CN105337885B - A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method - Google Patents

A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method Download PDF

Info

Publication number
CN105337885B
CN105337885B CN201510629458.8A CN201510629458A CN105337885B CN 105337885 B CN105337885 B CN 105337885B CN 201510629458 A CN201510629458 A CN 201510629458A CN 105337885 B CN105337885 B CN 105337885B
Authority
CN
China
Prior art keywords
time
sensitive
flows
flow
delay
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201510629458.8A
Other languages
Chinese (zh)
Other versions
CN105337885A (en
Inventor
谷晓燕
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Information Science and Technology University
Original Assignee
Beijing Information Science and Technology University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Information Science and Technology University filed Critical Beijing Information Science and Technology University
Priority to CN201510629458.8A priority Critical patent/CN105337885B/en
Publication of CN105337885A publication Critical patent/CN105337885A/en
Application granted granted Critical
Publication of CN105337885B publication Critical patent/CN105337885B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses a kind of multistages suitable for credit shaping network to be grouped the worst Delay computing method, and the process employs the computational methods that band width in physical constraint is grouped plus the multistage of logic bandwidth constraint.Increase boundary value according to the worst case of high low-priority traffic scheduling and corresponding credit, calculates flow disposably maximum sending time；Maximum flow burstiness is reached by judgement and the magnitude relationship of disposable maximum sending time increases logic bandwidth constraint on the basis of band width in physical constrains, and realizes the multistage grouping curve description for reaching flow.This method can effectively improve the accuracy of network delay calculating, belong to network communication technology field.

Description

A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method
Technical field
The present invention relates to the Delay computings of network communication technology field, refer to that one kind being applied to credit more particularly The worst Delay computing method of shaping network.
Background technology
IEEE AVB (Ethernet Audio/Video Bridging) network, be on the basis of traditional ethernet network, By guarantee bandwidth, the means such as limited delay is synchronous with precision clock improve time-sensitive (Time-Sensitive) information The real-time guarantees of transmission, to support the various networking technologys based on audio, the network-multimedia application of video.AVB (Ethernet sounds Video bridging technology) using the method realization flow control mesh based on credit shaping (Credited-based shaping) , it is embodied in 802.1 Qav standards of IEEE.
AVB supports two kinds of time-sensitive discharge patterns, i.e.,：SR_A time-sensitives flow and SR_B time-sensitive flows, SR_ A priority is higher than SR_B.SR_A and SR_B realizes real-time Transmission guarantee by bandwidth reservation (Stream Reservation), and Traffic shaping is realized using credit mechanism.Specifically：For SR_A or SR_B, if the credit corresponding to it is more than Or when equal to 0, and no high-priority packets are ready to be sent, then data packet is sent in SR_A SR_B queues License.The transmission slope of SR_A time-sensitive flows is denoted as sendslopeSR-A, the transmission slope note of SR_B time-sensitive flows For sendslopeSR-B.The recovery slope (also referred to as ensureing bandwidth rates) of SR_A time-sensitive flows is denoted as idleslopeSR-A, The recovery slope (also referred to as ensureing bandwidth rates) of SR_B time-sensitive flows is denoted as idleslopeSR-B.When data packet is sent Between in section, corresponding credit is according to sendslopeSR-AOr sendslopeSR-BSlope reduce, until credit is less than It sends and stops when 0；If there is buffered data packet in SR_A SR_B queues, but there is no the license of transmission, then credit Amount is according to idleslopeSR-AOr idleslopeSR-BSlope increases always；If do not buffered in SR_A SR_B queues Data packet, and credit is less than 0, then and credit is according to idleslopeSR-AOr idleslopeSR-BSlope stops when increasing to 0 Only, if credit is more than 0, credit resets to 0.
The worst delay is to weigh the important performance indexes of built-in network performance, and network calculus generally may be used The methods of (Network Calculus), method of loci (Trajectory Approach), pattern checking (Model Checking) Carry out the worst delayed evaluation.It is carried out in application, such as when AVB networks are placed in embedded environment：Avionics system needs Objective evaluation is carried out to the real-time performance of AVB networks.The worst delay that network is obtained using network calculus theory, is successfully answered For AFDX (Avionics Full Duplex Switch Ethernet) network, but AVB network credit shaping methods with AFDX networks Desired Min Tx Interval (Bandwidth Allocation Gap, BAG) flow shaping method is totally different, in the worst delay The AVB credit shaping features that need to combine closely when evaluation are rationally assessed.
In document " R.Queck.Analysis of Ethernet AVB for Automotive Network using Network Calculus.In Proceeding of IEEE international conference on Vehicular Electronics and Safety (ICVES 2012) .IEEE, 2012, discuss in pp.61-67 " AVB networks SR_A and The worst case of the flow scheduling of SR_B, and the calculation formula of flow corresponding with service Curve Technique delay is given, in conjunction with TAWith TB, the further precision description to SR_A and SR_B arrival curves may be implemented.
Invention content
In order to solve the worst Delay computing of any one node different priorities flow in existing AVB networks, the present invention Propose a kind of the worst Delay calculus method of the multistage packetization delay suitable for credit shaping network.The process employs physics The technological means that bandwidth constraint is grouped plus the multistage of logic bandwidth constraint.The worst item according to high low-priority traffic scheduling Part and corresponding credit increase boundary value, calculate flow disposably maximum sending time；It is prominent that maximum flow is reached by judgement The magnitude relationship of hair degree and disposable maximum sending time increases logic bandwidth constraint on the basis of band width in physical constrains, real Now reach the multistage grouping curve description of flow；Effectively increase the accuracy of network delay calculating.
A kind of multistage suitable for credit shaping network proposed by the present invention is grouped the worst Delay computing method, feature It is to include the following steps：
Step 1：Calculate the technology time delay of input port；
Time according to the corresponding service curves tachnical delay of SR_A time-sensitive flows Calculate any one input port portiUnder SR_A time-sensitive flows corresponding to service curves technology time delayi For the identification number of input port；Then step 2 is executed；
Time according to the corresponding service curves tachnical delay of SR_B time-sensitive flows Calculate any one input port portiUnder SR_B time-sensitive flows corresponding to service curves technology time delayi For the identification number of input port；Then step 2 is executed；
Step 2：Obtain first turning point in multistage curve；
FoundationSR_A time-sensitive flows are calculated with physical link rate The disposable maximum sending time of the data flow of CThen step 3 is executed；
FoundationSR_B time-sensitives flow is calculated with physical link speed The disposable maximum sending time of the data flow of rate CThen step 3 is executed；
Step 3：Obtain the turning point under physical link constraint；
Be arranged data packet burstiness b and in AVB networks message transmit source node S rc, destination node Dst；
To current data packet burstiness bCurrentlyUse the data packet burstiness b of previous nodePrevious nodeProlong with the worst flow LateThe sum of quantify, i.e.,ρ is to continue flow,It is message preceding The worst delay of one node；
FoundationUnder the single physical link constraint for calculating SR_A time-sensitive flows Turning point time；Then step 4 is executed；
FoundationUnder the single physical link constraint for calculating SR_B time-sensitive flows Turning point time；Then step 4 is executed；
Step 4：Obtain multistage curve establishment condition；
IfThen show that the burstiness of the SR_A time-sensitive flows reached is smaller, with link rate C's The transmission that all data packets of SR_A time-sensitive flows can be completed within the scope of maximum disposable sending time, records the grouping The last turning point of arrival curve isAnd it jumps into step 6 and is subsequently calculated；
IfThen show that the burstiness of the SR_A time-sensitive flows reached is larger, it cannot be with link speed The transmission that all data packets can be completed within the scope of the disposable sending time of maximum of rate C, will form SR_A time-sensitive flows Second segment curve, enter step five；
IfThen show reach SR_B time-sensitive bursts of traffic degree it is smaller, with link rate C most The transmission that all data packets of SR_B time-sensitive flows can be completed within the scope of big disposable sending time, records grouping arrival The last turning point of curve isAnd it jumps into step 6 and carries out follow-up calculating judgement；
IfThen show that the SR_B time-sensitive bursts of traffic degree reached is larger, it cannot be with link rate C The disposable sending time of maximum within the scope of can complete the transmissions of all data packets, the of SR_B time-sensitive flows will be formed Two sections of curves, enter step five；
Step 5：Obtain the last turning point ensured under bandwidth rates constraint；
It is bent that the curve that SR_A time-sensitives flow is formed in the case where ensureing bandwidth constraint constitutes SR_A time-sensitive flows second segment Line, the projection relation according to SR_A time-sensitive flow second segment curves The equivalent point for calculating the longitudinal axis, is denoted as
Consider SR_A time-sensitives flow to ensure bandwidth rates idleslopeSR-ASending time, closed according to turning point SystemCalculate SR_A time-sensitive flow multistages reach curve, second Turning pointIt then will be describedIt is assigned to last turning pointExecute step 6；
It is bent that the curve that SR_B time-sensitives flow is formed in the case where ensureing bandwidth constraint constitutes SR_B time-sensitive flows second segment Line, the projection relation according to SR_B time-sensitive flow second segment curves The equivalent point for calculating the longitudinal axis, is denoted as
Consider SR_B time-sensitives flow to ensure bandwidth rates idleslopeSR-BSending time, closed according to turning point SystemCalculate second turn that SR_B time-sensitive flow multistages reach curve BreakIt then will be describedIt is assigned to last turning pointExecute step 6；
Step 6：Compete output flow traversal；
Check the input port port corresponding to competition output flowiWhether traversal finishes, if traversal finishes, obtains out Maximum turning pointWith
It is describedIt is also the worst delay time point of SR_A time-sensitive flows, and according to traffic packet relationshipCalculate SR_A time-sensitive flows Cumulative bit stream executes step 7；
It is describedIt is also the worst delay time point of SR_B time-sensitive flows, and according to traffic packet relationshipCalculate SR_B time-sensitive flows Cumulative bit stream, execute step 7；
If traversal does not finish, one is gone to step, recalculate port under next input porti+1SR flows institute The technology time delay of corresponding service curves；
Step 7：Obtain the worst delay of output port；
The present invention is suitable for the advantages of multistage packetization delay the worst Delay calculus method of credit shaping network and is：
1. calculating first turning point using band width in physical constraint, be conducive to the equivalent projection for obtaining second segment curve Point.Then second segment curve is characterized according to equivalent subpoint.
2. calculating second turning point using logic bandwidth constraint, flow is characterized in the form of being segmented arrival curve, more Close to the true arrival process of flow.
3. the present invention is using segmentation arrival curve, in conjunction with packet mode, the worst delay for implementing credit shaping network is drilled Point counting is analysed, and the grouping calculation of band width in physical constraint, the worst delay result sigma compactness are used compared to grouping calculation, and list is not used It is stronger, it is closer to the worst true delay of network.
4. the present invention combines the logic bandwidth of credit shaping network to the cut-off current characteristics of arrival curve, and combines physical tape The segmentation description of flow arrival curve is implemented in width constraint, is constrained compared to single band width in physical, segmentation arrival mode is more life-like.
Description of the drawings
Fig. 1 is that the present invention describes credit shaping network arrival stream schematic diagram using sectional curve.
Fig. 2 is that the present invention calculates flow chart using multistage packetization delay.
Fig. 3 is topological structure of the present invention using multistage packetization delay calculation case.
Fig. 4 A are delay comparison diagram of the present invention using multistage packetization delay calculation case.
Fig. 4 B are delay comparison diagram of the present invention using multistage packetization delay calculation case.
Specific implementation mode
Below in conjunction with drawings and examples, the present invention is described in further detail.
A kind of multistage suitable for credit shaping network proposed by the present invention is grouped the worst Delay computing method, this method Document " R.Queck.Analysis of Ethernet AVB for Automotive Network using are borrowed Network Calculus.In Proceeding of IEEE international conference on Vehicular Electronics and Safety (ICVES 2012) .IEEE, 2012, the formula (5) of page 64 in pp.61-67 ".In order to Convenient for the unification to quoting letter in formula in the present invention, the letter in " formula (5) " is changed to by spy：
TSR-AIndicate the time of the corresponding service curves tachnical delay of SR_A time-sensitive flows；
TSR-BIndicate the time of the corresponding service curves tachnical delay of SR_B time-sensitive flows；
LSR-A,maxIndicate that the corresponding maximum data packet packet of SR_A time-sensitive flows is long；
LSR-B,maxIndicate that the corresponding maximum data packet packet of SR_B time-sensitive flows is long；
LmaxIndicate that the maximum data packet packet corresponding to the non-temporal sensitive flows of AVB is long；
C indicates physical link rate；
sendslopeSR-AIndicate the transmission slope of SR_A time-sensitive flows.
In credit shaping (Credited-based shaping) network, to using in the SR flows of AVB technologies The package identification number of SR_A time-sensitive flows is denoted as dataSR-A, and the package identification number of SR_B time-sensitive flows is remembered For dataSR-B。
Shown in Fig. 1, Fig. 2, the present invention proposes that a kind of multistage grouping suitable for credit shaping network is the worst to prolong Slow computational methods, this method the specific steps are：
Step 1：Calculate the technology time delay of input port；
In the present invention, foundationCalculate any one input port porti(i is Input port identification number) under SR_A time-sensitive flows corresponding to service curves technology time delayThen step is executed Rapid two.
In the present invention, foundationCalculate any one input port porti The technology time delay of the service curves corresponding to SR_B time-sensitive flows under (i is input port identification number)Then Execute step 2.
Step 2：Obtain first turning point in multistage curve；
FoundationSR_A time-sensitive flows are calculated with physical link rate The disposable maximum sending time of the data flow of CThen step 3 is executed.
FoundationSR_B time-sensitives flow is calculated with physical link speed The disposable maximum sending time of the data flow of rate CThen step 3 is executed.
In the present invention,First turning point of SR_A time-sensitive flow multistage arrival curves is defined, In first segment curve, arrival curve slope is physical link rate C, starting point MSR-A(single packet maximum burst degree), foundation It is defined in AVB, single packet maximum burst degree MSR-AFor the maximum packet length of SR_A flows, i.e. MSR-A=LSR-A,max。
In the present invention,First turning point of SR_B time-sensitive flow multistage arrival curves is defined, In first segment curve, arrival curve slope is physical link rate C, starting point MSR-B(single packet maximum burst degree), foundation It is defined in AVB, single packet maximum burst degree MSR-BFor the maximum packet length of SR_B flows, i.e. MSR-B=LSR-B,max。
Step 3：Obtain the turning point under physical link constraint；
Be arranged data packet burstiness b and in AVB networks message transmit source node S rc, destination node Dst.
In the present invention, purpose section is eventually arrived at through intermediate node Node from source node S rc due to the message in AVB networks Point Dst.Then to current data packet burstiness bCurrentlyUse the data packet burstiness b of previous nodePrevious nodeProlong with the worst flow LateThe sum of quantify, i.e.,ρ is to continue flow,It is message preceding The worst delay of one node.If present node is source node S rc, data packet burstiness b=MIF × L, MIF are a measurement It is spaced CMI (Class Measurement Interval) sum of data packet in the period, L is that data packet packet is long.
In the present invention, the message in AVB networks eventually arrives at destination node from source node S rc through intermediate node Node Dst.When there are the SR flows of multiple same priorities, the cumulative data packet burstiness of cumulative all SR flows is needed, is denoted as bIt is accumulative。
It is denoted as b for the cumulative data packet burstiness of SR_A time-sensitive flowsIt is accumulative, SR-A。
It is denoted as b for the cumulative data packet burstiness of SR_B time-sensitive flowsIt is accumulative, SR-B。
It is denoted as MIF for the sum of data packet within a CMI period of SR_A time-sensitive flowsSR-A。
It is denoted as MIF for the sum of data packet within a CMI period of SR_B time-sensitive flowsSR-B。
It is denoted as L for the corresponding maximum data packet packet length of SR_A time-sensitive flowsSR-A,max。
It is denoted as L for the corresponding maximum data packet packet length of SR_B time-sensitive flowsSR-B,max。
Single packet maximum burst degree M is set, M is denoted as the single packet maximum burst degree of SR_A time-sensitive flowsSR-A, It is denoted as M for the single packet maximum burst degree of SR_B time-sensitive flowsSR-B。
Setting continues flow ρ, and ρ is denoted as the lasting flow of SR_A time-sensitive flowsSR-A, for SR_B time-sensitives The lasting flow of flow is denoted as ρSR-B.When there is the SR flows of multiple same priorities, ρ needs the lasting stream of cumulative all SR streams Amount then has the cumulative of SR_A time-sensitive flows to continue flow, is denoted as ρIt is accumulative, SR-A.The cumulative lasting stream of SR_B time-sensitive flows Amount, is denoted as ρIt is accumulative, SR-B。
FoundationUnder the single physical link constraint for calculating SR_A time-sensitive flows Turning point time.
FoundationUnder the single physical link constraint for calculating SR_B time-sensitive flows Turning point time.
Step 4：Obtain multistage curve establishment condition；
In the present invention, it comparesWithRelative size,
Step 5：Obtain the last turning point ensured under bandwidth rates constraint；
It is bent that the curve that SR_A time-sensitives flow is formed in the case where ensureing bandwidth constraint constitutes SR_A time-sensitive flows second segment Line, the projection relation according to SR_A time-sensitive flow second segment curves The equivalent point for calculating the longitudinal axis, is denoted as
In the present invention, consider SR_A time-sensitives flow to ensure bandwidth rates idleslopeSR-ASending time, According to turning point relationshipIt calculates SR_A time-sensitive flow multistages and reaches bent Line, second turning pointIt then will be describedIt is assigned to last turning pointExecute step Six.
It is bent that the curve that SR_B time-sensitives flow is formed in the case where ensureing bandwidth constraint constitutes SR_B time-sensitive flows second segment Line, the projection relation according to SR_B time-sensitive flow second segment curves The equivalent point for calculating the longitudinal axis, is denoted as
In the present invention, consider SR_B time-sensitives flow to ensure bandwidth rates idleslopeSR-BSending time, According to turning point relationshipIt calculates SR_B time-sensitive flow multistages and reaches bent Second turning point of lineIt then will be describedIt is assigned to last turning pointExecute step Six.
Step 6：Compete output flow traversal；
Check the input port port corresponding to competition output flowiWhether traversal finishes, if traversal finishes, obtains out maximum Turning pointWithIt is describedAnd the worst delay time point of SR_A time-sensitive flows, and foundation Traffic packet relationshipWhen calculating SR_A Between sensitive flow cumulative bit stream, execute step 7；It is describedIt is also the worst delay time of SR_B time-sensitive flows Point, and according to traffic packet relationship The cumulative bit stream of SR_B time-sensitive flows is calculated, step 7 is executed；
If traversal does not finish, one is gone to step, recalculate port under next input porti+1SR flows (packet Include SR_A time-sensitives flow and SR_B time-sensitives flow) corresponding to service curves technology time delay.It indicates The worst delay time point of SR_A time-sensitive flows.Indicate the worst delay time of SR_B time-sensitive flows Point.
Step 7：Obtain the worst delay of output port；
In the present invention, step 1 to step 7 completes the worst delay of multistage grouping of this grade of competition output port SR flow Calculation analysis.Use the computational methods that band width in physical constraint is grouped plus the multistage of logic bandwidth constraint.This method can have Effect improves the accuracy that network delay calculates.
Embodiment 1
Sectional curve shown in Figure 1 describes credit shaping network and reaches stream schematic diagram, it is assumed that arbitrary in AVB The lasting uninterrupted of the high priority SR_A data flows of one node is denoted as ρSR-A=5.12Mbps, measurement interval CMI weeks There are 100 data packet data in phaseSR-AEtc. to be sent, each data packet dataSR-ASize be 64Bytes.The node has simultaneously One low priority SR_B data flow (Video) etc. is to be sent, data packet dataSR-BSize be 1522Bytes.SR_A data The idleslope of the corresponding reshaper of streamSR-AFor 65Mbps, sendslopeSR-AFor 35Mbps.
When the delay situation in view of SR_A data flows, under worst case, the data packet data of SR_B time-sensitive flowsSR-B It is first sent, is dependent on TSR-ACalculation formula, it can be deduced that TSR-A=121.76 milliseconds.In this sending time, when SR_A Between sensitive flow credit increase boundary be CHSR-A=TSR-A×idleslopeSR-A=7914.4.As data packet dataSR-BIt is complete At transmission, data packet dataSR-APreferential transmission route is obtained, according to sendslopeSR-ASetting, can calculate SR_A time-sensitives Disposably maximum sending time is the data flow of flow
Then existTime range in, SR_A time-sensitives flow at most may be used according to link rate C=100Mbps Number with transmission data packet isTherefore the 1st~45 data packet dataSR-AAccording to chain Road rate is forwarded to next-hop network node.
For the 46th~100 data packet dataSR-AThen according to idleslopeSR-AThe guarantee bandwidth of definition carries out data packet Forwarding, forwarding time isTherefore the 46th~100 data Wrap dataSR-AIt is forwarded to next-hop network node according to guarantee bandwidth rates.
In entire 1~100 data packet dataSR-ASending time Interior (including most starting waiting for SR_B data packets sending time), is dependent on the lasting flow of SR_A, can calculate Data packet dataSR-AIncreased data packet number is
Therefore, for the 101st~108 data packet dataSR-A, so that the follow-up data packet further overstock 109 ..., it will continue to be forwarded according to guarantee bandwidth rates.
The arrival curve of next-hop SR_A time-sensitive flows is further looked at, it can be found that the 1st~45 data packet dataSR-AReach according to link rate, the 46th~108 data packet dataSR-AIt is reached according to guarantee bandwidth, subsequent data packet It is reached according to lasting flow after 109, therefore further the arrival curve after precision has the spy of multistage curve combination Sign：I.e. first segment curve is 1-45 data packet group at second segment curve is 46-108 data packet group at third section curve is 109 data packet composition described above.
Therefore, in the present invention, go to carry out the worst Delay computing using segmentation arrival curve, it is stronger sigma compactness can be obtained Delay computing result.The relative size relationship wherein postponed is：
Idleslope constraint delays<Multistage packetization delay<Link rate constraint delay<Not packetization delay.
Specifically, the size of each delay time is that (flow caused by not considering the competition output of message source flow is most Big burstiness extends influence)：
Idleslope constrains delay=121.76us；
Multistage packetization delay=251.38us；
Link rate constrains delay=417.30us；
Not packetization delay=909.45us.
It is presented in Fig. 3 the concrete application case of an AVB network, AA, BB, CC, DD, EE and FF letter in figure point Interchanger is not represented, and A, B, C, D, E, F, G, H, I, J and K letter respectively represent end system.In the topology knot of the AVB networks of Fig. 3 In structure, link rate and port speed are all 100Mbps.The end system C (source node) being connected on switch b B produces SR_ A and SR_B time-sensitive flows；The end system D being connected on switch b B produces SR_A and SR_B time-sensitive flows； This four roads time-sensitive flow eventually arrives at the end system L (purposes being connected on interchanger CC by interchanger DD, interchanger EE Node).Fig. 3 gives the recovery corresponding to SR_A time-sensitives flow and SR_B time-sensitive flows in each sending port Slope, the recovery slope composition recovery Slope Parameters on multiple ports are to (idleslopeSR-A,idleslopeSR-B), unit Mbps。
The design parameter of 1 flow of table
Flow ID Priority Source node Destination node Message-length (byte) CMI(ms) MIF
1 SR_A End system C End system L 64 10 100
2 SR_B End system C End system L 1522 100 10
3 SR_A End system H End system L 64 10 100
4 SR_B End system H End system L 1522 100 10
SR_A time-sensitive flow simulations control signal, message-length be 64 bytes, CMI 10ms, MIF 100, thus The lasting flow of the single message of formation is 5.12Mbs, and maximum burst degree is 51200bit.
SR_B time-sensitive flow simulation audio signals, message-length be 1522 bytes, CMI 100ms, MIF 10, by The lasting flow of this single message formed is 1.217Mbps, and maximum burst degree is 121760bit.
The worst delay of each port is carried out using the worst Delay calculus method of multistage packetization delay proposed by the present invention It calculates, and result of calculation and not packetization delay and link rate constraint Delay computing result is compared, comparing result is as schemed Shown in 4A and Fig. 4 B.Using the method for the present invention acquired results both related method thereofs gained is constrained than not group technology and link rate As a result small, of the invention method acquired results is wanted to be closer to the worst true delay result of time-sensitive flow.
(1) not packetization delay compares：The worst delay that SR_A time-sensitive flows are reduced using the method for the present invention is reached 36-73%；The worst delay that SR_B time-sensitive flows are reduced using the method for the present invention reaches 40-82%.
(2) link rate constraint delay comparison：The worst the prolonging of SR_A time-sensitive flows is reduced using the method for the present invention Reach 13-40% late；The worst delay that SR_B time-sensitive flows are reduced using the method for the present invention reaches 34-76%.
Method proposed by the present invention is using segmentation packetization delay processing, than individually only being prolonged using link rate constraint It calculates and accurately to be become apparent from than not being grouped Delay computing method advantage late.From above-mentioned relation it can be found that multistage packetization delay meter It is smaller compared to not packetization delay and the delay of single-link rate constraint to calculate result, is closer to data flow the worst true delay knot Fruit, but more than list idleslope constraint delays, but idleslope constraint delays do not account for AVB and flow in practical repeating process The case where in the presence of being happened suddenly with link rate.
The present invention is a kind of the worst Delay computing method of multistage grouping suitable for credit shaping network, to be solved The technical issues of being how to improve Delay computing precision in network communication field, this method add logic by band width in physical constraint The technological means of the multistage grouping of bandwidth constraint, to realize that the worst delay to the competition output of any one node flow is estimated Meter can effectively improve the technique effect of the accuracy of network delay calculating.

Claims (5)

  1. It is respectively to SR_ in AVB networks 1. a kind of multistage suitable for credit shaping network is grouped the worst Delay computing method The processing of A time-sensitives flow and SR_B time-sensitive flows, it is characterised in that include the following steps：
    Step 1：The technology time delay of the calculating input port of SR_A time-sensitive flows；
    Time according to the corresponding service curves tachnical delay of SR_A time-sensitive flows Calculate any one input port portiUnder SR_A time-sensitive flows corresponding to service curves technology time delayi For the identification number of input port；Then step 2 is executed；
    TSR-AIndicate the time of the corresponding service curves tachnical delay of SR_A time-sensitive flows；
    LSR-B,maxIndicate that the corresponding maximum data packet packet of SR_B time-sensitive flows is long；
    LmaxIndicate that the maximum data packet packet corresponding to the non-temporal sensitive flows of AVB is long；
    C indicates physical link rate；
    Step 2：First turning point of SR_A time-sensitive flows obtained in multistage curve；
    FoundationSR_A time-sensitives flow is calculated with physical link rate C's The disposable maximum sending time of data flowThen step 3 is executed；Define SR_A time-sensitive flows First turning point of multistage arrival curve, in first segment curve, arrival curve slope is physical link rate C, and starting point is Single packet maximum burst degree MSR-A, it is dependent on AVB definition, single packet maximum burst degree MSR-AFor the maximum packet length of SR_A flows, That is MSR-A=LSR-A,max；
    Step 3：The turning point of SR_A time-sensitive flows obtained under physical link constraint；
    Be arranged data packet burstiness b and in AVB networks message transmit source node S rc, destination node Dst；
    To current data packet burstiness bCurrentlyUse the data packet burstiness b of previous nodePrevious nodeWith the worst traffic delayThe sum of quantify, i.e.,ρ is to continue flow,It is message previous The worst delay of node；
    FoundationCalculate the turnover under the single physical link constraint of SR_A time-sensitive flows The point time；Then step 4 is executed；
    It is denoted as b for the cumulative data packet burstiness of SR_A time-sensitive flowsIt is accumulative, SR-A；
    It is denoted as MIF for the sum of data packet within a CMI period of SR_A time-sensitive flowsSR-A；
    It is denoted as L for the corresponding maximum data packet packet length of SR_A time-sensitive flowsSR-A,max；
    Single packet maximum burst degree M is set, M is denoted as the single packet maximum burst degree of SR_A time-sensitive flowsSR-A；
    Setting continues flow ρ, and ρ is denoted as the lasting flow of SR_A time-sensitive flowsSR-A；When there is multiple same priorities When SR flows, ρ needs the lasting flow of cumulative all SR streams, then has the cumulative of SR_A time-sensitive flows to continue flow, be denoted as ρIt is accumulative, SR-A；
    Step 4：The acquisition multistage curve establishment condition of SR_A time-sensitive flows；
    ComparisonWithRelative size, obtain
    IfThen show that the burstiness of the SR_A time-sensitive flows reached is smaller, with the maximum of link rate C The transmission that all data packets of SR_A time-sensitive flows can be completed within the scope of disposable sending time records grouping arrival The last turning point of curve isAnd it jumps into step 6 and is subsequently calculated；
    IfThen show that the burstiness of the SR_A time-sensitive flows reached is larger, it cannot be with link rate C's The transmission that all data packets can be completed within the scope of maximum disposable sending time, will form the second of SR_A time-sensitive flows Section curve, enters step five；
    Step 5：The acquisition of SR_A time-sensitive flows ensures the last turning point under bandwidth rates constraint；
    The curve that SR_A time-sensitives flow is formed in the case where ensureing bandwidth constraint constitutes SR_A time-sensitive flow second segment curves, according to According to the projection relation of SR_A time-sensitive flow second segment curves The equivalent point for calculating the longitudinal axis, is denoted as
    Consider SR_A time-sensitives flow to ensure bandwidth rates idleslopeSR-ASending time, according to turning point relationshipCalculate second turnover that SR_A time-sensitive flow multistages reach curve PointIt then will be describedIt is assigned to last turning pointExecute step 6；
    Step 6：The competition output flow of SR_A time-sensitive flows traverses；
    Check the input port port corresponding to competition output flowiWhether traversal finishes, if traversal finishes, obtains out maximum Turning point
    It is describedIt is also the worst delay time point of SR_A time-sensitive flows, and according to traffic packet relationshipCalculate SR_A time-sensitive flows Cumulative bit stream executes step 7；
    If traversal does not finish, one is gone to step, recalculate port under next input porti+1SR flows corresponding to Service curves technology time delay；
    Step 7：The worst delay of the acquisition output port of SR_A time-sensitive flows；
    Since the competition of packet traffic exports, foundationCalculate output end The worst delay of the SR_A flows of mouthNext stage output for SR_A time-sensitive flows, the worst delay of this gradeThe worst delay of the previous node in next stage data packet burstiness calculating process is constituted, i.e.,
    Step 8：The technology time delay of the calculating input port of SR_B time-sensitive flows；
    Time according to the corresponding service curves tachnical delay of SR_B time-sensitive flows Calculate any one input port portiUnder SR_B time-sensitive flows corresponding to service curves technology time delayi For the identification number of input port；Then step 9 is executed；
    TSR-BIndicate the time of the corresponding service curves tachnical delay of SR_B time-sensitive flows；
    LSR-A,maxIndicate that the corresponding maximum data packet packet of SR_A time-sensitive flows is long；
    sendslopeSR-AIndicate the transmission slope of SR_A time-sensitive flows；
    Step 9：First turning point of SR_B time-sensitive flows obtained in multistage curve；
    FoundationSR_B time-sensitives flow is calculated with physical link rate C's The disposable maximum sending time of data flowThen step 10 is executed；Define SR_B time-sensitive streams First turning point for measuring multistage arrival curve, in first segment curve, arrival curve slope is physical link rate C, starting point For single packet maximum burst degree MSR-B, it is dependent on AVB definition, single packet maximum burst degree MSR-BFor the maximum packet length of SR_B flows Degree, i.e. MSR-B=LSR-B,max；
    Step 10：The turning point of SR_B time-sensitive flows obtained under physical link constraint；
    Be arranged data packet burstiness b and in AVB networks message transmit source node S rc, destination node Dst；
    To current data packet burstiness bCurrentlyUse the data packet burstiness b of previous nodePrevious nodeWith the worst traffic delayThe sum of quantify, i.e.,ρ is to continue flow,It is message previous The worst delay of node；
    FoundationCalculate the turnover under the single physical link constraint of SR_B time-sensitive flows The point time；Then step 11 is executed；
    It is denoted as b for the cumulative data packet burstiness of SR_B time-sensitive flowsIt is accumulative, SR-B；
    It is denoted as MIF for the sum of data packet within a CMI period of SR_B time-sensitive flowsSR-B；
    It is denoted as L for the corresponding maximum data packet packet length of SR_B time-sensitive flowsSR-B,max；
    Single packet maximum burst degree M is set, M is denoted as the single packet maximum burst degree of SR_B time-sensitive flowsSR-B；
    Setting continues flow ρ, and ρ is denoted as the lasting flow of SR_B time-sensitive flowsSR-B；When there is multiple same priorities When SR flows, ρ needs the lasting flow of cumulative all SR streams, then has the cumulative of SR_B time-sensitive flows to continue flow, be denoted as ρIt is accumulative, SR-B；
    Step 11：The acquisition multistage curve establishment condition of SR_B time-sensitive flows；
    ComparisonWithRelative size, obtain
    IfThen show that the SR_B time-sensitive bursts of traffic degree reached is smaller, with the maximum one of link rate C The transmission that all data packets of SR_B time-sensitive flows can be completed within the scope of secondary property sending time, records the grouping arrival curve Last turning point beAnd it jumps into step 13 and carries out follow-up calculating judgement；
    IfThen show reach SR_B time-sensitive bursts of traffic degree it is larger, cannot with link rate C most The transmission that all data packets can be completed within the scope of big disposable sending time, will form the second segment of SR_B time-sensitive flows Curve enters step 12；
    Step 12：The acquisition of SR_B time-sensitive flows ensures the last turning point under bandwidth rates constraint；
    The curve that SR_B time-sensitives flow is formed in the case where ensureing bandwidth constraint constitutes SR_B time-sensitive flow second segment curves, foundation The projection relation of SR_B time-sensitive flow second segment curvesMeter The equivalent point for calculating the longitudinal axis, is denoted as
    Consider SR_B time-sensitives flow to ensure bandwidth rates idleslopeSR-BSending time, according to turning point relationshipCalculate second turnover that SR_B time-sensitive flow multistages reach curve PointIt then will be describedIt is assigned to last turning pointExecute step 13；
    Step 13：The competition output flow of SR_B time-sensitive flows traverses；
    Check the input port port corresponding to competition output flowiWhether traversal finishes, if traversal finishes, obtains out maximum Turning point
    It is describedIt is also the worst delay time point of SR_B time-sensitive flows, and according to traffic packet relationship
    Calculate SR_B time-sensitive streams The cumulative bit stream of amount executes step 14；
    If traversal does not finish, eight are gone to step, recalculate port under next input porti+1SR flows corresponding to Service curves technology time delay；
    Step 14：The worst delay of the acquisition output port of SR_B time-sensitive flows；
    Since the competition of packet traffic exports, foundationCalculate output end The worst delay of the SR_B flows of mouthNext stage output for SR_B time-sensitive flows, the worst delay of this gradeConstitute the worst delay of the previous node in next stage data packet burstiness calculating process；I.e.
  2. 2. a kind of multistage suitable for credit shaping network according to claim 1 is grouped the worst Delay computing method, It is characterized in that：In step 3 and step 10, if present node is source node S rc, data packet burstiness b=MIF × L, MIF is the sum of data packet in a measurement interval CMI period, and L is that data packet packet is long.
  3. 3. a kind of multistage suitable for credit shaping network according to claim 1 is grouped the worst Delay computing method, It is characterized in that：In step 3 and step 10, the message in AVB networks is from source node S rc through intermediate node Node most Zhongdao Up to destination node Dst；When there are the SR flows of multiple same priorities, the cumulative data packet burst of cumulative all SR flows is needed Degree；When there is the SR flows of multiple same priorities, ρ needs the lasting flow of cumulative all SR streams.
  4. 4. a kind of multistage suitable for credit shaping network according to claim 1 is grouped the worst Delay computing method, It is characterized in that：The worst delay of SR_A time-sensitive flows can be made to reduce using multistage packetization delay comparison and reach 36- 73%；The worst delay reduction of SR_B time-sensitive flows reaches 40-82%.
  5. 5. a kind of multistage suitable for credit shaping network according to claim 1 is grouped the worst Delay computing method, It is characterized in that：Constraining delay comparison using link rate can be such that the worst delay reduction of SR_A time-sensitive flows reaches 13-40%；The worst delay reduction of SR_B time-sensitive flows reaches 34-76%.
CN201510629458.8A 2015-09-28 2015-09-28 A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method Active CN105337885B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510629458.8A CN105337885B (en) 2015-09-28 2015-09-28 A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510629458.8A CN105337885B (en) 2015-09-28 2015-09-28 A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method

Publications (2)

Publication Number Publication Date
CN105337885A CN105337885A (en) 2016-02-17
CN105337885B true CN105337885B (en) 2018-08-17

Family

ID=55288185

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510629458.8A Active CN105337885B (en) 2015-09-28 2015-09-28 A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method

Country Status (1)

Country Link
CN (1) CN105337885B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107070808B (en) * 2017-02-28 2020-03-10 北京航空航天大学 AVB network multi-class predetermined stream worst queuing delay processing method based on credit shaping
CN108199911B (en) * 2017-12-07 2021-08-03 中国航空工业集团公司西安航空计算技术研究所 Method for testing AFDX (avionics full Duplex switched Ethernet) shared same virtual link credit
CN107911265B (en) * 2017-12-18 2019-08-20 北京航空航天大学 A kind of device of the AVB network flow maximum delay test based on CBS flow-control mechanism
CN108418723B (en) * 2018-05-29 2021-01-12 合肥工业大学 Vehicle-mounted Ethernet AVB reserved bandwidth optimal configuration method
CN113347109B (en) * 2021-06-24 2024-03-15 中国科学院沈阳自动化研究所 Industrial network heterogeneous traffic shaper supporting 5G and TSN interconnection
CN113630335B (en) * 2021-07-13 2023-12-26 北京电子工程总体研究所 Optimized selection method for switching Ethernet data flow control strategy
CN115412513B (en) * 2022-07-15 2023-12-12 中国航空无线电电子研究所 ARINC664 network receiving terminal time integrity checking method

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1545286A (en) * 2003-11-21 2004-11-10 清华大学 ECN based congestion control method with prediction verification
CN102970156A (en) * 2012-10-11 2013-03-13 上海交通大学 Optimal modeling method for end-to-end delay in avionics full duplex switched Ethernet (AFDX) data networks

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030099199A1 (en) * 2001-11-27 2003-05-29 Amplify.Net, Inc., Bandwidth allocation credit updating on a variable time basis

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1545286A (en) * 2003-11-21 2004-11-10 清华大学 ECN based congestion control method with prediction verification
CN102970156A (en) * 2012-10-11 2013-03-13 上海交通大学 Optimal modeling method for end-to-end delay in avionics full duplex switched Ethernet (AFDX) data networks

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
Analysis of Ethernet AVB for Automotive Networks using Network Calculus;Rene Queck;《2012 IEEE International Conference on Vehicular Electronics and Safety》;20120724;全文 *

Also Published As

Publication number Publication date
CN105337885A (en) 2016-02-17

Similar Documents

Publication Publication Date Title
CN105337885B (en) A kind of multistage suitable for credit shaping network is grouped the worst Delay computing method
CN103346922B (en) The controller of determination network state based on SDN and determine method
CN102804714B (en) Controlling packet transmission
CN101217495A (en) Traffic monitoring method and device applied under T-MPLS network environment
CN102595503B (en) Congestion control method based on wireless multimedia sensor network
Lim et al. Performance comparison of ieee 802.1 q and ieee 802.1 avb in an ethernet-based in-vehicle network
CN114422448B (en) Time-sensitive network traffic shaping method
CN111628942B (en) Resource allocation method in time-sensitive network
TW200507560A (en) Method and system for open-loop congestion control in a system fabric
CN107948103A (en) A kind of interchanger PFC control methods and control system based on prediction
CN107070808A (en) A kind of the worst queueing delay processing method of the pre- constant current of AVB network multiclass based on credit shaping
CN101420461A (en) Adaptive token barrel marking method based on service sensing
CN101692648A (en) Method and system for queue scheduling
CN107579877A (en) A kind of exchange of electric power machine flux monitoring system and monitoring method based on FPGA
Soni et al. Optimizing network calculus for switched ethernet network with deficit round robin
CN113630335A (en) Method for optimizing and selecting flow control strategy of switched Ethernet data
CA2309527A1 (en) Method and apparatus for managing communications between nodes in a bi-directional ring network
CN103746878B (en) Simulation system for measuring transmission time of SpaceWire data packets
CN109327406A (en) A method of the service quality guarantee for difference queue service queuing data packet
CN107070815A (en) Sync message queuing delay control method and device in a kind of SDN
Liu et al. Design of an improved Ethernet AVB model for real-time communication in in-vehicle network
CN115604193B (en) Deterministic resource scheduling method and system in hot rolling control system
Hu et al. The delay bound analysis based on network calculus for asynchronous traffic shaping under parameter inconsistency
CN109391559A (en) The network equipment
Shalghum et al. Critical offset optimizations for overlapping-based time-triggered windows in time-sensitive network

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant