CN113141318A - Improved TFRC congestion control system and method based on video traffic prediction model - Google Patents
Improved TFRC congestion control system and method based on video traffic prediction model Download PDFInfo
- Publication number
- CN113141318A CN113141318A CN202110334863.2A CN202110334863A CN113141318A CN 113141318 A CN113141318 A CN 113141318A CN 202110334863 A CN202110334863 A CN 202110334863A CN 113141318 A CN113141318 A CN 113141318A
- Authority
- CN
- China
- Prior art keywords
- video
- sending
- sending rate
- rate
- data
- 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.)
- Granted
Links
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
- H04L47/25—Flow control; Congestion control with rate being modified by the source upon detecting a change of network conditions
-
- 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
-
- 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 requests to protect an improved TFRC congestion control system and method based on a video flow prediction model, which comprises the following steps: the device comprises a video flow data acquisition module, a video flow sending rate prediction module and a data packet sending rate adjustment module, wherein the video flow data acquisition module adopts packet capturing software to acquire the sending rate of video data packets per second; the video data packet sending rate prediction module predicts the video packet sending rate by adopting a combined prediction model of wavelet transformation and a neural network gated cyclic unit GRU, and predicts the video traffic sending rate by taking the historical sending rate and the influence parameters as the input of the prediction model; and the data packet sending rate adjusting module takes the sending rate predicted value as a reference factor for adjusting the sending rate of the video sending end by the TFRC protocol at the next moment, and the reference factor and the estimated value calculated by the original TFRC jointly act in the adjustment of the sending rate to adjust the video flow sending rate of the video sending end.
Description
Technical Field
The invention belongs to the technical field of congestion control in a communication network, and particularly relates to an improved TFRC congestion control system and method of a video flow prediction model.
Background
Data transmission over the Internet is mainly based on TCP and UDP. For network video transmission, the retransmission mechanism of TCP will increase the delay of the receiving end, and is not necessary; the TCP adopts a congestion back-off mechanism with a halved rate, which easily causes severe fluctuation of data transmission rate, resulting in delay and discontinuity of video pictures. UDP does not have any congestion control mechanism, and in a congested network environment, UDP streams will largely occupy the network bandwidth of TCP streams, and do not have the characteristic of TCP friendliness (TCP-friendly), and meanwhile, the packet loss rate of UDP itself will also be rapidly increased, and the risk of system congestion and breakdown may be brought. Neither TCP nor UDP can meet the requirements of network video transmission. To solve the above problems, Floyd et al proposed a TFRC (TCPflorandly rate control) protocol. The TFRC protocol employs an end-to-end, TCP throughput model-based rate control mechanism. Under the same condition, the TFRC flow can share the network bandwidth with the TCP flow fairly; on the other hand, TFRC flows have stable throughput changes and little jitter, and are therefore very suitable for applications requiring high requirements for smoothness and real-time performance of transmission rates.
The biggest benefit of TFRC is that it can preserve a relatively stable sending rate and it can adapt to changes in network bandwidth. However, when the network status changes rapidly, the throughput is low, and even if the transmission bandwidth remains in a stable state, before the TFRC receives the next parameter feedback, the TFRC adjusts the transmission rate by using the throughput calculated by the previous parameter as a target value, if a network fluctuation occurs during the period, and the TFRC still adjusts the throughput calculated by the previous situation that the network fluctuation does not occur, the increase of the transmission rate collides with the fluctuating network, which causes the situation that the transmission rate is reduced to avoid the congestion situation, thereby causing the fluctuation of the video. Based on the above limitations, the present invention provides an improved TFRC congestion control strategy based on a video traffic prediction model.
Disclosure of Invention
The present invention is directed to solving the above problems of the prior art. An improved TFRC congestion control system and method based on a video flow prediction model are provided. The technical scheme of the invention is as follows:
an improved TFRC congestion control system based on a video traffic prediction model, comprising: a video flow data acquisition module, a video flow sending rate prediction module and a data packet sending rate adjustment module, wherein,
the video flow data acquisition module adopts packet capturing software to acquire the sending rate of video data packets per second;
the video data packet sending rate prediction module predicts the video packet sending rate by adopting a combined prediction model of wavelet transformation and a neural network gated cyclic unit GRU, takes the historical sending rate and network throughput data, delay condition, packet loss rate and other influence parameters under the current network as the input of the prediction model, and predicts the video traffic sending rate;
the data packet sending rate adjusting module takes the sending rate predicted value as a reference factor for adjusting the sending rate of the video sending end by the TFRC protocol at the next moment, and the reference factor and the estimated value calculated by the original TFRC jointly act on the adjustment of the sending rate to adjust the video flow sending rate of the video sending end.
Further, the video traffic data obtaining module obtains a video traffic packet size, a video traffic sending rate, a network throughput, a time delay and a packet loss rate of a video sending end in the current network, wherein the video traffic packet size refers to a data packet size obtained by compressing and packaging video data by the video sending end; the video flow sending rate refers to the size of data volume sent by a video sending end per second; network throughput refers to the maximum data rate that a device can receive and forward without frame loss; the delay refers to the time used for transmission in the transmission medium, i.e. the time from the beginning of the message entering the network to the beginning of the message leaving the network; the packet loss rate is the ratio of the number of lost packets in the transmission process to the transmission data.
Furthermore, the packet capturing software adopted by the video traffic data acquisition module is that WireShark captures packet of data related to sending rate of a sending end, collects characteristic data, and then continuously decomposes the characteristic data into high-frequency and low-frequency components through wavelet transformation; wherein, the wavelet transformation function selects Symlets wavelet function to decompose and reconstruct data.
Further, the video traffic sending rate prediction module is built based on wavelet transformation and a GRU neural network, wherein the wavelet transformation preprocesses collected characteristic data and then serves as input data of the GRU neural network, and Symlets wavelets are selected through the wavelet transformation; the GRU neural network consists of an input layer, a hidden layer and an output layer, wherein the hidden layer is divided into two layers, MSE is selected as a loss function, and the formula is
Wherein N represents the number of samples,
represents the mean value of the samples, xtRepresenting the sample value and t the sample number. MSE is a measure of the degree of difference between the predicted and actual values, given that a is the predicted value and b is the actual value, (a-b)2For the mean square error called predicted value a, the optimizer selects adam; secondly, the input layer adopts the acquired characteristic data as input after wavelet transformation, the dimensionalities of the hidden layer are 128 and 64 respectively, the output result of the output layer is the video flow sending rate of the sending end at the future moment, the dropout rate is set to be 0.02, and the iteration number is set to be 1000.
Furthermore, the prediction process of the combined prediction model is divided into three steps, the first step is to decompose the data into two parts of high frequency and low frequency by using wavelet transformation, the second step is to train the low frequency data and the high frequency data respectively by using a GRU neural network and predict the result, the third step is to merge the prediction results of the high frequency component and the low frequency component by using the wavelet transformation to finally form a prediction result, the GRU neural network can make prediction at the current time step, and the prediction information is used as the input information of the next prediction, so that the prediction is continued forward.
Further, the data packet sending rate adjusting module is used for avoiding congestion when the network enters the congestion stateAfter the stage-free, the sending end will predict the sending rate muWSGInitial sending rate muinitAnd the transmission rate mu derived by the TFRC protocolTMaking a comparison if muT>μWSG>μinitThen the predicted transmission rate mu isWSGThe upper limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muWSG>μT>μinitThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate is adjusted as the upper limit of the transmission rate adjustment of the transmitting end; if μinit>μWSG>μTThen the predicted transmission rate mu isWSGThe lower limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muinit>μT>μWSGThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate adjustment is performed as a lower limit of the transmission rate adjustment of the transmitting end.
An improved TFRC congestion control method based on a video traffic prediction model, comprising the following steps: a video traffic data acquisition step, a video traffic transmission rate prediction step, and a packet transmission rate adjustment step, wherein,
video flow data acquisition: collecting the sending rate of video data packets per second by packet capturing software;
video data packet transmission rate prediction step: predicting the sending rate of a video packet by adopting a combined prediction model of wavelet transformation and a neural network gated loop unit GRU, and predicting the sending rate of video flow by taking the historical sending rate and the influence parameters of network throughput data, delay condition, packet loss rate and the like in the current network as the input of the prediction model;
data packet sending rate adjusting step: and taking the video flow sending rate predicted value as a reference factor for regulating the sending rate of the video sending end by the TFRC protocol at the next moment, and using the predicted value and the estimated value calculated by the original TFRC protocol together for regulating the sending rate to regulate the video flow sending rate of the video sending end.
Further, the video traffic sending rate predicting stepThe method comprises the following steps: preprocessing the collected characteristic data by adopting wavelet transformation, and then using the preprocessed characteristic data as input data of a GRU neural network, wherein Symlets wavelets are selected by the wavelet transformation; the GRU neural network consists of an input layer, a hidden layer and an output layer, wherein the hidden layer is divided into two layers, MSE is selected as a loss function, and the formula is
Wherein N represents the number of samples,
representing the mean value of the samples, xtRepresenting the sample value and t the sample number. MSE is a measure of the degree of difference between the predicted result and the actual value, given that a is the predicted result value and b is the actual value, (a-b)2For the mean square error, called the predicted value a, the optimizer selects adam; secondly, the input layer adopts the acquired characteristic data as input after wavelet transformation, the dimensionalities of the hidden layer are 128 and 64 respectively, the output result of the output layer is the video flow sending rate of the sending end at the future moment, the dropout rate is set to be 0.02, and the iteration number is set to be 1000.
Further, the step of adjusting the sending rate of the data packet is to predict the sending rate mu by the sending end after the network enters the congestion avoiding stageWSGInitial sending rate muinitAnd the transmission rate mu derived by the TFRC protocolTMaking a comparison if muT>μWSG>μinitThen the predicted transmission rate mu isWSGThe upper limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muWSG>μT>μinitThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate is adjusted as the upper limit of the transmission rate adjustment of the transmitting end; if μinit>μWSG>μTThen the predicted transmission rate mu isWSGThe lower limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muinit>μT>μWSGThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate adjustment is performed as a lower limit of the transmission rate adjustment of the transmitting end.
The invention has the following advantages and beneficial effects:
aiming at the problem that the response TCP speed of the TFRC protocol to the variable bandwidth is low; when the network environment is unstable, the TFRC scheme has the conditions of time delay and untimely adjustment on the reaction of the actual environment, by combining wavelet transform and GRU neural network, decomposing and reconstructing video flow data through wavelet transform is used as input of a prediction model, a prediction result is used as a reference factor of a TFRC protocol for adjusting sending rate of a sending end, an improved TFRC congestion control strategy based on a video flow prediction model is provided, the current video flow sending rate condition and the video flow sending rate condition of the next stage of prediction can be utilized, thereby the TFRC adjusts the sending rate more timely and accurately, has more TCP friendliness, achieves better congestion control effect, better ensures the throughput of the video sending end, and the stability of the time delay jitter is better ensured to a certain extent, and the packet loss rate is lower.
Drawings
FIG. 1 is a diagram of a video traffic transmission rate prediction model for video traffic prediction according to a preferred embodiment of the present invention;
fig. 2 is a flow chart of the improved TFRC congestion control strategy based on the video traffic prediction model of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described in detail and clearly with reference to the accompanying drawings. The described embodiments are only some of the embodiments of the present invention.
The technical scheme for solving the technical problems is as follows:
the invention relates to a method for predicting video flow sending rate based on a video flow prediction model of a GRU neural network after wavelet transformation processing of data, wherein a prediction result is used as a reference factor for regulating the sending rate of a video sending end by a TFRC protocol at the next moment, and the prediction result and a pre-estimated value calculated by an original TFRC jointly act in regulating the sending rate so as to achieve the improvement and optimization of a TFRC congestion control strategy and obtain better video transmission QoS.
An improved TFRC congestion control strategy based on a video flow prediction model mainly comprises the following steps: the device comprises a video flow data acquisition module, a video flow sending rate prediction module and a data packet sending rate adjustment module; wherein:
the acquisition module of the video flow data adopts packet capturing software such as WireShark to collect the sending rate of video data packets per second;
the prediction module of the video data packet sending rate predicts the video packet sending rate by using a combined prediction model of wavelet transformation and a neural network gated loop unit GRU (gated RecurrentUnit), and predicts the video flow sending rate by taking the historical sending rate and some influence parameters as the input of the prediction model.
The module for adjusting the sending rate of the data packet takes the predicted value as a reference factor for adjusting the sending rate of the video sending end by the TFRC protocol at the next moment, and the predicted value calculated by the original TFRC is jointly acted on the adjustment of the sending rate to adjust the sending rate of the video flow of the video sending end.
Further, the module for acquiring video traffic data mainly includes the size of a video traffic packet of the video sending end, the sending rate of the video traffic, the network throughput under the current network, the delay and the packet loss rate, wherein the size of the video traffic packet refers to the size of a data packet after the video sending end compresses and packages the video data; the video flow sending rate refers to the size of data volume sent by a video sending end per second; network throughput refers to the maximum data rate that a device can receive and forward without frame loss; the delay refers to the time used for transmission in a transmission medium, i.e. the time from the beginning of a message entering the network to the beginning of its leaving the network; the packet loss rate is the ratio of the number of lost packets in the transmission process to the transmission data.
Further, the prediction module of the video traffic sending rate is built based on wavelet transformation and a GRU neural network. Wherein the wavelet transform is paired upPreprocessing the collected characteristic data in the step, and then using the preprocessed characteristic data as input data of a GRU neural network, wherein Symlets wavelets are selected through wavelet transformation; the GRU neural network consists of an input layer, a hidden layer and an output layer, wherein the hidden layer is composed of two layers, MSE is selected as a loss function, and the formula is
MSE is a measure reflecting the degree of difference between the predicted result and the actual value, and assuming that a is the predicted result value and b is the actual value, (a-b)2The optimizer selects adam for the mean square error, referred to as the predictor a. Secondly, the input layer adopts the acquired characteristic data as input after wavelet transformation, the dimensionalities of the hidden layer are 128 and 64 respectively, the output result of the output layer is the video flow sending rate of the sending end at the future moment, the dropout rate is set to be 0.02, and the iteration times are set to be 1000.
Furthermore, the prediction flow of the prediction model is divided into three steps, the first step is to use wavelet transformation to decompose the data into two parts of high frequency and low frequency, the second step is to use GRU neural network to train the low frequency and high frequency data respectively and predict the result, and the third step is to use wavelet transformation to combine the prediction results of high frequency and low frequency components to finally form the prediction result. The GRU neural network makes a prediction at the current time step, and the prediction information is used as input information of the next prediction, so that the prediction is continued forwards.
Further, the module for adjusting the sending rate of the data packet is that after the network enters the congestion avoidance stage, the sending end predicts the sending rate muWSGInitial sending rate muinitAnd the transmission rate mu derived by the TFRC protocolTMaking a comparison if muT>μWSG>μinitThen the predicted transmission rate mu isWSGThe upper limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muWSG>μT>μinitThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate is adjusted as the upper limit of the transmission rate adjustment of the transmitting end; if μinit>μWSG>μTThen the predicted transmission rate mu isWSGThe lower limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muinit>μT>μWSGThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate adjustment is performed as a lower limit of the transmission rate adjustment of the transmitting end.
As shown in fig. 1, step 1 and step 2 still use the TFRC protocol to control the sending rate of the sending end, and the data sending process of the TFRC protocol is divided into two stages, namely slow start and congestion avoidance. Initially entering a slow start phase after a connection is established, the sender first sends packets at a very low initial rate and then multiplies the rate of sending approximately every RTT time period. When the receiving end detects the message loss event, the sending end enters the congestion avoiding stage after receiving the feedback data packet; (b) if the feedback timer is overtime, namely the sending end does not receive the feedback message after the timer is overtime, the sending rate is directly halved, and the congestion avoiding stage is entered.
And 3, adopting a professional packet grabbing tool WireShark to grab the packet of the data related to the sending rate of the sending end, collecting the characteristic data, and continuously decomposing the characteristic data into high-frequency and low-frequency components through wavelet transformation. Wherein, the wavelet transformation function selects Symlets wavelet function to decompose and reconstruct the data.
The step 4 is to predict the video traffic by using a GRU model as shown in fig. 1, and the predicting step is as follows:
1. establishment of prediction model
The GRU prediction model is composed of an input layer, a hidden layer and an output layer, wherein a four-layer GRU model is adopted, a gated circulation unit GRU (gated Recurrent Unit) is adopted, a gating mechanism is adopted to control information such as input and memory so as to make prediction at the current time step, the prediction information is used as input information of next prediction, the prediction is continued forwards, two hidden layers are selected, and the hidden layers have the dimensions of 128 and 64 respectively.
2. Model set-up
The model mainly takes the sending rate of a sending end as input, can be obtained by WireShark packet capturing software, adjusts the sending rate into the input of samples, step length and characteristics, adopts a gated cyclic unit GRU (gated Recurrent Unit), adopts a gating mechanism to control the input, memory and other information, and makes prediction at the current time step
MSE is used as a loss function, the dropout rate is set to be 0.02, the iteration times are set to be 1000, an adam optimizer is used, and the learning rate is an adam self-adaptive mechanism.
3. Model implementation
And constructing a model by adopting a keras platform, standardizing and normalizing the captured data and performing wavelet transformation, constructing a training model by utilizing the keras platform, training and predicting the current model, and finally performing wavelet transformation reconstruction on the prediction results of the high-frequency component and the low-frequency component to obtain a final prediction result.
The step 5 is to use the video flow prediction value as a reference factor for the TFRC protocol to adjust the sending rate of the video sending end at the next moment, and act together with the prediction value calculated by the original TFRC in adjusting the sending rate, and the specific rate adjustment strategy is as follows:
wherein μ is the adjusted rate, μinitFor the initial transmission rate of video, muTCalculated for the TFRC algorithm formula, the rate, muWSGIs the rate predicted by the model. The sending end will predict the sending rate muWSGInitial sending rate muinitAnd the sending rate mu derived by the TFRC protocolTMaking a comparison if muT>μWSG>μinitThen the predicted transmission rate mu isWSGThe sending rate is adjusted as the upper limit of sending rate adjustment of the sending end, if muWSG>μT>μinitThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate is adjusted as the upper limit of the transmission rate adjustment of the transmitting end; if μinit>μWSG>μTThen the predicted transmission rate mu isWSGThe lower limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muinit>μT>μWSGThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate adjustment is performed as a lower limit of the transmission rate adjustment of the transmitting end.
The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above examples are to be construed as merely illustrative and not limitative of the remainder of the disclosure. After reading the description of the invention, the skilled person can make various changes or modifications to the invention, and these equivalent changes and modifications also fall into the scope of the invention defined by the claims.
Claims (9)
1. An improved TFRC congestion control system based on a video traffic prediction model, comprising: a video flow data acquisition module, a video flow sending rate prediction module and a data packet sending rate adjustment module, wherein,
the video flow data acquisition module adopts packet capturing software to acquire the sending rate of video data packets per second;
the video data packet sending rate prediction module predicts the video packet sending rate by adopting a combined prediction model of wavelet transformation and a neural network gated cyclic unit GRU, takes the historical sending rate and the network throughput data, delay condition, packet loss rate and other influence parameters in the current network as the input of the prediction model, and predicts the video traffic sending rate;
and the data packet sending rate adjusting module takes the sending rate predicted value as a reference factor for adjusting the sending rate of the video sending end by the TFRC protocol at the next moment, and the reference factor and the estimated value calculated by the original TFRC jointly act in the adjustment of the sending rate to adjust the video flow sending rate of the video sending end.
2. The improved TFRC congestion control system based on the video traffic prediction model according to claim 1, wherein the video traffic data obtaining module obtains a video traffic packet size, a video traffic sending rate, a network throughput, a delay and a packet loss rate of a current network at a video sending end, wherein the video traffic packet size is a data packet size obtained by compressing and packing video data at the video sending end; the video flow sending rate refers to the size of data volume sent by a video sending end per second; network throughput refers to the maximum data rate that a device can receive and forward without frame loss; the delay refers to the time used for transmission in the transmission medium, i.e. the time from the beginning of the message entering the network to the beginning of the message leaving the network; the packet loss rate is the ratio of the number of lost packets in the transmission process to the transmission data.
3. The improved TFRC congestion control system based on the video traffic prediction model as claimed in claim 2, wherein the packet capturing software adopted by the video traffic data acquisition module is WireShark to capture the data related to the sending rate of the sending end, collect the characteristic data, and then continue decomposing the characteristic data into high frequency and low frequency components through wavelet transformation; wherein, the wavelet transformation function selects Symlets wavelet function to decompose and reconstruct the data.
4. The improved TFRC congestion control system based on the video traffic prediction model according to claim 1, 2 or 3, wherein the video traffic sending rate prediction module is constructed based on wavelet transform and GRU neural network, wherein the wavelet transform preprocesses collected feature data and uses the preprocessed feature data as input data of the GRU neural network, and Symlets wavelet is selected by the wavelet transform; the GRU neural network consists of an input layer, a hidden layer and an output layer, wherein the hidden layer is divided into two layers, MSE is selected as a loss function, and the formula is
Wherein N represents the number of samples,
represents the mean value of the samples, xtRepresenting the sample value and t the sample number. MSE is a measure of the degree of difference between the predicted and actual values, given that a is the predicted value and b is the actual value, (a-b)2For the mean square error, called the predicted value a, the optimizer selects adam; secondly, the input layer adopts the acquired characteristic data as input after wavelet transformation, the dimensionalities of the hidden layer are 128 and 64 respectively, the output result of the output layer is the video flow sending rate of the sending end at the future moment, the dropout rate is set to be 0.02, and the iteration number is set to be 1000.
5. The improved TFRC congestion control system based on the video traffic prediction model as claimed in claim 4, wherein the prediction process of the combined prediction model is divided into three steps, the first step is to use wavelet transform to decompose the data into two parts of high frequency and low frequency, the second part is to use GRU neural network to train the low frequency and high frequency data and predict the result, the third step is to use wavelet transform to combine the prediction results of high frequency and low frequency components to form the prediction result, the GRU neural network will make prediction at the current time step and use the prediction information as the input information of the next prediction to continue the prediction forward.
6. The improved TFRC congestion control system as claimed in claim 5, wherein the packet sending rate adjusting module is configured to predict the sending rate μ after the network enters the congestion avoidance phaseWSGInitial sending rate muinitAnd the sending rate mu derived by the TFRC protocolTMaking a comparison if muT>μWSG>μinitThen the predicted transmission rate mu isWSGThe sending rate is adjusted as the upper limit of sending rate adjustment of the sending end, if muWSG>μT>μinitThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate is adjusted as the upper limit of the transmission rate adjustment of the transmitting end; if μinit>μWSG>μTThen the predicted transmission rate mu isWSGThe lower limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muinit>μT>μWSGThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate adjustment is performed as a lower limit of the transmission rate adjustment of the transmitting end.
7. An improved TFRC congestion control method based on a video traffic prediction model is characterized by comprising the following steps: a video traffic data acquisition step, a video traffic transmission rate prediction step, and a packet transmission rate adjustment step, wherein,
video flow data acquisition: collecting the sending rate of video data packets per second by packet capturing software;
video data packet transmission rate prediction step: predicting the sending rate of a video packet by adopting a combined prediction model of wavelet transformation and a neural network gated loop unit GRU, and predicting the sending rate of video flow by taking the historical sending rate and the influence parameters of network throughput data, delay condition, packet loss rate and the like in the current network as the input of the prediction model;
data packet sending rate adjusting step: and taking the video flow sending rate predicted value as a reference factor for regulating the sending rate of the video sending end by the TFRC protocol at the next moment, and jointly acting the reference factor and the predicted value calculated by the original TFRC protocol in regulating the sending rate to regulate the video flow sending rate of the video sending end.
8. The improved TFRC congestion control method based on video traffic prediction model according to claim 7, wherein the video traffic sending rate prediction step specifically is: preprocessing the collected characteristic data by adopting wavelet transformation, and then using the preprocessed characteristic data as input data of a GRU neural network, wherein Symlets wavelets are selected by the wavelet transformation; the GRU neural network consists of an input layer, a hidden layer and an output layer, wherein the hidden layer is divided into two layers, MSE is selected as a loss function, and the formula is
Wherein N represents the number of samples,
represents the mean value of the samples, xtRepresenting the sample value and t the sample number. MSE is a measure of the degree of difference between the predicted and actual values, given that a is the predicted value and b is the actual value, (a-b)2For the mean square error, called the predicted value a, the optimizer selects adam; secondly, the input layer adopts the acquired characteristic data as input after wavelet transformation, the dimensionalities of the hidden layer are 128 and 64 respectively, the output result of the output layer is the video flow sending rate of the sending end at the future moment, the dropout rate is set to be 0.02, and the iteration number is set to be 1000.
9. The improved TFRC congestion control method based on video traffic prediction model as claimed in claim 7, wherein the packet transmission rate adjusting step,after the network enters the congestion avoidance stage, the sending end predicts the sending rate muWSGInitial sending rate muinitAnd the sending rate mu derived by the TFRC protocolTMaking a comparison if muT>μWSG>μinitThen the predicted transmission rate mu isWSGThe sending rate is adjusted as the upper limit of sending rate adjustment of the sending end, if muWSG>μT>μinitThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate is adjusted as the upper limit of the transmission rate adjustment of the transmitting end; if μinit>μWSG>μTThen the predicted transmission rate mu isWSGThe lower limit of the sending rate adjustment of the sending end is used for carrying out the sending rate adjustment, if muinit>μT>μWSGThen the sending rate mu obtained by the TFRC protocol is usedTThe transmission rate adjustment is performed as a lower limit of the transmission rate adjustment of the transmitting end.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110334863.2A CN113141318B (en) | 2021-03-29 | 2021-03-29 | Improved TFRC congestion control system and method based on video traffic prediction model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110334863.2A CN113141318B (en) | 2021-03-29 | 2021-03-29 | Improved TFRC congestion control system and method based on video traffic prediction model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113141318A true CN113141318A (en) | 2021-07-20 |
| CN113141318B CN113141318B (en) | 2023-03-31 |
Family
ID=76810104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110334863.2A Active CN113141318B (en) | 2021-03-29 | 2021-03-29 | Improved TFRC congestion control system and method based on video traffic prediction model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113141318B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114338834A (en) * | 2021-12-29 | 2022-04-12 | 重庆邮电大学 | Intelligent industrial protocol conversion system and method based on flow prediction |
| CN114339858A (en) * | 2021-12-30 | 2022-04-12 | 天翼物联科技有限公司 | Terminal packet sending parameter adjusting method and device and related equipment |
| CN115996195A (en) * | 2023-03-23 | 2023-04-21 | 腾讯科技（深圳）有限公司 | Data transmission method, device, equipment and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101075957A (en) * | 2007-06-26 | 2007-11-21 | 上海大学 | Method for controlling AVS fluid-medium transmission |
| US20100058147A1 (en) * | 2008-08-29 | 2010-03-04 | Hulya Seferoglu | Method and apparatus for reliable media transport |
| CN102592171A (en) * | 2011-12-30 | 2012-07-18 | 南京邮电大学 | Method and device for predicting cognitive network performance based on BP (Back Propagation) neural network |
| CN107483265A (en) * | 2017-09-11 | 2017-12-15 | 国网辽宁省电力有限公司 | A kind of network traffics dynamic prediction method and system based on wavelet analysis |
-
2021
- 2021-03-29 CN CN202110334863.2A patent/CN113141318B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101075957A (en) * | 2007-06-26 | 2007-11-21 | 上海大学 | Method for controlling AVS fluid-medium transmission |
| US20100058147A1 (en) * | 2008-08-29 | 2010-03-04 | Hulya Seferoglu | Method and apparatus for reliable media transport |
| CN102592171A (en) * | 2011-12-30 | 2012-07-18 | 南京邮电大学 | Method and device for predicting cognitive network performance based on BP (Back Propagation) neural network |
| CN107483265A (en) * | 2017-09-11 | 2017-12-15 | 国网辽宁省电力有限公司 | A kind of network traffics dynamic prediction method and system based on wavelet analysis |
Non-Patent Citations (6)
| Title |
|---|
| JEONG HA KIM ET AL.: ""Real-time image transmission on the TCP/IP network using wavelet transform and neural network"", 《IEEE》 * |
| 周鹏: ""视频传输QoS优化方案研究"", 《中国优秀硕士学位论文全文数据库》 * |
| 姚立霜: ""基于模态分解的网络流量预测模型研究"", 《中国优秀硕士学位论文全文数据库》 * |
| 林远华等: ""控制无线实时传输的改进TFRC算法"", 《计算机工程与设计》 * |
| 田妮莉等: "一种基于小波变换和FIR神经网络的广域网网络流量预测模型", 《电子与信息学报》 * |
| 邢钰莹等: "基于嵌入式的PSO-BP神经网络码率控制算法研究", 《电子测量技术》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114338834A (en) * | 2021-12-29 | 2022-04-12 | 重庆邮电大学 | Intelligent industrial protocol conversion system and method based on flow prediction |
| CN114338834B (en) * | 2021-12-29 | 2023-11-03 | 重庆邮电大学 | Intelligent industrial protocol conversion system and method based on flow prediction |
| CN114339858A (en) * | 2021-12-30 | 2022-04-12 | 天翼物联科技有限公司 | Terminal packet sending parameter adjusting method and device and related equipment |
| CN114339858B (en) * | 2021-12-30 | 2023-12-05 | 天翼物联科技有限公司 | Terminal packet sending parameter adjusting method and device and related equipment |
| CN115996195A (en) * | 2023-03-23 | 2023-04-21 | 腾讯科技（深圳）有限公司 | Data transmission method, device, equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113141318B (en) | 2023-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113141318B (en) | Improved TFRC congestion control system and method based on video traffic prediction model | |
| US20190222499A1 (en) | Network Data Flow Classification Method and System | |
| US10164893B2 (en) | Data transfer apparatus, data transfer controlling method and data stream | |
| CN104581422B (en) | A kind of method and apparatus transmitted for network data | |
| Chen | Network traffic modeling | |
| EP3528469A1 (en) | Adaptive restful real-time live media streaming | |
| EP3688938B1 (en) | Predictive scheduler | |
| CN106878192A (en) | A kind of data dispatching method of adaptive M PTCP | |
| Khan et al. | ISP can identify YouTube videos that you just watched | |
| Dai et al. | Statistical learning based congestion control for real-time video communication | |
| CN117114113B (en) | Collaborative reasoning acceleration method based on queuing theory | |
| CN107483990B (en) | Dynamic code rate adjusting method and device for streaming media transmission and transmission system | |
| WO2015044104A1 (en) | Video traffic management | |
| Togou et al. | An elastic DASH-based bitrate adaptation scheme for smooth on-demand video streaming | |
| Kofler et al. | Using MPEG-21 for cross-layer multimedia content adaptation | |
| CN114039931A (en) | Method, device, equipment and medium for controlling data transmission | |
| Saputri et al. | Learning Method of Performance-oriented Congestion Control (PCC) for Video Streaming Analysis | |
| Chellappa et al. | Adaptability between abr algorithms in dash video streaming and http/3 over quic: Research proposal | |
| Ye et al. | Wavelet-based VBR video traffic smoothing | |
| Khan et al. | QoE Evaluation of Legacy TCP Variants over DASH | |
| US10244205B1 (en) | Bandwidth estimation of a network connection | |
| Tadahal et al. | A Survey On Adaptive Bitrate Algorithms and Their Improvisations | |
| Wang et al. | Enhancing Low Latency Adaptive Live Streaming Through Precise Bandwidth Prediction | |
| PH12015000269A1 (en) | Data transfer apparatus, method of controlling data transfer, and a data stream | |
| CN118175101B (en) | Content distribution method for home intelligent terminal equipment without fixed IP |
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 |