CN116170380B - ECN marking strategy and queue management method and system based on congestion prediction - Google Patents
ECN marking strategy and queue management method and system based on congestion prediction Download PDFInfo
- Publication number
- CN116170380B CN116170380B CN202310432520.9A CN202310432520A CN116170380B CN 116170380 B CN116170380 B CN 116170380B CN 202310432520 A CN202310432520 A CN 202310432520A CN 116170380 B CN116170380 B CN 116170380B
- Authority
- CN
- China
- Prior art keywords
- queue
- reading period
- data packet
- current
- length
- 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
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/127—Avoiding congestion; Recovering from congestion by using congestion prediction
-
- 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
- H04L47/26—Flow control; Congestion control using explicit feedback to the source, e.g. choke packets
- H04L47/263—Rate modification at the source after receiving feedback
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/50—Queue scheduling
- H04L47/56—Queue scheduling implementing delay-aware scheduling
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/50—Reducing energy consumption in communication networks in wire-line communication networks, e.g. low power modes or reduced link rate
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to a congestion prediction-based ECN marking strategy and queue management method and system, wherein the method comprises the following steps: s1: reading the length of each queue on the switch according to a preset period Ti; s2: if the data packet arrives in the reading period, calculating the estimated value of the growth rate of each queue according to the length of the queue of the previous and current reading periods of each obtained queue; otherwise, using the estimated value of the increasing rate of the last reading period as the estimated value of the increasing rate of the current reading period; s3: and calculating linear prediction values of the lengths of the queues according to the estimated value of the growth rate, comparing the linear prediction values with a preset queue length threshold value, predicting the congestion condition of the queues, and marking the ECN of the data packet before network congestion according to the prediction result. The method provided by the invention can effectively inhibit the peak value of the queue and reduce the overflow of the queue caused by burst traffic.
Description
Technical Field
The invention relates to the field of computer networks, in particular to an ECN marking strategy and queue management method and system based on congestion prediction.
Background
With the development of networks, demands of users are increasing, and demands of services such as cloud computing and cloud storage are increasing, and these increasing applications and services require a large amount of computing and storage resources. In order to efficiently and quickly meet the demands, enterprises integrate large-scale computing storage nodes to form a data center. Traffic in a data center network is more massive in scale and is more sensitive to performance indexes such as bandwidth delay and the like.
Explicit congestion notification (Explicit Congestion Notification, ECN) is a function added to the internet protocol (Internet Protocol, IP) header. The method allows the switch to mark the data packet after the queue length exceeds the threshold value so as to inform the source end of congestion, and the source end can reduce the data rate before the intermediate switch queue overflows, thereby avoiding the defect of the traditional congestion detection based on packet loss. ECN is now widely used in data center networks. However, the existing ECN marking strategy based on the fixed threshold has the defects of poor adaptivity, uncontrollable maximum queue length caused by delay of an ECN feedback loop, and the like, and a large amount of packet loss still can be caused when the ECN marking strategy faces to burst traffic, so that the network performance is reduced.
Disclosure of Invention
In order to solve the technical problems, the invention provides an ECN marking strategy and queue management method and system based on congestion prediction.
The technical scheme of the invention is as follows: an ECN marking strategy and queue management method based on congestion prediction, comprising:
step S1: reading the length of each queue on the switch according to a preset period Ti;
step S2: if the data packet arrives in the reading period, calculating the estimated value of the growth rate of each queue according to the length of the queue of the previous and current reading periods of each obtained queue; otherwise, using the estimated value of the increasing rate of the last reading period as the estimated value of the increasing rate of the current reading period;
step S3: and calculating a linear prediction value of the length of each queue according to the estimated value of the growth rate, comparing the linear prediction value with a preset queue length threshold value, thereby predicting the congestion condition of the queue, and marking the ECN before network congestion of the data packet according to the prediction result.
Compared with the prior art, the invention has the following advantages:
1. the invention discloses an ECN marking strategy and a queue management method based on congestion prediction, which refer to the length of a queue and the growth rate of the queue at the same time when ECN marking decision is carried out, so that the invention can cope with the scene of frequent change of network flow rate, thereby solving the problem of insufficient self-adaptability of the traditional scheme based on fixed threshold.
2. The invention counteracts the delay of the feedback loop by predicting the congestion and feeding back the congestion information in advance, so that the network flow is controlled before the congestion, and the peak value of the queue is restrained, thereby solving the problem that the maximum queue length is uncontrollable.
Drawings
Fig. 1 is a flowchart of an ECN marking strategy and queue management method based on congestion prediction according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a change in a queue according to an embodiment of the invention;
fig. 3 is a block diagram of an ECN marking policy and queue management system based on congestion prediction according to an embodiment of the present invention.
Detailed Description
The invention provides an ECN marking strategy and a queue management method based on congestion prediction, which can effectively inhibit a peak value of a queue and reduce overflow of the queue caused by burst traffic.
The present invention will be further described in detail below with reference to the accompanying drawings by way of specific embodiments in order to make the objects, technical solutions and advantages of the present invention more apparent.
Example 1
As shown in fig. 1, the ECN marking policy and queue management method based on congestion prediction provided by the embodiment of the present invention includes the following steps:
step S1: reading the length of each queue on the switch according to a preset period Ti;
step S2: if the data packet arrives in the reading period, calculating the estimated value of the growth rate of each queue according to the length of the queue of the previous and current reading periods of each obtained queue; otherwise, using the estimated value of the increasing rate of the last reading period as the estimated value of the increasing rate of the current reading period;
step S3: and calculating linear prediction values of the lengths of the queues according to the estimated value of the growth rate, comparing the linear prediction values with a preset queue length threshold value, predicting the congestion condition of the queues, and marking the ECN of the data packet before network congestion according to the prediction result.
In one embodiment, step S1 described above: the method for reading the queue length of each queue on the switch according to the preset period Ti specifically comprises the following steps:
reading the queue length of each queue on the exchanger in a period Ti; wherein Ti is set according to the link bandwidth and the maximum transmission unit of the switch.
In order to more accurately predict network congestion, while at the same time increasing the adaptivity of ECN marking strategies, embodiments of the present invention require an estimation of the current network injection rate of the switch queues. Starting from the targets of easy deployment and reduced computational complexity, the rate estimation method is to periodically read the length of the queue, and the reading period is Ti. The read period should not be so small that there is no change in queue length over the interval, resulting in false quantification of network flow rate; too large a read cycle setting results in inaccurate instantaneous rate estimation and reduced sensitivity to bursty traffic. Therefore, in the embodiment of the invention, ti is set to be 1.1-1.5 times of the Maximum Transmission Unit (MTU) of the link of the switch divided by the link bandwidth.
In one embodiment, step S2 above: if the data packet arrives in the reading period, calculating the estimated value of the growth rate of each queue according to the length of the obtained queue of the previous reading period and the current reading period of each queue; otherwise, the estimated value of the increasing rate of the last reading period is used as the estimated value of the increasing rate of the current reading period, which specifically comprises:
if the data packet arrives at the moment of the reading period, the queue length of each queue of the current reading period is differentiated from the queue length of the previous reading period, and the average growth rate of each queue of the previous reading period is obtained by dividing the queue length by Ti, and the average growth rate is used as an estimated value of the growth rate of each queue of the current reading period;
if the data packet arrives within the read cycle, the growth rate estimate of the last read cycle is used as the growth rate estimate of the current read cycle.
For example, the read period is set to ti=2 seconds, i.e. the queue length is read at 2,4,6, … seconds, and if the packet arrives at 5 seconds, the average queue growth rate of the last period, i.e. the second period, is used as the estimated queue growth rate value of the current period. If the data packet arrives at the 6 th second, the current queue length is read first, and the difference is obtained between the current queue length and the queue length of the last reading period, and the difference is divided by the value of Ti to be used as the estimated value of the queue growth rate of the current period.
In one embodiment, the step S3: according to the estimated value of the growth rate, calculating a linear prediction value of each queue length, comparing the linear prediction value with a preset queue length threshold value, thereby predicting the congestion condition of the queue, and carrying out ECN marking on the data packet before network congestion according to the prediction result, wherein the method specifically comprises the following steps:
step S31: when a data packet arrives at a queue on a switch, multiplying the time Ta by an estimated value of the growth rate of the current queue and adding the current queue length to obtain a linear predicted value of the queue length; wherein Ta is set according to the end-to-end delay;
when the data packet arrives at the switch queue, multiplying Ta by the queue growth rate estimated value obtained in S2 and adding the current queue length to obtain a queue length linear predicted value, and then determining whether to make ECN marks according to the queue length linear predicted value. Ta is a preset duration, and excessive control of network flows can be caused by a larger Ta value, so that throughput is reduced, and therefore the Ta is set to be 1-5 times of end-to-end delay between servers.
Step S32: discarding the data packet if the current queue is full;
if the current queue is not full and the length linear predictive value of the current queue is greater than or equal to the preset queue length threshold, setting ECN mark bits for the data packets and then placing the data packets into the queue, and if the linear predictive value is less than the preset queue length threshold, directly placing the data packets into the queue.
If the current queue is full at the moment, the data packet is directly discarded; otherwise, if the length linear predictive value of the current queue is greater than or equal to the set queue length threshold, the data packet is marked and enqueued by ECN, otherwise, the data packet is enqueued directly.
Instead of using the current queue length as the ECN marking reference in the existing method, the present invention uses the length linear prediction value of the queue, so that the ECN marking time point is advanced by Ta time compared with the possible congestion occurrence time in time, so as to offset the delay of the ECN feedback loop. When ECN feedback is in effect, the queue length is controlled to drop, enabling the queue peak to be controlled below a preset threshold under normal conditions.
The ECN marking strategy and queue management method based on congestion prediction comprises three judging conditions:
judging: whether the current queue is full.
Judging: whether the read period Ti is reached.
Judging three: whether the queue length linear prediction value exceeds a threshold.
Taking fig. 2 as an example, when a plurality of data packets are sequentially injected into the same switch queue at different times, the ECN marking strategy and queue management method based on congestion prediction according to the present invention are shown:
1. when the packet arrives at the switch at position (1) in fig. 2, the following steps are performed:
step 1.1: according to the judgment one, when the queue is not full, the data packet is not discarded.
Step 1.2: and according to the judgment II, if the time is just at the moment of the reading period, acquiring the current queue length, and calculating the current queue growth rate estimated value= (current queue length-last period of queue length)/Ti, otherwise, taking the last period of queue growth rate estimated value as the current queue growth rate estimated value.
Step 1.3: individual queue length linear prediction value = current queue length + current queue growth rate estimation value x Ta is calculated.
Step 1.4: and according to the judgment III, if the linear predictive value of the queue length is smaller than the threshold value, directly enqueuing the data packet and waiting for transmission.
Thereafter the processing of the injected packets between position (1) and position (2) in fig. 2 is the same, and the queue continues to grow since no congestion feedback is performed in this process.
2. If a packet arrives at the switch at position (2) in fig. 2, i.e. the queue has already had a certain backlog at this time, and at the current rate, the threshold will be reached after Ta time.
Repeating the steps 1.1-1.3.
Step 2.4: and according to the judgment III, at the moment, the linear predictive value of the queue length is larger than or equal to the threshold value, and the data packet is enqueued to wait for transmission after ECN marking. At this point, the queue will reach the threshold after Ta time, so the marking decision currently made is Ta earlier than congestion occurrence time, i.e. the pre-ECN marking strategy employed by the present invention.
Thereafter, the processing procedure of the data packet between the position (2) and the position (3) in fig. 2 is the same, and since there is hysteresis in ECN congestion feedback, the data packet arriving at the position (2) will not immediately cause the queue to drop after being marked, the queue will be continuously backlogged, the queue length will gradually increase, and the data packet in the middle will be marked.
3. When the data packet is located at the position (3) in fig. 2, the marked data packet coming at the position (2) is fed back and the source end reduces the sending rate, and then the queue length is reduced, and the maximum queue length is usually controlled below a threshold value, namely the problem that the maximum queue length of the prior scheme is uncontrollable is solved by adopting the ECN marking strategy in advance.
The processing of the data packets between position (3) and position (4) in fig. 2 is then identical to the processing of the data packets at position (1), after which the input rate of the switch is less than the output rate and the queue length gradually decreases.
4. When in position (4) of fig. 2, since ECN feedback was not performed for a period of time, according to the current conventional source-side algorithm, the packet transmission rate is gradually increased, and the queue length is gradually increased.
If a packet arrives at the switch queue between location (4) and location (5) in fig. 2, the process is as follows:
step 4.1: the processing is the same as for the data packet at location (1).
Step 4.2: according to the second judgment, the processing procedure is the same as that of the data packet at the position (2). The network flow is faster than the previous peak due to the new packet injection at location (4). Therefore, the front-back change of the queue in the Ti time is larger, and the estimated value of the current queue growth rate calculated by the embodiment of the invention is larger, so that the invention is more sensitive to the change of the network flow rate than the prior scheme.
The rest processing steps are the same as the steps 1.3-1.4.
5. When the packet arrives at the switch at position (5) of fig. 2, the queue length is now shorter than at position (2), but the flow rate is now faster and at the current rate, the threshold will be reached after Ta time.
Repeating the steps 1.1-1.3.
Step 5.4: and according to the third judgment, when the linear prediction value of the queue length is larger than or equal to the threshold value, the data packet is enqueued for waiting to be sent after ECN marking. At this point the network flow rate is faster than at location (2), so the queue length at the time the marking decision is made is shorter. That is, the present invention is adaptive to network flow rate based on congestion prediction pre-ECN marking decisions.
Thereafter the processing of the data packets between location (5) and location (6) is the same as described above.
6. When in position (6) of fig. 2, the incoming packet marked ECN at position (5) of fig. 2 is fed back to the source and causes the source to reduce the sending rate, after which the queue length is reduced such that the maximum length of the queue is controlled below the threshold, i.e. the present invention still has maximum queue length control capability when the network flow rate is changed using the pre-ECN marking strategy.
The invention discloses an ECN marking strategy and a queue management method based on congestion prediction, which refer to the length of a queue and the growth rate of the queue at the same time when ECN marking decision is carried out, so that the invention can cope with the scene of frequent change of network flow rate, thereby solving the problem of insufficient self-adaptability of the traditional scheme based on fixed threshold. In addition, the invention counteracts the delay of the feedback loop by predicting the congestion and feeding back the congestion information in advance, so that the network flow is controlled before the congestion, and the peak value of the queue is restrained, thereby solving the problem that the maximum queue length is uncontrollable.
Example two
As shown in fig. 3, an embodiment of the present invention provides an ECN marking policy and queue management system based on congestion prediction, including the following modules:
an acquiring queue length module 41, configured to read a queue length of each queue on the switch in a preset period Ti;
a queue growth rate calculation module 42, configured to calculate a growth rate estimation value of each queue according to the obtained queue length of the previous and current reading periods of each queue if the data packet arrives in the reading period; otherwise, using the estimated value of the increasing rate of the last reading period as the estimated value of the increasing rate of the current reading period;
and the ECN module 43 is used for calculating linear prediction values of the lengths of the queues according to the estimated value of the growth rate, comparing the linear prediction values with a preset queue length threshold value so as to predict the congestion condition of the queues, and marking the ECN of the data packet before network congestion according to the prediction result.
The above examples are provided for the purpose of describing the present invention only and are not intended to limit the scope of the present invention. The scope of the invention is defined by the appended claims. Various equivalents and modifications that do not depart from the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (2)
1. An ECN marking policy and queue management method based on congestion prediction, comprising:
step S1: reading the length of each queue on the switch according to a preset reading period Ti;
step S2: if the data packet arrives in the reading period, calculating the estimated value of the growth rate of each queue according to the length of the queue of the previous and current reading periods of each obtained queue; otherwise, the estimated value of the increasing rate of the last reading period is used as the estimated value of the increasing rate of the current reading period, which specifically comprises:
if the data packet arrives at the moment of the reading period, the queue length of each queue of the current reading period is differentiated from the queue length of the previous reading period, and is divided by Ti to obtain the average growth rate of each queue of the previous reading period, and the average growth rate is used as the growth rate estimation value of each queue of the current reading period;
if the data packet arrives in the reading period, using the estimated value of the increasing rate of the last reading period as the estimated value of the increasing rate of the current reading period;
step S3: calculating linear prediction values of the length of each queue according to the estimated value of the growth rate, comparing the linear prediction values with a preset queue length threshold value so as to predict congestion conditions of the queues, and marking ECN before network congestion of the data packet according to a prediction result, wherein the method specifically comprises the following steps:
step S31: when a data packet arrives at a queue on the switch, multiplying the estimated value of the growth rate of the current queue by time Ta and adding the current queue length to obtain a linear prediction value of the queue length; wherein Ta is set according to the end-to-end delay;
step S32: discarding the data packet if the current queue is full;
if the current queue is not full and the length linear prediction value of the current queue is greater than or equal to the preset queue length threshold, setting an ECN mark bit for the data packet, then placing the data packet into the queue, and if the linear prediction value is less than the preset queue length threshold, directly placing the data packet into the queue.
2. An ECN marking policy and queue management system based on congestion prediction, comprising the following modules:
the queue length obtaining module is used for reading the queue length of each queue on the switch in a preset period Ti;
a queue growth rate estimation value calculating module, configured to calculate a growth rate estimation value of each queue according to the obtained queue length of the previous and current reading periods of each queue if the data packet arrives in the reading period; otherwise, the estimated value of the increasing rate of the last reading period is used as the estimated value of the increasing rate of the current reading period, which specifically comprises:
if the data packet arrives at the moment of the reading period, the queue length of each queue of the current reading period is differentiated from the queue length of the previous reading period, and is divided by Ti to obtain the average growth rate of each queue of the previous reading period, and the average growth rate is used as the growth rate estimation value of each queue of the current reading period;
if the data packet arrives in the reading period, using the estimated value of the increasing rate of the last reading period as the estimated value of the increasing rate of the current reading period;
the ECN marking module is used for calculating linear predicted values of the lengths of the queues according to the estimated value of the growth rate and comparing the linear predicted values with a preset queue length threshold value so as to predict the congestion condition of the queues, and performing ECN marking on the data packets before network congestion according to a predicted result, and specifically comprises the following steps:
step S31: when a data packet arrives at a queue on the switch, multiplying the estimated value of the growth rate of the current queue by time Ta and adding the current queue length to obtain a linear prediction value of the queue length; wherein Ta is set according to the end-to-end delay;
step S32: discarding the data packet if the current queue is full;
if the current queue is not full and the length linear prediction value of the current queue is greater than or equal to the preset queue length threshold, setting an ECN mark bit for the data packet, then placing the data packet into the queue, and if the linear prediction value is less than the preset queue length threshold, directly placing the data packet into the queue.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310432520.9A CN116170380B (en) | 2023-04-21 | 2023-04-21 | ECN marking strategy and queue management method and system based on congestion prediction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310432520.9A CN116170380B (en) | 2023-04-21 | 2023-04-21 | ECN marking strategy and queue management method and system based on congestion prediction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116170380A CN116170380A (en) | 2023-05-26 |
CN116170380B true CN116170380B (en) | 2023-08-29 |
Family
ID=86416668
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310432520.9A Active CN116170380B (en) | 2023-04-21 | 2023-04-21 | ECN marking strategy and queue management method and system based on congestion prediction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116170380B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116886622B (en) * | 2023-09-08 | 2023-11-24 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Network congestion control method, device, equipment and storage medium |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2001045331A1 (en) * | 1999-12-13 | 2001-06-21 | Nokia Corporation | Congestion control method for a packet-switched network |
CN1545286A (en) * | 2003-11-21 | 2004-11-10 | 清华大学 | ECN based congestion control method with prediction verification |
CN101414957A (en) * | 2007-10-17 | 2009-04-22 | 北京中电华大电子设计有限责任公司 | Buffering queue management method for wireless LAN |
WO2010089886A1 (en) * | 2009-02-06 | 2010-08-12 | 富士通株式会社 | Packet buffer device and packet discarding method |
US7983156B1 (en) * | 2004-11-12 | 2011-07-19 | Openwave Systems Inc. | System and method for controlling network congestion |
CN103999409A (en) * | 2011-10-04 | 2014-08-20 | 香港中文大学 | Method for link buffer size and queue length estimation for bandwidth-varying mobile data networks |
EP2784997A1 (en) * | 2013-03-28 | 2014-10-01 | British Telecommunications public limited company | Re-marking of packets for congestion control |
CN104270790A (en) * | 2014-10-23 | 2015-01-07 | 湘潭大学 | Congestion control method based on equitable distribution of communication channel |
CN106789701A (en) * | 2016-12-30 | 2017-05-31 | 北京邮电大学 | Self adaptation ECN labeling methods and device in a kind of data center |
EP3192299A1 (en) * | 2014-09-10 | 2017-07-19 | Telefonaktiebolaget LM Ericsson (publ) | Explicit congestion notification marking of user traffic |
CN107070804A (en) * | 2017-03-15 | 2017-08-18 | 清华大学 | Join the team and mark and go out the display congestion marking method and device that badge note is combined |
CN108270691A (en) * | 2018-01-11 | 2018-07-10 | 电子科技大学 | A kind of alternative ecn (explicit congestion notification) labeling method for data center |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB201003206D0 (en) * | 2010-02-25 | 2010-04-14 | Skype Ltd | Method of estimating congestion |
CN104272680B (en) * | 2012-03-09 | 2017-05-17 | 英国电讯有限公司 | Signalling congestion |
CN105075202B (en) * | 2013-03-28 | 2019-07-12 | 英国电讯有限公司 | For handling method, node and the packet network of grouping in packet network |
US9356868B2 (en) * | 2013-08-23 | 2016-05-31 | Broadcom Corporation | Congestion detection and management at congestion-tree roots |
US10084716B2 (en) * | 2016-03-20 | 2018-09-25 | Mellanox Technologies Tlv Ltd. | Flexible application of congestion control measures |
US10397124B2 (en) * | 2017-01-17 | 2019-08-27 | Argela Yazilim ve Bilisim Teknolojileri San. ve Tic. A.S. | System and method to prevent persistent full switch queues in software defined networks |
US20210328930A1 (en) * | 2020-01-28 | 2021-10-21 | Intel Corporation | Predictive queue depth |
US11818046B2 (en) * | 2021-01-26 | 2023-11-14 | Samsung Electronics Co., Ltd. | Coordinated congestion control in network-attached devices |
-
2023
- 2023-04-21 CN CN202310432520.9A patent/CN116170380B/en active Active
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2001045331A1 (en) * | 1999-12-13 | 2001-06-21 | Nokia Corporation | Congestion control method for a packet-switched network |
CN1545286A (en) * | 2003-11-21 | 2004-11-10 | 清华大学 | ECN based congestion control method with prediction verification |
US7983156B1 (en) * | 2004-11-12 | 2011-07-19 | Openwave Systems Inc. | System and method for controlling network congestion |
CN101414957A (en) * | 2007-10-17 | 2009-04-22 | 北京中电华大电子设计有限责任公司 | Buffering queue management method for wireless LAN |
WO2010089886A1 (en) * | 2009-02-06 | 2010-08-12 | 富士通株式会社 | Packet buffer device and packet discarding method |
CN103999409A (en) * | 2011-10-04 | 2014-08-20 | 香港中文大学 | Method for link buffer size and queue length estimation for bandwidth-varying mobile data networks |
EP2784997A1 (en) * | 2013-03-28 | 2014-10-01 | British Telecommunications public limited company | Re-marking of packets for congestion control |
EP3192299A1 (en) * | 2014-09-10 | 2017-07-19 | Telefonaktiebolaget LM Ericsson (publ) | Explicit congestion notification marking of user traffic |
CN104270790A (en) * | 2014-10-23 | 2015-01-07 | 湘潭大学 | Congestion control method based on equitable distribution of communication channel |
CN106789701A (en) * | 2016-12-30 | 2017-05-31 | 北京邮电大学 | Self adaptation ECN labeling methods and device in a kind of data center |
CN107070804A (en) * | 2017-03-15 | 2017-08-18 | 清华大学 | Join the team and mark and go out the display congestion marking method and device that badge note is combined |
CN108270691A (en) * | 2018-01-11 | 2018-07-10 | 电子科技大学 | A kind of alternative ecn (explicit congestion notification) labeling method for data center |
Non-Patent Citations (1)
Title |
---|
面向异构网络的多径传输性能增强机制研究;韩江萍;《中国科学技术大学》;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN116170380A (en) | 2023-05-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | Novel quick start (QS) method for optimization of TCP | |
CN107634911B (en) | Adaptive congestion control method based on deep learning in information center network | |
CN101414957B (en) | Buffering queue management method for wireless LAN | |
Sun et al. | PD-RED: to improve the performance of RED | |
CN107624232B (en) | Apparatus and method for controlling downlink throughput in communication system | |
US9686201B2 (en) | Predictive management of a network buffer | |
CN1327677C (en) | ECN based congestion control method with prediction verification | |
CN116170380B (en) | ECN marking strategy and queue management method and system based on congestion prediction | |
CN101984608A (en) | Method and system for preventing message congestion | |
CN110730469B (en) | Method for predicting bandwidth based on extended Kalman wireless network and congestion control thereof | |
CN111726301B (en) | Congestion control method and system for guaranteeing video quality in real-time video | |
CN111263102B (en) | ViLTE video call congestion control method and system based on delay gradient accumulation | |
JP4726978B2 (en) | Congestion control device and congestion control method | |
CN112995048A (en) | Blocking control and scheduling fusion method for data center network and terminal equipment | |
Mejri et al. | Hop-by-hop interest rate notification and adjustment in named data networks | |
CN114500394B (en) | Congestion control method for differentiated services | |
Abu-Shareha | Integrated Random Early Detection for Congestion Control at the Router Buffer. | |
CN113572655A (en) | Congestion detection method and system for loss-free network | |
CN109787861B (en) | Network data delay control method | |
CN116827867A (en) | Low-delay congestion flow identification method based on data center network | |
Çakmak et al. | A Review: Active queue management algorithms in mobile communication | |
CN112019443B (en) | Multipath data transmission method and device | |
Tang et al. | Adaptive queue management based on the change trend of queue size | |
CN114095126A (en) | Method, system and device for adjusting data sending rate | |
Almomani et al. | Simulation Based Performance Evaluation of Several Active Queue Management Algorithms for Computer Network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |