CN105049369A - Video transmission congestion control method based on MPTCP in heterogeneous wireless network - Google Patents

Abstract

The invention discloses an MPTCP-based video transmission congestion control method in a heterogeneous wireless network. The method of the present invention first establishes a congestion prediction model at the sending end, predicts whether link congestion is about to occur according to the change trend of link round-trip delay, and adjusts the congestion window size accordingly; then establishes a bandwidth estimation model at the sending end, according to the adjacent The bandwidth estimation value is obtained from the receiving time of the ACK confirmation packet and the number of data bytes successfully received by the sending end, and the link quality is judged based on this; finally, a congestion window adjustment model is established on the sending end, according to the estimated bandwidth of each link and the packet loss situation , adjust the size of the congestion window and the congestion threshold accordingly. The invention realizes the distinction of link packet loss causes, and effectively solves the problem of low throughput of video streams in wireless channels with high packet loss rates.

Description

Congestion Control Method for Video Transmission Based on MPTCP in Heterogeneous Wireless Networks

technical field

The invention relates to the technical field of mobile communication, in particular to an MPTCP-based video transmission congestion control method in a heterogeneous wireless network.

Background technique

With the continuous development of wireless communication technology, the types of application services are constantly enriched. Today, the main body of wireless network transmission is no longer simple text, voice, images, etc., but replaced by streaming media used in live video, video conferencing, etc. Compared with traditional data services, professional video streaming has higher requirements on bandwidth, packet loss rate, end-to-end delay and jitter. The single-path transmission control protocol (Transport Control Protocol, TCP) is difficult to meet these requirements, mainly reflected in: even if there are multiple transmittable paths in the network, the TCP connection can only use one of them, and as the network data flow continues to increase, it is bound to Congestion of the transmission path is caused, and once the link fails, the TCP connection may be interrupted, affecting user experience.

Multi-Path Transmission Control Protocol (Multi-PathTCP, MPTCP) is an extension of traditional TCP, using multiple links for concurrent transmission, aggregating the transmission performance of multiple links, and effectively improving the utilization of network resources. However, with the continuous increase of the amount of network transmission data, congestion is a problem that must be solved. Congestion is a continuous network overload state. When the network supply cannot meet the needs of users, the network will be congested. The result of congestion is the increase of end-to-end delay and jitter, the increase of packet loss rate, and even the possibility of system crash. In order to solve this problem, many scholars have proposed different multi-channel concurrent transmission congestion control methods. However, in these methods, as long as the transmission link loses packets, it is considered that the link is congested, and the congestion window is correspondingly reduced to reduce the transmission rate. However, due to the influence of the external environment, the heterogeneous wireless network will also When packet loss occurs, reducing the congestion window will only reduce throughput, so these methods are not suitable for wireless video transmission. Therefore, how to propose a congestion control method suitable for heterogeneous wireless network video stream transmission according to the actual link transmission quality is of great significance.

Contents of the invention

The object of the present invention is to propose a video transmission congestion control method based on MPTCP in a heterogeneous wireless network for the deficiency of the existing congestion control method. The one-way transmission delay and packet loss obtained by the receiving end are fed back to the sending end through the TCP protocol. The congestion prediction module of the sending end predicts the link congestion situation according to the change of the transmission delay of each link, and adjusts the link congestion window in advance. size, to a certain extent avoid congestion. Afterwards, the bandwidth estimation module at the sending end estimates the actual bandwidth of the link based on the number of data bytes sent by each link within a period of time, and adjusts the size of the congestion window according to the packet loss situation and the bandwidth estimation value to reduce the degree of congestion And the purpose of improving the transmission throughput.

To achieve the above object, the multi-channel congestion control method of the present invention includes a congestion prediction model, a bandwidth estimation model and a congestion window adjustment model;

The congestion prediction model is used to predict whether the transmission link is about to be congested, specifically: the round-trip time delay (Round-TripTime, RTT) of the data packet is obtained through the ACK confirmation packet returned by the receiving end, and then the RTT and the The minimum RTT of the link is compared to obtain the difference. When the difference is greater than a certain threshold, it is considered that the link is about to be congested. At this time, the size of the link congestion window is appropriately reduced to avoid congestion.

The bandwidth estimation model is used to estimate the actual transmission bandwidth of the transmission link, specifically: by calculating the number of bytes of data packets successfully received by the receiving end within the time T when the sending end receives two adjacent ACK confirmation packets The ratio of T to this period of time is used as the sampling value of the link bandwidth, and then the sampling value is smoothed by the weighted average method, and the processed sampling value is used as the bandwidth estimation value of the link.

The congestion window adjustment model is used to adjust the congestion window of each link, and adjust the congestion window of each link according to the packet loss situation of each link and the estimated bandwidth value fed back by the receiving end.

The present invention is realized through the following technical solutions, and the concrete steps are:

Step 1. Establish a video multi-channel concurrent transmission model. A sending end and a receiving end are connected through multiple network interfaces to form multiple data transmission links, and the bandwidth and delay of each link are different.

Step 2. Establish a congestion prediction model. The specific process is as follows:

2-1. Obtain the round-trip delay RTT _m of the transmission link m through the timestamp information of the ACK confirmation packet returned by the receiving end, and calculate the round-trip delay RTT _m and the minimum round-trip delay minRTT _m during the transmission of the link The difference between θ _m .

2-2. Compare the difference θ _m with the congestion judgment threshold b _m , when θ _m > b _m , it is considered that the link is about to be congested, and the congestion window C _m is adjusted to C _m -const*C _m , where const As a constant, the congestion window is appropriately reduced to avoid congestion, and at the same time, the throughput will not be greatly affected.

Step 3. Establish a bandwidth estimation model to obtain the actual bandwidth estimation value of each link; the specific process is as follows:

3-1. Suppose the time t _{k-1, m} of receiving the last ACK confirmation packet, the time t _k , m of receiving the current ACK confirmation packet, and the time difference between two adjacent ACK confirmation packets t _{k, m} -t _{k -1, the number of data bytes successfully received by the receiving end within m} d _{k, m} , to obtain the sampling value of the transmission link bandwidth:

3-2. The bandwidth sampling value is smoothed by the weighted average method to obtain the bandwidth estimation value:

${\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\frac{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; *</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}$

Among them, BW _{k, m} , BW _{k-1, m} are the bandwidth estimation values at the current time and the previous time on the link m respectively, sample_BW _{k, m} , sample_BW _{k-1, m} are respectively the current time and the previous time on the link m Time bandwidth sampling value, p is the weighting factor.

Step 4. Establish a congestion window adjustment model, and adjust the congestion window size according to the estimated bandwidth of each link and the packet loss situation. The specific process is as follows:

4-1. When receiving a new ACK confirmation packet, compare the size of the congestion window C _m with the congestion threshold S _m , if C _m ≤ S _m , then C _m =C _m +1;

4-2. If C _m >S _m , then where BW _{k, total} is the sum of all link bandwidth estimates at the current moment;

4-3. If three identical ACK confirmation packets are received, it is considered that packet loss occurs on the link, and the congestion threshold and congestion window value are adjusted accordingly: $\left\{\begin{array}{c}{S}_m=\frac{{\mathrm{BW}}_{k,m}}{{\mathrm{BW}}_{\max ,m}}*C_m\\ C_m=S_m\end{array}\right.,$ Among them, BW _{max, m} is the estimated maximum bandwidth value that occurs during the transmission of link m, and then data retransmission is performed.

4-4. If timeout occurs, adjust the congestion threshold and congestion window: $\left\{\begin{array}{c}{S}_m=\frac{{\mathrm{BW}}_{k,m}}{{\mathrm{BW}}_{\max ,m}}*C_m\\ C_m=1\end{array}\right.,$ Afterwards, the data is retransmitted.

Step 5. Repeat steps 2 to 4 until the transmission of the wireless video service is completed.

Compared with the existing multi-channel congestion control method, the advantages of the present invention are as follows:

1. Most of the existing multi-channel congestion control methods are to greatly reduce the size of the congestion window after the congestion occurs, and then retransmit the lost data packets, and the retransmission will inevitably lead to a decrease in network transmission throughput. The present invention predicts link quality changes according to RTT changes before congestion occurs, and appropriately reduces the size of the congestion window to avoid congestion and data retransmission if the predicted link is about to be congested.

2. In most of the existing multi-channel congestion control methods, as long as packet loss occurs, the link is considered to be congested, and the congestion window is greatly reduced. However, in a wireless network, packet loss is mostly caused by the instability of the wireless channel, and the link quality is better at this time. In the present invention, after packet loss, the congestion window is set to If it is wireless packet loss, the link quality is better at this time, then BW _{k, m} ≈ BW _{max, m} , and the congestion window basically does not change; if congestion occurs, the difference between BW _{k, m} and BW _{max, m} is large, The congestion window will be drastically reduced. Therefore, the present invention can better distinguish between congestion packet loss and wireless packet loss, and make reasonable congestion window adjustments.

Description of drawings

FIG. 1 is a system block diagram of a video stream transmission congestion control method based on MPTCP.

Fig. 2 is a schematic diagram of multi-channel concurrent transmission of video streams in a heterogeneous wireless network.

FIG. 3 is a flowchart of an MPTCP-based video stream transmission congestion control algorithm.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments.

Aiming at the deficiency of the existing multi-channel congestion control method, the invention proposes an MPTCP-based video transmission congestion control method in a heterogeneous wireless network. It can analyze the network status and predict whether the link is about to be congested, so as to avoid congestion to a certain extent. In addition, the actual bandwidth of each link is estimated, and the size of the congestion window is adjusted according to the actual transmission quality of the link, thereby improving the transmission throughput of the video stream in the wireless network. The system block diagram of the video stream transmission congestion control method based on MPTCP in the heterogeneous wireless network is shown in Figure 1, including the following models:

The congestion prediction model is used to predict whether the transmission link is about to be congested, specifically: the round-trip time delay (Round-TripTime, RTT) of the data packet is obtained through the ACK confirmation packet returned by the receiving end, and then the RTT and the The minimum RTT of the link is compared to obtain the difference. When the difference is greater than a certain threshold, it is considered that the link is about to be congested. At this time, the size of the link congestion window is appropriately reduced to avoid congestion.

The bandwidth estimation model is used to estimate the actual transmission bandwidth of the transmission link, specifically: by calculating the number of bytes of data packets successfully received by the receiving end within the time T when the sending end receives two adjacent ACK confirmation packets The ratio of T to this period of time is used as the sampling value of the link bandwidth, and then the sampling value is smoothed by the weighted average method, and the processed sampling value is used as the bandwidth estimation value of the link.

The congestion window adjustment model is used to adjust the congestion window of each link, and adjust the congestion window of each link according to the packet loss situation of each link and the estimated bandwidth value fed back by the receiving end.

This example is implemented through the following steps:

Step 1: Establish a video multi-channel concurrent transmission model. As shown in Figure 2: the video server at the sending end and the smart phone at the receiving end are connected through two network interfaces to form two transmission links, Wi-Fi and 4G, to transmit video services.

Step 2: Establish a congestion prediction model, obtain the round-trip delay RTT _m of the transmission link m through the timestamp information of the ACK confirmation packet returned by the receiving end, and calculate the minimum round-trip delay minRTT _m between it and the link transmission process The difference between θ _m . Compare θ _m with the congestion judgment threshold b _m . When θ _m > b _m , it is considered that the link is about to be congested, and the congestion window C _m is adjusted to C _m -const*C _m , otherwise the congestion window does not change. Where const is a smaller constant.

Step 3: Establish a bandwidth estimation model and obtain a bandwidth estimation value; according to the time t _{k-1, m} , t _{k, m} of receiving two adjacent ACK confirmation packets on the same link, and the number of successfully received The number of data bytes d _{k, m} to get the sampling value of transmission link bandwidth: The bandwidth sampling value is smoothed by the weighted average method to obtain the bandwidth estimation value:

${\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\frac{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; *</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}$

Among them, BW _{k, m} , BW _{k-1, m} are the bandwidth estimation values at the current moment and the previous moment on the link m respectively, and sample_BW _{k, m} , sample_BW _{k-1, m} are the bandwidth estimates at the current moment and the previous moment on the link m respectively. Bandwidth sampling value at one moment, p is the weighting factor.

Step 4: Establish a congestion window adjustment model, and adjust the size of the congestion window according to the estimated link bandwidth and packet loss. As shown in Figure 3: when receiving a new ACK confirmation packet, compare the size of the congestion window C _m and the congestion threshold S _m , if C _m <S _m , then C _m =C _m +1; otherwise Among them, BW _{k, total} is the sum of the estimated bandwidth of all links at the current moment; if three identical ACKs are received, it is considered that the link has lost packets, and the congestion threshold and congestion window value are adjusted: C _m ＝S _m , where BW _{max, m} is the estimated maximum bandwidth that occurs during the transmission of link m, and then retransmits data; if timeout occurs, adjust the congestion threshold and congestion window: C _m =1, then perform data retransmission.

Step 5. Repeat the above steps 2 to 4 until the transmission of the wireless video service is completed.

Claims (5)

1. the video transmission congestion control method based on MPTCP in the heterogeneous wireless network, it is characterized in that the method comprises the steps: Step 1. Establish a video multi-channel concurrent transmission model. A sending end and a receiving end are connected through multiple network interfaces to form multiple data transmission links, and the bandwidth and delay of each link are different. Step 2. Establish a congestion prediction model. The specific process is as follows: Step 3, establish a bandwidth estimation model, and obtain the actual bandwidth estimation value of each link; Step 4. Establish a congestion window adjustment model, and adjust the congestion window size according to the estimated bandwidth of each link and the packet loss situation; Step 5. Repeat steps 2 to 4 until the transmission of the wireless video service is completed. 2. the video transmission congestion control method based on MPTCP in the heterogeneous wireless network as claimed in claim 1, it is characterized in that the concrete process that congestion prediction model is set up is as follows: 2-1. Obtain the round-trip delay RTT _m of the transmission link m through the timestamp information of the ACK confirmation packet returned by the receiving end, and calculate the round-trip delay RTT _m and the minimum round-trip delay minRTT _m during the transmission of the link The difference between θ _m . 2-2. Compare the difference θ _m with the congestion judgment threshold b _m , when θ _m > b _m , it is considered that the link is about to be congested, and the congestion window C _m is adjusted to C _m -const*C _m , where const as a constant. 3. the video transmission congestion control method based on MPTCP in the heterogeneous wireless network as claimed in claim 2, it is characterized in that the establishment of bandwidth estimation model and the acquisition process of link actual bandwidth estimation are as follows: 3-1. Suppose the time t _{k-1, m} of receiving the last ACK confirmation packet, the time t _k , m of receiving the current ACK confirmation packet, and the time difference between two adjacent ACK confirmation packets t _{k, m} -t _{k -1, the number of data bytes successfully received by the receiving end within m} d _{k, m} , to obtain the sampling value of the transmission link bandwidth: 3-2. The bandwidth sampling value is smoothed by the weighted average method to obtain the bandwidth estimation value: ${\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\frac{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; *</span>{\mathrm{<span\; class="notranslate">\; BW</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}}$ Among them, BW _{k, m} , BW _{k-1, m} are the bandwidth estimation values at the current time and the previous time on the link m respectively, sample_BW _{k, m} , sample_BW _{k-1, m} are respectively the current time and the previous time on the link m Time bandwidth sampling value, p is the weighting factor. 4. the video transmission congestion control method based on MPTCP in the heterogeneous wireless network as claimed in claim 3 is characterized in that the establishment congestion window adjustment model described in step 4 adjusts congestion according to each link bandwidth estimate and packet loss The window size, the specific process is as follows: 4-1. When receiving a new ACK confirmation packet, compare the size of the congestion window C _m with the congestion threshold S _m , if C _m ≤ S _m , then C _m =C _m+1 ; 4-2. If C _m >S _m , then where BW _{k, total} is the sum of all link bandwidth estimates at the current moment; 4-3. If three identical ACK confirmation packets are received, it is considered that packet loss occurs on the link, and the congestion threshold and congestion window value are adjusted accordingly: $\left\{\begin{array}{c}{S}_m=\frac{{\mathrm{BW}}_{k,m}}{{\mathrm{BW}}_{\max ,m}}*C_m\\ C_m=S_m\end{array}\right.,$ Among them, BW _{max, m} is the estimated maximum bandwidth value that occurs during the transmission of link m, and then data retransmission is performed. 4-4. If timeout occurs, adjust the congestion threshold and congestion window: $\left\{\begin{array}{c}{S}_m=\frac{{\mathrm{BW}}_{k,m}}{{\mathrm{BW}}_{\max ,m}}*C_m\\ C_m=1\end{array}\right.,$ Afterwards, the data is retransmitted. 5. the video transmission congestion control method based on MPTCP in the heterogeneous wireless network as claimed in claim 4, is characterized in that: The congestion prediction module is used to predict whether the transmission link is about to be congested, specifically: obtain the round-trip time delay RTT of the data packet through the ACK confirmation packet returned by the receiving end, and then compare the RTT with the minimum RTT that occurs on the link For comparison, when the difference is greater than a certain threshold, it is considered that the link is about to be congested, and the congestion window size of the link is reduced to avoid congestion. The bandwidth estimating module is used to estimate the actual transmission bandwidth of the transmission link, specifically: by calculating the time when the sending end receives two adjacent ACK confirmation packets, the number of bytes of data packets successfully received by the receiving end and The ratio during this period is used as the sampling value of the link bandwidth, and the sampling value is smoothed by the weighted average method, and the processed value is used as the bandwidth estimation value of the link. The congestion window adjustment module is used to adjust the congestion window of each link, and adjust the congestion window of each link according to the packet loss situation of each link and the bandwidth estimation value fed back by the receiving end.