Method for solving avionics network node self-locking problem caused by traffic path overlapping in CBS network calculation

Abstract

The invention discloses a method for solving the problem of avionics network node self-locking caused by traffic path overlapping in CBS network calculation, which comprises the following steps: firstly, establishing an arrival curve model and a service curve model of traffic in CBS network calculation, and the models are used for describing communication behaviors in a network; then, analyzing and searching a self-locking node in the avionics network through traffic path comparison, and replacing an arrival curve of the traffic with a shaping curve of the traffic at a port of a previous node, so as to overcome the node self-locking problem; and finally, analyzing an end-to-end delay boundary of the flow by using a network calculation method. According to the method provided by the invention, the feasibility and effectiveness of network performance evaluation of the avionics network by using the traffic under the CBS shaping mechanism in the network calculation method can be enhanced, so that the avionics network performance evaluation of the end-to-end delay value under the worst condition is completed.

Description

Technical Field

[0001] The present invention relates to the technical field of network calculation evaluation of avionics systems, and is a network calculation system evaluation technology based on a credit shaping method (CBS), and more particularly, to a CBS network calculation evaluation method applicable to avionics networks that overcomes traffic self-locking. The method is used to overcome the node self-locking problem caused by traffic path overlap during network calculation and CBS processes of avionics networks. Background Art

[0002] The scope of avionics systems includes all systems and equipment related to electronics that support aircraft to complete their missions. Early avionics mainly involved systems that support aircraft takeoff, landing, guidance and navigation, as well as communication systems that connect aircraft crew members with the ground. With the development and maturity of aviation technology, aircraft have gradually assumed many tasks such as transportation, reconnaissance, early warning, air combat, and strike. In order to improve these mission capabilities, communication and navigation systems have been continuously enhanced and improved, and airborne target detection and identification have also been developed. The application of electronic computers and control technology has enabled the traditional mechanical linkage control to transition to fly-by-wire flight, and the management of airborne fire control and external weapons has also experienced a transition from mechanical to electronic control, making flight control systems, fire control and external management systems also an important part of avionics. As one of the main supporting technologies for avionics systems, airborne network technology has made great progress while accompanying the development of system architecture, and has become an important symbol of the generational evolution of avionics system architecture. In the modern avionics system design process, avionics integration technology is a means of effective integration of aircraft electronic systems. It is based on information theory and advanced electronic technology, and adopts the system engineering method to combine detection, communication, navigation, identification, electronic warfare, mission management, flight and fire control functions and corresponding electronic equipment into an organic whole in the physical structure space of the aircraft through airborne networks and software technologies (see Figure 1 As shown in the figure, the system can achieve a high degree of sharing of system resources and a significant improvement in overall efficiency, while balancing system performance, availability and life cycle costs.

[0003] Many flows in airborne networks have strict real-time requirements. In order to determine whether the designed network topology and shaper configuration can meet the requirements, it is necessary to analyze and evaluate the end-to-end delay of the flow in the worst case, so as to provide reference and basis for the planning and design of the network for airborne equipment. Network calculus is one of the commonly used evaluation methods, but when using network calculus to evaluate network communication performance, self-locking problems caused by overlapping traffic paths may be encountered.

[0004] Time-Sensitive Networking (TSN) is a network protocol that IEEE uses to further expand the application requirements and scope of deterministic Ethernet based on the real-time transmission solution of audio and video traffic in Audio Video Bridging (AVB) networks. TSN is committed to providing a universal time-sensitive mechanism at the data link layer of the network to solve the problem of time determinism guarantee when mixed transmission of network traffic with different quality of service requirements. The commonly used shaping method in TSN networks is credit-based shaping (CBS). This is a classic scheduling method defined in the AVB era, see Figure 2 The queuing and forwarding process of different traffic is shown in the figure. It is a bit like the single-rate token-bucket mechanism. For a queue shaped by CBS, if you want to schedule a message, it will be reduced at a constant rate (SendSlope); if and only if the credit is non-negative, the message can pass the CBS shaper; when the credit is negative, or there is a backlog of messages in the queue, the credit increases at a constant rate (IdleSlope); when the queue is empty, the credit is cleared. The effect of CBS is to provide a bandwidth equal to the rate for a specific queue while "scattering" the messages as much as possible, so that low-priority queues can also get a certain transmission opportunity.

[0005] The protocol identification tag (Tag Protocol Identifier) ​​is 0X8100, which indicates that Ethernet supports 802.1Q tags and can provide VLAN and QoS services; the 3-bit PCP (Priority Code Point) code defines 8 priorities: Level 0 is the lowest, used for traditional best-effort transmission traffic; Level 7 is the highest, which can be used for critical routing or network management functions (For the classification of traffic types, refer to the "Overview of Time-Sensitive Network Traffic Shaping Technology" published in Microelectronics and Computers, Volume 39, Issue 1, January 2022, by Zhang Lei and Wang Panpan). TSN networks will distinguish the importance of network traffic by priority, and perform traffic scheduling and network configuration according to different application scenarios.

[0006] Different types of traffic (Traffic Class) have different service quality requirements for the network. Network services always need to provide real-time network performance guarantees for various types of traffic transmission under limited network resources. For each type of traffic, the arrival curve and service curve of the traffic at the nodes along the way are modeled through network calculation methods. By measuring the maximum horizontal intercept of the two curves, the delay boundary of the traffic can be calculated, which can be used for matching optimization in design and improvement of network service efficiency, so as to achieve maximum energy efficiency of input resource cost and results under the premise of ensuring network services.

[0007] In order to overcome the above problems, the present invention proposes a CBS network calculation and evaluation method for overcoming flow self-locking in an avionics network. Summary of the invention

[0008] In order to solve the self-locking problem caused by the overlap of traffic paths in the CBS network calculation, the present invention adopts a network calculation method to evaluate the traffic communication performance of the CBS network, uses the arrival curve model and the service curve model to describe the communication behavior in the network, and for the self-locking problem caused by the overlap of traffic paths that may be encountered when using network calculation to evaluate the network communication performance, the shaping curve of the traffic at the previous node port is used at the self-locking port to replace the arrival curve of the traffic, thereby overcoming the self-locking problem, and finally calculating the maximum delay value of each type of traffic. Specifically, for the analysis of the self-locking problem, the present invention proposes a method for analyzing and finding self-locking node ports in the network based on traffic path comparison; the present invention uses the shaping curve of the traffic at the previous node port to replace the arrival curve of the traffic, thereby overcoming the self-locking problem, and using the network calculation method to analyze the end-to-end delay boundary of the traffic. Through the method of the present invention, the feasibility and effectiveness of the avionics network using the network calculation method to evaluate the network performance of the traffic under the CBS shaping mechanism can be enhanced.

[0009] In addition, the method of the present invention considers that the shaping curve under the CBS mechanism is also the upper boundary of the traffic transmission, and can be equivalently replaced with the arrival curve. Moreover, the shaping curve of the traffic at each output port is only related to the configuration of the port, and has nothing to do with other nodes and ports on the traffic path. Therefore, it is not affected by the self-locking problem. Therefore, the shaping curve is used instead of the arrival curve of the traffic, thereby overcoming the self-locking problem caused by the overlap of traffic paths in the CBS network calculation, and realizing the evaluation of network performance.

[0010] Thirdly, starting from the network calculation process of the time-sensitive network under the credit shaping mechanism, the present invention proposes a method for evaluating the network by using a shaping curve to replace part of the arrival curve that is difficult to calculate due to self-locking, thereby overcoming the self-locking problem.

[0011] A method of solving the self-locking problem of avionics network nodes caused by overlapping traffic paths in CBS network calculations is a network calculation evaluation of an avionics system; the method comprises the following steps:

[0012] Step 1: Obtain the avionics network topology and traffic information of each node, and find the self-locking port and traffic;

[0013] Step 101, obtaining the avionics network topology structure and each node;

[0014] Step 102, flow marking;

[0015] Step 103, analyzing and finding the node port with self-locking problem;

[0016] Step 103A, converting the flow paths of all flows into a combination of directed edges;

[0017] Step 103B, finding a flow combination having at least two identical directed edges in the path;

[0018] Step 2: Calculate the comprehensive traffic characteristics of the node;

[0019] Step 3: Calculate the delay value and burstiness of each type of traffic in each node on the traffic path;

[0020] Step 301, determine whether the current node is the output port of the first node;

[0021] Step 302, calculating the comprehensive burstiness of each type of traffic at the current output port;

[0022] Step 303, constructing a traffic arrival curve and a node service curve;

[0023] Step 304, calculating the maximum delay of the output port of each type of traffic node;

[0024] Step 305, calculating the maximum burstiness of the traffic at the output port of the next hop node;

[0025] Step 306, determining whether delay calculation is completed for all types of traffic at all nodes;

[0026] Step 4: Calculate the end-to-end delay of the traffic.

[0027] The technical effect achieved by the method of the present invention is:

[0028] ① According to the causes of the self-locking problem in CBS network calculation, the present invention proposes a method for finding self-locking node ports and traffic in the network based on traffic path comparison analysis.

[0029] ② In view of the self-locking problem in CBS network calculation, the present invention proposes to use the shaping curve of the traffic at the previous node port to replace the arrival curve of the traffic, thereby overcoming the self-locking problem and enhancing the feasibility and effectiveness of using network calculation methods to evaluate the network performance of the traffic under the CBS shaping mechanism.

[0030] ③ The calculation process of the present invention refers to the information carried by the traffic itself without adding external data resources, which can reduce the amount of calculation and improve the response speed of network calculation.

[0031] ④ The calculation process of the present invention is insensitive to the number of traffic types and can support network communication performance evaluation under any number of types and any amount of traffic within the priority range of 0 to 7. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a block diagram of the avionics system.

[0033] Figure 2 This is a schematic diagram of the traffic queuing and forwarding process under the CBS shaping mechanism.

[0034] Figure 3 It is a schematic diagram that briefly shows the end-to-end topological network structure in the avionics system.

[0035] Figure 4 This is an example diagram of calculating the worst-case delay of traffic based on the arrival curve and service curve.

[0036] Figure 5 It is a flow chart of a method for solving the self-locking problem of avionics network nodes caused by overlapping flow paths in CBS network calculations according to the present invention.

[0037] Figure 6 This is a network topology diagram of Example 1 provided by the present invention.

[0038] Figure 7 This is a diagram showing the end-to-end delay calculation results under the worst traffic conditions in Example 1. DETAILED DESCRIPTION

[0039] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. The various parameter demonstrations listed are only preferred embodiments of the present invention. It should be noted that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.

[0040] In the present invention, the network topology structure in the avionics network is recorded as networkTop, and the network topology structure networkTop includes a node set NODE and an edge set MES, that is, networkTop=<NODE,MES> .

[0041] NODE={node1,node2,…,node i ,…,node j ,…,node J}, the subscript j represents the identification number of the node in the network topology, and the subscript J represents the total number of nodes in the network topology, where:

[0042] node1 indicates the first node.

[0043] node2 indicates the second node.

[0044] node i represents the i-th node.

[0045] node j represents the jth node.

[0046] node J Indicates the last node.

[0047] For the sake of convenience, node j Also called any node. i With node j Not the same node.

[0048] In the present invention, when a node node on a traffic path j After an output port of is confirmed as a self-locking port, the node node j The nodes marked as self-locking are recorded as

[0049] In the present invention, any node node in the avionics network j Abstracted as an end system ES or switch SW, nodes can be distinguished using consecutive numbers, such as Figure 3 In the network topology, the edge set MES is composed of nodes with directed edges between nodes, that is, the directed edges are represented by a directed connection matrix [node i ,node j ], the [node i ,node j ] is set to 1 if there is a directed edge, and 0 if there is no directed edge. i ,node j ] is used to indicate whether there is a directed edge between two nodes.

[0050] In the present invention, there are multiple interfaces (or ports, which can be numbered and marked) on a device such as an end system ES or a switch SW. For example, an interface used for receiving signals is called an input port, and another interface used for outputting signals is called an output port. j The input port is denoted as node j,输入 , any node node j The output port is denoted as node j,输出 .

[0051] In the credit-based shaping method (CBS), the set of traffic types is denoted as MPCP, and The larger the number, the higher the priority.

[0052] PCP0 is the first priority type of traffic.

[0053] PCP1 is the second priority type of traffic.

[0054] PCP2 is the third priority type of traffic.

[0055] PCP3 is the fourth priority type of traffic.

[0056] PCP4 is the fifth priority type of traffic.

[0057] PCP5 is the sixth priority type of traffic.

[0058] PCP6 is the 7th priority type traffic.

[0059] PCP7 is the 8th priority type traffic.

[0060] For the convenience of explanation, any priority type of traffic is recorded as PCP h . It has a higher priority than PCP h PCP h,高 , with lower priority than PCP h PCP h,低 .

[0061] Any node j The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and

[0062] portName is node j In the present invention, a directed edge represented by a node combination [SW1, SW2] is used to uniquely represent an output port. For example, [SW1, SW2] represents a directed edge from the output port of SW1 to the input port of SW2.

[0063] CalFlag is a completion flag, and CalFlag (TRUE or FALSE), TRUE indicates that the calculation is completed, and FALSE indicates that the calculation is not completed. In the present invention, it refers to a flag that the traffic has completed the traffic characteristics, burstiness and delay calculation of all types of traffic at port portName.

[0064] flowRateSum is the total flow rate.

[0065] maxLength is the maximum frame length of the traffic.

[0066] initialDelay is the initial service delay of the port.

[0067] serviceRate is the traffic service rate.

[0068] toBusrt is the total burst rate of traffic.

[0069] toDelay is the maximum waiting time for traffic at the port.

[0070] In the network topology of the onboard network networkTop=<NODE,MES> The network flow set in is recorded as FLOW, and FLOW={flow1,flow2,…,flow f ,…,flow g ,…,flow F}, the subscript f represents the identification number of the flow, and the subscript F represents the total number of flows.

[0071] flow1 indicates the first network flow.

[0072] flow2 indicates the second network flow.

[0073] flow f Indicates the fth network traffic.

[0074] flow g Indicates the g-th network traffic.

[0075] flow F Indicates the last network flow.

[0076] For the sake of convenience, flow f Also called any network flow. f with flow g Different network traffic.

[0077] In the present invention, any network traffic flow f The traffic characteristic information is recorded as and

[0078] framelength is the frame length in bits.

[0079] Period is the traffic period, in μs.

[0080] type indicates the traffic type.

[0081] PCP is the traffic priority. The PCP corresponds to the type one by one. In the present invention, the order of the traffic priority PCP from high to low is PCP7, PCP6, PCP5, PCP4, PCP3, PCP2, PCP1 and PCP0.

[0082] path is the traffic path, which can be an array of node numbers along the way.

[0083] rate is the traffic rate, which is also the bandwidth occupied by traffic transmission, in Mbps, and

[0084] Bursts is the maximum burst rate of traffic transmitted along the path to the previous node of the last node at each node along the way, in bits. f Through the node i The maximum burst rate of the output port is recorded as flow f Through the node j The maximum burst rate of the output port is recorded as

[0085] Delays is the maximum delay of the traffic from the first node to the last node at each node along the way, referred to as the transmission delay on the same type of traffic path, in μs. The number information containing the number of nodes along the way (excluding the destination node) can be used to represent the delay value at each node along the way. f The nodes passed from the start node to the end node are node1, node2, node i 、node j 、node j+1 、node J-1 、node J , then the flow rate f The transmission delay statistics on the path is recorded as in, Indicates flow f The maximum delay at the output port of node node1, Indicates flow f The maximum delay at the output port of node node2, Indicates flow f In the node i Maximum delay at output port, Indicates flow f On the node j Maximum delay at output port, Indicates flow f On the node j+1 Maximum delay at output port, Indicates flow f On the node J-1 Maximum delay for output ports.

[0086] e2eDelay is the end-to-end delay of the traffic, in μs, which represents the sum of the delay values ​​at each node along the way.

[0087] The problem of traffic self-locking in network calculation

[0088] Here is an example to explain the problem of traffic self-locking in network calculation, such as Figure 1 , Figure 3 As shown, the circles represent the end systems in the avionics system (i.e., the first end system ES1, the second end system ES2, the third end system ES3, and the fourth end system ES4), and the squares represent the switches in the avionics system (i.e., the first switch SW1, the second switch SW2, the third switch SW3, and the fourth switch SW4). f is any flow sent from end system ES1 to end system ES3. g It is any flow sent from end system ES4 to end system ES2. The network calculation is based on any flow f The order in which traffic is propagated in the network topology has an important impact on the flow f In the flow path The arrival curve at each node (end system and switch are both network nodes) and service curve The traffic from the same port will interfere with each other, so the traffic information from the same port should be substituted into the calculation. Figure 3 middle:

[0089] For example, flow f The flow path is The nodes passed in the order of traffic are recorded as According to the equipment record

[0090] For example, flow g The flow path is The nodes passed in the order of traffic are recorded as According to the equipment record

[0091] (AA) Based on traffic path When you need to calculate flow f Arrival curve at switch SW1 and service curve When f Arrival curve at end system ES1 and service curve Flow is also needed gArrival curve at switch SW3 and service curve

[0092] (AB) When you need to calculate flow f Arrival curve at switch SW2 and service curve When both flow and f Arrival curve at switch SW1 and service curve Flow is also required g Arrival curve at switch SW1 and service curve That is, the arrival and service status of the previous node on the traffic path;

[0093] (AC) When you need to calculate flow f Arrival curve at switch SW4 and service curve When both flow and f Arrival curve at switch SW2 and service curve Flow is also required g Arrival curve at end system ES4 and service curve

[0094] (AD) When you need to calculate flow f Arrival curve at switch SW3 and service curve When both flow and f Arrival curve at switch SW4 and service curve Flow is also required g Arrival curve at switch SW4 and service curve

[0095] (BA) Based on traffic path When you need to calculate flow g Arrival curve at switch SW4 and service curve When both flow and g Arrival curve at end system ES4 and service curve Flow is also needed f Arrival curve at switch SW2 and service curve

[0096] (BB) When you need to calculate flow g Arrival curve at switch SW3 and service curve When both flow and g Arrival curve at switch SW4 and service curve Flow is also required f Arrival curve at switch SW4 and service curve

[0097] (BC) When you need to calculate flow g Arrival curve at switch SW1 and service curve When both flow and g Arrival curve at switch SW3 and service curve Flow is also required f Arrival curve at end system ES1 and service curve

[0098] (BD) When you need to calculate flow g Arrival curve at switch SW2 and service curve When both flow and g Arrival curve at switch SW1 and service curve Flow is also required f Arrival curve at switch SW1 and service curve

[0099] exist Figure 3 The flow paths that overlap are: (A) f When the switch SW1 transmits to the switch SW2, the traffic flow g When the switch SW1 transmits to the switch SW2, there are overlapping paths for the same-direction transmission. (B) Traffic flow f When the switch SW4 transmits to the switch SW3, the traffic flow g When the switch SW4 transmits to the switch SW3, there are overlapping paths for the same-direction transmission. (C) Traffic flow g When the switch SW4 transmits to the switch SW3, the traffic flowf When the switch SW4 transmits to the switch SW3, there are overlapping paths for the same-direction transmission. (D) Traffic flow g When the switch SW1 transmits to the switch SW2, the traffic flow f When the switch SW1 transmits to the switch SW2, overlapping paths of the same-direction transmission appear. Figure 3 Medium flow f and flow g There are two overlapping traffic paths, namely, the transmission from switch SW1 to switch SW2 and the transmission from switch SW4 to switch SW3, and thus the self-locking node set It can be seen from this that Figure 3 In the end-to-end topology of the avionics system network shown, during the network calculation process, the self-locking problem refers to the overlap of two or more traffic paths, sharing two or more output ports, and the order of output at these common output ports is different, resulting in the calculation of multiple output ports being interdependent when the delay boundary value of the output port is calculated progressively, so that the calculation in the network calculation process falls into an infinite loop. The network calculation method refers to "Network Calculus: Theory of Internet Deterministic Queuing Systems" published in March 2022, by Jean-Yves Le Boudec [Switzerland] and Patrick Thiran [Belgium], translated by Li Qiao et al., pages 29-35.

[0100] Worst latency of traffic;

[0101] See also Figure 4 As shown in the figure, the flow arrival curve (line ①) is a (rate, burst) oblique line model, where rate refers to the rate of arrival of the curve, which is reflected in the figure as the slope (r a ), burst refers to the burstiness, which is shown as the vertical intercept b in the figure. The service curve of the output port (line ②) is a (rate, time) slope model, where rate refers to the speed of the service curve, which is shown as the slope (r b ), time refers to the initial delay, which is shown as the horizontal intercept T in the figure. The arrival curve of the traffic and the service curve of the output port are plotted on a graph. The maximum horizontal intercept of the two curves is the worst-case delay of the traffic at the port.

[0102] like Figure 3 The end-to-end network topology shown in the figure shows that the process of calculating the end-to-end delay value is progressive. The burstiness of the comprehensive flow at the current output port is used to calculate any flow. fThe maximum delay value of the traffic is calculated, and then the burst degree of the traffic output to the next node is calculated based on the delay value, which is used as the basis for calculating the delay value of the traffic at the next output port. After the worst-case maximum delay value of each type of traffic in all output ports is calculated, any traffic flow f Along its flow path The delay values ​​of the corresponding output ports are summed up, and the sum of the delay values ​​can represent the upper boundary of the end-to-end delay of the traffic.

[0103] The present invention provides a method for calculating the end-to-end delay of CBS traffic based on network calculation, which overcomes the self-locking problem. The present invention establishes any traffic flow based on the network calculation method. f The arrival curve and service curve Model, and in the self-locking node Use flow f At the previous node j-1 The shaping curve at To replace flow f The arrival curve Thus overcoming the self-locking node The self-locking problem at The arrival curve at and service curve After that, the flow can be evaluated f In the flow path Worst-case end-to-end delay on

[0104] The present invention follows the protocol to classify traffic according to priority 0 to 7, and supports network communication performance evaluation under any type and traffic quantity within this range. In Examples 1 and 2, the traffic in the TSN network is divided into three priority types according to the different requirements of network traffic on network service quality. The traffic priority types used in the present invention are PCP7, PCP4 and PCP0 from high to low.

[0105] See also Figure 5 As shown, the present invention uses network calculation to calculate the maximum delay value and the maximum backlog of traffic queues under the CBS shaping mechanism, so as to solve the node self-locking in the avionics network, and uses the shaping curve of the traffic at the previous node port to replace the arrival curve of the traffic at the self-locking port to overcome the self-locking problem, and finally calculates the maximum delay value of each type of traffic. Specifically, it includes the following four steps:

[0106] Step 1: Obtain the avionics network topology and traffic information of each node, and find the self-locking port and traffic;

[0107] Step 101, obtaining the avionics network topology structure and each node;

[0108] In the present invention, the network topology structure in the avionics network is recorded as networkTop=<NODE,MES> , the network flow set is recorded as FLOW = {flow1, flow2,…, flow f ,…,flow g ,…,flow F Any flow f The traffic characteristic information is recorded as

[0109] Network topology networkTop =<NODE,MES> It contains the node set information NODE = {node1, node2, ..., node i ,…,node j ,…,node J} and edge set information MES. The nodes in the avionics network are abstracted into end systems ES and switches SW, and the nodes are numbered with consecutive numbers. In the network topology, a directed connection matrix is ​​used to represent the directed edges, which are recorded as [ES, SW]. The value of [ES, SW] is 1 if there is a directed edge, and 0 if there is no edge. The value of [ES, SW] is assigned to indicate whether there is a directed edge between two nodes (end system ES to switch SW).

[0110] Step 102, flow marking;

[0111] Any flow f Contains a variety of characteristic information, so the present invention uses a structure to construct traffic characteristic information and

[0112] Similarly, the flow characteristic information of flow1 is and

[0113] Similarly, the flow characteristic information of flow2 is and

[0114] Similarly, flow g Traffic characteristics information and

[0115] Similarly, flow F Traffic characteristics information and

[0116] Then there is: flow-flow feature set

[0117] Step 103, analyzing and finding the node port with self-locking problem;

[0118] After obtaining the network topology networkTop=<NODE,MES> All flows FLOW={flow1,flow2,…,flow f ,…,flow g ,…,flow F}, and the flow characteristics corresponding to the flow Afterwards, based on Analyze all self-locking nodes NODE that have self-locking problems during CBS network calculation 锁 .

[0119] The steps of analyzing and finding the self-locking nodes with self-locking problems in the present invention are as follows:

[0120] Step 103A, converting the flow paths of all flows into a combination of directed edges;

[0121] In the present invention, for example Figure 3 In the flow f Traffic path Transformation into path sequence representation

[0122] Similarly, the flow g Traffic path Transformation into path sequence representation

[0123] Step 103B, finding a flow combination having at least two identical directed edges in the path;

[0124] If the same directed edges appear in a different order in a directed edge combination, a self-locking problem will definitely occur, and the self-locking nodes corresponding to the flow will be recorded.

[0125] like Figure 3 Medium flow f and flow g In the path, there are directed edges (SW_1, SW_2) and (SW_4, SW_3), but for the flow f (SW_1, SW_2) appears before (SW_4, SW_3); and for flow g In the example, (SW_1, SW_2) appears after (SW_4, SW_3), so when calculating the end-to-end delay bounds of these two flows, a self-locking problem will inevitably occur.

[0126] exist Figure 3 Medium flow f and flow g There are two overlapping traffic paths, namely, the transmission from switch SW1 to switch SW2 and the transmission from switch SW4 to switch SW3. The self-locking ports include SW1, SW2, SW3, and SW4, and then the self-locking node set is selected, which is recorded as NODE. 锁 ,and By NODE 锁 Filter out the corresponding self-locking flow set FLOW on the node 锁 ,and

[0127] Indicates the first self-locking node. Indicates the second self-locking node. Indicates the third self-locking node. Indicates the 4th self-locking node. Indicates the fth self-locking network traffic. Indicates the g-th self-locking network traffic.

[0128] By finding the flow combination with at least two identical directed edges in the path, the self-locking flow set FLOW is confirmed 锁 And the self-locking node set NODE 锁 .

[0129] Step 2: Calculate the comprehensive traffic characteristics of the node;

[0130] Network calculus by calculating PCP h Priority type traffic at node j The upper bound of the delay at and will Assign to and Then along the flow path Through the output port of each node on the path Add the sum to calculate the upper bound of the end-to-end delay of the traffic. and Combined with the CBS class-by-class shaping mechanism, in the calculation Previously, we needed to calculate the node j Each type of traffic at the port The comprehensive flow characteristics.

[0131] Network topology in the avionics network networkTop=<NODE,MES> In the calculation node set NODE = {node1, node2, ..., node i,…,node j ,…,node J}Comprehensive traffic characteristics of each node in .

[0132] The node-comprehensive traffic characteristics of the same priority type traffic of node node1 are recorded as and

[0133] The node-comprehensive traffic characteristics of the same priority type traffic of node node2 are recorded as and

[0134] node i The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and

[0135] node j The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and

[0136] node J The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and

[0137] flowRateSum is the total flow rate. In this step,

[0138] maxLength is the maximum frame length of the traffic. In this step,

[0139] initialDelay is the initial delay of the port, which is equal to the set value.

[0140] serviceRate is the traffic service rate, which is equal to the credit growth rate configured in the CBS mechanism.

[0141] toBusrt is the total burstiness of traffic, which is initialized to 0 in this step.

[0142] toDelay is the maximum waiting time for traffic at the port, which is initialized to 0 in this step.

[0143] Step 3: Calculate the delay value and burstiness of each type of traffic in each node on the traffic path;

[0144] When there are self-locking nodes in the avionics network, the shaping curve of the traffic at the previous node port is used to replace the arrival curve of the traffic, so that the delay value toDelay and the burstiness toBusrt of each type of traffic in each node on the traffic path can be calculated.

[0145] In order to explain the self-locking at the node, the flow f The flow path is recorded as The node j Mark as the current node, node i is a node j The previous node, node j+1 is a node j The next node.

[0146] Step 301, determine whether the current node is the output port of the first node;

[0147] Case 1: Initially, if the current output port is the output port of the first node, node 1,输出 , then the flow rate f In node 1,输出 The maximum burst degree is denoted as and Indicates flow f The frame length;

[0148] Case 2: If the current output port is not the output port of the first node 1,输出 , then execute step 302 to calculate node j,输出 The comprehensive burstiness of each type of traffic;

[0149] Step 302, calculating the comprehensive burstiness of each type of traffic at the current output port;

[0150] Compute Node j,输出 The comprehensive burstiness of each type of traffic is recorded as and in flow f Through the node port node j,输出 And the type corresponds to PCP h The sum of the maximum burstiness of

[0151] Step 303, constructing a traffic arrival curve and a node service curve;

[0152] Case 1: If the output port of the current node is j,输出 If there is no self-locking problem, then according to the node j,输出 Comprehensive flow characteristics Calculation of PCP h Arrival curve of priority type traffic and the service curve of the node

[0153]

[0154] Indicates the output port node j,输出 PCP h The corresponding total traffic rate.

[0155] t represents the traffic transmission time.

[0156] Indicates the output port node j,输出 PCP h The corresponding total traffic burst rate.

[0157] C represents the link bandwidth of the avionics network.

[0158] Indicates the output port node j,输出 PCP h The corresponding maximum frame length of the traffic.

[0159] Indicates the output port node j,输出 PCP h The corresponding traffic service rate.

[0160] Indicates the output port node j,输出 PCP h The initial delay of the corresponding port.

[0161] Case 2: If the output port of the current node is j,输出 If there is a self-locking problem, use node j Previous node i The shaping curve As a PCP h Priority type traffic arrival curve and in Indicates that the CBS shaping mechanism processes PCP h For priority type traffic, the credit value is t from the maximum credit value 最大信用量 To the minimum credit amount t 最小信用量 The amount of data continuously transmitted within a certain period of time; at the same time, the node service curve is constructed and recorded as and

[0162] Represents a node i Corresponding PCP h The traffic service rate.

[0163] Represents a node j Corresponding PCP h The traffic service rate.

[0164] Represents a node j Corresponding PCP h The initial delay of the port.

[0165] Step 304, calculating the maximum delay of the output port of each type of traffic node;

[0166] The flow arrival curve of the node output port constructed in step 303 is and node service curve Node can be calculated j,输出 PCP h Priority type of traffic flow f The maximum delay result is recorded as and

[0167] For PCP h The traffic at node j Maximum waiting time on ;

[0168] δ represents the maximum horizontal intercept size used to represent the arrival curve and service curve in network calculation, which is considered to be the maximum delay value.

[0169] Step 305, calculating the maximum burstiness of the traffic at the output port of the next hop node;

[0170] flow f At the next node j+1,输入 The maximum burst degree at and At the same time, according to the CalFlag flag, the node j,输出 Whether all types of traffic in the calculation have the maximum delay and update the current node node j,输出 The CalFlag logo.

[0171] Indicates flow f Through the node j The maximum burst rate of the output port.

[0172] Indicates flow f flow rate.

[0173] Indicates flow f In the node j The maximum delay of the output port.

[0174] Step 306, determining whether delay calculation is completed for all types of traffic at all nodes;

[0175] Case 1: According to the CalFlag flag, if all current nodes j,输出 If the CalFlag values ​​in are all TRUE, it means that the delay calculation of all types of traffic on all nodes has been completed, then execute step 4;

[0176] Case 2: According to the CalFlag flag, if all current nodes j,输出 If the CalFlag value in is FALSE, it indicates that the maximum delay calculation of all types of traffic of all nodes has not been completed, then the process jumps to step 301.

[0177] Step 4: Calculate the end-to-end delay of the traffic;

[0178] For any flow f , according to the flow f Transmission path Will be along The maximum delay of all transmissions on the node output port is added up to get the flow rate. f Maximum end-to-end delay The calculation formula is

[0179] Example 1

[0180] like Figure 6 The figure shows a calculation diagram of the end-to-end transmission delay of traffic at a self-locking node. In the figure, each node performs traffic shaping through CBS and uses network calculation methods to evaluate the network performance, thereby providing a feasibility and effectiveness reference for avionics network design.

[0181] according to Figure 6 It can be seen that the flow f It is sent from end system ES1, passes through switch SW1, switch SW2, switch SW4 and switch SW3, and is transmitted to end system ES3.

[0182] according to Figure 6 It can be seen that the flow gIt is sent from end system ES4, passes through switch SW4, switch SW3, switch SW1 and switch SW2, and is finally transmitted to end system ES2.

[0183] exist Figure 6 For network traffic flow f and flow g The transmission paths converge at the output ports of SW1 and SW4 respectively, and overlap twice, resulting in a self-locking problem of traffic in the performance evaluation based on network calculation. The attribute configuration parameters of the traffic in the network are shown in Table 1, and the link rate is 100Mbps.

[0184] Table 1 Network traffic attribute configuration parameters of Example 1

[0185]

[0186] Based on the flow configuration parameters in Table 1, the calculation steps for solving the flow self-locking problem in network calculation in this embodiment will be specifically described according to the method of the present invention:

[0187] (1) According to the flow in Table 1 f The attribute parameters of flow are calculated f The traffic arrival curve at the ES1 node is

[0188] (2) According to the node attribute parameters in Table 1, the ES1 node service curve is calculated as follows:

[0189] (3) Calculate the flow rate based on the network calculation method f The maximum latency at the ES1 output is

[0190] (4) Flow f After output from ES1, the flow arrival curve in the next node SW1 is

[0191] (5) Due to the self-locking problem in the SW1 output port, the shaping curve of the previous port needs to be used to replace the flow arrival curve.

[0192] (6) According to the node attribute parameters in Table 1, the service curve of SW1 node is calculated as follows:

[0193] (7) Calculate the flow according to the network calculation method f with flow g The maximum delay at SW1 output is

[0194] (8) Flow f After outputting from SW1, the flow arrival curve in the next node SW2 is

[0195] (9) According to the node attribute parameters in Table 1, the flow is calculated f The corresponding SW2 node service curve is

[0196] (10) According to the network calculation method, the flow is calculated f The maximum delay at SW2 output is

[0197] (11) Flow f After outputting from SW2, the flow arrival curve in the next node SW4 is

[0198] (12) Similar to the network performance evaluation content of steps (1)(2)(3)(4), according to the traffic flow in Table 1 g and node attribute parameters, flow g The maximum delay when outputting from ES4 is 216μs, and the traffic arrival curve in the subsequent node SW4 is

[0199] (13) Flow f with flow g After aggregation in SW4, the traffic flow can be obtained f with flow g The arrival curve after aggregation is

[0200] (14) According to the node attribute parameters in Table 1, the flow is calculated f with flow g The corresponding SW4 node service curve is

[0201] (15) According to the network calculation method, the flow is calculated f with flow g The maximum delay at SW4 output is

[0202] (16) Flow f After output from SW4, the flow arrival curve in the next node SW3 is

[0203] (17) According to the node attribute parameters in Table 1, the service curve of SW3 node is calculated as follows:

[0204] (18) According to the network calculation method, the flow is calculated f The maximum delay at SW3 output is

[0205] (19) According to steps (1) to (18), the flow rate can be calculated f The end-to-end delay is Similarly, using the above calculation steps, the flow rate can also be calculated g The end-to-end delay results are shown in Figure 1, but they are not calculated one by one in this example.

[0206] According to the flow obtained in step (19) f The end-to-end delay results show that by adopting the solution to the self-locking problem in the present invention, the calculation of the end-to-end delay boundary of the flow can be successfully realized, and the evaluation results show that 1622.5μs is obviously greater than the cycle time of the flow 1000μs, which means that the flow f The deadline may be missed, so it is believed that the method of the present invention can effectively identify the unreasonable design and network parameter configuration that may exist in the avionics network.

[0207] Figure 7 This is a calculation result diagram of using a shaping curve to replace the flow arrival curve to solve the self-locking problem. The figure shows the relationship between the credit change of CBS (the vertical axis in the lower direction) and the flow transmission modeling based on network calculation (the vertical axis in the upper direction) by taking the SW1 node in the embodiment as an example. It shows that CBS shaping will limit the maximum amount of data that can be continuously transmitted by the constrained flow, so that the maximum amount of data can be used as the burst size of the flow arrival curve, thereby solving the flow in the example. g The problem of uncalculated burst due to self-locking.

Claims

1. A method for solving the self-locking problem of avionics network nodes caused by overlapping traffic paths in CBS network calculation is a network calculation evaluation of avionics systems; it is characterized by There are the following steps: Step 1: Obtain the avionics network topology and traffic information of each node, and find the self-locking port and traffic; Step 101, obtaining the avionics network topology structure and each node; The network topology structure in the avionics network is recorded as networkTop=<NODE,MES> , the network flow set is recorded as FLOW = {flow1, flow2,…, flow f ,…,flow g ,…,flow F }、Any flow f The traffic characteristic information is recorded as Node set NODE = {node1, node2, ..., node i ,…,node j ,…,node J }, the subscript j represents the identification number of the node in the network topology, and the subscript J represents the total number of nodes in the network topology, where: node1 represents the first node; node2 represents the second node; node i represents the i-th node; node j represents the jth node; node J Indicates the last node; The edge set MES is composed of nodes with directed edges between nodes; The nodes in the avionics network are abstracted into end systems ES and switches SW; flow1 represents the first network flow; flow2 represents the second network flow; flow f represents the fth network traffic; flow g represents the g-th network traffic; flow F Indicates the last network traffic; framelength is the frame length; period is the traffic period; type is the traffic type; PCP is the traffic priority; path is the traffic path; rate is the flow rate; bursts is the maximum burst degree of the node before the traffic reaches the last node along the path; delays is the maximum delay of the traffic from the first node to the node before the last node at each node along the way; e2eDelay is the end-to-end delay of the traffic; Step 102, flow marking; Any flow f The traffic characteristics information included is Then there is a network flow set FLOW = {flow1, flow2, ..., flow f ,…,flow g ,…,flow F The flow-flow characteristic set of} is The flow characteristic information of flow1 is: The flow characteristic information of flow2 is: flow g The traffic characteristic information is flow F The traffic characteristic information is Step 103, analyzing and finding the node port with self-locking problem; After obtaining the network topology networkTop=<NODE,MES> All flows FLOW={flow1,flow2,…,flow f ,…,flow g ,…,flow F }, and the flow characteristics corresponding to the flow Afterwards, based on Analyze all self-locking nodes NODE that have self-locking problems during CBS network calculation 锁 ; Specifically, the steps to analyze and find the self-locking nodes with self-locking problems are: Step 103A, converting the flow paths of all flows into a combination of directed edges; The flow f Traffic path Transformation into path sequence representation Similarly, the flow g Traffic path Transformation into path sequence representation Step 103B, finding a flow combination having at least two identical directed edges in the path; If the same directed edges appear in different orders in a directed edge combination, a self-locking problem will occur, and the self-locking nodes corresponding to the flow will be recorded; the confirmed self-locking flow set is recorded as FLOW 锁 And the self-locking node set is recorded as NODE 锁 ; Step 2: Calculate the comprehensive traffic characteristics of the node; Network calculus by calculating PCP h Priority type traffic at node j The upper bound of the delay at and will As flow Then along the flow path Through the output port of each node on the path Add the sum to calculate the upper bound of the end-to-end delay of the traffic. Combined with the CBS class-by-class shaping mechanism, in the calculation Previously, we needed to calculate the node node j Each type of traffic at the port Comprehensive flow characteristics of Network topology in the avionics network networkTop=<NODE,MES> In the calculation node set NODE = {node1, node2, ..., node i ,…,node j ,…,node J Comprehensive traffic characteristics of each node in}; The node-comprehensive traffic characteristics of the same priority type traffic of node node1 are recorded as and The node-comprehensive traffic characteristics of the same priority type traffic of node node2 are recorded as and node i The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and node j The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and node J The node-comprehensive traffic characteristics of the same priority type traffic are recorded as and flowRateSum is the total flow rate; in this step, maxLength is the maximum frame length of the traffic; in this step, initialDelay is the initial delay of the port, which is equal to the set value; serviceRate is the traffic service rate, which is equal to the credit growth rate configured in the CBS mechanism; toBusrt is the total burstiness of traffic, which is initialized to 0 in this step; toDelay is the maximum waiting time of traffic at the port, which is initialized to 0 in this step; Step 3: Calculate the delay value and burstiness of each type of traffic in each node on the traffic path; When there are self-locking nodes in the avionics network, the shaping curve of the traffic at the previous node port is used to replace the arrival curve of the traffic, so as to calculate the delay value toDelay and the burstiness toBusrt of each type of traffic in each node on the traffic path; The flow f The flow path is recorded as The node j Mark as the current node, node i is a node j The previous node, node j+1 is a node j The next node of Step 301, determine whether the current node is the output port of the first node; Case 1: Initially, if the current output port is the output port of the first node, node 1,输出 , then the flow rate f In node 1,输出 The maximum burst rate is Case 2: If the current output port is not the output port of the first node 1,输出 , then execute step 302 to calculate node j,输出 The comprehensive burstiness of each type of traffic; Step 302, calculating the comprehensive burstiness of each type of traffic at the current output port; Compute Node j,输出 The comprehensive burstiness of each type of traffic Step 303, constructing a traffic arrival curve and a node service curve; Case 1: If the output port of the current node is j,输出 If there is no self-locking problem, then according to the node j,输出 Comprehensive flow characteristics Calculation of PCP h Arrival curve of priority type traffic and the service curve of the node Indicates the output port node j,输出 PCP h The corresponding total flow rate; t represents the traffic transmission time; Indicates the output port node j,输出 PCP h The corresponding total traffic burst; C represents the link bandwidth of the avionics network; Indicates the output port node j,输出 PCP h The corresponding maximum frame length of the traffic; Indicates the output port node j,输出 PCP h The corresponding traffic service rate; Indicates the output port node j,输出 PCP h The initial delay of the corresponding port; Case 2: If the output port of the current node is j,输出 If there is a self-locking problem, use node j Previous node i The shaping curve As a PCP h The arrival curve of the priority type traffic is in Indicates that the CBS shaping mechanism processes PCP h For priority type traffic, the credit value is from the maximum credit value t 最大信用量 To the minimum credit amount t 最小信用量 The amount of data transmitted continuously within a certain period of time; at the same time, the node service curve is constructed as Represents a node i Corresponding PCP h Traffic service rate; Represents a node j Corresponding PCP h Traffic service rate; Represents a node j Corresponding PCP h The initial delay of the port; Step 304, calculating the maximum delay of the output port of each type of traffic node; The flow arrival curve of the node output port constructed in step 303 is and node service curve Node can be calculated j,输出 PCP h Priority type of traffic flow f The maximum delay result is For PCP h The traffic at node j Maximum waiting time on ; δ represents the maximum horizontal intercept size used to represent the arrival curve and service curve in network calculation, which is considered to be the maximum delay value; Step 305, calculating the maximum burstiness of the traffic at the output port of the next hop node; flow f At the next node j+1,输入 The maximum burst degree at At the same time, according to the CalFlag flag, the node j,输出 Whether all types of traffic in the calculation have the maximum delay and update the current node node j,输出 CalFlag logo; Indicates flow f Through the node j The maximum burst rate of the output port; Indicates flow f The flow rate; Indicates flow f On the node j The maximum delay of the output port; Step 306, determining whether delay calculation is completed for all types of traffic at all nodes; Case 1: According to the CalFlag flag, if all current nodes j,输出 If the CalFlag values ​​in are all TRUE, it means that the delay calculation of all types of traffic on all nodes has been completed, then execute step 4; Case 2: According to the CalFlag flag, if all current nodes j,输出 If the CalFlag value in is FALSE, it indicates that the maximum delay calculation of all types of traffic of all nodes has not been completed, then the process jumps to step 301; Step 4: Calculate the end-to-end delay of the traffic; For any flow f , according to the flow f Transmission path Will be along The maximum delay of all transmissions on the node output port is added up to calculate the flow f Maximum end-to-end delay The calculation formula is 2. The method for solving the self-locking problem of avionics network nodes caused by overlapping traffic paths in CBS network calculation according to claim 1 is characterized in that: Network calculation is used to calculate the maximum delay value and maximum backlog of traffic queue under CBS shaping mechanism to solve the node self-locking problem in avionics network.

3. The method for solving the self-locking problem of avionics network nodes caused by overlapping traffic paths in CBS network calculation according to claim 1 is characterized in that: The self-locking problem is overcome by replacing the flow arrival curve at the self-locking port with the shaping curve of the flow at the previous node port.

Images