Hybrid scheduling network performance evaluation method and device based on network calculation

Abstract

The invention discloses a hybrid scheduling network performance evaluation method and device based on network calculation. The method comprises the following steps: acquiring a network traffic information set and a traffic transmission network information set; processing the network traffic information set and the traffic transmission network information set to obtain a blocking effect value caused by the CQF traffic to the non-CQF traffic; and processing the blocking effect value caused by the CQF traffic to the non-CQF traffic and the traffic transmission network information set to obtain the worst end-to-end delay of other traffic under the interference of the CQF traffic under the hybrid scheduling architecture.

Description

Technical Field

[0001] This invention relates to the fields of time-sensitive networking and data processing, specifically to a method and apparatus for evaluating the performance of hybrid scheduling networks based on network calculus. Background Technology

[0002] Time-Sensitive Networking (TSN) is an evolving IEEE 802.1 standard designed to provide deterministic low-latency communication over Ethernet to meet the stringent requirements of hybrid mission-critical systems such as industrial automation, automotive electronics, and aerospace. To support different types of traffic flows, the TSN standard defines various traffic scheduling and shaping mechanisms and allows them to be deployed together on the same network switch. These scheduling mechanisms primarily include: Time-Aware Shaper (TAS): Based on a globally synchronized clock, it reserves a dedicated transmission window for time-triggered (TT) traffic through a gate-control list (GCL), providing a deterministic guarantee of ultra-low latency and zero jitter.

[0003] Strict Priority (SP): The most basic scheduling mechanism, which always prioritizes sending event-triggered (ET) traffic from the high-priority queue.

[0004] Credit-Based Shaper (CBS): A shaping mechanism designed for Audio / Video Bridging (AVB) traffic. It uses a credit value algorithm to reserve bandwidth for low-priority traffic, preventing high-priority traffic from "starving" low-priority traffic.

[0005] Cyclic Queuing and Forwarding (CQF) is a scheduling mechanism that uses alternating queues and fixed transmission periods on switches to provide a strict upper bound on end-to-end latency for traffic flowing through multi-hop networks. CQF can be implemented in two ways based on its traffic triggering method: Time-based CQF and Event-based CQF.

[0006] In practical applications, networks typically employ hybrid scheduling architectures (e.g., TAS+CBS+CQF+SP) to accommodate the needs of various traffic types. However, the CQF scheduler's non-preemptive and non-work-conserving characteristics make it a complex source of interference in hybrid architectures. On one hand, the non-preemptive transmission of CQF frames blocks higher-priority traffic (such as CBS or SP traffic); on the other hand, the non-work-conserving nature of CQF (i.e., even if the link is idle, it must wait for a specified period before sending) makes its bandwidth utilization patterns unpredictable, thus causing complex queuing interference for low-priority (LP) traffic.

[0007] Currently, while relatively mature formal analysis methods exist for scheduling mechanisms such as TAS, CBS, and SP, there is a lack of a unified model in the existing technology that can accurately and formally quantify the worst-case interference exerted by CQF schedulers (including both TT-based and ET-based implementations) on other coexisting scheduling mechanisms in hybrid architectures. This technological gap makes it impossible to accurately verify the end-to-end timing performance of hybrid TSN networks containing CQF, hindering their deployment and certification in safety-critical systems. Summary of the Invention

[0008] This invention primarily addresses the problem of accurately verifying the end-to-end timing performance of a hybrid TSN network incorporating CQF. It discloses a hybrid scheduling network performance evaluation method and apparatus based on network calculus.

[0009] In a first aspect, this invention discloses a method for evaluating the performance of a hybrid scheduling network based on network calculus, comprising: S1, collects the network traffic information set and the traffic transmission network information set; S2, process the network traffic information set to obtain the arrival rate of each traffic at the source and the switch, as well as the burstiness at the source. S3 processes the arrival rates of all traffic at the source and the switch, the burstiness of traffic at the source, and the traffic transmission network information set to obtain the worst end-to-end delay of other traffic under CQF traffic interference in the hybrid scheduling architecture.

[0010] The network traffic information set includes information about each traffic transmitted in the traffic transmission network; the traffic information includes traffic ID number, frame length, end-to-end deadline, propagation path, minimum frame interval, scheduling mode information, and aggregated output arrival curve set; the propagation path includes the source end, switch, and end end that the traffic passes through in sequence during its transmission in the traffic transmission network; the aggregated output arrival curve set includes the aggregated output arrival curve of the traffic at each output port. The source, switch, and end are all nodes in the propagation path; each node includes a corresponding output port. The traffic transmission network information set includes a hybrid scheduling architecture information set and a network information set; the hybrid scheduling architecture information set includes the scheduling mode adopted by the hybrid scheduling architecture and the priority information of each scheduling mode; the network information set includes the link rate and the default parameters of all scheduling mode shapers. The default parameters for a CQF shaper include the trigger mechanism mode, CQF cycle length, CQF synchronization guard band length, and the fixed number of CQF bits transmitted per cycle. For time-triggered CQF shapers, the default parameters include the time-over-cycle parameter. and maximum period The triggering mechanism modes include event-triggered (ET) and time-triggered (TT).

[0011] flow f The burstiness at the source Equal to its frame length ,Right now ,flow f The arrival rate at the source is expressed as ,in for f Minimum frame interval; traffic f In the switch The arrival rate is .

[0012] The process of processing the arrival rates of all traffic at the source and the switch, the burstiness of traffic at the source, and the traffic transmission network information set yields the worst-case end-to-end delay of other traffic under CQF traffic interference in the hybrid scheduling architecture, including: S31, for each type of CQF traffic, sort it according to its traffic priority from high to low to obtain a corresponding priority sequence; the priority sequence includes several priorities; set the output port. The output port is the source of the traffic. S32, calculate the output port Upper bound information on the interference of CQF traffic on other types of traffic; S33, at the output port In the context of CQF traffic, if its priority is lower than the i-th priority of the traffic sequence of type SP, but higher than the (i+1)-th priority, then the expression for calculating the minimum service curve of the traffic of the i-th priority of that type is: , in, , This represents the arrival curve of the j-th priority flow in a flow sequence of type SP, where s is a time variable. Let be the frame length of the j-th priority traffic in a traffic sequence of type SP. For output port The upper bound of the blocking effect of CQF traffic on high-priority traffic; At the output port In this context, for CQF traffic, if its priority is higher than the i-th priority of the traffic sequence of type SP, but lower than the (i-1)-th priority, then the expression for calculating the minimum service curve of the traffic of the i-th priority of that type is: , in, For output port The upper bound of the blocking effect of CQF traffic on low-priority traffic; S34 calculates the traffic based on the minimum service curve. f In a single output port Worst case queue delay ; S35, calculate the outflow of traffic f from the current port. Output arrival curve Its calculation expression is: , in, For the preset flow function, if , ,otherwise for ; The minimal additive convolution operator in network computation is formally defined as follows: , For output port The input flow rate arrival curve of the flow rate f; S36, utilizing The current output port is where the flow f enters. The input flow arrival curve of the next port is used to determine the output port. Whether it is the output port at the end of the traffic propagation path, a first determination result is obtained. If the first determination result is yes, execute S37; if the first determination result is no, the output port... For the output port of the current output port at the next node in the propagation path, execute S32; S37, for all output ports The worst-case queuing delays are accumulated to obtain the worst-case end-to-end delays of other traffic under CQF traffic interference in the hybrid scheduling architecture. .

[0013] The output port Upper bound information on the interference of CQF traffic on other types of traffic, including: output port. Upper bound information and output port of the blocking effect of CQF traffic on high-priority traffic. The upper bound of the blocking effect of CQF traffic on low-priority traffic; The output port The expression for calculating the upper bound of the congestion effect caused by CQF traffic on high-priority traffic is: , in, For output port The maximum frame length of all CQF traffic at that location. For output port The upper bound of the blocking effect of CQF traffic on high-priority traffic.

[0014] The output port The expression for calculating the upper bound of the congestion effect caused by CQF traffic on low-priority traffic is as follows: , , in, For output port The upper bound information on the blocking effect of CQF traffic on low-priority traffic. It depends on the output port. The type of CQF traffic in the data is adopted. or The expression, This represents the additional maximum burst of traffic caused by interference from CQF traffic and misalignment between the CQF cycle length and the analysis time window boundary. For output port The maximum frame length for all traffic with a priority lower than CQF. Traffic with higher priority than CQF traffic is routed to the output port. The aggregated output reaches the curve at that point. The length of the CQF synchronous protection belt. C For link speed, The CQF period length, Operation representation .

[0015] The worst-case queuing delay The calculation expression is: , in, It is an operator for calculating the maximum horizontal deviation between two curves, and its formal definition is: Where sup represents the supremum, inf represents the infremum, and d is the computational variable. The traffic of the i-th priority in the traffic sequence of type S at the output port The aggregated output reaches the curve.

[0016] A second aspect of this invention discloses a hybrid scheduling network performance evaluation device based on network calculus, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the hybrid scheduling network performance evaluation method based on network calculus.

[0017] In a third aspect, the present invention discloses a computer-readable storage medium storing computer instructions, which, when invoked by a computer, are used to execute the hybrid scheduling network performance evaluation method based on network calculus.

[0018] In a fourth aspect of this invention, an information data processing terminal is disclosed, which is used to implement the hybrid scheduling network performance evaluation method based on network computation.

[0019] The beneficial effects of this invention are as follows: This invention provides for the first time a formal model that can uniformly analyze TT-based and ET-based CQF interference; this interface model has modularity and can be seamlessly integrated into the network computation and analysis framework of existing schedulers such as SP and CBS, solving the problem of difficulty in verifying end-to-end delay under hybrid scheduling architecture; this invention realizes composable, scalable, end-to-end deterministic timing performance evaluation of complex TSN networks containing CQF. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention; Figure 2 This is a network topology diagram related to Embodiment 3 of the present invention. Detailed Implementation

[0021] To better understand the content of this invention, an embodiment is provided here.

[0022] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention. Figure 2 This is a network topology diagram related to Embodiment 3 of the present invention.

[0023] Example 1: In a first aspect, this invention discloses a method for evaluating the performance of a hybrid scheduling network based on network calculus, comprising: S1, collects the network traffic information set and the traffic transmission network information set; S2, process the network traffic information set to obtain the arrival rate of each traffic at the source and the switch, as well as the burstiness at the source. S3 processes the arrival rates of all traffic at the source and the switch, the burstiness of traffic at the source, and the traffic transmission network information set to obtain the worst end-to-end delay of other traffic under CQF traffic interference in the hybrid scheduling architecture.

[0024] The network traffic information set includes information about each traffic transmitted in the traffic transmission network; the traffic information includes traffic ID number, frame length, end-to-end deadline, propagation path, minimum frame interval, scheduling mode information, and aggregated output arrival curve set; the propagation path includes the source end, switch, and end end that the traffic passes through in sequence in the traffic transmission network; the aggregated output arrival curve set includes the aggregated output arrival curve of the traffic at each port; The source, switch, and end are all nodes in the propagation path; each node includes a corresponding output port. The traffic transmission network information set includes a hybrid scheduling architecture information set and a network information set; the hybrid scheduling architecture information set includes the scheduling mode adopted by the hybrid scheduling architecture and the priority information of each scheduling mode; the network information set includes the link rate and the default parameters of all scheduling mode shapers. The default parameters of a CQF shaper include the trigger mechanism mode, CQF cycle length (in µs), CQF synchronization guard band length (in bits), and the fixed number of CQF bits transmitted per cycle. For time-triggered CQF shapers, i.e., TT-based CQF, the default parameters include the time-over-cycle parameter. (Unit: µs) and maximum period (Unit: µs); The triggering mechanism modes include event-triggered (ET) and time-triggered (TT); The traffic transmission network can be a TSN network, which adopts a hybrid scheduling architecture when transmitting information.

[0025] The traffic corresponding to each scheduling mechanism is referred to as scheduling mechanism traffic, such as CQF traffic, CBS traffic, and TAS traffic.

[0026] flow f The burstiness at the source Equal to its frame length ,Right now ,flow f The arrival rate at the source is expressed as ,in for f Minimum frame interval; traffic f In the switch The arrival rate is .

[0027] The process of processing the arrival rates of all traffic at the source and the switch, the burstiness of traffic at the source, and the traffic transmission network information set yields the worst-case end-to-end delay of other traffic under CQF traffic interference in the hybrid scheduling architecture, including: S31, for each type of CQF traffic, sort it according to its traffic priority from high to low to obtain a corresponding priority sequence; the priority sequence includes several priorities; set the output port. The output port is the source of the traffic. S32, calculate the upper bound information of the interference of CQF traffic on other types of traffic; S33, at the output port In the context of CQF traffic, if its priority is lower than the i-th priority of the traffic sequence of type SP, but higher than the (i+1)-th priority, then the expression for calculating the minimum service curve of the traffic of the i-th priority of that type is: , in, , This represents the arrival curve of the j-th priority flow in a flow sequence of type SP, where s is a time variable. Let be the frame length of the j-th priority traffic in a traffic sequence of type SP. For output port The upper bound of the blocking effect of CQF traffic on high-priority traffic; At the output port In this context, for CQF traffic, if its priority is higher than the i-th priority of the traffic sequence of type SP, but lower than the (i-1)-th priority, then the expression for calculating the minimum service curve of the traffic of the i-th priority of that type is: , in, For output port The upper bound of the blocking effect of CQF traffic on low-priority traffic; S34 calculates the traffic based on the minimum service curve. f In a single output port Worst case queue delay , The calculation expression is: , in, It is an operator for calculating the maximum horizontal deviation between two curves, and its formal definition is: Where sup represents the supremum, inf represents the infremum, and d is the computational variable. The traffic of the i-th priority in the traffic sequence of type S at the output port The aggregated output reaches the curve; S35, calculate the outflow of traffic f from the current port. Output arrival curve Its calculation expression is: , in, For the preset flow function, if , ,otherwise for ; The minimal additive convolution operator in network computation is formally defined as follows: , For output port The input flow rate arrival curve of the flow rate f; S36, utilizing The current output port is where the flow f enters. The input flow arrival curve of the next port is used to determine the output port. Whether it is the output port at the end of the traffic propagation path, a first determination result is obtained. If the first determination result is yes, execute S37; if the first determination result is no, the output port... For the output port of the current output port at the next node in the propagation path, execute S32; S37, for all output ports The worst-case queuing delays are accumulated to obtain the worst-case end-to-end delays of other traffic under CQF traffic interference in the hybrid scheduling architecture. ; The worst end-to-end delay The expression is: , in It's traffic. f The upper bound of the end-to-end latency, It's traffic. f The set of all output ports on the transmission path. On a single port Calculate the flow rate f The worst-case scenario is queuing delays.

[0028] The calculation of the upper bound information of CQF traffic based on the blocking effect value of CQF traffic on non-CQF traffic includes: S321, for TT-based CQF traffic, a supercycle contains One CQF cycle, , will the n The number of CQF bits transmitted in a fixed number of CQF cycles is expressed as follows: The expression for its input flow arrival curve is: in, , Indicates the time length t at the output port Maximum number of arriving bits for TT-based CQF traffic at the location Indicates the first The fixed number of CQF bits transmitted within a CQF cycle, where % represents the modulo value, k represents the number of cycles index, and n represents the number of CQF cycles index.

[0029] S322, for ET-based CQF traffic, the expression for its input traffic arrival curve is: , in, Indicates rounding down. This indicates the time length t at the output port. Maximum number of arriving bits for ET-based CQF traffic at the location , For traffic The intermediate function, whose expression is: , Indicates flow rate Frame length, Indicates flow rate The minimum frame interval, Indicates output port The CQF traffic set in the middle.

[0030] S323, based on the flow arrival curve, calculates the upper bound information of CQF flow; The output port Upper bound information on the interference of CQF traffic on other types of traffic, including: output port. Upper bound information and output port of the blocking effect of CQF traffic on high-priority traffic. The upper bound of the blocking effect of CQF traffic on low-priority traffic; The output port The expression for calculating the upper bound of the congestion effect caused by CQF traffic on high-priority traffic is: , in, For output port The maximum frame length of all CQF traffic at that location. For output port The upper bound of the blocking effect of CQF traffic on high-priority traffic; The expression for calculating the upper bound of the blocking effect of CQF traffic on low-priority traffic is: , , in, For output port The upper bound information on the blocking effect of CQF traffic on low-priority traffic. It depends on the output port. The type of CQF traffic in the data is adopted. or The expression, This represents the additional maximum burst of traffic caused by interference from CQF traffic and misalignment between the CQF cycle length and the analysis time window boundary. For output port The maximum frame length for all traffic with a priority lower than CQF. Traffic with higher priority than CQF traffic is routed to the output port. The aggregated output reaches the curve at that point. The length of the CQF synchronous protection belt. For link speed, The CQF period length, Operation representation . It is the arrival curve of the CQF cyclic alignment interval length.

[0031] The calculation of the upper bound information on the interference of CQF traffic on other types of traffic also includes: When the non-CQF traffic to be calculated has a higher priority than the CQF traffic, the statistics are obtained at the output port. In the process, before the start of the busy period for the non-CQF traffic to be calculated, the frame length of a CQF traffic currently being transmitted is determined, and the frame length is used for the output port. The blocking effect of CQF traffic on non-CQF traffic in the output port. The expression for the blocking effect of CQF flow on non-CQF flow is: , in, Indicates output port In this context, the frame length of a CQF traffic being transmitted before the start of a high-priority traffic busy period.

[0032] In all embodiments of the present invention, the values ​​of the independent variables in the input of all computational expressions or mathematical functions meet the reasonable requirements of the input range of the computational expressions or mathematical functions, and can ensure that the computational expressions or mathematical functions can be calculated smoothly without violating physical laws or mathematical rules.

[0033] A second aspect of this invention discloses a hybrid scheduling network performance evaluation device based on network calculus, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the hybrid scheduling network performance evaluation method based on network calculus.

[0034] In a third aspect, the present invention discloses a computer-readable storage medium storing computer instructions, which, when invoked by a computer, are used to execute the hybrid scheduling network performance evaluation method based on network calculus.

[0035] In a fourth aspect of this invention, an information data processing terminal is disclosed, which is used to implement the hybrid scheduling network performance evaluation method based on network computation.

[0036] Example 2: This invention discloses a hybrid scheduling network performance evaluation method based on network calculus, comprising: Step 1: Read flow information and shaper parameters Step 1-1: Read traffic information Based on the input file, read the information of each time-critical traffic, including traffic ID, maximum frame length (in bytes), end-to-end deadline (in microseconds), propagation path, minimum frame interval (in microseconds), and scheduling mode information.

[0037] Traffic transmission networks consist of nodes, which include source nodes, switches, and endpoints. Step 1-2: Read TSN hybrid architecture information The input file reads detailed information about the selected hybrid architecture. The hybrid architecture can consist of different scheduling modes, such as TT-based CQF, ET-based CQF, TAS, CBS, and SP, as well as different numbers of priorities (maximum of 8).

[0038] Steps 1-3: Read link and shaper parameters Based on the input file, read the link rate information and the default parameters of the scheduling mode shaper. The link rate is consistent across the entire network (in Mbit / s). The default parameters of the TAS shaper include the integration mode selection (GuardBand / Preemption) and the credit volume mode selection (Frozen / NonFrozen) for CBS during GuardBand due to TAS interference. CBS Shaper's default parameters include different priorities. i idle slope (Priority) i The bandwidth represents the proportion of the total network bandwidth; the default parameters of the CQF shaper include the trigger mechanism mode (ET / TT), the CQF cycle length (in µs), the CQF synchronization guard band length (in bits), and the fixed number of CQF bits transmitted per cycle; in particular, for TT-based CQF, an additional parameter, CQF overcycle, is also included. (Unit: µs), CQF maximum period (unit: µs). Steps 1-4: Preprocessing The read information is preprocessed, and template data structures for output ports, queues, and time-critical traffic under different scheduling modes (CQF, TAS, CBS, SP) are constructed to store the corresponding information. Store the default parameters of the corresponding shaper and calculate the burstiness and rate of each time-critical data stream at the source.

[0039] For data streams f In terms of its suddenness at the source... Equal to its maximum frame length ,Right now The arrival rate at the source can be expressed as ,in for f The minimum frame interval; its burstability at the switch is unknown, but the arrival rate is still [missing information]. .

[0040] For time-triggered scheduling mode (TT-based CQF, TAS), the port information also needs to record their GCL information. The GCL of TAS records the gating start and end times of different TAS windows within the time-out period; the GCL of TT-based CQF records the CQF period ID and corresponding frame length of all CQF data frames within the time-out period. The aggregation output arrival curve for traffic with higher priority than CQF, i.e., the maximum number of bits of CQF traffic arriving at h within the time range t.

[0041] Step 2: Construct a real-time interface abstraction for circular queuing and forwarding (CQF) The goal of this step is to build a mathematical model to abstract the impact of the CQF scheduler on other coexisting traffic in the network.

[0042] Step 2-1: Construct a CQF real-time interface model for high-priority traffic In the TSN / HP+CQF+LP hybrid scheduling architecture, the interference of CQF on higher-priority traffic (such as traffic scheduled by SP or CBS) mainly stems from its non-preemptive transmission characteristics. Once a CQF frame begins transmission, even if a higher-priority frame arrives, it must wait for the current CQF frame to complete transmission. This interference is abstracted into a real-time interface function. It indicates the output port of the node. The value of the blocking effect of CQF on high-priority traffic. The formal definition of this interface is: in, Indicates output port In this model, the length of a CQF frame being transmitted before the start of a high-priority traffic busy period is considered. This model captures the worst-case congestion latency caused by a single CQF frame.

[0043] Step 2-2: Construct a real-time CQF interface model for low-priority traffic CQF's interference with lower-priority traffic manifests as its bandwidth consumption. Since CQF is a non-power-guaranteed scheduler, the actual transmission behavior of its traffic more accurately reflects its actual bandwidth consumption than its arrival behavior. Therefore, this invention innovatively uses the cumulative departure function of CQF traffic. To construct its real-time interface function for low-priority traffic, it represents the output port of the node. At this point, the blocking effect of CQF on low-priority traffic is [value]. Its definition is: in, This is the minimum addition and deconvolution operator in network computation. This interface can safely and compactly characterize the maximum amount of service consumed by CQF within any time window.

[0044] Step 3: Calculate the worst-case upper bound of the CQF real-time interface This step, based on network calculus theory, derives the closed upper bound of the real-time interface defined in step two.

[0045] Step 3-1: Calculate the upper bound of the real-time interface for high-priority traffic. Since CQF interference is limited to a single frame in transmission, its real-time interface... The worst-case upper bound of the blocking effect is the network port. Maximum frame length in all CQF streams ,Right now: in, . Indicates output port The CQF traffic set in Indicates flow rate Maximum frame length.

[0046] Step 3-2: Calculate the arrival curves of CQF under different implementation mechanisms. For TT-based CQF: its transmission behavior follows a fixed supercycle. .

[0047] Assuming it contains within a supercycle The CQF cycle, and the The number of CQF bits transmitted within a fixed CQF cycle is Then its flow arrival curve is: in, , Indicates the time length t at the output port Maximum number of arriving bits for CQF traffic at the location Indicates the first The fixed number of CQF bits transmitted within a CQF cycle, where % represents the modulo value, k represents the number of cycles index, and n represents the number of CQF cycles index.

[0048] For ET-based CQF: its transmission behavior does not have a fixed periodic pattern, and its analysis depends on the CQF stream. Output arrival curve of the source system (ES) in its transmission path By recursively applying the CQF transmission paradigm along the path, its application at any downstream port can be derived. The flow arrival curve is as follows: in, Indicates rounding down. .

[0049] Step 3-3: Calculate the upper bound of the real-time interface for low-priority traffic. Through in-depth analysis of the non-power-guaranteed scheduling behavior of CQF, this invention derives a low-priority real-time interface. The worst-case upper bound of the blockage effect consists of the sum of two parts: one part is the periodic base arrival, and the other part is the additional burst caused by non-periodic factors. Its upper bound formula is: in, This is the input arrival curve of CQF traffic aggregated within an arbitrary CQF period alignment interval, calculated in step 3-2. The corresponding function is selected based on the traffic scheduling mode. This part captures the maximum cumulative data volume of CQF traffic under ideal alignment conditions. This represents the additional maximum burst caused by interference from high-priority and low-priority traffic, as well as the misalignment between the CQF cycle and the analysis time window boundaries. This part is a quantitative compensation for the worst-case misalignment effect and external interference. Its calculation formula is: in, For port The maximum frame length for traffic with a lower priority than CQF. For the aggregated output arrival curve of traffic with higher priority than CQF, The length of the CQF synchronous protection belt. C For link speed, The CQF period length, For CQF traffic on the port The aggregation arrival curve; Operation representation .

[0050] Step 4: Calculate the end-to-end performance of other critical traffic flows under the hybrid scheduling architecture. Step 4-1: Integrate CQF interference and calculate service curve The purpose of this step is to apply the CQF interference quantification model calculated in the preceding steps to the performance analysis of other coexisting traffic in the network (such as SP and CBS traffic). The presence of CQF will preempt some network service resources, thereby reducing the effective service that other traffic can obtain. The CQF real-time interface upper bound of this invention can be accurately modeled as the worst-case interference generated by CQF.

[0051] Taking SP service as an example, assuming that SP has n priorities ( Indicates priority i Higher priority j There are two situations: (1) If the priority of CQF is lower than the priority of SP i Priority higher than SP i +1, then at the output port Priority is i The minimum service curve that the SP queue can obtain , in C For link speed, , Indicates priority as j Arrival curves of SP aggregated traffic, CQF priority relative to SP i In other words, it presents an upper bound on the interference of high-priority events. (2) If the priority of CQF is higher than the priority of SP. i Priority lower than SP i -1, then at the output port Priority is i The minimum service curve that the SP queue can obtain The expression is: At this point, CQF has priority relative to SP. i In contrast, it presents an upper bound on interference with low-priority events. For other service modes (such as CBS), the worst-case interference boundary of CQF is integrated into the existing network calculus and analysis framework in a similar manner. For details, please refer to the paper "Quantitative Performance Comparison of Various Traffic Shapers in Time-Sensitive Networking".

[0052] This invention provides an accurate CQF interference model, enabling these existing frameworks to be directly applied to complex hybrid scheduling scenarios that include CQF.

[0053] Step 4-2: Iteratively calculate end-to-end performance metrics After obtaining the service curve of the target traffic on a single network node, this step will complete the calculation from single-node performance to end-to-end performance.

[0054] First, calculate the flow. f In a single output port Worst case queue delay This is achieved by calculating the aggregated output arrival curve of that flow. (Specific details depend on the type) and the service curve obtained in step 4-1 (e.g.) This is achieved by the maximum horizontal deviation between ( ) in It is an operator for calculating the maximum horizontal deviation between two curves, and its formal definition is: Where sup represents the supremum and inf represents the infimum. This formula calculates the longest waiting time required to ensure that the service volume can catch up with the arrival volume.

[0055] Next, in order to perform the next hop analysis, it is necessary to calculate the traffic leaving the current port. Output arrival curve This curve represents the result of the input arrival curve after it has been served by the current node, and its upper bound is given by the following minimum addition-deconvolution formula: in like Otherwise ; The minimal additive convolution operator in network computation is formally defined as follows: This output reaches the curve. This will be used as the input arrival curve when traffic enters the next network node, thus enabling hop-by-hop propagation of the analysis.

[0056] Finally, through traffic f Routing The above calculation is repeated for each output port traversed, and the upper bound of the delay for each hop is accumulated to obtain the worst-case end-to-end delay. : in It's traffic. f The upper bound of the terminal-to-end delay. It's traffic. f The set of all output ports on the path. On a single port The queuing delay is calculated above. Through this iterative process, the end-to-end performance evaluation of any specified traffic in a complex network environment containing CQF can be completed.

[0057] Example 3: This implementation case aims to illustrate the application process of the CQF interference analysis method described in this invention under a specific TSN network configuration. The network topology used in this example has 6 end systems and 1 switch, such as... Figure 2 As shown.

[0058] Step 1: Read flow information and shaper parameters Step 1-1: Read traffic information Based on the input file msg.txt, read detailed information about all time-critical traffic in the network, including traffic ID, maximum frame length (in bytes), end-to-end deadline (in microseconds), propagation path, and minimum frame interval (period, in microseconds). The traffic information is shown in the table below: Table 1 Time-critical traffic information Step 1-2: Read TSN hybrid architecture information Based on the input file architecture.txt, determine that the hybrid scheduling architecture used for all switch output ports is as follows: , , , This means that the priorities from highest to lowest are: 1 TAS queue (priority 0), 1 CBS queue (priority 1), 1 CQF queue (priority 2), and 1 SP queue (priority 3).

[0059] Steps 1-3: Read link and shaper parameters Based on the input file rate.txt, read the network configuration parameters: overall network link rate. The TAS shaper uses the GuardBand integrated mode, and the CBS credits are in NonFrozen mode during the GuardBand period. CBS shaper parameters: CBS idle slope for priority 1. Priority 2 CBS idle slope CQF shaper parameters: ET (Event Triggered) mode; CQF cycle. CQF synchronous protection belt length .

[0060] Steps 1-4: Preprocessing Based on the read traffic information, calculate the initial burst of each traffic at the source. and rate Regarding traffic f Its burstiness at the source end is equal to its maximum frame length. The rate is equal to the frame length divided by the period. For example, f2(CBS): , f6(SP): , All traffic information, path information, architecture information, and shaper parameters are stored in the corresponding data structures for subsequent calculations.

[0061] Step 2: Construct a real-time interface abstraction for circular queuing and forwarding (CQF) This implementation follows the method described in step two of this invention to model CQF interference.

[0062] Step 2-1: Construct a CQF real-time interface model for high-priority traffic In this example, traffic with a higher priority than CQF (priority 2) is TAS (f1, priority 0) and CBS (f2, priority 1). We focus on the impact of CQF on CBS traffic f2. The model represents the interference of CQF on higher priority traffic f2 as a maximum value. Byte blocking.

[0063] Step 2-2: Construct a real-time CQF interface model for low-priority traffic In this example, traffic with a lower priority than CQF (priority 2) is SP (f7-f10, priority 3). The model indicates that the interference of CQF on lower priority traffic f7-f10 is determined by its cumulative departure process.

[0064] Step 3: Calculate the worst-case upper bound of the CQF real-time interface This step calculates the specific upper bound of the interface defined in step two.

[0065] Step 3-1: Calculate the upper bound of the real-time interface for high-priority traffic. This step calculates the maximum congestion caused by CQF non-preemption for higher-priority traffic. This calculation is performed independently for each output port. For a specific high-priority traffic (such as CBS traffic f2 in this example), we need to analyze each output port on its path to find the maximum frame length among the CQF traffic sharing that port.

[0066] The path for traffic f2 is: [ES1,SW1]; [SW1,ES6]. The output ports involved are [ES1,SW1] and [SW1,ES6]. For port [ES1,SW1]: the traffic flowing through this port includes f2 (CBS), f3 (CQF), f8 (SP), and f10 (SP). Among these, the CQF traffic is {f3}. The frame length of f3 is 988 bytes. Therefore, on this port, Similarly, for ports [SW1, ES6], we can obtain... .

[0067] In summary, for the ports that high-priority traffic f2 passes through on its path, the upper limit of non-preemptive blocking caused by CQF is 988 bytes.

[0068] Step 3-2: Calculate the arrival curves of CQF under different implementation mechanisms. This case study uses ET-based CQF. First, the traffic for each CQF is calculated. Rate at the source and sudden .

[0069] Table 2 ET-based CQF source-side flow parameters For any output port in the network The periodically aligned aggregation reaches the curve It is calculated using the following formula: in , It is determined by the source parameters Defined traffic The arrival curve (usually) ).because It is a linear function, and the time variable is... t Quantified as multiples of Therefore, the item of a single flow It is a step function. It processes all flows through the port. The aggregate arrival curve is obtained by summing the step functions corresponding to the CQF flows. It is also a non-decreasing step function. This function precisely defines the step function with length . In the worst case, the port is reached within an integer multiple of the time window. Total CQF flow.

[0070] For example, for port [SW1, ES4], the CQF traffic flowing through this port is {f4, f6}. Its periodically aligned aggregated arrival curve... It is a specific step function, which can be calculated by substituting the source-end arrival curves of f4 and f6 into the above summation formula. (Note: The specific function form needs to be calculated using network calculation tools).

[0071] Step 3-3: Calculate the upper bound of the real-time interface for low-priority traffic. According to the formula Calculate the upper bound. Where... It is the stepped arrival curve function calculated in step 3-2. According to the formula... , The calculation requires the following parameters: Table 3 Calculate the required parameters (example ports [SW1, ES4]) Substitute the above parameters into Formula (taking [SW1, ES4] ports as an example): Calculation yields specific It is 7912 bits.

[0072] Similarly, list the other ports. Calculate the required parameters and The calculated values ​​are shown in the table below.

[0073] Table 4. Port Parameter Data (Unit: bits) Finally, the step function obtained in step 3-2 is... With the calculated constant Adding (equivalent to shifting the entire step function upwards) ), obtain low-priority real-time interface The worst-case upper bound. This upper bound is still a step-like arrival curve function.

[0074] Step 4: Calculate the end-to-end performance of other critical traffic flows under the hybrid scheduling architecture. This case focuses on the performance of critical traffic affected by CQF interference, including CBS traffic f2 (priority 1) which has a higher priority than CQF, and SP traffic f7-f10 (priority 3) which has a lower priority than CQF.

[0075] Step 4-1: Integrate CQF interference and calculate service curve For CBS traffic f2: When calculating the service curve for f2, interference from higher-priority TAS traffic (f1) and non-preemptive blocking interference from lower-priority CQF traffic (f3-f6) need to be considered. The maximum blocking length calculated in step 3-1... As one of the interference factors, substituting it into the service curve calculation model applicable to the TAS+CBS architecture in reference [A], we obtain the CBS traffic f2 per port. The obtained service curve .

[0076] For SP traffic f7-f10: When calculating the service curve of SP traffic, the interference from higher priority TAS traffic (f1), CBS traffic (f2), and CQF traffic (f3-f6) needs to be considered. The upper bound of CQF interference to low-priority SP traffic calculated in step 3-3, together with the interference models of TAS and CBS, is used as a high-priority interference term. Based on the network calculus model applicable to hybrid scheduling architecture in reference [A], the SP queue (priority 3) at each output port is calculated. The obtained service curve .

[0077] Step 4-2: Iteratively calculate end-to-end performance metrics For each target traffic f (Including f2 and f7 to f10), starting from its source node, perform the following calculations hop by hop: Determine the flow f In the current port Input aggregation arrival curve .

[0078] Use the corresponding service curve obtained in step 4-1 (f2) F7-F10 use ), calculate flow f Queuing delay at this port : (Where k is the queue priority, and Shaper is CBS or SP) Calculate flow f Exit Port Output arrival curve , as input for the next hop: Repeat the above steps until the traffic reaches the destination node. Finally, delay the queuing of all ports along the path. Accumulate to obtain the end-to-end delay. : By performing the complete iterative calculation process described above (which requires the use of network calculus tools for specific numerical calculations), the worst-case end-to-end delay (WCD) results for the target traffic f2 and f7-f10 can be obtained.

[0079] Table 5. Worst-case end-to-end delay calculation results for target traffic. The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A performance evaluation method for hybrid scheduling networks based on network calculus, characterized in that, include: S1, collects the network traffic information set and the traffic transmission network information set; S2, process the network traffic information set to obtain the arrival rate of each traffic at the source and the switch, as well as the burstiness at the source. S3 processes the arrival rates of all traffic at the source and the switch, the burstiness of traffic at the source, and the traffic transmission network information set to obtain the worst end-to-end delay of other traffic under CQF traffic interference in the hybrid scheduling architecture.

2. The hybrid scheduling network performance evaluation method based on network calculus as described in claim 1, characterized in that, The network traffic information set includes information about each traffic transmitted in the traffic transmission network; The traffic information includes traffic ID number, frame length, end-to-end deadline, propagation path, minimum frame interval, scheduling mode information, and aggregated output arrival curve set; The propagation path includes the source, switch, and end points through which traffic passes in sequence during its transmission in the traffic transmission network; the aggregated output arrival curve set includes the aggregated output arrival curve of traffic at each output port. The source, switch, and end are all nodes in the propagation path; each node includes a corresponding output port. The traffic transmission network information set includes a hybrid scheduling architecture information set and a network information set; the hybrid scheduling architecture information set includes the scheduling mode adopted by the hybrid scheduling architecture and the priority information of each scheduling mode; the network information set includes the link rate and the default parameters of all scheduling mode shapers. The default parameters for a CQF shaper include the trigger mechanism mode, CQF cycle length, CQF synchronization guard band length, and the fixed number of CQF bits transmitted per cycle. For time-triggered CQF shapers, the default parameters include the time-over-cycle parameter. and maximum period The triggering mechanism modes include event-triggered (ET) and time-triggered (TT).

3. The hybrid scheduling network performance evaluation method based on network calculus as described in claim 1, characterized in that, flow f The burstiness at the source Equal to its frame length ,Right now ,flow f The arrival rate at the source is expressed as ,in for f Minimum frame interval; traffic f In the switch The arrival rate is .

4. The hybrid scheduling network performance evaluation method based on network calculus as described in claim 3, characterized in that, The process of processing the arrival rates of all traffic at the source and the switch, the burstiness of traffic at the source, and the traffic transmission network information set yields the worst-case end-to-end delay of other traffic under CQF traffic interference in the hybrid scheduling architecture, including: S31, for each type of CQF traffic, sort it according to its traffic priority from high to low to obtain a corresponding priority sequence; the priority sequence includes several priorities; set the output port. The output port is the source of the traffic. S32, calculate the output port Upper bound information on the interference of CQF traffic on other types of traffic; S33, at the output port In the context of CQF traffic, if its priority is lower than the i-th priority of the traffic sequence of type SP, but higher than the (i+1)-th priority, then the expression for calculating the minimum service curve of the traffic of the i-th priority of that type is: ， in, , This represents the arrival curve of the j-th priority flow in a flow sequence of type SP, where s is a time variable. Let be the frame length of the j-th priority traffic in a traffic sequence of type SP. For output port The upper bound of the blocking effect of CQF traffic on high-priority traffic; At the output port In this context, for CQF traffic, if its priority is higher than the i-th priority of the traffic sequence of type SP, but lower than the (i-1)-th priority, then the expression for calculating the minimum service curve of the traffic of the i-th priority of that type is: ， in, For output port The upper bound of the blocking effect of CQF traffic on low-priority traffic; S34 calculates the traffic based on the minimum service curve. f In a single output port Worst case queue delay ; S35, calculate the outflow of traffic f from the current port. Output arrival curve Its calculation expression is: ， in, For the preset flow function, if , ,otherwise for ; The minimal additive convolution operator in network computation is formally defined as follows: , For output port Traffic f The input flow arrival curve; S36, utilizing The current output port is where the flow f enters. The input flow arrival curve of the next port is used to determine the output port. Whether it is the output port at the end of the traffic propagation path, a first determination result is obtained. If the first determination result is yes, execute S37; if the first determination result is no, the output port... For the output port of the current output port at the next node in the propagation path, execute S32; S37, for all output ports The worst-case queuing delays are accumulated to obtain the worst-case end-to-end delays of other traffic under CQF traffic interference in the hybrid scheduling architecture. .

5. The hybrid scheduling network performance evaluation method based on network calculus as described in claim 4, characterized in that, The output port Upper bound information on the interference of CQF traffic on other types of traffic, including: output port. Upper bound information and output port of the blocking effect of CQF traffic on high-priority traffic. The upper bound of the blocking effect of CQF traffic on low-priority traffic; The output port The expression for calculating the upper bound of the congestion effect caused by CQF traffic on high-priority traffic is: ， in, For output port The maximum frame length of all CQF traffic at that location. For output port The upper bound of the blocking effect of CQF traffic on high-priority traffic.

6. The hybrid scheduling network performance evaluation method based on network calculus as described in claim 5, characterized in that, The output port The expression for calculating the upper bound of the congestion effect caused by CQF traffic on low-priority traffic is as follows: ， ， in, For output port The upper bound information on the blocking effect of CQF traffic on low-priority traffic. It depends on the output port. The type of CQF traffic in the data is adopted. or The expression, This represents the additional maximum burst of traffic caused by interference from CQF traffic and misalignment between the CQF cycle length and the analysis time window boundary. For output port The maximum frame length for all traffic with a priority lower than CQF. Traffic with higher priority than CQF traffic is routed to the output port. The aggregated output reaches the curve at that point. The length of the CQF synchronous protection belt. For link speed, The CQF period length, Operation representation .

7. The hybrid scheduling network performance evaluation method based on network calculus as described in claim 4, characterized in that, The worst-case queuing delay The calculation expression is: ， in, It is an operator for calculating the maximum horizontal deviation between two curves, and its formal definition is: Where sup represents the supremum, inf represents the infum, and d is the computational variable. The traffic of the i-th priority in the traffic sequence of type S at the output port The aggregated output reaches the curve.

8. A performance evaluation device for a hybrid scheduling network based on network calculus, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the hybrid scheduling network performance evaluation method based on network calculus as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which, when invoked by a computer, are used to execute the hybrid scheduling network performance evaluation method based on network calculus as described in any one of claims 1 to 7.

10. An information data processing terminal, characterized in that, The information data processing terminal is used to implement the hybrid scheduling network performance evaluation method based on network calculus as described in any one of claims 1 to 7.

Images