p _avg,j is the average packet size, and n _j is the number of packets.

Preferably, in the FQ-CoDel algorithm, an optimal quantum for determining a weight for each flow includes calculating by Equation (10).

Equation (10)

는 플로우 j에서 사용되는 최적 퀀텀값, n_j는 플로우 j에서 처리해야 되는 패킷 개수, p_avg,j 는 플로우 j에서 평균 패킷 크기, r_j은 플로우 j에서 예측되는 라운드, k는 우선 순위 k에서 플로우 수를 나타낸다.where q _j is the default quantum value used in flow j,

is the optimal quantum value used in flow j, n _j is the number of packets to be processed in flow j, p _avg,j is the average packet size in flow j, r _j is the predicted round in flow j, k is in priority k Indicates the number of flows.

은 주어진 제약조건에 맞는 최적의 퀀텀

를 이용하여 모든 패킷을 처리하는데 필요한 라운드이다.Preferably the objective function

is the optimal quantum for a given constraint

This is the round required to process all packets using

Preferably, the optimized quantum that minimizes the delay time at which the last packet is finally processed in all flows is defined by Equation (11).

Equation (11)

는 플로우 j에서 사용되는 최적 퀀텀값, n_j는 플로우 j에서 처리해야 되는 패킷 개수, p_avg,j 는 플로우 j에서 평균 패킷 크기, r_j은 플로우 j에서 예측되는 라운드를 나타낸다.where q _j is the default quantum value used in flow j,

is the optimal quantum value used in the flow j, n _j is the number of packets to be processed in the flow j, p _avg,j is the average packet size in the flow j, and r _j is the predicted round in the flow j.

은 디폴트 퀀텀 값을 이용하여 예측한 라운드 보다 작아야 하고, 라운드 간 편차를 최소화하는 최적의 퀀텀 값을 찾는다.Preferably, the objective function represents an L1 norm that can be generated in all flows, and

must be smaller than the round predicted using the default quantum value, and find the optimal quantum value that minimizes the deviation between rounds.

The parameter optimization method of the FQ-CoDel algorithm according to an embodiment of the present invention allows the maximum amount of packets to arrive within a target delay time as the packet is discarded at an appropriate time when a queue congestion situation occurs, and the queue stabilization time is determined. It has the effect of speeding up the network and stabilizing the network by reducing the waiting time for each packet. In particular, the present invention has an advantage in dynamically managing individual queues to satisfy queuing delay requirements of application services such as voice, video, and message.

In addition, the present invention has the effect of changing the weight of individual flows in order to classify information importance and military priorities for operational operation and prioritize processing, or adjust the weights of all flows to reduce delay time.

1A is a basic conceptual diagram of a CoDel protocol.
1B is an FQ-CoDel network structure diagram according to an embodiment of the present invention.
2 is an exemplary diagram of parameters used for target delay optimization according to an embodiment of the present invention.
3 is a diagram illustrating parameters used for quantum optimization according to an embodiment of the present invention.
4 is an exemplary diagram of an FQ-CoDel algorithm according to an embodiment of the present invention.
5 is a structural diagram of FQ-CoDel for performance analysis of a parameter optimization method of the FQ-CoDel algorithm according to an embodiment of the present invention.
6 is an exemplary environment diagram for performance analysis of the parameter optimization method of the FQ-CoDel algorithm according to an embodiment of the present invention.
7 is a diagram showing the results of the target delay and interval rate optimization performance analysis of the parameter optimization method of the FQ-CoDel algorithm according to an embodiment of the present invention.
8 is a characteristic diagram showing the amount of packets inside a queue compared to the present invention and the conventional method.
9 is a characteristic diagram showing the waiting time for each packet compared to the present invention and the conventional method.
10 is an exemplary diagram of a first-order service quantum optimization performance analysis result in the present invention.
11 is an exemplary diagram of a second-order service quantum optimization performance analysis result in the present invention.

Hereinafter, the embodiments disclosed in the present specification will be described in detail with reference to the accompanying drawings, but the same or similar components are assigned the same reference numbers regardless of reference numerals, and redundant description thereof will be omitted. The suffixes "part" and "group", "module" and "part", "unit" and "part", "device" and "system", "terminal" and "node" for components used in the description below. and "digital walkie-talkie" are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves.

In addition, in describing the embodiments disclosed in the present specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in this specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical idea disclosed herein is not limited by the accompanying drawings, and all changes included in the spirit and scope of the present invention , should be understood to include equivalents or substitutes.

Terms including an ordinal number such as 1st, 2nd, etc. may be used to describe various elements, but the elements are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

When an element is referred to as being “connected” or “connected” to another element, it is understood that it may be directly connected or connected to the other element, but other elements may exist in between. it should be On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that the other element does not exist in the middle.

The singular expression includes the plural expression unless the context clearly dictates otherwise.

In the present application, terms such as "comprises" or "have" are intended to designate that a feature, number, step, operation, component, part, or a combination thereof described in the specification exists, but one or more other features It should be understood that this does not preclude the existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. It is apparent to those skilled in the art that the present invention may be embodied in other specific forms without departing from the spirit and essential characteristics of the present invention.

Since the CoDel (Controlled Delay) algorithm operates based on the delay time, the queue internal deletion policy is determined using the target delay value, which is the allowable waiting delay time, and the policy update frequency is determined using the interval value.

The CoDel protocol tracks the delay time by recording a time stamp at the point in time when a packet is received in the queue as shown in FIG. 1A. Since the oldest packet in the queue stored at the head of the queue has the longest delay time, the delay time of this packet is checked, and if more than the target delay time has passed, it is discarded. [T. Hoeiland- Joergensen, P. McKenney, D. Taht, J. Gettys, and E. Dumazet, “ The flow

queue CoDel packet scheduler and active queue management algorithm ,” RFC 8290, Jan. 2018.]

The FQ-CoDel protocol is a method of scheduling multiple CoDel queues in a Deficit Weighted Round Robin (DWRR) method. Create logical queues for each service flow and schedule them in the DWRR method. [E. Grigorescu, C. Kulatunga, and G. Fairhurst, “Evaluation of Priority Scheduling and Flow Starvation for Thin Streams with FQ-CoDel,” in Proc . EuCNC , Jun. 2015.]

In the FQ-CoDel algorithm, the weight for each flow is determined using the quantum value, which is the amount of packets processed in different flows every round.

On the other hand, CoDel and FQ-CoDel algorithms assume a general network situation and set the standard values of target delay, interval, and quantum as follows.

- target delay : 5 ms

- interval: Initial value 100 ms, the next discard timing decreases sequentially in inverse proportion to the square root of the number of packet discards

- quantum : fixed value 1514 bytes, Ethernet MTU (1,500 bytes) + header (14 bytes) size

However, in an actual network environment, it is necessary to set the appropriate target delay and interval values in consideration of the amount of traffic generated, the link capacity in the bottleneck section, and the round-trip delay time. A related study is the study of Kulatunga [C. Kulatunga, N. Kuhn, G. Fairhurst, and D. Ros, “Tackling bufferbloat in capacity-limited networks,” in Proc . EuCNC , Jun. 2015.] and a study of Ye [J. Ye, K.-C. Leung, and Victor OK Li, “Optimal delay control for combating bufferbloat in the internet,” in Proc . IEEE ICCS , Dec. 2016.].

Kulatunga's study is to find a combination of target delay and interval to increase the usage rate of bottleneck section and limit queue delay time in a fixed network environment. Kulatunga's study has a disadvantage in that it has low practical utility because it has to find a new parameter combination for each situation in which the network environment changes.

Ye's research designs a feedback controller based on round-trip latency, bottleneck bandwidth, and TCP flow count using the TCP fluid model. It maximizes TCP goodput according to network conditions and optimizes and derives the target delay that satisfies the delay requirements within the stable range of the CoDel system. Ye's study considers only one target delay among parameters, and there is a limitation in classifying and processing TCP sessions for each flow.

Therefore, in the present invention, a design method for adaptively optimizing three parameters of FQ-CoDel that affects network congestion and service performance according to tactical traffic is performed in stages.

In particular, in the tactical communication network, which is a field to be applied in the present invention, various application services such as guaranteed/non-guaranteed type, voice/video, and file/short message transmission are provided according to the military operating system. In addition, it is necessary to classify flows for each application service for the purpose of achieving military priority according to the importance of information in operational operation. In addition, in general, since the properties of traffic generated by each application service are different, it is necessary to dynamically manage individual queues in order to meet different network requirements such as data rate, delay, and reliability.

1B shows a network structure diagram of FQ-CoDel according to an embodiment of the present invention.

The FQ-CoDel algorithm of the present invention is designed with x level priorities, y service classes, and z flows (dynamic flows). In the embodiment, two priorities (level priority), five service classes (service class), and 64 flows (dynamic flow) are designed.

In the present invention, for the differential processing of each tactical priority, the first priority (Highest priority) processes the assured application service (Assured Voice, Assured Multimedia Conferencing) with the highest priority. The second priority (lowest priority) processes three application services (Short Message, Non-Assured Voice, Broadcast Video) later. Based on the designed structure, a logical queue is created for each service flow, and the FQ-CoDel protocol to which parameters obtained through the algorithm of FIG. 4 are applied to each queue is implemented. A flow corresponding to each application service operates in a known Deficit Weighted Round Robin (DWRR) method, and finally, a complex scheduler technology combined with a known Priority Queuing (PQ) method is applied.

At this time, the present invention determines a target delay that is an allowable waiting delay time, an interval rate that determines the frequency at which the policy is updated, and a weight for each flow in order to determine the queue internal deletion policy of the FQ-CoDel algorithm. In order to optimize the three parameters of the quantum, the first calculates the target delay in units of service classes and the interval rate in units of flows. Second, the quantum is calculated in units of flows with different purposes for each priority.

The present invention optimizes target delay. The FQ-CoDel algorithm records the time stamp when the packet is entered into the queue, tracks the delay time that the corresponding packet stays, and compares it with the target delay value, which is the target delay time. In the present invention, it is assumed that the bandwidth size of the bottleneck in the tactical communication network is known. Using the bandwidth size C of the bottleneck section and the target delay time t _target , the allowable queue length β _target per time for not exceeding the target delay time can be calculated as in Equation (1).

Equation (1)

β _target = C(1 + t _target )

β _target is calculated as the sum of the number of packets that can be processed in the future considering the amount of packets that can be processed in the queue and the allowable delay time. In the FQ-CoDel protocol in which multiple flows are entered into one queue, a method for determining the optimal target delay value according to the network environment is proposed as in Equation (2) in Ye's study.

Equation (2)

Equation (2) represents the maximum value of the stable range of the target delay according to the network environment. The network environment considered here includes the bottleneck bandwidth C, the number of flows N _s , the propagation delay time T _p , and the initial packet arrival rate λ ₀ exceeding the minimum allowable queue length β _min .

In the present invention, in order to apply Equation (2) to the FQ-CoDel protocol, it is assumed that all flows belonging to the same service class use the same target delay value, and the target delay value is the following Equation (3), It is calculated as in Equation (4).

Equation (3)

Equation (4)

Equation (3) represents the sum of target delays of all flows within the same service class. Equation (4) represents the optimal target delay of an individual flow. The parameters used in the equation are shown as shown in FIG. 2, and parameters that vary according to the service class s are N _s and t _max,s .

t _max,s is calculated as the product of the number of flows and the EI-CER (End Instrument-Customer Edge Router) delay time. The optimal target delay applied to each flow is compared with the lower limit to determine the final value.

Next, the present invention optimizes the interval rate. The FQ-CoDel algorithm discards all packets that exceed the allowable queue length when the packet discard time arrives. The packet discard interval affects the total amount of packets discarded and the packet loss rate. Also, depending on the time when packets are discarded, the congestion level of the queue and the time of stabilization are affected.

Equation (5)

Equation (5) is the point at which a packet is discarded in the existing CoDel algorithm. Here, n denotes the number of packet discards, and I(n) denotes a packet discard interval according to the number of packet discards. The first packet discard time is 100 ms, and the next packet discard time sequentially decreases in inverse proportion to the square root of the number of packet discards.

In general, the CoDel algorithm is used to prevent an abnormal increase in expected processing time due to excessive packets occurring at once in the bottleneck section. Therefore, even when the packet discard time arrives, if there is no packet to be discarded, the turn is skipped. Conversely, even if the network is very congested due to excessive traffic, the packet must be discarded after waiting until the packet discard time. In the bottleneck section with a small bandwidth, packets are accumulated in the queue exponentially, causing performance degradation that increases latency.

Therefore, according to the amount of incoming packets and the network environment, an algorithm that discards packets at an appropriate point in time for rapid queue stabilization is required. In the present invention, the amount of packets lost at the initial packet discard time is checked, and future traffic is predicted to determine the next packet discard time point.

Equation (6)

p _100,j represents the number of discarded packets at 100 ms, which is the first packet discard time in flow j. β _opt is the allowable queue length calculated by substituting the optimal target delay derived from Equation (4) into Equation (1). Equation (6) represents a relative packet loss ratio compared to β _opt in p _100,j .

Equation (7)

Equation (8)

)을 적절한 비율로 매핑한 optimal interval rate 지수이다. 수학식(8)은 패킷 손실 비율(

)을 적용한 새로운 optimal interval rate 이다.Equation (7) is the packet loss ratio (

) is the optimal interval rate index mapped to an appropriate ratio. Equation (8) is the packet loss ratio (

) is the new optimal interval rate applied.

Next, the present invention optimizes quantum. The FQ-CoDel algorithm uses a Deficit Weighted Round Robin (DWRR) scheduling method to create a logical queue for each service flow and process the queues individually. In order to guarantee different QoS requirements for each application service in the tactical communication network, quantum values, which are weights for each flow, should be appropriately distributed. If the relationship between quantum and delay time in the FQ-CoDel protocol is defined as an equation, it is as shown in Equation (9) below.

Equation (9)

Equation (9) is a value predicted by the total delay time required to process all packets in the flow j. q _j is the quantum value, p _avg,j is the average packet size, and n _j is the number of packets. When all flows are processed once by the RR (Round Robin) scheduling method, round is defined, and the unit of delay time is round. The parameters used in quantum optimization are summarized as shown in FIG. 3 . In the present invention, two quantum optimization methods are proposed according to the purpose using Equation (9).

Equation (10)

은 주어진 제약조건에 맞는 optimal quantum

를 이용하여 모든 패킷을 처리하는데 필요한 round 이다. 한 번의 round 에서 처리되는 quantum 합은 유지하면서 특정 플로우의 지연시간을 단축시킨다.Equation (10) is an optimization equation that prioritizes a specific flow j. objective function

is the optimal quantum for a given constraint

It is a round required to process all packets using . It reduces the delay time of a specific flow while maintaining the quantum sum processed in one round.

Equation (11)

은 default quantum 값을 이용하여 예측한 round 보다 작아야 하고, round 간 편차를 최소화하는 optimal quantum 값을 찾는다.Equation (11) is an optimization equation that minimizes the delay time at which the last packet is finally processed in all flows. The objective function represents the L1 norm that can be generated in all flows. every

must be smaller than the predicted round using the default quantum value, and find the optimal quantum value that minimizes the deviation between rounds.

Existing FQ-CoDel uses fixed parameters regardless of application service and network environment. However, in the present invention, a logical queue is created for each service to accommodate different queuing delay requirements for each application service, and parameters are applied individually.

4 is a design procedure of the entire FQ-CoDel algorithm according to an embodiment of the present invention using target delay, interval rate, and quantum parameter optimization.

The algorithm of the present invention is largely divided into two steps. In the first step, the target delay is calculated for each service class, and the interval rate is calculated for each flow. In the second step, the purpose is different for each priority, and quantum is calculated in units of flow. In the algorithm shown in Fig. 4, k denotes a priority, s denotes a service class, and j denotes the number of flows.

In the present invention, for step-by-step performance verification, in the first step, the total packet discard amount, the queue stabilization time, and the amount of outdated packets are analyzed, and in the second step, round times are compared.

In the present invention, for performance analysis of the FQ-CoDel algorithm, a system structure is configured as shown in FIG. 1B. The FQ-CoDel algorithm is designed with x level priorities, y service classes, and z flows (dynamic flows).

And, as shown in FIG. 5, it is designed with two priorities (level priority), five service classes (service class), and 64 flows (dynamic flow).

For differential processing of each tactical priority, first priority (Highest priority) prioritizes assured application services (Assured Voice, Assured Multimedia Conferencing). The second priority (lowest priority) processes three application services (Short Message, Non-Assured Voice, Broadcast Video) later. Based on the designed structure, a logical queue is created for each service flow, and the FQ-CoDel protocol to which parameters obtained through the algorithm of FIG. 4 are applied to each queue is implemented.

Simulation is performed with MARLAB, and optimization is performed using the CVX tool. The flow corresponding to each application service operates in a Deficit Weighted Round Robin (DWRR) method, and finally, a composite scheduler technology combined with a Priority Queuing (PQ) method is applied.

Additional parameters for performance analysis are shown in FIG. 6 . In each application service, it is assumed that traffic occurs with a Poisson distribution that is 1.3 to 1.8 times that of the bottleneck section. It is assumed that the average size of packets arriving at the bottleneck is 1000 bytes. The total number of incoming packets per flow is limited to 5,000. The amount of traffic is checked in units of packets, and the time that packets are received and processed is in ms. The EI-CER delay time represents a value for one flow in each service.

7 is a result of completing the first step in the algorithm of the present invention. The existing method used fixed values of the target delay and interval rate indices of 5 ms and 0, respectively, and in the present invention, the optimal target delay and optimal interval rate indices were applied according to the application service and network environment. The definitions of the three performance indicators are as follows.

- Number of discarded packets: The total number of packets that are lost until the last packet leaves the queue among 5,000 packets that have entered each queue.

- Queue stabilization time: The time when all future packets arrive within the target delay time is judged as the stabilization time, and the index of the last packet that exceeds the target delay time

-Outdated number of packets: Total number of packets that arrived at the receiving end after the target delay time and did not satisfy the service requirements.

Compared with the conventional method, the present invention has two characteristics.

First, the target delay of the present invention is slightly increased, so the allowable queue length is

Increase the β _target . Second, in the present invention, the interval is appropriately advanced so that the queue is checked at more frequent times. Therefore, the present invention discards the packet at frequent intervals, but discards the packet based on the appropriate β _target .

Accordingly, although the total amount of discarded packets slightly increases, the queue stabilization time is greatly advanced and the number of outdated packets is reduced. That is, according to the present invention, only packets that inevitably require loss because of queue congestion exceed β _target are discarded. These results are common to all five application services.

8 and 9 are results of comparing the existing method of the Non-Assured Voice service among the five application services and the method of the present invention.

8 shows the amount of packets inside the queue according to time. In FIG. 8 , a blue bar indicates the amount of packets discarded, a solid red line indicates the amount of packets before processing or discarding, and a solid black line indicates the amount of packets afterward.

Conventionally, the length of the cue is not stable due to severe fluctuations in the graph, but it can be seen that the present invention is in a state of saturation to some extent after about 300 ms. In the case of the present invention, the total amount of discarded packets is increased by about 200 compared to the conventional one.

9 shows the waiting time for each packet. The latency refers to the time difference between when a packet enters the queue and when it is processed and out. Conventionally, about 37% of packets do not reach the target waiting time because packets cannot be discarded at an appropriate time. However, according to the present invention, all packets reach a stable waiting time after the 1,000th packet.

10 and 11 are results of completing the second step in the algorithm of the present invention. Conventionally, a value fixed to 1,514 bytes is used for quantum, and the present invention uses optimal quantum.

In FIG. 10, optimization equation (10) is applied to prioritize the first flow of the Assured Multimedia Conferencing application service in the first priority. While maintaining the sum of quantum required in all flows, the delay time of the first flow is reduced to 1/3 level.

In FIG. 11, the optimization equation (11) is applied to reduce the processing delay time of the entire application service in the second priority. The standard deviation is greatly reduced by reducing the delay time required until the last packet is processed in 50 flows and minimizing the difference in delay time between application services.

As described above, in the present invention, a parameter optimization method of the FQ-CoDel algorithm that can be applied to the multi-layer integrated tactical backbone network using the FQ-CoDel protocol in which the queue management and scheduler techniques are combined is proposed. The present invention designed an algorithm for adaptively optimizing FQ-CoDel parameters in order to guarantee QoS of individual application services and manage network congestion control in order to obtain tactical utility in various application services. The queue management method of the present invention allows the maximum amount of packets to arrive within a target delay time by discarding packets at an appropriate time when a queue congestion occurs, and advances the queue stabilization time. In particular, the present invention is advantageous in managing individual queues as operations to satisfy the queuing delay requirements of application services such as voice, video, and message. In addition, the scheduling method of the present invention has the advantage of being able to change the weights of individual flows in order to classify information importance and military priorities in operational operation and process them first, or adjust the weights of all flows to reduce delay time.

The present invention described above is implemented in a computing environment including a computing device. The computing device may be any type of computing device that transmits/receives signals to and from other terminals.

The computing device includes at least one processor, a computer-readable storage medium, and a communication bus. The processor may control the computing device to operate according to the aforementioned embodiments. The processor may execute one or more programs stored in a computer-readable storage medium. The one or more programs may include one or more computer-executable instructions, which, when executed by a processor, may be configured to cause a computing device to perform operations in accordance with an exemplary embodiment.

The computer-readable storage medium is configured to store computer-executable instructions or computer-executable instructions to program code, program data and/or other suitable form of information. A program stored in a computer-readable storage medium includes a set of instructions executable by a processor. In one embodiment, the computer-readable storage medium includes memory (volatile memory, such as random access memory, non-volatile memory, or a suitable combination thereof), one or more magnetic disk storage devices, optical disk storage devices, flash memory device , other types of storage media that can be accessed by the computing device and can store desired information, or a suitable combination thereof.

A communication bus interconnects various other components of the computing device, including processors and computer-readable storage media.

The computing device may also include one or more input/output interfaces and one or more communication interfaces that provide interfaces for one or more input/output devices. The input/output interface and the communication interface are coupled to the communication bus. An input/output device (not shown) may be connected to other components of the computing device through an input/output interface. Exemplary input/output devices include input devices such as pointing devices (such as a mouse or trackpad), keyboards, touch input devices (such as touchpads or touchscreens), voice or sound input devices, various types of sensor devices and/or imaging devices; and/or output devices such as display devices, printers, speakers and/or network cards. The exemplary input/output device may be included in the computing device as a component constituting the computing device, or may be connected to the computing device as a separate device distinct from the computing device.

The operations according to the present embodiments may be implemented in the form of program instructions that can be performed through various computer means and recorded in a computer-readable medium. Computer-readable media refers to any medium that participates in providing instructions to a processor for execution. Computer-readable media may include program instructions, data files, data structures, or a combination thereof. For example, there may be a magnetic medium, an optical recording medium, a memory, and the like. A computer program may be distributed over a networked computer system so that computer readable code is stored and executed in a distributed manner. Functional programs, codes, and code segments for implementing the present embodiment may be easily inferred by programmers in the art to which this embodiment belongs.

The above detailed description should not be construed as restrictive in all respects and should be considered as exemplary. The scope of the present invention should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the present invention are included in the scope of the present invention.

Claims (22)

A method for optimizing parameters of the FQ-CoDel algorithm in a processor for queue management of a military-operated communication network, the method comprising:
The FQ-CoDel algorithm is designed with x priorities (level priority), y service classes (service class), and z flows (dynamic flow),
To determine the dequeue policy of the FQ-CoDel algorithm, the target delay, which is the allowable waiting delay, the interval rate, which determines the frequency at which the policy is updated, and the quantum that determines the weight for each flow To optimize the three parameters of
Step 1 of calculating the target delay in units of service classes and calculating the interval rate in units of flows;
A method for optimizing parameters of the FQ-CoDel algorithm, which includes two steps of calculating the quantum in units of flow with different objectives for each priority. The method according to claim 1,
The FQ-CoDel algorithm is a parameter optimization method of the FQ-CoDel algorithm that is designed with 2 level priorities, 5 service classes, and 64 flows (dynamic flow). 3. The method according to claim 2,
For differential processing of each tactical priority, the first priority is a parameter optimization method of the FQ-CoDel algorithm that prioritizes assured application services (Assured Voice, Assured Multimedia Conferencing). 4. The method according to claim 3,
The second priority is a parameter optimization method of the FQ-CoDel algorithm that processes three application services (Short Message, Non-Assured Voice, Broadcast Video) later. 5. The method according to claim 4,
The FQ-CoDel algorithm is a parameter optimization method of the FQ-CoDel algorithm that creates a logical queue for each service flow and operates in a Deficit Weighted Round Robin (DWRR) method to process queues individually. 6. The method of claim 5,
A parameter optimization method of the FQ-CoDel algorithm in which the flow corresponding to each application service operates in the Deficit Weighted Round Robin (DWRR) method and is controlled by the PQ (Priority Queuing) method at the final stage. 7. The method of claim 6,
For step-by-step performance verification of the parameter optimization method of the FQ-CoDel algorithm, in the first step, the parameter optimization method of the FQ-CoDel algorithm analyzes the total packet discard amount, queue stabilization time, and the amount of outdated packets. 8. The method of claim 7,
The parameter optimization method of the FQ-CoDel algorithm is to compare round times in the second step to verify the step-by-step performance of the parameter optimization method of the FQ-CoDel algorithm. A method for optimizing parameters of the FQ-CoDel algorithm in a processor for queue management of a military-operated communication network, the method comprising:
The FQ-CoDel algorithm is designed with x priorities (level priority), y service classes (service class), and z flows (dynamic flow),
In order to optimize the target delay, which is the allowable waiting delay time for determining the dequeue policy of the FQ-CoDel algorithm,
Step 1 of recording a time stamp when a packet is entered into a queue, tracking the delay time at which the packet stays, and comparing it with a target delay value, which is a target delay time;
Step 2 of calculating an allowable queue length not to exceed the target delay time using the size of the bandwidth of the bottleneck and the target delay time; and
In the FQ-CoDel protocol in which a plurality of flows are introduced into one queue, it includes the third step of determining an optimal target delay value according to the network environment by Equation (2),
In step 3, in order to apply Equation (2) to the FQ-CoDel protocol, all flows belonging to the same service class use the same target delay value, and the parameter of the FQ-CoDel algorithm that calculates the target delay value in units of service classes Optimization method.
Equation (2)

(Bottleneck bandwidth C, number of flows N _s , initial interval I _{0 ,} propagation delay time T _p , first packet arrival rate λ ₀ exceeding the minimum allowable queue length β _min ) 10. The method of claim 9,
The optimal target delay is a parameter optimization method of the FQ-CoDel algorithm calculated by Equation (4).
Equation (4)

Here, t _s is the sum of the target delays of all flows in the same service class, t _min is the minimum target delay, and N _s is the number of flows in the service class s. 11. The method of claim 10,
Optimal target delay is a parameter optimization method of the FQ-CoDel algorithm that represents the optimal target delay of an individual flow. 12. The method of claim 11,
The parameter optimization method of the FQ-CoDel algorithm in which the sum of target delays of all flows within the same service class is calculated by Equation (3).
Equation (3)

Here, t _stable,s is the maximum value of the stable range of the target delay according to the network environment, t _min is the minimum target delay value, and t _max,s is the maximum target delay value in each service class. 13. The method of claim 12,
The parameter optimization method of the FQ-CoDel algorithm where the parameters that vary according to the service class s are N _s and t _max,s . 10. The method of claim 9,
In order to optimize an interval rate that determines the frequency at which a policy is updated in the FQ-CoDel algorithm, the optimal interval rate is calculated by Equation (8) .
Equation (8)

Here, α _j denotes an optimal interval rate index, and n denotes the number of packet discards. 15. The method of claim 14,
The optimal interval rate index (α _j ) is a parameter optimization method of the FQ-CoDel algorithm calculated by Equation (7).
Equation (7)

here

represents the packet loss rate. 16. The method of claim 15,
Packet loss rate (

) is a parameter optimization method of the FQ-CoDel algorithm calculated by Equation (6).
Equation (6)

Here, p _100,j denotes the number of packets discarded at 100 ms, which is the first packet discard time in flow j. β _opt represents the allowable queue length using the optimal target delay. 17. The method of claim 16,
Packet loss rate (

) is a parameter optimization method of the FQ-CoDel algorithm representing the relative packet loss ratio compared to β _opt at p _100,j . 18. The method of claim 17,
The total delay time required to process all packets in flow j is a parameter optimization method of the FQ-CoDel algorithm defined by Equation (9).
Equation (9)

Here, q _j is the quantum value, p _avg,j is the average packet size, and n _j is the number of packets. 19. The method of claim 18,
A method for optimizing parameters of the FQ-CoDel algorithm, comprising calculating an optimal quantum for determining a weight for each flow in the FQ-CoDel algorithm by Equation (10).
Equation (10)

where q _j is the default quantum value used in flow j,

is the optimal quantum value used in flow j, r _j is the predicted round in flow j, k is the number of flows in priority k,

is the optimal quantum for a given constraint

is used to indicate the round required to process all packets. 20. The method of claim 19,
The optimized quantum that minimizes the delay time at which the last packet is finally processed in all flows is a parameter optimization method of the FQ-CoDel algorithm defined by Equation (11).
Equation (11)

here

is the optimal quantum value used in flow j, n _j is the number of packets to be processed in flow j, p _avg,j is the average packet size in flow j, r _j is the predicted round in flow j,

is the optimal quantum for a given constraint

is used to indicate the round required to process all packets. 21. The method of claim 20,
every

must be smaller than the round predicted using the default quantum value, and a parameter optimization method of the FQ-CoDel algorithm that finds the optimal quantum value that minimizes the deviation between rounds. As a computer program stored in a computer-readable recording medium,
When a computer program is executed by a processor,
Design the FQ-CoDel algorithm with x level priorities, y service classes, and z flows (dynamic flows),
To determine the dequeue policy of the FQ-CoDel algorithm, the target delay, which is the allowable waiting delay, the interval rate, which determines the frequency at which the policy is updated, and the quantum that determines the weight for each flow To optimize the three parameters of
Step 1 of calculating the target delay in units of service classes and calculating the interval rate in units of flows;
A computer program including instructions for causing a processor to perform a parameter optimization method of the FQ-CoDel algorithm, which includes the second step of calculating the quantum for each flow with different objectives for each priority.

Images