CN114979010A - Active selection system and method of TCP congestion control algorithm based on decision tree - Google Patents
Active selection system and method of TCP congestion control algorithm based on decision tree Download PDFInfo
- Publication number
- CN114979010A CN114979010A CN202210519657.3A CN202210519657A CN114979010A CN 114979010 A CN114979010 A CN 114979010A CN 202210519657 A CN202210519657 A CN 202210519657A CN 114979010 A CN114979010 A CN 114979010A
- Authority
- CN
- China
- Prior art keywords
- packet loss
- loss rate
- congestion control
- decision tree
- control algorithm
- 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.)
- Pending
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 90
- 238000003066 decision tree Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 21
- 230000010354 integration Effects 0.000 claims abstract description 8
- 238000012544 monitoring process Methods 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 9
- 238000009826 distribution Methods 0.000 claims description 4
- 238000010187 selection method Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000005034 decoration Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
Images
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
- H04L47/19—Flow control; Congestion control at layers above the network layer
- H04L47/193—Flow control; Congestion control at layers above the network layer at the transport layer, e.g. TCP related
-
- 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 discloses an active selection system and method of a decision tree-based TCP congestion control algorithm, which comprises the following steps of S1: the round trip time RTT and the packet loss rate of a TCP link network are monitored, round trip time RTT and packet loss rate data of the link are acquired in a time slicing mode, and meanwhile network bandwidth parameters are acquired; step S2, data prediction: predicting the predicted round trip time RTT and the packet loss rate of the next fixed time slice by using a differential integration moving average autoregressive model according to the round trip time RTT, the packet loss rate data and the network bandwidth which are obtained in the step S1; step S3, output selection: through a decision tree model, taking the predicted round trip time RTT and the predicted packet loss rate as input data, outputting the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and if the decision results of the current link congestion control algorithm and the decision tree are different, switching; otherwise, the switching is not carried out, thereby improving the throughput rate of the TCP link.
Description
Technical Field
The invention belongs to the technical field of TCP congestion control algorithms, and particularly relates to an active selection system and method of a TCP congestion control algorithm based on a decision tree.
Background
In the same network, the situation that wired and wireless networks with different characteristics exist simultaneously and the network round trip delay RTT and packet loss rate fluctuate on one TCP network link easily occurs, and since the Tahoe algorithm of TCP is proposed, TCP congestion control algorithms customized for certain characteristic network environments are continuously proposed, such as High Speed TCP with High bandwidth and Hybla with long RTT. However, among the proposed TCP congestion control algorithms, it is not possible to select an optimal congestion control algorithm for different network environments because different TCP congestion control algorithms are required to adapt to the characteristics of different network environments.
For a dynamically changing network environment, if the same congestion control algorithm is adopted, good network transmission performance cannot be maintained under different network environments, for example, a Reno algorithm suitable for a low packet loss and low delay scenario shows far worse than Hybla under a high delay network environment, similarly, a Hybla algorithm designed for a long delay network shows better performance than Reno in the long delay network, and when the network delay of a TCP link continuously changes along with network traffic fluctuation, the Reno algorithm used in the TCP link all the time cannot maintain high throughput for the link when the round-trip delay of the TCP link network increases. None of the existing congestion control algorithms can exhibit good execution efficiency in all network environments, so it is necessary to dynamically switch the congestion control algorithm according to the network environment.
Disclosure of Invention
The invention provides an active selection system and method of a TCP congestion control algorithm based on a decision tree, aiming at the problem that the congestion control algorithm can not be selected aiming at different network environments in the prior art, and the active selection system comprises a step S1, data monitoring: the round trip time RTT and the packet loss rate of a TCP link network are monitored, round trip time RTT and packet loss rate data of the link are acquired in a time slicing mode, and meanwhile network bandwidth parameters are acquired; step S2, data prediction: predicting the predicted round trip time RTT and the packet loss rate of the next fixed time slice by using a differential integration moving average autoregressive model according to the round trip time RTT, the packet loss rate data and the network bandwidth which are obtained in the step S1; step S3, output selection: through a decision tree model, taking the predicted round trip time RTT and the predicted packet loss rate as input data, outputting the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and if the decision results of the current link congestion control algorithm and the decision tree are different, switching; otherwise, the switching is not carried out, thereby improving the throughput rate of the TCP link.
In order to achieve the purpose, the invention adopts the technical scheme that: the active selection method of the decision tree-based TCP congestion control algorithm comprises the following steps:
s1, data monitoring: the round trip time RTT and the packet loss rate of a TCP link network are monitored, and round trip time RTT and packet loss rate data of a link are acquired in a time slicing mode;
s2, data prediction: predicting the predicted round trip time RTT and the packet loss rate of the next fixed time slice by using a differential integration moving average autoregressive (ARIMA) model according to the round trip time RTT and the packet loss rate data obtained in the step S1; in the ARIMA (p, d, q) model, a time series with the average value mu is generated by inputting N groups of monitoring data according to the following formula:
wherein y is t And a t Representing the real variable value at time t and the corresponding random error;
are polynomials of order p and q of B;
and theta j (j ═ 1,2, … …, q) are model parameters;
b is the move backward operator; the orders p and q of the model and the differential order d are integers; random error a t Are assumed to be independent and identical distributions with a mean of zero and a constant variance of σ 2 ；
S3, decision selection: through a decision tree model, taking the predicted round trip time RTT and the predicted packet loss rate as input data, outputting the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and if the decision results of the current link congestion control algorithm and the decision tree are different, switching; otherwise, no handover is performed.
In order to achieve the purpose, the invention also adopts the technical scheme that: the active selection system of the TCP congestion control algorithm based on the decision tree comprises a monitoring module, a prediction module and a decision module,
the monitoring module is used for keeping monitoring the round trip time RTT and the packet loss rate of the TCP link network and acquiring the round trip time RTT and the packet loss rate data of the link in time slices;
the prediction module predicts the predicted round trip time delay RTT and the packet loss rate of the next fixed time slice by using a difference integration moving average autoregressive model according to the round trip time delay RTT and the packet loss rate data acquired by the monitoring module;
the decision module takes the round trip time RTT and the predicted packet loss rate acquired by the prediction module as input data through a decision tree model, outputs the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and switches if the decision results of the current link congestion control algorithm and the decision tree are different; otherwise, no handover is performed.
Compared with the prior art: for a TCP link, the decision tree model is used for selecting the most appropriate congestion control algorithm for a fluctuating network environment instead of using a fixed congestion control algorithm, so that the link throughput rate can be improved by better utilizing the characteristics of various congestion control algorithms; in addition, by adding a prediction module, the invention can overcome the hysteresis influence caused by using network monitoring data in the prior art.
Drawings
FIG. 1 is a schematic structural diagram of an active selection system of a decision tree-based TCP congestion control algorithm according to the present invention;
FIG. 2 is a flow chart of the steps of the decision tree based active selection method of TCP congestion control algorithm of the present invention;
FIG. 3 is a data flow diagram of the active selection method of the decision tree based TCP congestion control algorithm of the present invention;
fig. 4 is a diagram of the format of the timestamp option in the TCP option used in step S1 according to embodiment 2 of the present invention;
fig. 5 is a diagram of variation in round trip time RTT of an actual TCP link network according to embodiment 3 of the present invention;
fig. 6 is a graph illustrating a packet loss ratio variation of an actual TCP link according to embodiment 3 of the present invention;
fig. 7 is a graph comparing throughput rates of the model and other existing TCP congestion control algorithms provided in embodiment 3 of the present invention.
Detailed Description
The present invention will be further illustrated with reference to the accompanying drawings and specific embodiments, which are to be understood as merely illustrative of the invention and not as limiting the scope of the invention.
Example 1
As shown in fig. 1, the active selection system of a TCP congestion control algorithm based on a decision tree includes a monitoring module, a prediction module and a decision module, wherein the prediction module is respectively connected to the monitoring module and the decision module, and the monitoring module is configured to keep monitoring round trip time RTT and packet loss rate of a TCP link network, and acquire round trip time RTT and packet loss rate data of a link in time slices; the prediction module predicts the predicted round trip time RTT and the packet loss rate of the next fixed time slice by using a difference integration moving average autoregressive model according to the round trip time RTT and the packet loss rate data acquired by the monitoring module, and can overcome the hysteresis influence caused by using network monitoring data in the prior art; the decision module takes the round trip time RTT and the predicted packet loss rate acquired by the prediction module as input data through a decision tree model, outputs the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and switches if the decision results of the current link congestion control algorithm and the decision tree are different; otherwise, no handover is performed. The system selects the optimal congestion control algorithm according to different network environments, and can better utilize the characteristics of various congestion control algorithms to improve the link throughput rate.
Example 2
A method for actively selecting a TCP congestion control algorithm based on a decision tree is disclosed, as shown in FIG. 2, and comprises the following steps:
step S1, data monitoring:
in the monitoring stage, the packet loss rate of the TCP link is detected through the ICMP message, a fixed number (for example, 20) of ICMP messages are sent within a fixed time (for example, 10 seconds), and the packet loss rate of the link is calculated according to whether a return result is obtained or not.
The TCP link round trip delay RTT is measured using the TCP option in the TCP message format. The timestamp option in the TCP option is shown in fig. 4 and consists of four parts, which are a kind of kid field, a length option length field, a TSValue field, and a TS Echo Reply field. When the kid of the TCP selectable item is 8, it indicates that the content of the selectable item is a TCP timestamp, and at this time, the Value in length is 10, which indicates ten bytes in total of a TCP timestamp field, one byte in the kid field, one byte in the total length, four bytes in TS Value, which is used for indicating the timestamp for sending the ACK packet, and four bytes in TS Echo Reply, which is used for indicating the timestamp for sending the packet in response to the ACK packet. The TS Echo Reply field (TSecr) is validated only when the ACK in the TCP header is set to valid, and its Value is the TS Value in the latest packet, i.e. the transmission time of the packet. When the TCP message captured by the host is an ACK message (an ACK flag bit in TCP is 1), the last eight bytes of the TCP selectable item are TS Value and TSecr fields of the TCP timestamp selectable item, the real time of the TSecr fields can be obtained through timestamp conversion, meanwhile, the current system time now () is obtained, and the network round-trip delay of the group of messages is obtained through the calculation of the now () Tser. Because a plurality of ACK messages are available in a fixed time, a plurality of groups of averaging values are adopted to calculate the round trip time RTT of the network.
Step S2, data prediction:
the invention uses ARIMA (differential integration moving average autoregressive model) model to predict the RTT of the network delay. The model is trained with historical RTT data. And inputting RTT data obtained by monitoring after the model training is finished, and outputting the predicted RTT of the next fixed time slice. When constructing the ARIMA (p, d, q) model, 50 sets of monitoring data were input to generate the mean value μThe formula for the subsequence:
wherein y is t And a t Representing the true variable value at time t and the corresponding random error.
Are polynomials of order p and order q of B.
And theta j (j ═ 1,2, … …, q) are model parameters.
B is the move back operator. The orders p and q of the model and the difference order d are integers. Random error a t Are assumed to be independent and identical distributions with a mean of zero and a constant variance of σ 2 . The predicted round trip delay RTT and packet loss rate of the next fixed time slice are predicted by the model formula generated by inputting the 50 groups of data detected in S1.
The invention uses the traditional estimation method to predict the packet loss rate: SLOSS + (1- α) LOSS, where SLOSS is an estimated packet LOSS rate, and LOSS is an actually measured packet LOSS rate, that is, an estimated value of the network packet LOSS rate in the next fixed time slice is obtained through linear processing of a parameter α (for example, 0.8) by using the estimated value of the network packet LOSS rate in the previous time and the measured value of the actual packet LOSS rate in the previous time.
Step S3, decision selection:
the invention takes the predicted RTT and the predicted packet loss rate as input data through the decision tree and outputs the most appropriate congestion control algorithm in the next fixed time slice of the TCP link. The process of constructing the CART decision tree comprises two parts of constructing a sample set and training the CART decision tree. When a sample set is constructed, network conditions (network bandwidth, network delay RTT and network packet loss rate) are continuously changed through a tc (traffic control) tool of the ubuntu system, and congestion control is adjustedThe algorithm comprises the steps of testing the TCP throughput rate through an iperf tool, then comparing the actual throughput rates of six congestion control algorithms in the same network environment, marking the congestion control algorithm with the highest throughput rate in each network environment, and using the marked data as training data of a CART decision tree. When training the CART decision tree, for the sample set S, the kini coefficient is defined
For the attribute A, the Gain _ GINI after the data set is divided into two parts by respectively calculating any attribute value, and the minimum value min (Gain GINI) is selected A,i (S)) i ∈ A as the optimal dichotomy obtained by the attribute A. For the sample set S, calculating the optimal bipartite scheme of all attributes, and selecting the minimum value min (GainGINI) of the optimal bipartite scheme A,i (S)) i ∈ A) A ∈ Attributes as the optimal bipartite solution for the sample set S. The obtained attribute a and the attribute value i thereof are the optimal splitting attribute and the optimal splitting value of the sample set S. The attribute set includes three elements, namely round trip delay RTT, packet loss rate and network bandwidth. When the CART algorithm constructs the decision tree, the process of selecting the attributes and the attribute values is repeated, and the current optimal scheme is continuously selected as the classification rule of the decision nodes until the example data set can not be divided again.
After the CART decision tree is constructed, the optimal congestion control algorithm of the TCP link in the corresponding time slicing can be decided through the link bandwidth, the predicted RTT, the packet loss rate and the network bandwidth, the congestion control algorithm is compared with the congestion control algorithm used by the link, if the congestion control algorithm is different, switching is carried out, and otherwise, switching is not carried out. The overall data flow diagram of this embodiment is shown in fig. 3.
Example 3
The congestion control algorithm used by the invention is a reno congestion control algorithm, a bbr congestion control algorithm, a westwood congestion control algorithm, a hybla congestion control algorithm, a vegas congestion control algorithm and a cubic congestion control algorithm, and is installed through the configuration script provided by the invention. The decision tree uses a CART decision tree. The training data of the decision tree is derived from the measured results. By continuously changing network conditions (network bandwidth, network delay RTT and network packet loss rate), adjusting a congestion control algorithm, testing the TCP throughput rate through an iperf tool, and then comparing and marking actual throughput rates of the six algorithms in the same network environment, the training data of the decision tree can be obtained.
The invention compares the congestion control algorithm selected by the decision tree with the congestion control algorithm of the current link, if the algorithm is different, the switching is carried out, and if the algorithm is the same, the algorithm does not need to be changed.
On a link with the bandwidth of 100Mbps, an intercepted segment of network monitoring data is time sliced every 10s, the variation of the network round trip delay RTT is shown in FIG. 5, and the unit of the ordinate is ms; the packet loss rate change is shown in fig. 6, which shows that the packet loss rate value is a percentage value, the network delay starts to increase suddenly at a certain time point due to the sudden change of factors such as the traffic volume and the like in the link state, and the packet loss rate also changes, and a TCP link throughput rate graph using the model provided by the present invention and the fixed TCP congestion control algorithm on the TCP link in the network condition is shown in fig. 7.
Before the generation of the data shown in fig. 5 and fig. 6, the construction of the prediction model ARIMA (p, d, q) model is completed according to the collected link network data, and a formula of a time series with the average value of μ is generated:
wherein y is t And a t Representing the true variable value at time t and the corresponding random error.
Are polynomials of order p and q of B.
And theta j (j ═ 1,2, … …, q) are model parameters.
B is the move back operator. The orders p and q of the model and the difference order d are integers. Random error a t Are assumed to be independent and identical distributions, with an average value of zero,constant variance is σ 2 . Therefore, the model provided by the invention can continuously predict the network time slicing. The training of the CART decision tree is also completed by the construction in the manner mentioned in example 2.
In the network change process, the model provided by the invention monitors the RTT and the packet loss rate of the link in a time slice of every 10s, and put into the trained ARIMA model for prediction, put the output predicted RTT and packet loss rate of the next time slice and network bandwidth into the decision tree model, select the most suitable congestion control algorithm for the next time slice, as shown in fig. 7, before the network condition changes, the model provided by the invention (the method shown as o in fig. 7) selects the Cubic algorithm and Reno as the congestion control algorithm most suitable for the link, because the network delay and the packet loss rate of the segment are both very low and have little change, when the network condition changes suddenly, the packet loss rate and the network delay are increased and changed continuously, at the moment, the model provided by the invention predicts the trend of the change of the network environment through a prediction model, and selects the most appropriate congestion control algorithm for the link after the change.
The model provided by the invention correctly selects the most appropriate congestion control algorithm as the BBR algorithm under the network environment with higher network delay of 100-300ms and lower packet loss rate of 5 percent as shown in fig. 5 and 6, thereby achieving extremely high throughput rate before and after network environment mutation. If the BBR algorithm is used for the link all the time, the best throughput rate cannot be achieved before network fluctuation, and if the Cubic or Reno algorithm is used for the link all the time, the highest throughput rate cannot be achieved after the network fluctuation.
It should be noted that the above-mentioned contents only illustrate the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and it is obvious to those skilled in the art that several modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations fall within the protection scope of the claims of the present invention.
Claims (8)
1. The active selection method of the decision tree-based TCP congestion control algorithm is characterized by comprising the following steps:
s1, data monitoring: the method comprises the steps of keeping monitoring on the round trip time RTT and the packet loss rate of a TCP link network, and acquiring the round trip time RTT and the packet loss rate data of a link in time slices;
s2, data prediction: predicting the predicted round trip time RTT and the packet loss rate of the next fixed time slice by using a difference integration moving average autoregressive (ARIMA) model according to the round trip time RTT and the packet loss rate data obtained in the step S1; in the ARIMA (p, d, q) model, by inputting N groups of monitoring data, a time sequence with the average value of mu is generated according to the following formula:
wherein y is t And a t Representing the real variable value at time t and the corresponding random error;
are polynomials of order p and q of B;
and theta j (j ═ 1,2, … …, q) are model parameters;
b is the move-back operator; the orders p and q of the model and the differential order d are integers; random error a t Are assumed to be independent and identical distributions with a mean of zero and a constant variance of σ 2 ；
S3, decision selection: through a decision tree model, taking the predicted round trip time RTT and the predicted packet loss rate as input data, outputting the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and if the decision results of the current link congestion control algorithm and the decision tree are different, switching; otherwise, no handover is performed.
2. The method for proactive decision tree-based TCP congestion control algorithm according to claim 1, characterized by: in step S3, the decision tree model includes a set of constructed samples and a CART-trained decision tree:
when a sample set is constructed, comparing throughput rates of TCP links of different congestion control algorithms in the same network environment, marking the congestion control algorithm with the highest throughput rate in each network environment, and taking the marked data as training data of a CART decision tree;
when training the CART decision tree, for the sample set S, the kini coefficient is defined
For the attribute A, the Gain _ GINI after the data set is divided into two parts by respectively calculating any attribute value, and the minimum value min (Gain GINI) is selected A,i (S)) i belongs to the optimal bipartite scheme obtained by taking A as the attribute A; for the sample set S, calculating the optimal bipartite scheme of all attributes, and selecting the minimum value min (GainGINI) of the optimal bipartite scheme A,i (S)) i belongs to A) A belongs to Attributes and serves as an optimal bipartite scheme of the sample set S; the obtained attribute A and the attribute value i thereof are the optimal splitting attribute and the optimal splitting value of the sample set S;
when the CART algorithm constructs the decision tree, the process of selecting the attributes and the attribute values is repeated, and the current optimal scheme is continuously selected as the classification rule of the decision nodes until the example data set can not be divided again.
3. The method for proactive decision tree-based TCP congestion control algorithm according to claim 1, characterized by: in step S1, the packet loss rate of the TCP link is detected by using the ICMP message, a fixed number of ICMP messages are sent per second by using the ping procedure, and the network packet loss rate of the TCP link is calculated according to the received response.
4. A method for proactive decision tree based TCP congestion control algorithm according to claim 2 or 3, characterized by: in step S2, the predicted value of the network packet loss rate in the next fixed time slice is obtained through linear processing of the parameter α by using the predicted value of the network packet loss rate and the previous measured value of the actual packet loss rate, where the predicted packet loss rate SLOSS specifically is:
SLOSS＝α*LOSS+(1-α)*LOSS，
wherein SLOSS is the predicted packet LOSS rate, and LOSS is the actual packet LOSS rate.
5. The method of proactive decision tree-based TCP congestion control algorithm according to claim 4, characterized by: in step S2, a plurality of sets of averaging methods are used to calculate the round trip time RTT.
6. The method for proactive decision tree-based TCP congestion control algorithm according to claim 6, characterized by: the proactively selectable or switchable congestion control algorithms include at least reno, bbr, westwood, hybla, vegas, and cubic algorithms.
7. The method for proactive decision tree-based TCP congestion control algorithm according to claim 6, characterized by: the time slice is 10 seconds.
8. The active selection system of the decision tree-based TCP congestion control algorithm is characterized by comprising a monitoring module, a prediction module and a decision module,
the monitoring module is used for keeping monitoring the round trip time RTT and the packet loss rate of the TCP link network and acquiring the round trip time RTT and the packet loss rate data of the link according to fixed time slices;
the prediction module predicts the predicted round trip time delay RTT and the packet loss rate of the next fixed time slice by using a difference integration moving average autoregressive model according to the round trip time delay RTT and the packet loss rate data acquired by the monitoring module;
the decision module takes the round trip time RTT and the predicted packet loss rate acquired by the prediction module as input data through a decision tree model, outputs the most appropriate congestion control algorithm in the next fixed time slice of the TCP link, and switches if the decision results of the current link congestion control algorithm and the decision tree are different; otherwise, the switching is not carried out.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210519657.3A CN114979010A (en) | 2022-05-13 | 2022-05-13 | Active selection system and method of TCP congestion control algorithm based on decision tree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210519657.3A CN114979010A (en) | 2022-05-13 | 2022-05-13 | Active selection system and method of TCP congestion control algorithm based on decision tree |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114979010A true CN114979010A (en) | 2022-08-30 |
Family
ID=82982780
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210519657.3A Pending CN114979010A (en) | 2022-05-13 | 2022-05-13 | Active selection system and method of TCP congestion control algorithm based on decision tree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114979010A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105703954A (en) * | 2016-03-17 | 2016-06-22 | 福州大学 | Network data flow prediction method based on ARIMA model |
| CN113300969A (en) * | 2021-05-20 | 2021-08-24 | 清华大学深圳国际研究生院 | Congestion control switching method based on scene change, storage medium and electronic equipment |
| CN113872873A (en) * | 2021-09-29 | 2021-12-31 | 西安交通大学 | Multi-scene cross-layer congestion control method suitable for 5G new application |
| CN114422443A (en) * | 2022-01-24 | 2022-04-29 | 西安电子科技大学 | Satellite network TCP congestion control method based on bandwidth estimation and congestion prediction |
-
2022
- 2022-05-13 CN CN202210519657.3A patent/CN114979010A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105703954A (en) * | 2016-03-17 | 2016-06-22 | 福州大学 | Network data flow prediction method based on ARIMA model |
| CN113300969A (en) * | 2021-05-20 | 2021-08-24 | 清华大学深圳国际研究生院 | Congestion control switching method based on scene change, storage medium and electronic equipment |
| CN113872873A (en) * | 2021-09-29 | 2021-12-31 | 西安交通大学 | Multi-scene cross-layer congestion control method suitable for 5G new application |
| CN114422443A (en) * | 2022-01-24 | 2022-04-29 | 西安电子科技大学 | Satellite network TCP congestion control method based on bandwidth estimation and congestion prediction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10171332B2 (en) | Probing technique for predictive routing in computer networks | |
| Koksal et al. | Quality-aware routing metrics for time-varying wireless mesh networks | |
| WO2012117549A1 (en) | Failure analysis device, and system and method for same | |
| Pokhrel et al. | Adaptive admission control for IoT applications in home WiFi networks | |
| US11019151B2 (en) | Message schema control | |
| Zhuang et al. | Data collection with accuracy-aware congestion control in sensor networks | |
| CN114629802B (en) | Service awareness-based power communication backbone network quality assessment method | |
| US10142201B2 (en) | Data acquisition using remote platform gateways | |
| US20170195238A1 (en) | Maximize Network Capacity Policy With Heavy-Tailed Traffic | |
| Steinert et al. | Long-term adaptation and distributed detection of local network changes | |
| Boucetta et al. | QoS in IoT networks based on link quality prediction | |
| Balaraju et al. | Dynamic adaptation of the user data flow to the changing data rates in VHF networks: An exploratory study | |
| US10764219B2 (en) | Message schema control | |
| CN114726723A (en) | Real-time performance adjustment of network devices | |
| CN114979010A (en) | Active selection system and method of TCP congestion control algorithm based on decision tree | |
| Sun et al. | Effect of large buffers on TCP queueing behavior | |
| Bansal et al. | An extension of the TCP flow control algorithm for wireless networks | |
| CN112583818A (en) | Adaptive transport protocol selection method and device for mobile Web service | |
| Abdul-Jabbar et al. | A new wavenet-based network congestion predictor-wbcp | |
| Iri et al. | Congestion-adaptive data collection with accuracy guarantee in cyber-physical systems | |
| Tse et al. | Estimating contention of ieee 802.11 broadcasts based on inter-frame idle slots | |
| Li et al. | Using multiple metrics for rate adaptation algorithms in IEEE 802.11 WLANs | |
| Curran et al. | Sample: Statistical network link modelling in an on-demand probabilistic routing protocol for ad hoc networks | |
| CN116260695B (en) | Comprehensive evaluation method and system for computer network health degree | |
| US11973635B2 (en) | Method and system for detecting incidents in at least one local area communication 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220830 |