KR102624295B1 - Time control scheduling method and system for supporting non-isochronous emergency traffic in time-sensitive networks - Google Patents

Abstract

A city control scheduling method and system in a civic network environment are disclosed. A city control scheduling system in a civic network environment according to an embodiment of the present invention includes a traffic transmission unit that transmits civic traffic and other traffic according to a pre-booked TAS schedule; an emergency traffic determination unit that determines whether emergency traffic requiring immediate transmission is input; a queuing rule adjuster that redefines a priority-to-traffic class mapping table when the emergency traffic is input; and a dynamic scheduling control unit that temporarily and dynamically extends the scheduled traffic (ST) time window in accordance with the transmission time of the emergency traffic.

Description

{Time control scheduling method and system for supporting non-isochronous emergency traffic in time-sensitive networks}

The present invention relates to a civic network, and more specifically, to a time-controlled scheduling method and system for immediate transmission of non-isochronous emergency event traffic in a civic network environment.

With the advancement of network technology, interest in Industrial Internet of Things (IIoT) technology is increasing beyond the Internet of Things (IoT), and civic (time-sensitive or time-critical) systems (e.g. : The need for networking technology to improve service quality and ensure real-time (autonomous driving, aviation, medical, smart factory, etc.) is emerging.

In the civic network, various communication devices coexist, simultaneously generating and transmitting data with different requirements (time-critical, audio/video, best-effort, etc.).

Taking self-driving cars as an example, intelligent driver assistance systems (Advanced Driver Assistance Systems, ADAS) periodically collect data necessary for autonomous driving, such as information about the current surrounding environment and the vehicle, and transmit the collected data to multiple devices to respond to the situation. It is requested that it be transmitted immediately.

Data that must guarantee real-time (hereinafter referred to as 'citizen traffic') must be transmitted first and foremost, and must have reliability, low end-to-end latency, and low delay. It has requirements such as low jitter.

To ensure these requirements, the IEEE Time-Sensitive Networking Task Group has standardized Time-Sensitive Networking (hereinafter referred to as 'TSN') technology that is compatible with the existing IEEE 802.1 and IEEE 802.3 standards. , standardization work is still actively underway.

More specifically, IEEE 802.1 Qby is one of the sub-standards of TSN. It is a time-aware shaper (Time Awareness Shaper) to guarantee the requirements of civic traffic by scheduling and controlling transmission time based on prior information of periodic civic traffic. -Aware Shaper (hereinafter referred to as 'TAS') technology is provided.

In an actual network environment, non-isochronous emergency event traffic (hereinafter referred to as 'emergency traffic') may occur sporadically to propagate and respond to incidents/accidents, etc., and this has the highest priority. It must be transmitted instantly with minimal delay.

However, TAS cannot guarantee the requirements of emergency traffic that is not defined in advance, and has serious adverse effects on previously reserved civic traffic and systems.

Korea Patent Publication No. 10-2021-0047748 (published on April 30, 2021) - 5G-based industrial communication system

The present invention defines new scheduling rules for immediate delivery of emergency traffic, and at the same time provides a time-controlled scheduling method and system in which scheduling is dynamically controlled to guarantee/protect the requirements of existing civil traffic in the network. .

The present invention is intended to provide a time control scheduling method and system that can minimize the impact experienced by existing civil traffic when emergency traffic occurs in a congested network environment and ensure stable performance of the real-time system through immediate transmission.

Other objects of the present invention may be easily understood through the following description.

According to one aspect of the present invention, there is a time control scheduling system in a civic network environment, comprising: a traffic transmission unit that transmits civic traffic and other traffic according to a pre-booked TAS schedule; an emergency traffic determination unit that determines whether emergency traffic requiring immediate transmission is input; a queuing rule adjuster that redefines a priority-to-traffic class mapping table when the emergency traffic is input; and a dynamic scheduling control unit that temporarily and dynamically extends a scheduled traffic (ST) time window corresponding to the transmission time of the emergency traffic.

The queuing rule adjuster may assign the highest traffic class to the emergency traffic.

The queuing rule adjuster may allocate the emergency traffic to an independent queue.

The queuing rule adjuster may maintain the gate state of the queue to which the emergency traffic is assigned in a always open state in all time windows including the guard band (GB) during scheduling.

The dynamic scheduling control unit may calculate the transmission time required to transmit the emergency traffic, perform time window expansion control according to the type of the current time window, and perform window delay control according to the type of the next time window.

The dynamic scheduling control unit accumulates the transmission time in the extended request time if the current time window is the ST time window, and updates the extended request time if the current time window is not the ST time window and records the current time in lastTimestamp to determine the ST time window. You can record the time when the window should be extended.

If the next time window is an NST time window, the dynamic scheduling control unit temporarily delays the next time window by the extended request time, otherwise, subtracts the start time of the next time window from the sum of the extended request time and lastTimestamp. You can update with the remaining time and change the state to the next time window.

Meanwhile, according to another aspect of the present invention, there is provided a time-controlled scheduling method performed by a time-controlled scheduling system in a civic network environment, comprising: arriving at an emergency traffic frame in a queue; A transmission selection module selecting the emergency traffic frame; transmitting a time window extension (TWE) message including the size of the emergency traffic frame to a gate control list; transmitting the emergency traffic frame; Calculating the transmission time of the emergency traffic frame in the gate control list and accumulating it in a time variable τ _ext where the ST time window may require expansion; and when it is time for the gate control list to change the gate state to the next state, and the next state is an NST time window, the gate control list delays the transition by τ _ext .

Other aspects, features and advantages in addition to those described above will become apparent from the following drawings, claims and detailed description of the invention.

According to an embodiment of the present invention, there is an effect of defining a new scheduling rule for immediate delivery of emergency traffic and simultaneously guaranteeing/protecting the requirements of existing civil traffic in the network.

In addition, it minimizes the impact on existing civic traffic when emergency traffic occurs in a congested network environment, and has the effect of ensuring stable performance of the real-time system through immediate transmission.

1 is a diagram showing exemplary operation of TAS in a TSN switch;
Figure 2 is an example when emergency traffic transmission is allowed within the NST time window,
Figure 3 is an example when emergency traffic transmission is allowed within the ST time window,
4 is a diagram illustrating possible problems that may occur when emergency traffic transmission is allowed in both ST and NST time windows;
5 is a schematic configuration diagram of a time control scheduling system according to an embodiment of the present invention;
6 is a flowchart of a time control scheduling method according to an embodiment of the present invention;
Figure 7 is an example of emergency traffic being transmitted in all time windows including the guard band;
Figure 8 is a detailed flowchart of the dynamic scheduling control process among the time-controlled scheduling methods;
Figure 9 is a diagram showing the operating procedure for dynamic scheduling time window expansion;
Figure 10 is an example of emergency traffic transmission when dynamic scheduling time window expansion is applied;
11 is a diagram showing the network topology of a TSN-based ADAS scenario;
12 is a graph showing delay, delay deviation and throughput results for all scenarios including w/o-ET, ET-in-ST, ET-in-NST, ET-in-ST & NST and eTAS;
Figure 13 is a graph showing the delay, delay deviation and throughput performance of ET in simulation when introducing burst ET with various burst and frame sizes;
Figure 14 is a graph showing the performance of ST using eTAS according to the burst of ET;
Figure 15 is a graph showing results for NST flows that are not protected by TAS or eTAS by design.

Since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all changes, equivalents, and substitutes included in the spirit and technical scope of the present invention.

When a component is said to be "connected" or "connected" to another component, it is understood that it may be directly connected to or connected to the other component, but that other components may exist in between. It should be. On the other hand, when it is mentioned that a component is “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

Terms such as first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

The terms used herein are only used to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as “comprise” or “have” are intended to indicate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

In addition, the components of the embodiments described with reference to each drawing are not limited to the corresponding embodiments, and may be implemented to be included in other embodiments within the scope of maintaining the technical spirit of the present invention, and may also be included in separate embodiments. Even if the description is omitted, it is natural that a plurality of embodiments may be re-implemented as a single integrated embodiment.

In addition, when describing with reference to the accompanying drawings, identical or related reference numbers will be assigned to identical or related elements regardless of the drawing symbols, and overlapping descriptions thereof will be omitted. In describing the present invention, if it is determined that a detailed description of related known technologies may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. In the drawings attached to this specification, colors are assigned to help distinguish components. However, even for the same component, its color may vary in perspective and cross-sectional views. And even different components may be given the same or similar colors.

In addition, terms such as “… unit,” “… unit,” “… module,” and “… unit” used in the specification refer to a unit that processes at least one function or operation, which refers to hardware or software or hardware and software. It can be implemented by combining .

The present invention relates to a time control scheduling technology for ensuring real-time performance of a real-time system through immediate transmission of emergency traffic for disseminating and responding to emergency situations in a civic network environment and at the same time minimizing interference that may occur in civic traffic. will be.

Previously, traffic was separated into civic traffic and other traffic based on predefined traffic information, and the transmission space (ST time window) for civic traffic was created through gate controlling techniques. TAS technology exists to ensure requirements by allocating transmission space (NST time window) for other traffic and separating and reserving transmission time (hereinafter referred to as scheduling).

1 is a diagram showing an exemplary operation of TAS in a TSN switch.

A switch's egress port can have up to eight queues, each corresponding to a traffic class. When a frame enters the switch, the switch identifies its priority from the priority code point (PCP) of the VLAN identifier in the IEEE 802.1Q Ethernet header. Then, each frame is mapped to a traffic class according to priority based on the mapping recommended by IEEE 802.1Q, and then the frame is inserted into the queue corresponding to the traffic class. The PCP value of a frame may vary depending on the type of traffic, but in the audio video bridging (AVB) standard, audio and video traffic have PCP values 3 and 2, corresponding to SR classes AVB-A and AVB-B, the highest traffic class. It is recommended that it be mapped to . For example, if the number of queues is 8, frames of type AVB-A and AVB-B are assigned traffic classes 7 and 6, and are inserted into the 8th and 7th queues, respectively.

Each queue may have a transmission selection algorithm, which determines whether the queue should provide a frame for transmission. Credit-based shaper (CBS) and asynchronous traffic shaper (ATS) are examples of such algorithms, which adjust queues and redirect transmissions based on some stream reservation criteria to protect low-priority traffic. It plays a regulatory role.

Additionally, each queue has a transmission gate. The gate has two states: open (o) and closed (c), and the frame of each queue can be transmitted only when the gate of that queue is open. If multiple gates are open simultaneously and there are frames available for transmission in one or more of these queues, the transmission selection module transmits the frames using strict priority criteria (e.g., decreasing order of traffic classes).

Gates are controlled by the gate control list (GCL). Gate control lists define when each gate opens and closes, ensuring that specified classes of traffic are transmitted at specific times. The opening and closing times of gates are calculated based on several constraints such as network topology, configuration, traffic type, number of flows, frame size and spacing of each flow. To calculate the gate control list, TAS first assigns time-sensitive periodic traffic to a scheduled traffic (ST) time window. The time of the ST time window is allocated enough to accommodate the total transmission time of the frames that must be transmitted during that time window, and the remaining time is allocated to less time-sensitive non-scheduled traffic (NST) (e.g. AVB). or best-effort). The schedule is determined based on ST because, by definition, the amount and schedule of NST are unknown and need not be known and prefer to maximize bandwidth utilization.

During the ST time window, only the gates related to the ST queue are opened and only the frames of that queue are transmitted to the network. In addition, to prevent ST delay due to interference caused by long NST frame transmission just before the start of the ST time window, a guard band (GB) is added in front of the ST time window, as shown at the bottom of FIG. 1. The length of the guardband is set to be as long as the transmission time of the largest Ethernet frame in the network, and all gates are closed during the guardband. Therefore, even if the longest frame is transmitted just before the guardband starts, transmission is completed within the GB time window and does not interfere with the ST.

Finally, all times except ST and GB are allocated to the NST time window for transmitting other traffic. During this time window, unlike the ST time window, all gates related to ST traffic are closed and only other gates are open. This is the basic exclusive gating method recommended by the IEEE 802.1Q standard. However, the standard allows deviation from exclusive gating, and this opportunity is utilized in an embodiment of the present invention described later.

Whenever the route of civic traffic is changed due to a change in the network environment or new civic traffic with periodicity is added, there is a prior technology to allocate a new ST time window to the pre-booked schedule or reconfigure the existing schedule.

However, the aforementioned TAS technology can guarantee the requirements for civic traffic through scheduling based on information on predefined civic traffic, but the requirements for civic traffic that occurs sporadically with an isochronism are not There are limits to what cannot be guaranteed. For example, when an emergency event (e.g. a fire) is detected, the system immediately propagates the corresponding message in real time across the network to alert people to solve the problem and respond to the need to activate response devices (e.g. sprinklers). There will be. Traffic related to these emergency events (emergency traffic) is time-sensitive and critical, as any delay or loss can cause catastrophic damage to the system.

Therefore, if emergency traffic is transmitted at a low priority, it may lose the ongoing transmission competition with other traffic, resulting in extreme delays or starvation, preventing a quick response to problems and worsening system problems. there is.

Problems related to the arrival time of emergency traffic will be explained through drawings and examples.

Figure 2 is an example diagram when emergency traffic transmission is allowed within the NST time window.

[Case 1] Emergency traffic (ET-in-NST) within the NST time window: Since TAS does not consider emergency traffic when scheduling, emergency traffic will naturally and inevitably be transmitted during the NST time window in standard TSN. Then, if emergency traffic has the highest priority, it is generally transmitted before other NST frames (e.g. AVB, best effort) waiting at the switch. However, even the highest priority frame must wait until completion if there is currently transmission in progress. This can cause significant delays in emergency traffic if an ET frame arrives in the queue near the end of the NST time window while there is an NST frame currently being transmitted. For example, in Figure 2, an ET frame arrives while the NST frame (N2) is transmitted. Even if N2 transmission is completed, ETs waiting in the queue cannot start transmission because all gates are closed until the next GB. Therefore, the ET must wait until the next NST time window at every switch along the flow path, which can cause significant delays in the ET.

Figure 3 is an example diagram when emergency traffic transmission is allowed within the ST time window.

[Case 2] Emergency traffic (ET-in-ST) within the ST time window: The TSN standard is flexible in that it allows other configurations beyond what is broadly permitted as basic/default TNS operation within the scope of the standard. The network administrator can configure the gate control list so that the TAS gates to ET are open during the ST time window without assigning a dedicated (and potentially wasteful) time schedule to ET. However, Figure 3 shows problems that can occur for both ST and ET in the ET-in-ST case.

First, assume that the ET frame arrives at the queue within the ST time window (e.g., S ¹ ₃ ) as shown in (a) of FIG. 3 before transmission of the last ST frame within the ST time window. Once the transmission of S ¹ ₂ is completed, ET occupies the time reserved for S ¹ ₃ even if it is waiting for the transmission of S ¹ ₃ . S ¹ ₃ is then deprived of the opportunity to be transmitted and is delayed until the next ST time window. Additionally, when S ¹ ₃ is pushed and moves to the next ST window, a chain reaction of cascading delays occurs in the next ST frame (e.g. S ² ₁ , S ² ₃ ), which is transmitted in the subsequent ST time window. It has to be.

In addition, as shown in (b) of Figure 3, if an ET frame arrives after the start of the last ST frame within the ST time window, the ET frame is significantly delayed until the next ST time window, and as shown in (a) of Figure 3, the subsequent ST frame The same problem of continuous delay for frames follows. This problem can get worse as more ET frames steal the ST frame's reserved time window.

In other words, if emergency traffic has the highest priority and is transmitted in the ST time window, the civic traffic scheduled for transmission in that window will be delayed until the next ST time window due to the transmission of the emergency traffic, which will be delayed until the next ST time window. This causes continuous delay problems for other civic traffic that must be transmitted in the time window.

Figure 4 is a diagram showing problems that may occur when emergency traffic transmission is allowed in both ST and NST time windows.

[Case 3] Emergency traffic (ET-in-ST & NST) in both ST and NST time windows: The average delay time of ET is reduced, but ET may still be delayed by GB period. In the worst case, when there is sustained transmission of the maximum Ethernet frame size near the end of the NST time window, when an ET frame arrives, ET may be delayed by up to 246.72 μs, twice the GB period (on a 100 Mbps link, the GB period is 123.36 μs; transmission time of the maximum Ethernet frame size). More importantly, the same successive delay issues will follow for subsequent ST frames as in the ET-in-ST case.

Although it is possible to allocate a separate time window for emergency traffic during scheduling, it is very difficult to allocate an appropriate window size and location due to the sporadic nature of emergency traffic. Even if a random window is assigned, if emergency traffic does not occur, bandwidth for that window is wasted, and the waiting time for emergency traffic to be transmitted to that window may result in a relatively large delay.

Reconfiguring the existing reserved schedule according to changes in the network environment and traffic requires a lot of time and complex calculations. In particular, it is necessary to reconfigure the schedule as soon as non-isochronous emergency traffic occurs and to reconfigure the schedule for each network device (e.g., switch/ Rapid redeployment to bridges (including switches/bridges) requires a lot of money and resources, and is very difficult and complex.

Therefore, the present invention is compatible with the existing TSN (IEEE 802.1Q) standard and TSN through a new scheduling rule for immediate delivery of emergency traffic and a dynamic scheduling technique to protect network reserved civil traffic and ensure requirements. We want to enable dynamic operation without additional modification of the TAS schedule previously reserved for the underlying network device.

Figure 5 is a schematic configuration diagram of a time control scheduling system according to an embodiment of the present invention, Figure 6 is a flow chart of a time control scheduling method according to an embodiment of the present invention, and Figure 7 is an all time control schedule including a guard band. This is an example of emergency traffic being transmitted in a window, Figure 8 is a detailed flowchart of the dynamic scheduling control process among the time control scheduling methods, Figure 9 is a diagram showing an operating procedure for dynamic scheduling time window expansion, and Figure 10 is a dynamic scheduling This is an example of emergency traffic transmission when time window expansion is applied.

The time-controlled scheduling system 100 according to an embodiment of the present invention includes a traffic transmission unit 110, an emergency traffic determination unit 120, a queuing rule adjustment unit 130, and a dynamic scheduling control unit 140.

The traffic transmission unit 110 causes civic traffic and other traffic to be transmitted according to a pre-booked TAS schedule (step S240).

The emergency traffic determination unit 120 determines whether emergency traffic requiring immediate transmission is input in addition to the previously reserved civic traffic and other traffic (step S200). When emergency traffic is input, the emergency traffic determination unit 120 may activate the queuing rule adjustment unit 130 and the dynamic scheduling control unit 140 to ensure immediate delivery of the emergency traffic.

The queuing rule adjuster 130 may define a new scheduling rule for emergency traffic to be transmitted immediately with minimal delay and redefine the priority-to-traffic class mapping table. The mapping table of priority to traffic class may be a mapping table proposed in the IEEE 802.1Q standard.

Traffic with different priorities can be mapped to the same traffic class and assigned to the same queue. At this time, emergency traffic must wait for the transmission of traffic ahead of it. Additionally, if multiple gates are in an open state, emergency traffic may experience delays due to strict priority rules of the transmission selection module connected to each gate.

For example, in the current IEEE 802.1Q standard, AVB traffic (SR classes A and B) is defined to have the highest traffic class, with priority values of 3 and 2. If the number of queues available on the switch is 8, AVB-A and AVB-B traffic are assigned classes 7 (highest) and 6, respectively. And if multiple priority values are mapped to the same traffic class, frames with corresponding priority values are placed in the same transmission queue. In this case, low-priority frames can interfere with and delay other high-priority frames because they are in the same traffic class queue.

Therefore, so that emergency traffic can always be transmitted preferentially in transmission competition, the queuing rule adjustment unit 130 assigns the highest traffic class to the emergency traffic (step S205) and allocates the emergency traffic to an independent queue for this (step S205). Step S210).

[Table 1]

Table 1 redefines the priority-to-traffic class mapping table in the IEEE 802.1Q standard for the time-controlled scheduling system according to this embodiment.

The blue shaded portion (priority 7) indicates the traffic class (i.e., queue number) of the ET to which the ET priority is mapped according to the number of queues (traffic class) supported by the switch. The red shaded areas (priority 2, 3) represent traffic classes for which the priority should be remapped to AVB traffic types without modifying the priority for AVB.

Unlike the standard that maps the highest traffic class to AVB, in this embodiment AVB is mapped to the second highest traffic class after ET. This is to assign independent traffic classes (queues) to ET without overlapping different priorities, regardless of the number of queues supported by the switch. For example, if the number of queues supported by the switch is 5 and the priority value of ET is 7, ET is mapped to traffic class 4 (highest). AVB-A and AVB-B are mapped to the next highest values, 3 and 2, respectively.

In addition, in order to minimize the delay experienced by emergency traffic and enable transmission at all times, the queuing rule adjuster 130 always sets the gate status of the queue to which emergency traffic is assigned to open in all time windows including the guard band (GB) during scheduling. It can be maintained (step S215).

The TAS gate corresponding to ET must always be open, including the GB time window. As mentioned above by way of example, if ET can only be transmitted in the ST or NST time window, the ET will experience significant delay while waiting for that time window, and conversely affect other traffic, as shown in Figures 2 and 3. can affect As shown in Figure 4, even if ET is allowed to transmit in both ST and NST time windows, a worst-case delay of twice the size of GB can occur across all switches.

Therefore, in this embodiment, in order to reduce the delay of ET, the gate for ET is always open and transmission is allowed in all time windows including GB. The only remaining source of delay for ET is the time waiting for the previous transmission to complete, which can be handled by a frame preemption technique.

By allowing ET transmission in all time windows, latency is reduced because ET can be transmitted immediately without waiting even if it arrives during the GB time window, as shown in Figure 7. However, since it may still interfere with ST traffic, this problem is solved through the dynamic scheduling control unit 140, which will be described later.

If an ET frame begins transmission near the end of the GB window, it may invade the ST time window, as illustrated in Figure 7. In this case, the ST time window may expire before S ² ₃ transmission due to S ² ₁ delay. Therefore, S ² ₃ is pushed to the next ST window, resulting in significant sequential delay for subsequent ST frames, as in the ET-in-ST and ET-in-ST & NST cases.

Therefore, in this embodiment, the ST time window can be temporarily and dynamically expanded to correct the requirements and solve the extreme delay problem experienced by civic traffic due to the transmission of emergency traffic through the dynamic scheduling control unit 140 (dynamic scheduling control unit 140). ST time window expansion technique).

The dynamic scheduling control unit 140 includes a transmission selection module and a gate control list. When emergency traffic arrives at the queue of the network device, the transmission selection module begins preparations to preferentially transmit the emergency traffic according to the above-mentioned rules.

At this time, the transmission selection module transmits a time window extension (TWE) message including the frame size of the emergency traffic to the gate control list (GCL) and then transmits the emergency traffic to the network (step S220).

Referring to FIG. 8, the gate control list calculates the time required to transmit emergency traffic (step S2201).

Time window expansion control is performed according to the type of the current time window (step S225). In detail, if the current time window is the ST time window (step S2251), the calculated time is accumulated in the extension request time (τ _ext ) (step S2252). If not, the extension request time (τ _ext ) is updated, and the current time is recorded in lastTimestamp to record the time at which the ST time window should be confirmed (step S2253).

Afterwards, window delay control is performed according to the type of the next time window (step S230). In detail, the gate control list temporarily delays the window by the expansion request time (τ _ext ) when the next time window is the NST time window (step S2301) (step S2302). If not, it is updated with the remaining time obtained by subtracting the start time of the next time window from the sum of the extension request time (τ _ext ) and lastTimestamp and changes the state to the next time window (step S2303).

Referring to Figure 9, an operating procedure for dynamic scheduling time window expansion is shown. 1) When an ET frame arrives at the queue, 2) the transmission selection module selects ET over other frames according to the rules defined as described above. Then, 3-1) signal the gate control list by transmitting a TWE message containing the size of the ET frame to be transmitted, and 3-2) transmit the ET frame. Next, 4) the gate control list calculates the transmission time of the ET frame and accumulates it in the variable τ _ext , which represents the time at which the ST window can request expansion. After that, 5) it is time for the gate control list to change the gate state to the next state, and if the next state is an NST time window, the gate control list delays the transition by τ _ext . That is, since the state of the gate does not change for another τ _ext , the ST time window period is temporarily extended by the time using the ET frame.

Algorithm 1 is pseudo-code for a method of expanding the scheduled time window within the gate control list in the dynamic scheduling control unit 140.

Let us consider the frame arrival example shown in Figure 10. At the bottom of FIG. 10, the pre-schedule line and the eTAS line represent a predefined (fixed) schedule and a schedule dynamically adjusted by the dynamic scheduling control unit 140, respectively.

When an ET frame arrives in the queue and there is no transmission in progress, the transmission selection module selects the ET frame and passes a TWE message to the gate control list immediately before starting transmission. When the gate control list receives a TWE message in the transmission selection, the gate control list calculates the transmission time of the ET frame and updates τ _ext (lines 1-9 of Algorithm 1). When transmission of an ET frame starts in the ST time window (e.g., ET1 in Figure 10), the gate control list accumulates the ET transmission time in τ _ext (row 6). Conversely, if ET transmission starts in a GB or NST time window (e.g., ET2 or ET3 in Figure 10), the gate control list replaces τ _ext with the transmission time of ET and records the current time in lastTimestamp. (Lines 8-9). The dynamic scheduling control unit 140 repeats this process until it moves to the next time window.

In the dynamic scheduling control unit 140, the gate control list determines whether to extend the time window when receiving a time window transition (TWT) message from the gate scheduler (lines 12-29). The gate scheduler is a component that repeats the gate control list using a cycle timer according to a predefined schedule, and sends a TWT message to the gate control list when it is time to transition to the next state in the original schedule. Then, the dynamic scheduling control unit 140 first copies the information about the next time window (the next row of the gate control list) and maintains the predefined schedule as is (line 13). If the current time window is for ST, and there is an ET transmitted during this window, the gate control list delays the start time of the next time window (i.e. the NST time window) by τ _ext and reschedules the next transition, with τ _ext Initialize to 0 (lines 15-19). This means that the gate state has not changed at this time because the gate control list has not transitioned to the next time window.

If the current time window is GB or NST, the gate control list starts transmission in the current time window and checks whether there is an ET frame to complete in the next window (line 20). Taking ET2 in Figure 10 as an example, if the sum of lastTimestamp (ET2 transmission start time) and τ _ext (ET2 transmission time) is greater than the start time of the next window, it means that ET2 will use a certain time of the next window time. Therefore, the gate control list recalculates τ _ext to get only the times when the ET2 transmission overlaps the next time window and initializes lastTimestamp to 0 (lines 21-23). Otherwise (e.g. ET3), τ _ext and lastTimestamp are reinitialized (lines 24-25). After the above process, the gate control list changes the state of the gate to the next time window (line 29).

Therefore, as can be seen from the eTAS timeline in Figure 10, the ST time window affected by the occurrence of ET1 and ET2 is temporarily expanded, and conversely, the NST time window is decreased by that amount. Through this, the ST frame delayed due to ET can be transmitted within the ST time window extended with a delay equal to the size of the ET frame. Accordingly, the ST delay problem described above can be alleviated, and the continuous delay problem can be solved. If ET does not occur, the dynamic scheduling control unit 140 does not call the above process and operates in the same way as standard TAS (step S240) to strictly change the gate state to the next time window according to the pre-scheduled gate control list. (Line 12 → Line 29).

In this embodiment, emergency traffic can be transmitted in any time window, and by assigning the highest traffic class and assigning it to an independent queue, it is always selected first in transmission competition between traffic with different priorities, and is immediately selected with minimal delay. Transmission can be made possible.

By delaying the NST time window by the emergency traffic transmission time τ _ext , transmission space for civic traffic is secured in the ST time window, thereby solving the continuous delay problem of civic traffic due to emergency traffic and meeting the requirements and performance of the civic system. can be corrected.

It is compatible with the existing TSN (IEEE 802.1Q) standard and instantaneously expands the ST time window space by the time τ _ext used only when transmitting emergency traffic, allowing reconfiguration and reconfiguration of the TAS schedule pre-booked for each TSN-based network device. /or can operate dynamically without redistribution.

According to this embodiment, even if 2000 emergency traffic bursts per second occur in the network, high reliability, low end-to-end delay, and uniform and low delay deviation are achieved for both emergency traffic and civil traffic, thereby providing a real-time system even in emergency situations. stable performance can be guaranteed.

In the following, we first demonstrate that emergency traffic (ET) has a significant negative impact on scheduled traffic (ST) in civic networks (TSNs) using TAS, and through an extensive set of simulations under various scenarios and configurations and burst conditions, we demonstrate that the standard We will evaluate the effectiveness of the time control scheduling system and method according to this embodiment by comparing it with TAS. For evaluation, we simulate a TSN-based intelligent driver assistance system (ADAS) scenario for autonomous driving.

Figure 11 is a diagram showing the network topology of a TSN-based ADAS scenario.

Table 2 shows the configuration of traffic flow for simulation.

[Table 2]

The ADAS network consists of 11 transmitters, 9 receivers and 4 switches, and all nodes in the network are interconnected by each meter-long 100 Mbps bandwidth Ethernet link. Each flow is assigned a fixed value of priority and traffic class using eight queues according to the mapping proposed in Table 1, such that ET has the highest traffic class as shown in Table 2.

Since it is not difficult to achieve low latency in a lightly loaded network, we aim to simulate a very congested network. For a fair and controlled experiment, the total link utilization between switches should be approximately To adjust to 80%, three BE (best effort) traffic transmitters and one BE traffic receiver are added, and the transmission interval (data rate) of each BE flow is adjusted as shown in Table 3.

[Table 3]

The non-isochronous emergency traffic (i.e., ET) transmitter generates randomly uniform ET frames within each T seconds, once per T seconds during the simulation time. Here, T can be varied (T = 1 second in the baseline simulation). The simulation time is set to 60 seconds.

Finally, the cycle time of the gate control list is set to the least common multiple of all frame intervals of the ST flow (i.e., the cycle is set to 500 μs). The gate control list for each switch is calculated considering transmission, propagation, and processing delays. In the framework of this experiment, the propagation delay (Delayprop) is basically set to length(m) = (2 x 10 ⁸ m/sec) and the processing delay (Delayproc) is set to 8 ㎲ in the switch. Therefore, the expected end-to-end delay time (E-latency _e2e ) can be calculated as follows.

[Equation 1]

Among the 500㎲ cycle time in the gate control list, the ST time window is set equal to the sum of the transmission times of two ST frames (106.72㎲), and the GB time window is set to the transmission time of the largest frame in the network (i.e. BE, 123.04㎲). ) is set. The remaining time in GCL is allocated to the NST time window (270.24 μs).

We first evaluate the baseline performance of the ADAS scenario without ET (without ET) and then compare this to the results for each case discussed previously.

Baseline without ET (w/o-ET): Table 4 shows the latency, delay deviation, and throughput performance of each traffic type in the ADAS scenario without ET.

[Table 4]

In the simulation setup, since both ST flows start transmission at the same time, frames from both ST flows arrive at SW1 at the same time. Therefore, one ST frame can be transmitted immediately without queuing delay, but the other will wait for the transmission time of the ST frame (i.e., 53.36 μs). As a result, in Equation 1, the expected minimum and maximum end-to-end latency (E-latency _e2e ) of the ST frame is 294.025 ㎲ and 347.385 ㎲, respectively. The simulation results in Table 4 are consistent with these estimates, with only a slight difference of +1.79 μs. ST's throughput also matches the intended data rates in Table 2. This means there is no loss or significant queuing delay.

Figure 12 shows delay, delay deviation and throughput results for all scenarios including w/o-ET, ET-in-ST, ET-in-NST, ET-in-ST & NST and eTAS.

ET in the ST time window (ET-in-ST): Figure 12(a) and Figure 12(b) show that ST suffers the greatest delay in the ET-in-ST case than in other cases where ET occurs . The worst delay time and delay deviation of ST are respectively 15349㎲ and It is measured at 3755㎲, which is about 44 times higher than w/o-ET, and the delay deviation is incomparably higher. This is because ET takes away the ST's transmission opportunity, causing a cascading delay problem as shown in Figure 3, and the ST's queuing delay continues to accumulate every time an ET occurs. For ET, max. Delay time and delay deviation are respectively 736㎲ and It is measured at 121㎲, which is higher than other scenarios. This is because, as shown in Figure 3 (b), if ET arrives at the queue while the last ST is transmitting, it must wait until the next ST time window. Therefore, the queuing delay of ET is Since it can be as large as 447㎲ (the sum of the transmission time of the ST frame and the period of the NST and GB time windows), the maximum E-latency _e2e of the ET frame is in the worst case up to It can be 741㎲.

The throughput results in Figure 12(c) show that for the ET-in-ST case, the throughput achieved by ET (5.176 kbps) is exactly the same as that lost by ST compared to the w/o-ET case. The throughput of other NST flows did not change at all. This result means that ET was transmitted and only stole the opportunity of the ST time window.

ET in the NST time window (ET-in-NST): In this case, the ET does not interfere with the transmission of the ST and therefore, by definition, does not affect the ST. However, ET may still experience significant queuing delays due to ongoing NST transmissions at switches along the ET flow path, as well as ST and GB periods. Therefore, in the worst case, the maximum E-latency _e2e of ET is Minimum 893㎲, 3 hops x 123.04㎲ (maximum BE frame size) It can be greater than 524㎲ (sum of NST and GB time windows + transmission/processing/propagation delay of ET frame). Maximum in simulation. The delay time and delay deviation of ET are respectively 656㎲ and Measured at 69㎲, it is longer than the ET-in-ST case. 11% and 43% lower than theoretical minimum ( It is measured to be approximately twice as large as the value that can be obtained with 294㎲). The delay and delay deviation of other NST flows also slightly increase compared to the case without ET (w/o-ET case).

For throughput, all flows meet the requirements. That is, there is no throughput loss. This is because the ADAS simulation setup was configured to appropriately configure the data rate of the BE flow to have a baseline link utilization of 80% (Table 3). The remaining available time is by definition in the NST time window, and ET uses that time for transmission.

ET in both ST and NST time windows (ET-in-ST & NST): ET can be transmitted in both ST and NST time windows, so ET is expected to have lower latency than the previous two cases, but ET can be transmitted in both ST and NST time windows. may interfere with ET may be delayed by ongoing transfers and GB, resulting in an expected worst-case maximum E-latency _e2e of up to It is 786㎲. In the simulation results, ET is maximum. The delay time is 549㎲, delay deviation is At 51 μs, it is lower than other cases using standard TAS, as expected and shown in Figures 12(a) and 9(b). However, ST still experiences significant delays due to cascading delay issues, as in the case of ET-in-ST. The worst delay time and delay deviation of ST are respectively 9349㎲ and It is measured at 2116㎲. than ET-in-ST case 39% and Although it is 44% lower, the delay time is about 27 times higher than the w/o-ET and ET-in-NST cases, and the delay deviation is too high to meet the requirements of TSN.

The throughput of ET and NST is the same as the other cases, but as shown in Figure 12(c), the throughput of ST is higher than that of w/o-ET. 3kbps is low. This is because ET was allowed to transmit in either the ST or NST window, so ST lost some opportunities due to ET, but NST had additional space (80%) to utilize the link.

So far, standard TAS has shown serious performance problems not only with ET, but also with ST when ET is introduced into TSN networks. Now, the performance of the time control scheduling system 100 (hereinafter also referred to as eTAS) according to this embodiment is compared with the standard TAS.

Unlike standard TAS, eTAS allows ET transfers in any time window, including GB. Therefore, assuming that the ET is transmitted as soon as it arrives at the queue (zero queuing delay), the theoretical minimum end-to-end delay time (E-latency _e2e in Equation 1) of the ET in the topology is 294.025 μs. The delay that ET may experience due to ongoing transmission is 369.12 μs (3 hops in this scenario), so the expected maximum E-latency _e2e is up to It is 663㎲.

The simulation results for eTAS in Figure 12 show that the minimum and maximum delay times of ET are respectively 294㎲ and It shows that it is 486㎲ and the median value is 346.152㎲. The minimum matches the estimate and the maximum is better than the worst case. It is 177㎲ lower and is about 34%/26%/11% lower than ET-in-ST/ET-in-NST/ET-in-ST & NST. The median is closer to the minimum. Additionally, the delay deviation of ET is about 44㎲, which is 63%/36%/13% lower than each scenario, respectively. In other words, eTAS can provide ET with a delay time close to the theoretical minimum with low delay deviation.

eTAS can minimize the impact of ET on ST by temporarily extending the ST time window. The maximum delay time of ST is in eTAS It is measured at 403 μs, which is about 97% and 96% lower than the ET-in-ST and ET-in-ST & NST cases respectively, but 15% higher than the w/o ET and ET-in-NST cases. The average delay time was 348.096㎲, 17.7% (17.7%) longer than the minimum (295.785㎲). 52㎲) is higher. Also max. The ST delay increase is exactly 53.36 μs, the ET frame size. This means that the ST frame is transmitted immediately after the ET frame transmission is completed, which means that eTAS is operating as designed. The average delay deviation of ST is It is measured at 0.02㎲, which is small enough to meet the requirements of TSN. Additionally, the throughput of ST was not affected by ET. This means that eTAS can provide ET with minimal impact on ST and minimal latency, and the results show that it achieves better performance than all other cases using standard TAS.

Finally, the adequacy of NST performance is worth mentioning. Using eTAS, the worst-case latency for AVB-A and AVB-B is measured to be approximately 66% and 78% higher than the w/o-ET case, respectively. However, the AVB standard defines target maximum worst-case latency for AVB-A and AVB-B as 2ms and 50ms, respectively, over 7 hops when using a 100Mbps Ethernet link. TSN requirements are met when estimating the latency proposed in the AVB standard. The delay variance of AVB traffic increases slightly with eTAS, but the standard does not require the delay variance of AVB to meet the maximum delay requirement. Additionally, the throughput of all NST flows using eTAS is the same as the w/o-ET case, so there is no impact at all.

According to a document from the Industrial Internet Consortium, there are three types of sporadic event traffic (alarms, operator commands and controls) with different requirements for industrial systems. For the 'Alarm' type, a burst of up to 2,000 frames within 1 second must be guaranteed, after which some loss may be tolerated. Therefore, to determine the effectiveness of eTAS in bursty ET flows, simulations are performed with different ET burst sizes B (frames per second, 1 to 2000) and payload sizes P (46 to 1500, minimum to maximum Ethernet payload size). The simulation time is set to 60 seconds and each scenario is repeated 10 times. During the simulation, the ET transmitter randomly selects burst occurrence times every 20 seconds from the start of the simulation. It then randomly generates B frames of size P at selected times using a uniform real distribution. The flows for all other traffic types are the same as before in Table 2. When P is 1500 bytes and B is 2000, the link utilization exceeds 100% in the worst case, and some loss is inevitable.

ET performance using burst ET: Figure 13 shows the delay, delay deviation and throughput performance of ET in simulations when introducing burst ET with various burst and frame sizes. Based on the same analysis performed so far, the predicted worst-case end-to-end latency (E-latency _e2e ) for ET when P is 1500 bytes is 1013.465 μs. ET's predicted worst-case E-latency _e2e decreases by 200 μs for every 500 bytes of decrease in P. Figure 13 (b) shows that when B is greater than 1, the maximum delay time of ET measured slightly increases, but is still lower than the theoretical worst case E-latency _e2e of Equation 1. In addition, even if B increases from 500 to a maximum of 2000, the delay time (Figure 13 (a)) and delay deviation (Figure 13 (c)) are relatively uniform, which means an increase in delay time (at B = 1). It is only due to self-interference between ET frames, and eTAS can support ET without interference from other traffic types. It can also be seen from Figure 13(d) that the average throughput of ET increases appropriately in proportion to P and B for each scenario, as desired. These results show that eTAS can handle ET bursts of 2000 frames per second and guarantee low latency without performance degradation regardless of burst size.

ST performance using burst ET: Figure 14 shows the performance of ST using eTAS according to the burst of ET. The results show that the worst delay and delay deviation of ST gradually increase as P and B increase (Figures 14(b) and 14(c)). This is because ST frames may need to wait for successive frames of burst ET flows, and their intensity increases as P and B increase. But max. The delay time of ST is when P is 1500 bytes and B is 2000 (when link utilization is 100%) measured up to 634 ㎲, but this is still significantly longer than for ET-in-ST & NST, where only one ET is transmitted per second ( 83%) low. Worst case delay deviation ( 66㎲) is also higher than the ET-in-ST & NST case. 97% lower. For the throughput of ST (Figure 14(d)), the maximum when P exceeds 1000 bytes and B exceeds 1000 It decreases slightly to 26bps (see y-axis scale). However, this causes a slight momentary decrease due to the burst of ET, and the ST regains performance after the burst of ET has passed. Ultimately, all frames in the ST are delivered to their destination without being accumulated in the queue.

NST performance using burst ET: So far, we have seen that eTAS guarantees better performance for ET and ST compared to what can be achieved with standard TAS. Finally, Figure 15 shows results for NST flows that are not protected by TAS or eTAS by design. First, the latency of AVB-A is affected by the increase of P and B, but the worst-case latency of all configurations is sufficiently lower than the requirements of the AVB standard (red horizontal line in Fig. 15(a)). The throughput of AVB-A may decrease slightly as shown in Fig. 15(d), but the difference with w/o-ET is maximum in all scenarios. It's only 35bps, which is negligible in the context of audio traffic. On the other hand, the performance of AVB-B and BE traffic is greatly affected by both P and B. In particular, when P is more than 1000 bytes, the delay time of AVB-B is higher than the target value of AVB-B's maximum E-latency _e2e (red horizontal line in Figure 15(a)). The throughput of AVB-B and BE also shows a linear decrease, as shown in Figures 15(e) and 15(f).

The reason is clear. We increased the total bandwidth utilization by introducing large bursts of ET frames, resulting in significant congestion. Link utilization exceeds 100% in the worst case (P = 1500 bytes, B = 2000). Increasing link utilization reduces the likelihood that lower priority traffic, such as AVB or BE, will be selected for next transmission. This is a common starvation problem in priority-based queuing and transmission selection methods. Among AVB-A, AVB-B, and BE, AVB-A has the least impact because it has the highest traffic class among NST flows. Nevertheless, the AVB standard requires performance guarantees only when bandwidth utilization is less than 75% and there are no further requirements. eTAS achieves very low latency to ET while meeting the requirements of TSN for ST and AVB-A traffic even in highly congested (almost 100%) scenarios.

The time-controlled scheduling method described above can also be implemented in the form of a recording medium containing instructions executable by a computer, such as an application or program module executed by a computer. Computer-readable media can be any available media that can be accessed by a computer and includes both volatile and non-volatile media, removable and non-removable media. Additionally, computer-readable media may include computer storage media. Computer storage media includes both volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules, or other data.

The above-described time control scheduling method can be executed by an application installed by default on the terminal (this may include programs included in the platform or operating system, etc. installed by default on the terminal), and the user can use the application store server, application, or It may also be executed by an application (i.e. program) installed directly on the master terminal through an application providing server such as a service-related web server. In this sense, the above-described time control scheduling method may be implemented as an application (i.e., program) installed by default in the terminal or directly installed by the user and recorded on a computer-readable recording medium such as the terminal.

Although the present invention has been described above with reference to embodiments, those skilled in the art can make various modifications and changes to the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. You will understand that you can do it.

100: Time control scheduling system 110: Traffic transmission unit
120: Emergency traffic determination unit 130: Queuing rule adjustment unit
140: Dynamic scheduling control unit

Claims (8)

As a city control scheduling system in a civic network environment,
A traffic transmission unit that transmits civic traffic and other traffic according to a pre-booked TAS schedule;
an emergency traffic determination unit that determines whether emergency traffic requiring immediate transmission is input;
a queuing rule adjuster that redefines a priority-to-traffic class mapping table when the emergency traffic is input; and
A dynamic scheduling control unit that temporarily and dynamically extends the ST (scheduled traffic) time window corresponding to the transmission time of the emergency traffic,
The queuing rule adjuster assigns the highest traffic class to the emergency traffic,
The queuing rule adjuster allocates the emergency traffic to an independent queue,
The queuing rule adjuster is a time-controlled scheduling system characterized in that the gate state of the queue to which the emergency traffic is assigned is always maintained in an open state in all time windows including the guard band (GB) during scheduling.
delete delete delete As a city control scheduling system in a civic network environment,
A traffic transmission unit that transmits civic traffic and other traffic according to a pre-booked TAS schedule;
an emergency traffic determination unit that determines whether emergency traffic requiring immediate transmission is input;
a queuing rule adjuster that redefines a priority-to-traffic class mapping table when the emergency traffic is input; and
A dynamic scheduling control unit that temporarily and dynamically expands the ST (scheduled traffic) time window according to the transmission time of the emergency traffic,
The dynamic scheduling control unit calculates the transmission time required for transmission of the emergency traffic,
Performs time window expansion control according to the type of current time window,
Then perform window delay control according to the type of time window,
The dynamic scheduling control unit accumulates the transmission time in the extended request time if the current time window is the ST time window, and updates the extended request time if the current time window is not the ST time window and records the current time in lastTimestamp to determine the ST time window. A time-controlled scheduling system characterized by recording the time at which a window should be extended.
delete According to clause 5,
If the next time window is an NST time window, the dynamic scheduling control unit temporarily delays the next time window by the extended request time, otherwise, subtracts the start time of the next time window from the sum of the extended request time and lastTimestamp. A time-controlled scheduling system characterized by updating with the remaining time and changing the state to the next time window.
A time-controlled scheduling method performed by a time-controlled scheduling system in a civic network environment, comprising:
An emergency traffic frame arrives at the queue;
A transmission selection module selecting the emergency traffic frame;
transmitting a time window extension (TWE) message including the size of the emergency traffic frame to a gate control list;
transmitting the emergency traffic frame;
Calculating the transmission time of the emergency traffic frame in the gate control list and accumulating it in a time variable τ _ext where the ST time window may require expansion; and
When it is time for the gate control list to change the gate state to the next state, and the next state is an NST time window, the gate control list delays the transition by τ _ext .