CN105873096B - A Method for Optimizing Effective Throughput of Multipath Parallel Transmission System - Google Patents
A Method for Optimizing Effective Throughput of Multipath Parallel Transmission System Download PDFInfo
- Publication number
- CN105873096B CN105873096B CN201610178755.XA CN201610178755A CN105873096B CN 105873096 B CN105873096 B CN 105873096B CN 201610178755 A CN201610178755 A CN 201610178755A CN 105873096 B CN105873096 B CN 105873096B
- Authority
- CN
- China
- Prior art keywords
- time delay
- delay
- value
- data packet
- path
- 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
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/0289—Congestion control
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention relates to a kind of optimization methods of multipath parallel transmission system effective throughput, belong to field of communication technology.Method includes the following steps: step 1) in multipath parallel transmission system, is estimated using propagation delay of the Kalman filtering algorithm to the concurrent chain road of each item；Step 2) derives data packet order transfer constraint condition in multipath parallel transmission system；Step 3) is surrounded by the constraint condition of sequence propagation according to the propagation delay estimated result and data derived, is adjusted to the congestion window of transmitting terminal.A kind of optimization method of multipath parallel transmission system effective throughput provided by the invention, it can be estimated according to propagation delay and data are surrounded by sequence and propagate constraint condition, it is adaptively adjusted the congestion window size of each subflow, to the load of balanced each chain road, the maximum delay reduced between concurrent link is poor, data packet disorder is reduced, multipath parallel transmission system effective throughput is improved.
Description
Technical field
The invention belongs to fields of communication technology, are related to a kind of optimization side of multipath parallel transmission system effective throughput
Method.
Background technique
With the development of mobile communication technology, nowadays in the market the mobile terminal of mainstream be provided with greatly two or two with
On radio network interface.On this basis, how people utilizes multiple and different network interface progress multipaths if beginning one's study
Transmission promotes Streaming Media, video conference equiband intensity business so as to the available bandwidth polymerizeing on multiple wireless networks
QoS, wherein many research is on how to promote multipath parallel transmission system effective throughput.
In wireless communication technique highly developed today, the transmission rate of single access network already close to shannon limit,
Want to become more and more difficult by bottom layer signal processing optimisation technique raising rate.It, can be with however under heterogeneous network environment
By way of polymerizeing multiple network interface bandwidths the throughput performance of terminal is greatly improved, to meet business need
It asks.In the process, the Business Stream of transmitting terminal will be divided into multiple business subflows parallel transmission on multiple and different networks, and
It completes to converge in multiplex roles receiving end.In the ideal case the handling capacity of aggregated links can achieve multilink handling capacity it
With.
When multipath parallel transmission system is applied in heterogeneous network environment, due to not right between heterogeneous network difference path
Title property, so that data packet disorder phenomenon can occur for receiving end, and then limits mentioning for multipath parallel transmission system throughput performance
It rises.Data packet disorder (packet reordering) problem is due to the propagation delay time difference on different paths, and data packet reaches
The sequence of receiving end and transmission sequence be not identical, and the data packet sent afterwards may arrive earlier than the data packet sent before
Up to receiving end.However, providing in Stream Control Transmission Protocol, the data packet only sequentially reached could up submit application layer and be handled.
When data packet disorder phenomenon than it is more serious when, this, which allows for out-of-order data packet, can be trapped in the caching of receiving end, can not be timely
It submits upper layer application to be handled, increases the propagation delay time of grouping, reduce effective throughput performance of multipath parallel transmission.
Although existing correlative study can effectively improve the effective throughput of multipath parallel transmission, there is also one
A little problems.It includes: that incorrect end is arrived that multipath parallel transmission system performance under heterogeneous network environment, which is lower than expected reason,
Hold the asymmetry etc. between round-trip delay (Round-Trip Time, RTT) estimation and path.Incorrect RTT estimation makes total
There is error in implementing result according to packet shunting, data packet retransmission scheduling algorithm, and the asymmetry between path is then to cause to be transmitted across
Occur one of data packet disorder, basic reason that effective throughput reduces in journey.Existing research cannot take into account end-to-end very well
Asymmetry problem between round-trip delay estimation and path, so that multipath parallel transmission system can not be made in heterogeneous network environment
In the performance that is optimal.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of optimization sides of multipath parallel transmission system effective throughput
Method, this method are made of the time delay estimation based on Kalman filtering and the congestion avoidance algorithm two parts controlled based on delay inequality,
Data disorder phenomenon can be effectively reduced, the throughput performance of Transmission system is improved.
In order to achieve the above objectives, the invention provides the following technical scheme:
A kind of optimization method of multipath parallel transmission system effective throughput, in the method, multipath parallel transmission
There are the mutually independent transmission path of N item, respectively L between transmitting terminal and receiving end for system1,L2,...,Li,...LN, N >=
2；On each paths end to end propagated time delay meet d (1)≤d (2)≤... d (i) ...≤d (N)；This method include with
Lower step:
S1: in multipath parallel transmission system, with Time Delay Estimation Algorithms, the optimal priori for estimating current time is estimated
Evaluation dk；
S2: according to the constraint condition d (i) of data packet order transfer >=d (i-1)-ΔT(i) and in step 1) time delay is estimated
The estimated result of algorithm adjusts the congestion window of each chain road, and the end-to-end time delay for reducing different chain roads is poor, improves different
Multipath parallel transmission effective throughput under structure network environment.
Further, specifically includes the following steps:
1) optimal estimation, the shape of the linear random differential equation are carried out using state of the Kalman filtering algorithm to discrete system
Formula are as follows:
X (k)=AX (k-1)+BU (k-1)+W (k-1), the measured value of system mode can be described as: Z (k)=HX (k)+V
(k), wherein X (k) and X (k-1) are the system mode at k moment and k-1 moment, the control of etching system when U (k-1) is k-1 respectively
Amount；A, B is the parameter of goal systems, if goal systems is Multi-model System, A and B are the forms of matrix；When Z (k) is k
The measured value of etching system state, H is the parameter of measuring system, if more measuring systems, then similarly H be matrix form；W(k-
1) be systematic procedure noise, and V (k) indicate measurement noise, the two is white Gaussian noise, and variance is Q and R respectively；
2) assume that end-to-end propagation delay time is the sum of the component an of constant signal and the variation of a high frequency, the high fdrequency component
It is a white Gaussian noise, then end-to-end time delay can indicate are as follows: X (k)=X (k-1)+W (k-1), Z (k)=X (k)+V (k),
Wherein, the measured value that X (k) and Z (k) respectively indicate the true value of end-to-end time delay and obtained by SACK message measurement；W(k)
The high frequency noise components for representing end-to-end time delay meet the Gaussian Profile that variance is Q, i.e. W (k)~N (0, Q)；V (k) indicates end
To the noise of terminal delay time measured value, meet the Gaussian Profile that variance is R, i.e. V (k)~N (0, R)；
3) update time delay estimated value: X (k | k-1)=X (k-1 | k-1), P (k | k-1)=P (k-1 | k-1)+Q；More new estimation
Error:X (k | k)=X (k | k-1)+Kg (k) (Z (k)-X (k | k-1)),
P (k | k)=(1-Kg (k)) P (k | k-1)；Wherein, X (k | k-1) is the k obtained according to the status predication at k-1 moment
The prior estimate of moment end-to-end time delay, X (k-1 | k-1) are the optimal estimations at k-1 moment；P (k | k-1) is that X (k | k-1) is corresponding
Evaluated error, similar, P (k-1 | k-1) is the corresponding evaluated error of X (k-1 | k-1)；And update the evaluated error stage according to
The time delay estimated value at current time is modified priori estimates and its evaluated error, and the time delay for obtaining current time optimal is estimated
Evaluation, as the foundation estimated next time；Kg is kalman gain (Kalman Gain), and Z (k) is the delay measurements at k moment
(being calculated by the information for including in SACK message)；
4) it takes newest delay measurements as time delay estimation initial value X (0 | 0), and inputs the initial value P (0 | 0) of evaluated error
(can take any nonzero value) obtains the time delay optimal estimation value at current time according to X (k | k-1)=X (k-1 | k-1)；
5) according to the time delay optimal estimation value at current time, time delay coefficient θ is updated,
Wherein, dmaxFor maximum value in the time delay optimal estimation value at current time, dminFor current time time delay optimal estimation
Minimum value in value；
6) two threshold θs are defined0And θmax, threshold θ0It is defined as θ > θ0When, indicate on current different paths delay inequality away from
It is larger, it is possible to the generation that will lead to data disorder phenomenon, due to the transmitting path and used data distribution plan of data packet i
It slightly closes, therefore in order to guarantee to meet data packet order transfer constraint condition, it is desirable that the transmission interval on all paths will expire
Foot: d (i) >=d (i-1)-(T (i)-T (i-1))=d (i-1)-ΔT(i), so that
Threshold θmaxDefinition be, as θ > θmaxWhen, indicate that current path delay variation is very big, data in transmission process
Disorder phenomenon is serious, it is more likely that will lead to receiving end caching obstruction, therefore selects the cache size of receiving end as θmaxIt calculates
Reference standard, the data packet number that can accommodate is N in the caching of note receiving endBuffer,
7) judge θ and θ0And θmaxRelationship:
If θ > θ0, search the maximum path P of current transmission time delay estimated valuei；
If 0 < θ0< θmax, then by PiOn congestion window cwndiReduce are as follows:
If θ > θmax, then by PiOn congestion window cwndiReduce are as follows:
Wherein, cwnd: congestion control window (Congestion control window), it is primary for controlling transmitting terminal
The data packet number that property can at most be sent into network；Transmitting terminal is adaptively adjusted according to the congestion condition of current network
The size of cwnd to carry out congestion control to network；
8) compare the cwnd after reducingiWith the ssthresh SSthresh in the pathiIf cwndi< SSthreshi,
Then enable SSthreshi=cwndi, wherein ssthresh: ssthresh (Slow-start threshold) distinguishes slow turn-on
The critical value in stage and congestion avoidance phase executes slowstart algorithm as cwnd < ssthresh；Conversely, working as cwnd >
When ssthresh, congestion avoidance algorithm is executed.
The beneficial effects of the present invention are: optimization method provided by the invention can accurately carry out path delay of time estimation, and
The time delay estimated is applied to window congestion control, transmitting terminal congestion window size is adaptively adjusted, can effectively reduce
Data disorder phenomenon improves the throughput performance of Transmission system.
Detailed description of the invention
In order to keep the purpose of the present invention, technical scheme and beneficial effects clearer, the present invention provides following attached drawing and carries out
Illustrate:
Fig. 1 is end-to-end time delay estimation schematic diagram of the multipath parallel transmission based on Kalman filtering；
Fig. 2 is network load and throughput concerns figure；
Fig. 3 is the multi-path transmission general scene schematic diagram with two concurrent paths；
Fig. 4 is the best schematic diagram of a scenario of multi-path transmission with two concurrent paths；
Fig. 5 is multipath parallel transmission system effective throughput optimization method flow diagram.
Specific embodiment
Below in conjunction with attached drawing, a preferred embodiment of the present invention will be described in detail.
A kind of optimization method of multipath parallel transmission system effective throughput provided by the invention, this method can be accurate
Path delay of time estimation is carried out, and the time delay estimated is applied to window congestion control, is adaptively adjusted transmitting terminal congestion window
Mouth size, can effectively reduce data disorder phenomenon, improve the throughput performance of Transmission system.
Fig. 1 is end-to-end time delay estimation schematic diagram of the multipath parallel transmission based on Kalman filtering；
Kalman filtering algorithm can carry out optimal estimation, the shape of the linear random differential equation to the state of discrete system
Formula: X (k)=AX (k-1)+BU (k-1)+W (k-1), the measured value of system mode can be described as:
Z (k)=HX (k)+V (k), wherein X (k) and X (k-1) are the system mode at k moment and k-1 moment, U (k- respectively
1) when being k-1 etching system control amount.A, B is the parameter of goal systems, if goal systems is Multi-model System, A and B are
The form of matrix.Z (k) is the measured value of k moment system mode, and H is the parameter of measuring system, if more measuring systems, then together
Reason H is the form of matrix.W (k-1) is the noise of systematic procedure, and V (k) indicates the noise of measurement, and the two is that white Gaussian is made an uproar
Sound, variance are Q and R respectively.
Two stages of Kalman filtering algorithm can be described with following two groups of formula respectively:
Update time delay estimated value:
X (k | k-1)=X (k-1 | k-1)
P (k | k-1)=P (k-1 | k-1)+Q
Update evaluated error:
X (k | k)=X (k | k-1)+Kg (k) (Z (k)-X (k | k-1))
P (k | k)=(1-Kg (k)) P (k | k-1)
Updating the time delay estimated value stage completes according to the time delay optimal estimation value and time delay evaluated error of previous moment to working as
The prior estimate of the end-to-end time delay at preceding moment.Wherein, X (k | k-1) is the k moment obtained according to the status predication at k-1 moment
The prior estimate of end-to-end time delay, X (k-1 | k-1) are the optimal estimations at k-1 moment.P (k | k-1), which is that X (k | k-1) is corresponding, to be estimated
Error is counted, similar, P (k-1 | k-1) is the corresponding evaluated error of X (k-1 | k-1).And the evaluated error stage is updated according to current
The time delay estimated value at moment is modified priori estimates and its evaluated error, obtains the time delay optimal estimation at current time
Value, as the foundation estimated next time.Kg is kalman gain (Kalman Gain), and Z (k) is the delay measurements at k moment
(being calculated by the information for including in SACK message).
In practice, take newest delay measurements as time delay estimation initial value X (0 | 0), and input evaluated error
Initial value P (0 | 0) (any nonzero value can be taken), time delay estimation can automatic cycling operating, complete to end-to-end time delay
Prediction.According to the time delay priori estimates at Formula X (k | k-1)=X (k-1 | k-1) available current time.
Time delay estimation based on Kalman filtering can be divided into two stages constantly recycled: update time delay estimated value
It (predicted state) and updates evaluated error (amendment state).
Fig. 2 is network load and throughput concerns figure, with the rapid development of internet, the type of business transmitted in network
It is more and more diversified, a large amount of appearance of new business especially multimedia service, so that the flow transmitted in network is increasing.
However, the transmittability of link is limited for single transmission link, if lacking necessary congestion control mechanism pair
The data flow of the chain road is controlled, and when link load is lighter, with the increase of the flow of injection, handling capacity still can be gradually
It is linearly increasing；After handling capacity is more than some critical value, increase load again, the throughput of transmissions of the chain road will not continue
Increase, can gradually reduce instead, until generating " deadlock ".In multipath parallel transmission, if a certain concurrent chain road occurs " extremely
Lock ", then the end-to-end time delay of link will increased dramatically, and cause to delay to reach by the data packet that the link is sent to receive
Serious data packet disorder is caused in receiving end in end, so that the effective throughput of multipath parallel transmission sharply deteriorates, therefore
It is had in multipath parallel transmission agreement comprising corresponding congestion control mechanism, with the throughput performance of guarantee agreement.
Fig. 3 is the multi-path transmission general scene schematic diagram with two concurrent paths, and SCTP-CMT agreement passes through will not
Resource consolidation and unified management with network promote the purpose of QoS of survice to reach the available bandwidth of polymerization heterogeneous networks.
However, the agreement is not suitable for heterogeneous wireless network environment.In heterogeneous network, since the asymmetry between path causes not
It is huge with the end-to-end propagation delay time difference on path, when the data packet of parallel transmission on different paths will be unable to according to sending
Sequence reach receiving end in an orderly manner, so as to cause data packet disorder phenomenon.Out-of-order data packet will be trapped in receiving end caching
In, after waiting the lesser data packet of sequence number to be transmitted to all arrive at, it could be submitted together to upper layer application, seriously constrain multichannel
The raising of the diameter parallel transmission effectively amount of gulping down.Particularly, for real-time video meeting etc. to the business that the data has high real-time requirements,
Even if final data packet can successfully be submitted to upper layer application, the data in data packet may also have been over term of validity and
It is abandoned by application, this deteriorates effective throughput performance of multipath parallel transmission system further.
The present invention is by the effective throughput of multipath parallel transmission is defined as: receiving end is properly received and presses in the unit time
Sequence is submitted to the data packet number of application layer.When the maximum delay difference on each parallel route of multipath parallel transmission system is enough
Hour, it can ideally not influenced completely by random ordering.Therefore, can by each item in multipath parallel transmission system simultaneously
Maximum delay difference on line link is regulated and controled, it is made to meet d (i) >=d (i-1)-ΔT(i), data packet unrest occurs to reduce
A possibility that sequence, improves the effective throughput of system.Parameter declaration is as follows:
τi--- the data packet on each path of transmitting terminal sends interval, and wherein i is path number (i=1,2).
di--- the propagation delay (i.e. data packet from leave transmitting terminal taken time to receiving end is arrived at) on the i of path, and d1
< d2。
The propagation delay in Δ d --- path 1 and path 2 is poor, Δ d=| d2-d1|。
T --- receive a complete orderly data block the time it takes.
S --- the size of the ordered data block received in total in time T.
G --- the effective throughput of multipath parallel transmission system,
Assuming that continuous data packet of a total of 4 transmission sequence number TSN etc. is to be sent, the TSN of data packet is respectively 1,2,3
It is a sequentially unit referred to here as this 4 data packets with 4.Wherein there are 3 data coatings to be assigned on path 1 to be transmitted, remain
Under a data packet passage path 2 transmission.After transmission starts, data packet 1 and data packet 2 are sent by path 1 and path 2 respectively
It goes out.
Δ d > τ1When, when the data packet by sending on the biggish path 2 of propagation delay reaches receiving end, receiving end
A complete ordered data block can just be received and consign to application layer upwards, receive what the ordered data block was spent at this time
Time T=Δ d, therefore
Fig. 4 is the best schematic diagram of a scenario of multi-path transmission with two concurrent paths；
Assuming that continuous data packet of a total of 4 transmission sequence number TSN etc. is to be sent, the TSN of data packet is respectively 1,2,3
It is a sequentially unit referred to here as this 4 data packets with 4.Wherein there are 3 data coatings to be assigned on path 1 to be transmitted, remain
Under a data packet passage path 2 transmission.After transmission starts, data packet 1 and data packet 2 are sent by path 1 and path 2 respectively
It goes out.
Δd≤τ1When, it is sent since SCTP regulation receiving end is completely received after an ordered data block to transmitting terminal
SACK response message, transmitting terminal receives the data transmission of beginning next round after the message, therefore receiving end receives in this case
Approximatively interval can be sent with the data packet on path 2 to a complete ordered data block the time it takes to replace, therefore
Compared by above-mentioned analysis and learnt, when the transmission time delay difference of two paths is bigger, it is continuous that receiving end receives TSN
Data block and the time spent in consigning to application layer is longer, effective throughput is with regard to smaller.Therefore, we can be by sending
Data package transmission velocity is adjusted in end, the load on balanced difference path, when regulating and controlling the maximum between concurrent path indirectly
Prolong difference, mitigate data packet disorder, improves the handling capacity of multipath parallel transmission system.
Fig. 5 is multipath parallel transmission system effective throughput optimization method flow diagram, the time delay of Kalman filtering
Estimation accurately estimates the end-to-end time delay of each concurrent path with the principle of Kalman filtering, estimates compared to tradition SRTT
Algorithm has higher accuracy, is suitable for the changeable feature of chain-circuit time delay in heterogeneous wireless network, avoids because time delay is estimated
The inaccuracy of meter influences the performance of subsequent optimization algorithm.Jamming control method based on delay inequality control is according to current each concurrent chain
Maximum delay between road is poor, is adaptively adjusted the congestion window of link, realizes the load balancing between link, and it is maximum to reach reduction
The purpose of delay inequality.It is in transmitting terminal that the maximum delay difference regulation between link is reasonable at one by combining two above step
In the range of, to reduce data random ordering, improve the effective throughput of multipath parallel transmission system.
Finally, it is stated that preferred embodiment above is only used to illustrate the technical scheme of the present invention and not to limit it, although logical
It crosses above preferred embodiment the present invention is described in detail, however, those skilled in the art should understand that, can be
Various changes are made to it in form and in details, without departing from claims of the present invention limited range.
Claims (1)
1. a kind of optimization method of multipath parallel transmission system effective throughput, it is characterised in that: in the method, multipath
There are the mutually independent transmission path of N item, respectively L between transmitting terminal and receiving end for parallel transmission system1,L2,...,
Li,...LN, N >=2；On each paths end to end propagated time delay meet d (1)≤d (2)≤... d (i) ...≤d (N)；
Method includes the following steps:
S1: in multipath parallel transmission system, with Time Delay Estimation Algorithms, the optimal estimation value at current time is estimated；
S2: according to the constraint condition d (i) of data packet order transfer >=d (i-1)-ΔT(i) and in step S1 it estimates current
The optimal estimation value at moment, adjusts the congestion window on each paths, and the end-to-end time delay reduced on different paths is poor, improves
Multipath parallel transmission effective throughput under heterogeneous network environment；Wherein d (i) indicates the propagation delay on the i of path, ΔT(i)
Indicate that the data packet on the i of transmitting terminal path sends interval；
The step S1 specifically includes the following steps:
S11: optimal estimation, the form of the linear random differential equation are carried out using state of the Kalman filtering algorithm to discrete system
Are as follows: X (k)=AX (k-1)+BU (k-1)+W (k-1), the measured value description of system mode are as follows: Z (k)=HX (k)+V (k), wherein
X (k) and X (k-1) are the true value at k moment and k-1 moment end-to-end time delay, the control of etching system when U (k-1) is k-1 respectively
Amount；A, B is the parameter of goal systems, if goal systems is Multi-model System, A and B are the forms of matrix；When Z (k) is k
The delay measurements of etching system state, H is the parameter of measuring system, if more measuring systems, then similarly H be matrix form；W
(k-1) be end-to-end time delay high frequency noise components, and V (k) indicate end-to-end time delay measured value noise, the two is Gauss
White noise, variance are Q and R respectively；
S12: assuming that end-to-end propagation delay time is the sum of the component an of constant signal and the variation of a high frequency, high frequency variation
Component is a white Gaussian noise, then end-to-end time delay indicates are as follows: X (k)=X (k-1)+W (k-1), Z (k)=X (k)+V (k),
Wherein, X (k) and Z (k) respectively indicate the true value of k moment end-to-end time delay and the delay measurements of system mode；W (k) is represented
The high frequency noise components of end-to-end time delay meet the Gaussian Profile that variance is Q, i.e. W (k)~N (0, Q)；V (k) indicates end-to-end
The noise of delay measurements meets the Gaussian Profile that variance is R, i.e. V (k)~N (0, R)；
S13: the time delay estimated value at current time is updated: X (k | k-1)=X (k-1 | k-1), P (k | k-1)=P (k-1 | k-1)+Q；
Update evaluated error:X (k | k)=X (k | k-1)+Kg (k) (Z (k)-X (k | k-1)), P (k |
K)=(1-Kg (k)) P (k | k-1)；Wherein, X (k | k-1) be k moment for being obtained according to the status predication at k-1 moment it is end-to-end when
The prior estimate prolonged, X (k-1 | k-1) are the optimal estimations at k-1 moment；P (k | k-1) is the corresponding evaluated error of X (k | k-1),
Similar, P (k-1 | k-1) is the corresponding evaluated error of X (k-1 | k-1)；And the evaluated error stage is updated according to current time
Time delay estimated value is modified priori estimates and its evaluated error, obtains the time delay optimal estimation value at current time, as
The foundation estimated next time；Kg is kalman gain (Kalman Gain), and Z (k) is the delay measurements of k moment system mode,
It is calculated by the information for including in SACK (Selective ACK) message；
S14: taking newest delay measurements as time delay estimation initial value X (0 | 0), and inputs the initial value P (0 | 0) of evaluated error,
Any nonzero value is taken, the time delay optimal estimation value at current time is obtained according to X (k | k-1)=X (k-1 | k-1)；
The step S2 specifically includes the following steps:
S21: according to the time delay optimal estimation value at current time, updating time delay coefficient θ,
Wherein, dmaxFor maximum value in the time delay optimal estimation value at current time, dminFor the time delay optimal estimation value at current time
Middle minimum value；
S22: two threshold θs are defined0And θmax, threshold θ0Is defined as: as θ > θ0When, indicate on current different paths delay inequality away from
It is larger, it is possible to the generation that will lead to data disorder phenomenon, due to the transmitting path and used data distribution plan of data packet i
It slightly closes, therefore in order to guarantee to meet data packet order transfer constraint condition, it is desirable that the transmission interval on all paths will expire
Foot: d (i) >=d (i-1)-(T (i)-T (i-1))=d (i-1)-ΔT(i), so that
Wherein, T (i) indicates that path i receives a complete orderly data block the time it takes, ΔT(i) transmitting terminal is indicated
Data packet on the i of path sends interval, min (ΔT) indicate that the smallest data packet sends interval in all paths；Threshold θmax's
Definition is, as θ > θmaxWhen, indicate that current path delay variation is very big, data disorder phenomenon is serious in transmission process, has very much
It may result in receiving end caching obstruction, therefore select the cache size of receiving end as θmaxThe reference standard of calculating, note receive
The data packet number that can be accommodated in the caching of end is NBuffer,
S23: judge θ and θ0And θmaxRelationship:
If θ > θ0, search the maximum path P of current transmission time delay estimated valuei；
If 0 < θ0< θmax, then by PiOn congestion window cwndiReduce are as follows:
If θ > θmax, then by PiOn congestion window cwndiReduce are as follows:
Wherein, cwnd: congestion control window (Congestion control window), for controlling transmitting terminal disposably most
The data packet number that can be mostly sent into network；Transmitting terminal is adaptively adjusted cwnd according to the congestion condition of current network
Size come to network carry out congestion control；
S24: compare the cwnd after reducingiWith the ssthresh SSthresh in the pathiIf cwndi< SSthreshi, then
Enable SSthreshi=cwndi, wherein ssthresh: ssthresh (Slow-start threshold) distinguishes slow turn-on rank
The critical value of section and congestion avoidance phase executes slow start stage as cwnd < ssthresh；Conversely, working as cwnd >
When ssthresh, congestion avoidance phase is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610178755.XA CN105873096B (en) | 2016-03-24 | 2016-03-24 | A Method for Optimizing Effective Throughput of Multipath Parallel Transmission System |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610178755.XA CN105873096B (en) | 2016-03-24 | 2016-03-24 | A Method for Optimizing Effective Throughput of Multipath Parallel Transmission System |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105873096A CN105873096A (en) | 2016-08-17 |
| CN105873096B true CN105873096B (en) | 2019-05-10 |
Family
ID=56625958
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610178755.XA Active CN105873096B (en) | 2016-03-24 | 2016-03-24 | A Method for Optimizing Effective Throughput of Multipath Parallel Transmission System |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105873096B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110808859A (en) * | 2019-11-01 | 2020-02-18 | 海南大学 | A regression model calculation method and system for buffer consumption of MPTCP multi-path transmission |
Families Citing this family (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106878192B (en) * | 2016-12-26 | 2020-02-14 | 北京邮电大学 | Data scheduling method of self-adaptive MPTCP |
| CN108337160A (en) * | 2017-01-20 | 2018-07-27 | 华为技术有限公司 | Data transmission method, destination node and source node |
| CN108512708B (en) * | 2017-02-24 | 2023-01-06 | 中兴通讯股份有限公司 | Cache calculation method and device |
| CN108924872B (en) * | 2017-04-13 | 2022-03-18 | 中兴通讯股份有限公司 | Data transmission method, terminal and core network equipment |
| CN107026716B (en) * | 2017-05-12 | 2019-07-26 | 中南大学 | A transmission control method based on concurrency awareness in data center network |
| CN107733811B (en) * | 2017-09-18 | 2020-02-21 | 中国电子科技集团公司第五十四研究所 | A Method of Dynamically Adjusting Encoding Strength |
| CN109672701B (en) * | 2017-10-16 | 2020-12-11 | 中国科学院信息工程研究所 | A differential TCP link management method and device |
| CN109729007B (en) | 2017-10-28 | 2022-07-22 | 华为技术有限公司 | Method, device and equipment for transmitting data |
| CN107809662B (en) * | 2017-11-06 | 2019-08-16 | 陕西师范大学 | A kind of scalable video transmission method and device based on heterogeneous wireless ad hoc network |
| CN107682235A (en) * | 2017-11-24 | 2018-02-09 | 重庆邮电大学 | A kind of propagation delay time prediction data dispatching method based on Kalman filtering |
| CN109120534B (en) * | 2018-08-23 | 2020-07-28 | 北京理工大学 | Method for solving stream deviation of maximum concurrent stream of multiple commodity streams |
| CN110113142A (en) * | 2019-04-28 | 2019-08-09 | 天通畅达(深圳)科技有限公司 | The transmission method and system of carrying large-data operation are concurrently bundled based on multiple data channel |
| CN110635988B (en) * | 2019-10-24 | 2020-10-27 | 北京大学 | A data forwarding method and device for multi-path transmission |
| CN110996279B (en) * | 2019-12-16 | 2022-04-26 | 东南大学 | Multi-channel parallel transmission data scheduling method and system applied to unmanned aerial vehicle communication |
| CN111464437B (en) * | 2020-05-11 | 2021-10-29 | 南京邮电大学 | A forward-delay-based multipath transmission path optimization method in vehicular heterogeneous networks |
| CN113746745B (en) * | 2020-05-30 | 2025-02-25 | 华为技术有限公司 | Multipath communication method, device and system |
| CN112019443B (en) * | 2020-09-02 | 2023-09-12 | 首都师范大学 | Multipath data transmission method and device |
| CN113347096B (en) * | 2021-08-06 | 2021-11-09 | 湖南三湘银行股份有限公司 | Transmission processing method of third party payment data |
| CN114710807B (en) * | 2022-04-28 | 2024-06-07 | 云南师范大学 | A Kalman filter dynamic congestion control method for 5G millimeter wave communication |
| CN115134304B (en) * | 2022-06-27 | 2023-10-03 | 长沙理工大学 | Adaptive load balancing method to avoid data packet disorder in cloud computing data center |
| CN115484098A (en) * | 2022-09-15 | 2022-12-16 | 中国银行股份有限公司 | Login authentication processing method, system, storage medium and electronic equipment |
| CN116155830A (en) * | 2023-03-03 | 2023-05-23 | 广东电网有限责任公司 | Data packet multipath transmission scheduling method, system, terminal and medium |
| CN116600381B (en) * | 2023-07-14 | 2023-09-19 | 武汉能钠智能装备技术股份有限公司四川省成都市分公司 | Multi-channel parallel generation system and method for broadcast signals |
| CN120034449B (en) * | 2025-04-24 | 2025-06-20 | 中国民航大学 | Network delay prediction model training method, device and network status detection method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014127790A (en) * | 2012-12-26 | 2014-07-07 | Univ Of Electro-Communications | Information processing device, method and program |
| CN104954206A (en) * | 2015-07-31 | 2015-09-30 | 重庆邮电大学 | Disorder analyzing method for multipath parallel transmission system |
-
2016
- 2016-03-24 CN CN201610178755.XA patent/CN105873096B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014127790A (en) * | 2012-12-26 | 2014-07-07 | Univ Of Electro-Communications | Information processing device, method and program |
| CN104954206A (en) * | 2015-07-31 | 2015-09-30 | 重庆邮电大学 | Disorder analyzing method for multipath parallel transmission system |
Non-Patent Citations (2)
| Title |
|---|
| CMT-QA: Quality-Aware Adaptive Concurrent Multipath Data Transfer in Heterogeneous Wireless Networks;Changqiao Xu et al.;《IEEE TRANSACTIONS ON MOBILE COMPUTING》;20131130;第12卷(第11期);全文 |
| 基于时延差控制的MPTCP有效吞吐量提高方法;陶洋等;《计算机应用研究》;20150930;第32卷(第9期);全文 |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110808859A (en) * | 2019-11-01 | 2020-02-18 | 海南大学 | A regression model calculation method and system for buffer consumption of MPTCP multi-path transmission |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105873096A (en) | 2016-08-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105873096B (en) | A Method for Optimizing Effective Throughput of Multipath Parallel Transmission System | |
| CN107171842B (en) | Multipath transmission protocol congestion control method based on reinforcement learning | |
| CN115208828B (en) | Rate-optimized congestion management | |
| CN110505099B (en) | Service function chain deployment method based on migration A-C learning | |
| CN107624232B (en) | Apparatus and method for controlling downlink throughput in a communication system | |
| US7778265B2 (en) | Method and apparatus for local adaptive provisioning at a node | |
| Varma | Internet congestion control | |
| WO2018225039A1 (en) | Method for congestion control in a network | |
| CN113271256B (en) | Information age multi-path transmission method and system | |
| CN113595684B (en) | MPTCP dynamic coding scheduling method and system based on deep reinforcement learning | |
| Liao et al. | A distributed end-to-end overload control mechanism for networks of SIP servers | |
| CN104486163A (en) | Single end available bandwidth measurement method and terminal adopting same | |
| CN104935524B (en) | The soft load-balancing method that a kind of multipath network is controlled based on chain-circuit time delay | |
| CN107070802A (en) | Congestion Control Technology for Wireless Sensor Networks Based on PID Controller | |
| CN104618258A (en) | Control method of data transmission rate | |
| CN109698925B (en) | Data-driven real-time video congestion control method and device | |
| KR20200019727A (en) | Optimization of network parameters to enable network coding | |
| Pokhrel et al. | Fair and efficient distributed edge learning with hybrid multipath TCP | |
| Jung et al. | Reveno: Rtt estimation based multipath tcp in 5g multi-rat networks | |
| CN108848415A (en) | A kind of the network real-time video transmission method and device of adaptive learning | |
| Ruan et al. | PTCP: A priority-based transport control protocol for timeout mitigation in commodity data center | |
| Talau et al. | Improving TCP performance over a common IoT scenario using the Early Window Tailoring method | |
| Patel | Comparative analysis of cumulative distribution function for TCP congestion window size and triple‐duplicate period | |
| CN109787856A (en) | A kind of HAS bandwidth prediction method based on LTE network link state | |
| Noroozi et al. | Age-of-information in tandem queues with delayed feedback: zero-wait vs. pipelining |
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 |