CN102404208B - Stable network congestion control method with high-efficiency - Google Patents
Stable network congestion control method with high-efficiency Download PDFInfo
- Publication number
- CN102404208B CN102404208B CN201110358192.XA CN201110358192A CN102404208B CN 102404208 B CN102404208 B CN 102404208B CN 201110358192 A CN201110358192 A CN 201110358192A CN 102404208 B CN102404208 B CN 102404208B
- Authority
- CN
- China
- Prior art keywords
- value
- feedback
- packet
- bandwidth
- congestion
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a stable network congestion control method with high-efficiency, which belongs to the technical field of network, comprising the following steps: 1) a transmitting terminal A arranges a congestion expansion head in a data packet to be transmitted; 2) a router updates the length of the next control space according to the end-to-end delay di flowing through the data packet; 3) the router calculates the total congestion feedback value Phi K+1 of the next control space, and transmits the Phi K+1 to the data packet of each traffic flow in the next control space; 4) the router updates the bandwidth incremental which flows the data packet and waits for determining as Deltai, j+1 according to the average window space Xi of the received data packet; 5) a receiving terminal B accumulates the received bandwidth incremental of the data packet transmitted by the transmitting terminal A and waits for determining, and adds the bandwidth incremental to the feedback value of the bandwidth incremental of the feedback data packet; 6) the transmitting terminal A adjusts the congestion window value W according to the feedback value of the bandwidth incremental in the received data packet. The invention has fast response speed of network load change and improves the resource utilization rate.
Description
Technical field
The present invention relates to the Research of Congestion Control Techniques of packet switching network, be specifically related to a kind of internet congestion control method based on explicit congestion information.
Background technology
At present in the Internet the jam control function of router a little less than, only provide network congestion information by packet loss signal to subscriber terminal equipment, cause network bandwidth resources utilance low, queueing delay and end to end network delay are all larger.Explicit congestion control protocol (XCP eXplicit Control Protocol) is a kind of network resource usage efficiency that strengthens, and reduces the congestion avoidance algorithm of network packet loss rate.It increases congestion control information in packet; Adopt and first measure the strategy distributing afterwards, router is first measured the idling-resource quantity of its output interface, then these resources is distributed to each session of flowing through, and transmitting terminal, again according to the instruction of network-feedback, is adjusted its transmission speed.Rate Control Protocol RCP(Rate Control Protocol) similar to XCP, but RCP adopted and first distributed the strategy of adjusting afterwards, and router is first measured and is estimated current fair bandwidth, and distributes to all sessions of flowing through.Like this, in the time having a large amount of new sessions to arrive, RCP system just there will be the state of extreme congestion.
Emulation experiment shows, XCP, at the larger net environment of end-to-end transmission delay difference, there will be unsettled situation.Although be stable from the stream mode of theory analysis XCP, but in practical application, need stream mode discretization, need to introduce the control interval and measure and estimate state and the parameters such as its average end-to-end delay, and the length in its control interval is also set as average end-to-end delay.The delay that XCP realizes system is the twice of its stream mode, cannot obtain theoretical guarantee so XCP realizes the stability of system.
Summary of the invention
In view of this, the present invention is devoted to provide a kind of internet congestion control method.
Above-mentioned purpose of the present invention is achieved by the following technical solutions:
A kind of internet congestion control method, comprising:
A1) in packet, increase congested extension header, comprise end-to-end delay, bandwidth increments to be confirmed, bandwidth increments value of feedback and average window pitch that transmitting terminal is estimated, wherein average window pitch is that end-to-end delay is divided by current congestion window value;
B1) the average end-to-end delay that router flows by session on its output interface of Estimation and Measurement, and upgrade the length in its control interval, make its control interval much smaller than average end-to-end delay;
C1) router is within the control interval, measure the congestion state of its output interface, calculate total congestion feedback value, and this total congestion feedback value is distributed in the packet of each session stream in the next control interval, upgrade the bandwidth increments to be confirmed of the packet of flowing through;
D1) receiving terminal, cumulative the bandwidth increments to be confirmed of receiving packet, in the time having message to return to transmitting terminal, copies the bandwidth increments value of feedback of this accumulated value to return data bag;
E1) the bandwidth increments value of feedback that transmitting terminal is received according to it, adjusts the size of its congestion window value W;
F1) transmitting terminal is also controlled the transmission interval of packet, makes it be not less than average packet interval, and wherein average packet interval equals end-to-end delay and be multiplied by the length of packet, then divided by current congestion window value.
Further comprise after described step f1: router is taking each control interval as border, measure the average end-to-end delay of all sessions streams on the average arrival rate of its output interface and minimum queue length and this interface, upgrade and control and allocation of parameters.
Router utilizes the bandwidth of its output interface to deduct average arrival rate and obtains average idle bandwidth, then is multiplied by idle bandwidth feedback factor, just obtains idle bandwidth value of feedback.Router utilizes queue length feedback factor to be multiplied by minimum queue length, just obtains queue length value of feedback.
To in this control interval, do not divide the total congestion feedback value sending out, part is transferred in the next control interval and gone.Router idle bandwidth value of feedback deducts queue length value of feedback, adds this control interval and does not divide the part sending out total congestion feedback value, just obtains total congestion feedback value in next control interval.
Idle bandwidth feedback factor and queue length feedback factor that this method is used can ensure the stability of this congestion control system, and to make the transient process of this congestion control system be critical damping response.
In the next control interval, router is assigned to total congestion feedback value liberally in each session by this output interface and goes.In the time that router receives that new session is set up message, the in the situation that of its total congestion feedback value abundance, point balanced session bandwidth is given newly arrived session.In the time that router is received session close message, the shared bandwidth resources of this session will be reclaimed immediately.
The present invention has following technique effect.
1, the control interval of native system is less than XCP system.Like this, the delay of this control system is less, and the response that offered load is changed is faster, and the utilance of Internet resources is also higher.
2, in native system, data source is in the time sending packet, and the transmission interval of packet is more even.Like this, the queue length of router is shorter, and delay is end to end less, and the loss of packet is also less.
3, in native system, selection of control parameter is more reasonable, has ensured the stability of system.Even in the situation that end-to-end delay differs greatly, also still performance is stable for system.
4,, in native system, router divides balanced session bandwidth to newly arrived session.So just for the short stream business such as WEB browses provide good service quality.
5, in native system, router is received session close message, just reclaims immediately the shared bandwidth resources of this session.Idle link circuit resource can be used as soon as possible like this.
6, in native system, the previous control interval is not divided the total congestion feedback value sending out, and part is transferred in the next control interval and gone.So just reduce the error of total congestion feedback value at assigning process.
Brief description of the drawings
Fig. 1 is congested prolate-headed cut-away view in the present invention;
Fig. 2 is the operating mechanism figure of the present invention in IPv4 network environment;
The calculating of the total congestion feedback value of Fig. 3;
The distribution of the total congestion feedback value of Fig. 4.
Embodiment
The present invention, taking IPv4 network environment as basis, realizes the congestion control system of stability and high efficiency.N/mono-that in this realization, control interval value is average end-to-end delay, wherein N is 20.Below in conjunction with the drawings and specific embodiments, the present invention is described in detail.
Protocol extension
In order to meet the requirement of congestion control, need to expand existing IPv4 standard, in IP packet header, increase a congested extension header, as shown in Figure 1.This congested extension header defines in the draft draft-falk-xcp-spec-03 of IETF.The present invention follows the regulation of this draft, and does suitable expansion.Below provide this prolate-headed form.
First field is the type in next packet header.
Second field is extension header length, and this value is set as 20 bytes, is encoded to 0x14.
The 3rd field is this prolate-headed version, and the present invention is set as 4.
The 4th field is this prolate-headed form, can get 3 values: 1,2 and 3.
1 represents that three the average window pitch of field, end-to-end delay and bandwidth increments to be confirmed are thereafter all effective;
2 represent that three the average window pitch of field, end-to-end delay and bandwidth increments to be confirmed are thereafter all invalid;
3 represent that two average window pitch of field, end-to-end delay are thereafter invalid.
The 5th field is reserved field, do not use temporarily.
The 6th field is average window pitch (X), and value is nonnegative number, and 28 bits represent fractional part thereafter, and front 4 bits represent integer.
The 7th field is end-to-end delay (d), and value is nonnegative number, and 28 bits represent fractional part thereafter, and front 4 bits represent integer.
The 8th field is bandwidth increments value of feedback (Δ), and value is integer, and unit is bps.This value can just can be born, for timing represents the throughput that permission increases; Represent the throughput that must reduce when negative.
The 9th field is bandwidth increments to be confirmed (δ), and value is integer, and unit is bps.This value can just can be born, for timing represents the throughput that permission increases; Represent the throughput that must reduce when negative.
The operating mechanism of this congestion control system as shown in Figure 2.
Each packet that transmitting terminal A sends has comprised congested extension header.Congested prolate-headed end-to-end delay is set to the end-to-end delay d that transmitting terminal estimation obtains
i(transmitting terminal adopts the end-to-end delay evaluation method in existing TCP host-host protocol.In Transmission Control Protocol, receiving terminal is received after packet, can send acknowledge message to transmitting terminal.Like this, receiving terminal, according to sending the time of packet and receiving the confirmation the interval between message, just can estimate the end-to-end delay of transmitting terminal to receiving terminal); Average window pitch X
ibe set to end-to-end delay d
idivided by current congestion window value W; Bandwidth increments to be confirmed is set as sender and expects the bandwidth increments δ obtaining
i, 0; The aggregate-value of the bandwidth increments to be confirmed from receiving terminal B that bandwidth increments value of feedback is set as receiving.Bandwidth increments value of feedback be sometime section in aggregate-value, use to receiving terminal B.In actual network application, a common session is two-way, and existing A, to the data flow of B, also has the data flow of B to A.
Packet arrives after router R1, and router R1 collects the end-to-end delay of each packet of flowing through, and obtains the average end-to-end delay D in this control interval
k, 1, and determine that the length in next control interval is D
k, 1/ N, N value is 20 herein; According to the average window pitch information of packet (average window pitch information is arranged in data packet head by transmitting terminal, router read this information just can), upgrading bandwidth increments to be confirmed is δ
i, 1(δ
i, 1<=δ
i, 0); According to its output interface bandwidth resources situation (comprising average idle bandwidth and queue length), calculate total congestion feedback value in next control interval; Collect the average window pitch information of each packet, calculate the feedback parameter in next control interval.
Packet arrives after router R2, and router R2 does the processing similar to R1, and the congested prolate-headed bandwidth increments to be confirmed of packet is updated to δ
i, 2(δ
i, 2<=δ
i, 1).
Packet arrives after receiving terminal B, and receiving terminal B is the bandwidth increments δ to be confirmed in congested packet extension header
i,
2be added in variable z.Until receiving terminal B, during to transmitting terminal A return data bag, bandwidth increments value of feedback Δ is just set to z, and variable z is reinitialized to 0.
Transmitting terminal A just upgrades its transmission speed after receiving that bandwidth increments value of feedback is the packet of Δ, thereby completes the closed-loop control process of congestion control system.
The computational methods of average end-to-end delay D are as follows:
Router has forwarded M packet within the control interval, the end-to-end delay d of each packet
i, average end-to-end delay D is calculated as follows:
D＝∑d
i/M
The computational methods of average session bandwidth v are as follows:
The length of control interval k is η
k, receive that a length is L
i, average window pitch is X
ipacket, it is X to the contribution of number of sessions
i* L
i/ η
k.In the time that the control interval, k finished, the number of sessions Y of estimation
kfor
Y
k＝∑(X
i*L
i)/η
k
If the average arrival rate in this control interval is u
k, can calculate its average session bandwidth:
v
k=u
k*η
k/∑(X
i*L
i)
The calculating of total congestion feedback value
As shown in Figure 3, the length of control interval k is η
k, the link bandwidth of output interface is C, measuring output interface average arrival rate is u
k, can calculate output interface idle bandwidth value of feedback for (0.4/ (1+1/N)) * η
k* (C-u
k)/N.In control interval k, the minimum queue length of its output interface is Q
k, its queue length value of feedback is-1.414* (0.4/ (1+1/N))
2* Q
k/ N.If remaining total congestion feedback value is r_ Φ when the control interval, k finished
k, the transfer congestion feedback value in so next control interval is 0.95*r_ Φ
k.Total congestion feedback value in next control interval is Φ
k+1can be calculated by following formula, wherein N value is 20:
Φ
k＋1＝0.95*r_Φ
k+0.019*η
k*(C-u
k)–0.01*Q
k
The decomposition of total congestion feedback value
In the time that the control interval, k finished, also need total congestion feedback value in next control interval to be divided into positive feedback p_ Φ
k+1with negative feedback n_ Φ
k+1
When | Φ
k+1| > γ * u
ktime, setting flow heavy distribution factor γ is 0.1, get floating resources 10% for the data flow to poor throughput the data flow resource transfers of high-throughput,
p_Φ
k＋1＝(|Φ
k＋1|+Φ
k＋1)/2
n_Φ
k＋1＝(|Φ
k＋1|-Φ
k＋1)/2
Otherwise,
p_Φ
k＋1＝γ*u
k＋1＋(|Φ
k＋1|+Φ
k＋1)/2
n_Φ
k＋1＝－γ*u
k＋1＋(|Φ
k＋1|-Φ
k＋1)/2
The calculating of feedback parameter:
The distribution principle of feedback is: positive feedback is averagely allocated to each session stream; Negative feedback is pro rata distributed to each session stream, and the negative feedback that the session flow point that throughput is larger obtains is also more.The feedback parameter p of positive feedback
k+1for:
p
k＋1＝p_Φ
k＋1/Y
k
Degenerative feedback parameter n
k+1for:
n
k＋1＝n_Φ
k＋1/u
k
The distribution of total congestion feedback value
As shown in Figure 4, at control interval k+1, receive input packet pkt
itime, remaining positive feedback value is r_p_ Φ
k+1, the positive feedback value of distributing to this packet is p_pkt
i:
p_pkt
i=p
k＋1*(X
i*L
i/η
k+1)
Remaining negative feedback value is r_n_ Φ
k+1, the negative feedback value of distributing to this packet is n_pkt
i:
n_pkt
i＝n
k＋1*X
i/(X
i*L
i/η
k+1)
n_pkt
i＝n
k＋1*η
k+1/L
i
The congestion feedback value of distributing to this packet is t_pkt
i:
t_pkt
i＝min(p_pkt
i,r_p_Φ
k＋1)–min(n_pkti,r_n_Φ
k＋1)
r_p_Φ
k＋1-=min(p_pkt
i,r_p_Φ
k＋1)
r_n_Φ
k＋1-=min(n_pkt
i,r_n_Φ
k＋1)
In the time that the control interval, k+1 finished, its remaining total congestion feedback value r_ Φ
k+1for
r_Φ
k＋1＝r_p_Φ
k＋1－r_n_Φ
k＋1
Session establishment process
At first packet of session establishment, transmitting terminal cannot be known the end-to-end delay that arrives receiving terminal B, so the congested prolate-headed end-to-end delay of this packet is set as 0, average window spacing setting is 0, and format fields is set as 3.
Router R1 receives the initial data packets pkt of session establishment
iafter, at its positive feedback p_ Φ
1in the situation of resource abundance, point balanced session bandwidth v
1give this session, and reduce the surplus resources of positive feedback,
δ
1=min(δ
i,0,v
1,p_Φ
1)
p_Φ
1-=δ
i,1
Router R2 receives the initial data packets pkt of session establishment
iafter, at its positive feedback p_ Φ
2in the situation of resource abundance, point balanced session bandwidth v
2give this session, and reduce the surplus resources of positive feedback,
δ
i,2=min(δ
i,1,v
2,p_Φ
2)
p_Φ
2-=δ
i,2
Receiving terminal D receives after this session establishment request, if receive this session, just returns to immediately acknowledge message, and the bandwidth increments value of feedback that this packet is set is δ
i, 2; If refuse this session, just return to immediately refuse information, and the bandwidth increments value of feedback that this packet is set is 0.
Session dispose procedure
When router R1 receives the message pkt of conversation end
itime, router is regained the bandwidth resources that this session takies, and increases the surplus resources of positive feedback.
p_Φ
1+=0.95/X
i
X
ifor the congested prolate-headed average window pitch of this packet.
When router R2 receives the message pkt of conversation end
itime, router is regained the bandwidth resources that this session takies, and increases the surplus resources of positive feedback.
p_Φ
2+=0.95/X
i
X
ifor the congested prolate-headed average window pitch of this packet.
Claims (9)
1. an internet congestion control method, the steps include:
1) transmitting terminal A increases a congested extension header in packet to be sent; Described congested extension header comprises: the end-to-end delay d that transmitting terminal A estimates
i, bandwidth increments δ to be confirmed
i, 0, bandwidth increments value of feedback Δ
iwith average window pitch X
i; Wherein, bandwidth increments value of feedback Δ
ibe initialized as 0, average window pitch is that end-to-end delay is divided by current congestion window value;
2) router is according to the end-to-end delay d of the packet of flowing through
i, calculate the average end-to-end delay D that in the current control interval, on output interface, session is flowed
k, and according to current D
kvalue is upgraded the length in its next control interval;
3) router calculates total congestion feedback value Φ in next control interval according to the congestion state of output interface in the current control interval
k+1, and by Φ
k+1be distributed in the packet of each session stream in next control interval,
4) router is according to the average window pitch X that receives packet
i, the bandwidth increments to be confirmed of upgrading the packet of flowing through is δ
i, j+1; Wherein, δ
i, j+1≤ δ
i,j, δ
i,jthe bandwidth increments of confirming after j router for session data bag i;
5) that the transmitting terminal A receiving is sent in packet to bandwidth increments to be confirmed is cumulative for receiving terminal B, in the time that receiving terminal B has message to return to transmitting terminal A, this bandwidth increments accumulated value to be confirmed is added in the bandwidth increments value of feedback of return data bag;
6) transmitting terminal A receives the bandwidth increments value of feedback in packet according to it, adjusts its congestion window value W;
Wherein, calculate described total congestion feedback value Φ
k+1method be: described router utilizes the bandwidth C of its output interface to deduct packet average arrival rate u in the current control interval
kobtain average idle bandwidth, then be multiplied by idle bandwidth feedback factor, obtain idle bandwidth value of feedback; And described router utilizes the minimum queue length Q of output interface in the current control interval
kbe multiplied by minimum queue length feedback factor, obtain queue length value of feedback; Then described router utilizes idle bandwidth value of feedback to deduct queue length value of feedback, adds current control interval residue and does not divide the total congestion feedback value sending out, and obtains total congestion feedback value Φ in next control interval
k+1.
2. the method for claim 1, is characterized in that described router is according to current D
kvalue is upgraded the length in its next control interval, N/mono-that control interval value is average end-to-end delay, and N is 20.
3. the method for claim 1, is characterized in that Φ
k+1the method being distributed in next control interval in the packet of each session stream is:
A) to total congestion feedback value Φ
k+1be decomposed into positive feedback p_ Φ
k+1with negative feedback n_ Φ
k+1, when | Φ
k+1| > γ * u
ktime, p_ Φ
k + 1=(| Φ
k+1|+Φ
k+1)/2, n_ Φ
k+1=(| Φ
k+1|-Φ
k+1)/2, otherwise, p_ Φ
k+1=γ * u
k+1+ (| Φ
k+1|+Φ
k+1)/2, n_ Φ
k+1=-γ * u
k+1+ (| Φ
k+1|-Φ
k+1)/2;
B) establish a positive feedback parameter p
k+1, a negative feedback parameter n
k+1; Wherein p
k+1=p_ Φ
k+1/ Y
k, n
k+1=n_ Φ
k+1/ u
k;
C) in next control interval, i.e., in the k+1 control interval, receive input packet pkt
itime, residue positive feedback value is r_p_ Φ
k+1, residual negative value of feedback is r_n_ Φ
k+1, the positive feedback value of distributing to this packet is p_pkt
i=p
k+1* (X
i* L
i/ η
k+1), the negative feedback value of distributing to this packet is n_pkt
i=n
k+1* η
k+1/ L
i;
D) the congestion feedback value of distributing to this packet is t_pkt
i=min (p_pkt
i, r_p_ Φ
k+1) – min (n_pkt
i, r_n_ Φ
k+ 1), and make r_p_ Φ
k+1-=min (p_pkt
i, r_p_ Φ
k+1), r_n_ Φ
k+1-=min (n_pkt
i, r_n_ Φ
k+1);
Wherein, Y
kfor the number of sessions in current control interval, η
kfor the length of control interval k, L
ifor packet pkt
ilength, X
ifor packet pkt
iaverage window pitch, γ is the heavy distribution factor of flow.
4. method as claimed in claim 3, is characterized in that described transmitting terminal A, in the time often receiving the bandwidth increments value of feedback of non-zero, upgrades average window pitch, and average window pitch is that end-to-end delay is divided by current congestion window value.
5. method as claimed in claim 4, is characterized in that described transmitting terminal also controls the transmission interval of packet, makes it be not less than average packet interval, and wherein average packet interval equals end-to-end delay and be multiplied by the length of packet, then divided by current congestion window value.
6. the method for claim 1, is characterized in that the method for upgrading described bandwidth increments to be confirmed is: δ
i, j+1=min (δ
i,j, v
i, j+1); Wherein, v
i, j+1for this router is confirmed bandwidth increments.
7. method as claimed in claim 1 or 2, while it is characterized in that described router is received conversation end message, regains the bandwidth resources that this session takies, and increases the surplus resources of positive feedback.
8. the method for claim 1, is characterized in that described congested prolate-headed form is: the type in next packet header, extension header length, this prolate-headed version, average window pitch, end-to-end delay, bandwidth increments value of feedback, bandwidth increments to be confirmed.
9. the method for claim 1, is characterized in that described congestion state comprises: average idle bandwidth and the queue length of output interface in the current control interval of router.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110358192.XA CN102404208B (en) | 2011-11-11 | 2011-11-11 | Stable network congestion control method with high-efficiency |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110358192.XA CN102404208B (en) | 2011-11-11 | 2011-11-11 | Stable network congestion control method with high-efficiency |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102404208A CN102404208A (en) | 2012-04-04 |
| CN102404208B true CN102404208B (en) | 2014-07-02 |
Family
ID=45886011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110358192.XA Expired - Fee Related CN102404208B (en) | 2011-11-11 | 2011-11-11 | Stable network congestion control method with high-efficiency |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102404208B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016106702A1 (en) * | 2014-12-31 | 2016-07-07 | 华为技术有限公司 | Service traffic distribution method and apparatus |
| CN108011834A (en) * | 2016-10-28 | 2018-05-08 | 华为技术有限公司 | The definite method and apparatus of TCP congestion windows |
| CN111542089B (en) * | 2020-05-14 | 2021-11-09 | 海南大学 | Multipath congestion control method and device based on time delay and window jitter |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101023455A (en) * | 2004-08-17 | 2007-08-22 | 加州理工大学 | Method and apparatus for network congestion control using queue control and one-way delay measurements |
-
2011
- 2011-11-11 CN CN201110358192.XA patent/CN102404208B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101023455A (en) * | 2004-08-17 | 2007-08-22 | 加州理工大学 | Method and apparatus for network congestion control using queue control and one-way delay measurements |
Non-Patent Citations (1)
| Title |
|---|
| Specification for the Explicit Control Protocol;A.Falk,Y.Pryadkin;<IETF:Network Working Group>;20070118;正文第2.1部分第3至8段，第3.2部分第13段，第4.1.1部分第5段，第4.1.2部分第2段，第4.1.2.1部分第1段，第4.1.3部分第1段，第4.2.1部分第1段，第4.2.2部分第2至3段，第5段，第10段和第16段，第4.2.3部分第7段 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102404208A (en) | 2012-04-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1672142B (en) | System and method of quality of service management in network gateways | |
| CN100518148C (en) | Token-based Internet traffic control method | |
| JP2004538719A (en) | Method for providing a non-linear, highly scalable increase-decrease congestion control mechanism | |
| JP2009055327A (en) | Network system | |
| Bonald | Comparison of TCP Reno and TCP Vegas: efficiency and fairness | |
| JP2022549683A (en) | Method, device and system for determining required bandwidth for data stream transmission | |
| CN108768880B (en) | Congestion control mechanism method based on adjustment of notification window in SDN environment | |
| CN102404208B (en) | Stable network congestion control method with high-efficiency | |
| CN115460156A (en) | Data center lossless network congestion control method, device, equipment and medium | |
| Pazos et al. | Using back-pressure to improve TCP performance with many flows | |
| CN101557273A (en) | Method simultaneously suitable for wired network real-time streaming media transport protocol and wireless network real-time streaming media transport protocol | |
| Low et al. | Understanding TCP vegas: Theory and practice | |
| Sayenko et al. | Comparison and analysis of the revenue-based adaptive queuing models | |
| Wydrowski | Techniques in Internet congestion control | |
| Yang et al. | Improving XCP to achieve max–min fair bandwidth allocation | |
| Shi | Token-based congestion control: Achieving fair resource allocations in P2P networks | |
| Welzl | Traceable congestion control | |
| Andrew et al. | MaxNet: Theory and implementation | |
| Manikandan et al. | Active queue management based congestion control protocol for wireless networks | |
| Shy et al. | Router active queue management for both multimedia and best-effort traffic flows | |
| Athuraliya et al. | An empirical validation of a duality model of TCP and queue management algorithms | |
| WO2010063063A1 (en) | Method and apparatus for network traffic control | |
| Wakeman et al. | A combined admission and congestion control scheme for variable bit rate video | |
| Mao | A Classified AIMD Congestion Control Scheme in IP Networks | |
| CN106850452B (en) | Distributed optimal flow control method with limited network cache |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140702 Termination date: 20161111 |