CN111726301A - Congestion control method and system for ensuring video quality in real-time video - Google Patents

Abstract

The invention discloses a congestion control method and system for ensuring video quality in real-time video, which can quickly respond to bandwidth changes, overcome the shortcomings of excessive code rate increase, improve the tracking speed of the code rate to the bandwidth, and improve the bandwidth utilization rate. , overcoming the shortcoming of network congestion caused by excessive code rate increase. The specific scheme is: receive video data packets in real time, and calculate the delay gradient. Perform overload detection on the current network state according to the delay gradient to obtain the current network state signal; the current network state signal is the overload state signal Overuse, the normal state signal Normal or the low load state signal Underuse. Build a finite state machine to perform state transitions based on current network state signals. According to the current state of the state machine, the code rate calculation is performed. The transmission code rate of the video data packet is set according to the calculated code rate at the next moment, that is, time t _i .

Description

Congestion control method and system for ensuring video quality in real-time video

technical field

The invention relates to the technical field of network transmission, in particular to a congestion control method and system for ensuring video quality in real-time video.

Background technique

Network congestion is a common network transmission problem in IP-based datagram switching networks, which has a serious impact on the quality of network transmission. Network congestion is one of the main reasons for reduced network throughput and network packet loss. The problem makes the upper-layer application unable to effectively use the network bandwidth to obtain high-quality network transmission effect. Especially in the field of communication, problems such as packet loss, delay, and jitter caused by network congestion seriously affect the communication quality. If these problems cannot be solved well, a communication product cannot be used normally in the real environment.

The GCC (Google Congestion Control) congestion control algorithm is mainly divided into two parts: one is congestion control based on packet loss, and the other is congestion control based on delay. The two congestion control algorithms are implemented at the sender and receiver respectively. The estimated bandwidth calculated by the congestion control algorithm at the receiver will be fed back to the sender through the RTCP remb message, and the sender will synthesize the results of the two control algorithms. Get a final transmit bit rate, and send data packets at this bit rate.

The basic idea of congestion control based on packet loss is to judge the degree of network congestion according to the number of lost packets. The more packets lost, the more congested the network is, so the transmission rate needs to be reduced to relieve network congestion; if there is no packet loss, this indicates that the network The condition is very good. At this time, you can increase the transmission code rate and detect whether there is more bandwidth available. Delay-based congestion control, GCC uses delay gradient to determine the degree of network congestion, its algorithm is divided into several parts: arrival time filter, overload detector, rate controller.

After the original GCC congestion control algorithm is applied to the system, when the bandwidth situation changes from poor to better, the bit rate remains at a low level, resulting in wasted bandwidth and poor video quality for users. Based on this shortcoming, the algorithm is improved.

The existing GCC congestion control algorithm has the following shortcomings: when the bandwidth starts to increase from a low value, the code stream calculated by the algorithm is much lower than the bandwidth, and the code rate does not quickly track the change of the bandwidth. The reason is that when the input signal of the state machine is underuse, no matter which state the current state is in, the next state is hold. When it is in the hold state, the code rate change strategy remains unchanged, resulting in no real-time increase in the code rate; once the code rate starts to increase If the increase is too high, it exceeds the bandwidth range, causing network congestion, increased packet loss, and video freezes.

Figure 1 shows the bandwidth tracking effect of the traditional GCC congestion control algorithm. Among them, when the bandwidth fluctuates for a short time, the sensitivity of detecting low bandwidth is high, and it can be tracked within 1-2s, but it is not sensitive to the increase of bandwidth, and the code rate increases very slowly. The curve in the figure below reflects the code rate change curve of the congestion control algorithm before the optimization when the bandwidth changes once every 12 seconds. It can be seen from the figure that when the bandwidth increases, the code rate is maintained at a low code rate, and the increase in bandwidth is delayed by 10 seconds to respond, increasing bandwidth by more than 300kbps.

It can be seen that the strategy of the GCC congestion control algorithm is conservative, unable to quickly respond to the available bandwidth resources, and is not suitable for real-time video communication where video quality is important; the bit rate increase strategy increases the bit rate too much, aggravating network congestion and packet loss . The existing congestion control algorithm has a slow tracking speed of the bandwidth in the video transmission process, and the growth rate is not suitable, which cannot meet the demand.

SUMMARY OF THE INVENTION

In view of this, the present invention provides a congestion control method and system for ensuring video quality in real-time video, which can quickly respond to changes in bandwidth, overcome the shortcomings of excessive increase in code rate, improve the tracking speed of code rate to bandwidth, and improve It improves the bandwidth utilization rate and overcomes the shortcoming of network congestion caused by excessive code rate increase.

In order to achieve the above-mentioned purpose, the technical scheme of the present invention is: a congestion control method for ensuring video quality in a real-time video, comprising the following steps:

Step 1: Receive video data packets in real time and calculate the delay gradient.

Step 2: Perform overload detection on the current network state according to the delay gradient to obtain the current network state signal; the current network state signal is the overload state signal Overuse, the normal state signal Normal or the low load state signal Underuse.

Step 3: Construct a finite state machine: including the code rate reduction state Decrease, the code rate maintaining state Hold and the code rate increasing state Increase, the initial state of the finite state machine is preset, and the current network state signal is input into the finite state machine, If the input is Overuse, the finite state machine is directly converted to the Decrease state; if the input is Normal, the finite state machine is directly converted to the Hold state; if the input is Underuse and the current state of the finite state machine is Decrease, the finite state machine is converted to Hold State; if the input is Underuse, and the current state of the finite state machine is not Decrease, the finite state machine transitions to the Incease state.

Step 4. According to the current state of the state machine, calculate the code rate as follows:

1) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is greater than the set threshold, the code rate at the next moment, that is, moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(0.2×A _r (t _i-1 ), 1000), 1.4×incoming_bitrate), incoming_bitrate)

where Ar (t _i ) is the bit rate at time t _{i ; A r (} t _i-1 ) is the bit rate at the current moment, i.e. time t _i-1 ; incoming_bitrate is the received data calculated according to the statistics of the received video data packets bit rate; according to the rate at which the video data packet is received, calculate whether the distance detection bandwidth difference is greater than the set threshold;

2) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is not greater than the set threshold, then the code rate at the next moment, i.e. at moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate)

Among them, averOfPacket is the average data volume of video data packets, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets;

averOfPacket=bits_per_frame/packets_per_frame

Where bits_per_frame is the amount of data per frame, packets_per_frame is the number of packets contained in a frame;

3) When the current state of the state machine is the Decrease state, the code rate at the next moment, i.e. at moment t _i , is:

_Ar (t _i )=min(0.75×incoming_bitrate, Ar ( _{t i-1} ))

Among them, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets;

4) When the current state signal of the state machine is Hold, the code rate remains unchanged, that is, the code rate at the next moment t _i is equal to the code rate at the current moment t _i-1 ;

Step 5: Set the sending code rate of the video data packet according to the calculated code rate at the next moment, that is, time t _i .

Further, in step 1, video data packets are received in real time, and the delay gradient is calculated, specifically:

The time interval of the data packet group is preset, and all video data packets received within the time interval of a data packet group are grouped; for the current group of video data packets, the arrival time of the last video data packet is subtracted from the first video data packet. The sending time of each video data packet is taken as the time delay; the time delay difference between the current group of video data packets and the previous group of video data packets is taken as the delay gradient of the current group of video data packets.

Further, step 2 is specifically:

The overload threshold Th ₁ and the low load threshold Th ₀ are preset. If the delay gradient is greater than the overload threshold Th ₁ , the current network status signal is Overuse.

If the delay gradient is less than the low load threshold Th ₀ , the current network status signal is Underuse.

If the delay gradient is in the range of [Th ₀ , Th ₁ ], the current network state signal is Normal.

Further, in step 4, calculate whether the distance detection bandwidth difference is greater than the set threshold according to the rate of the received video data packets, specifically:

The reception rate of the video data packet is incoming_bitrate, and the initial value of the set average maximum code rate avg_max_bitrate is the reception rate of the first group of video data packets, and the average maximum code rate avg_max_bitrate is updated every time a group of video data packets is received:

avg_max_bitrate=(1-α)×avg_max_bitrate_last+α×incoming_bitrate

α is a preset coefficient constant, which can be set according to experience in the present invention. For example, in the embodiment of the present invention, the coefficient constant α can be set to be 0.05.

Update the maximum bit rate variance var_max_bitrate as:

var_max_bitrate=(1-α)×var_max_bitrate_last+α×(avg_max_bitrate-incoming_bitrate) ²

Update the maximum bitrate standard deviation std_max_bitrate to:

In the Increase state, if the incoming_bitrate of the received bit rate is greater than the sum of avg_max_bitrate and std_max_bitrate, it is considered that the distance detection bandwidth has a large difference. Each time it enters the Decrease state, it is considered that the difference from the detection bandwidth is small.

Further, bits_per_frame is the amount of data per frame, and bits_per_frame is updated after each bit rate calculation is completed:

bits_per_frame=current_bitrate/frame_per_second.

Where frame_per_second is the number of frames per second; current_bitrate is the current bit rate.

Further, the code rate adjustment time interval is set to be within 4s, and in each code rate adjustment time interval, steps 1 to 5 are performed to adjust the code rate.

Another embodiment of the present invention provides a congestion control system for ensuring video quality in real-time video, the system includes a video transmitter, a video receiver, a time filter, an overload detector, a state converter, and a bit rate calculator;

The video sending end is used to receive video data packets according to the sending bit rate adjusted in real time.

The video receiver is used to receive video data packets in real time.

Temporal filter for computing delay gradients based on the sending and receiving times of video packets.

The overload detector is used to perform overload detection on the current network state according to the delay gradient to obtain the current network state signal; the current network state signal is the overload state signal Overuse, the normal state signal Normal or the low load state signal Underuse.

State conversion machine, used to construct a finite state machine: including the code rate reduction state Decrease, the code rate maintaining state Hold and the code rate increasing state Increase, the initial state of the finite state machine is preset, and the current network state signal is input to the finite state In the machine, if the input is Overuse, the finite state machine is directly converted to the Decrease state; if the input is Normal, the finite state machine is directly converted to the Hold state; if the input is Underuse, and the current state of the finite state machine is Decrease, then the finite state machine Convert to Hold state; if the input is Underuse, and the current state of the finite state machine is not Decrease, the finite state machine is converted to Incease state.

The code rate calculator is used to calculate the code rate according to the current state of the state machine as follows:

1) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is greater than the set threshold, the code rate at the next moment, that is, moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate)

where Ar (t _i ) is the bit rate at time t _{i ; A r (} t _i-1 ) is the bit rate at the current moment, i.e. time t _i-1 ; incoming_bitrate is the received data calculated according to the statistics of the received video data packets bit rate; according to the rate at which the video data packet is received, calculate whether the distance detection bandwidth difference is greater than the set threshold;

2) When the current state of the state machine is the Increase state, and the distance from the detection bandwidth is not greater than the set threshold, the code rate at the next moment, that is, moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate)

Among them, averOfPacket is the average data volume of video data packets, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets;

averOfPacket=bits_per_frame/packets_per_frame

Where bits_per_frame is the amount of data per frame, packets_per_frame is the number of packets contained in a frame;

3) When the current state of the state machine is the Decrease state, the code rate at the next moment, i.e. at moment t _i , is:

_Ar (t _i )=min(0.75×incoming_bitrate, Ar ( _{t i-1} ))

Among them, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets;

4) When the current state signal of the state machine is Hold, the code rate remains unchanged, that is, the code rate at the next time t _i is equal to the code rate at the current time t _i-1 .

The bit rate calculator calculates the real-time adjusted sending bit rate and sends it to the video sending end.

Further, the video sending end adjusts the code rate according to the real-time adjusted sending code rate according to the set adjustment time interval.

Beneficial effects:

The invention provides a congestion control method and system for ensuring video quality in real-time video. The solution is based on the traditional GCC congestion control algorithm solution, and the original congestion control algorithm is optimized and adapted according to the requirements and technical characteristics of the system. , which improves the tracking speed of the code rate to the bandwidth, improves the bandwidth utilization rate, and overcomes the shortcoming of network congestion caused by too much increase in the code rate. The invention adopts a modified and optimized state transition mechanism, and based on the strategy of rapidly responding to bandwidth changes, it overcomes the problem that the code rate cannot be quickly tracked when the bandwidth increases, so that the calculated code rate quickly tracks the bandwidth change, and improves the bandwidth utilization rate, so as to achieve The target demand for improving the video picture quality within the range allowed by the bandwidth. The invention adopts an optimized code rate change function, overcomes the disadvantage of excessive increase in code rate, and avoids aggravating network congestion.

Description of drawings

Fig. 1 is the bandwidth tracking effect diagram of the traditional GCC congestion control algorithm;

2 is a state change trend diagram of a finite state machine constructed in an embodiment of the present invention;

3 is a flowchart of a method for congestion control for ensuring video quality in real-time video according to an embodiment of the present invention;

4 is a block diagram of a congestion control system for ensuring video quality in real-time video according to an embodiment of the present invention;

Fig. 5 is a bandwidth tracking effect diagram in which the bandwidth takes 4 seconds as a change interval in the embodiment of the present invention;

FIG. 6 is a bandwidth tracking effect diagram in which the bandwidth is changed at intervals of 6 seconds according to an embodiment of the present invention.

Detailed ways

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

The present invention provides a congestion control method for ensuring video quality in a real-time video, the process of which is shown in Figure 3 and is characterized in that it includes the following steps:

Step 1: Receive video data packets in real time and calculate the delay gradient.

In this embodiment of the present invention, the bit rate adjustment time interval is preset, and a group of video data packets can be received within the bit rate adjustment time interval. For the current group of video data packets, the arrival time of the last video data packet is subtracted from the arrival time of the last video data packet. The arrival time of the first video data packet is taken as the arrival time difference, the sending time of the last video data packet minus the sending time of the first video data is taken as the sending time difference, and the difference between the arrival time difference and the sending time difference is taken as the current group of video Delay gradient for packets.

In the embodiment of the present invention, the code rate adjustment time interval is set to be within 4s, and steps 1 to 5 are performed to adjust the code rate within each code rate adjustment time interval.

Step 2: Perform overload detection on the current network state according to the delay gradient to obtain the current network state signal; the current network state signal is the overload state signal Overuse, the normal state signal Normal or the low load state signal Underuse.

The overload threshold Th ₁ and the low load threshold Th ₀ are preset. If the delay gradient is greater than the overload threshold Th ₁ , the current network status signal is Overuse.

If the delay gradient is less than the low load threshold Th ₀ , the current network status signal is Underuse.

If the delay gradient is in the range of [Th ₀ , Th ₁ ], the current network state signal is Normal.

Step 3: Construct a finite state machine: including the code rate reduction state Decrease, the code rate maintaining state Hold and the code rate increasing state Increase, the initial state of the finite state machine is preset, and the current network state signal is input into the finite state machine, If the input is Overuse, the finite state machine is directly converted to the Decrease state; if the input is Normal, the finite state machine is directly converted to the Hold state; if the input is Underuse and the current state of the finite state machine is Decrease, the finite state machine is converted to Hold State; if the input is Underuse, and the current state of the finite state machine is not Decrease, the finite state machine transitions to the Incease state.

In the embodiment of the present invention, the state change trend of the constructed finite state machine is shown in FIG. 2 .

Step 4. According to the current state of the state machine, calculate the code rate as follows:

1) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is greater than the set threshold, the code rate at the next moment, that is, moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(0.2×A _r (t _i-1 ), 1000), 1.4×incoming_bitrate), incoming_bitrate)

where Ar (t _i ) is the bit rate at time t _{i ; A r (} t _i-1 ) is the bit rate at the current moment, i.e. time t _i-1 ; incoming_bitrate is the received data calculated according to the statistics of the received video data packets code rate.

In the embodiment of the present invention, it is calculated according to the rate of receiving video data packets whether the difference between the distance detection bandwidth is greater than the set threshold.

The reception rate of the video data packet is incoming_bitrate, and the initial value of the set average maximum code rate avg_max_bitrate is the reception rate of the first group of video data packets, and the average maximum code rate avg_max_bitrate is updated every time a group of video data packets is received:

avg_max_bitrate=(1-α)×avg_max_bitrate_last+α×incoming_bitrate

α is a preset coefficient constant, which can be set according to experience in the present invention. For example, in the embodiment of the present invention, the coefficient constant α can be set to be 0.05.

Update the maximum bit rate variance var_max_bitrate as:

var_max_bitrate=(1-α)×var_max_bitrate_last+α×(avg_max_bitrate-incoming_bitrate) ²

Update the maximum bit rate standard deviation std_max_bitrate to:

In the Increase state, if the incoming_bitrate of the received bit rate is greater than the sum of avg_max_bitrate and std_max_bitrate, it is considered that the distance detection bandwidth difference is greater than the set threshold.

2) When the current state of the state machine is the Increase state, and the distance from the detection bandwidth is not greater than the set threshold, the code rate at the next moment, that is, the moment t _i is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate)

Among them, averOfPacket is the average data volume of video data packets, and incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets.

averOfPacket=bits_per_frame/packets_per_frame

Wherein bits_per_frame is the amount of data per frame, and packets_per_frame is the number of data packets included in one frame; in the embodiment of the present invention, according to the setting, the maximum packet size is 1200 bytes.

packets_per_frame=bits_per_frame/(8×1200).

In this embodiment of the present invention, bits_per_frame is the amount of data per frame, and bits_per_frame is updated after each bit rate calculation is completed:

bits_per_frame=current_bitrate/frame_per_second;

Where frame_per_second is the number of frames per second; current_bitrate is the current bit rate.

3) When the current state of the state machine is the Decrease state, the code rate at the next moment, i.e. at moment t _i , is:

_Ar (t _i )=min(0.75×incoming_bitrate, Ar ( _{t i-1} ))

The incoming_bitrate is the received data bitrate calculated according to the statistics of the received video data packets.

4) When the current state signal of the state machine is Hold, the code rate remains unchanged, that is, the code rate at the next moment t _i is equal to the code rate at the current moment t _i-1 ;

Step 5: Set the sending code rate of the video data packet according to the calculated code rate at the next moment, that is, time t _i .

Another embodiment of the present invention also provides a congestion control system for ensuring video quality in real-time video. As shown in FIG. 4 , the system includes a video transmitter, a video receiver, a time filter, an overload detector, and a state transition machine. and a bitrate calculator.

The video sending end is used to receive video data packets according to the sending bit rate adjusted in real time.

The video receiver is used to receive video data packets in real time.

The time filter is used to calculate the delay gradient according to the sending time and the receiving time of the video data packet; it can be calculated according to the calculation method given in the above embodiment.

The overload detector is used to perform overload detection on the current network state according to the delay gradient to obtain the current network state signal; the current network state signal is the overload state signal Overuse, the normal state signal Normal or the low load state signal Underuse; The given overload detection method is calculated.

State conversion machine, used to construct a finite state machine: including the code rate reduction state Decrease, the code rate maintaining state Hold and the code rate increasing state Increase, the initial state of the finite state machine is preset, and the current network state signal is input to the finite state In the machine, if the input is Overuse, the finite state machine is directly converted to the Decrease state; if the input is Normal, the finite state machine is directly converted to the Hold state; if the input is Underuse, and the current state of the finite state machine is Decrease, then the finite state machine Convert to Hold state; if the input is Underuse, and the current state of the finite state machine is not Decrease, the finite state machine is converted to Incease state.

The code rate calculator is used to calculate the code rate according to the current state of the state machine as follows:

1) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is greater than the set threshold, the code rate at the next moment, that is, moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(0.2×A _r (t _i-1 ), 1000), 1.4×incoming_bitrate), incoming_bitrate)

where Ar (t _i ) is the bit rate at time t _{i ; A r (} t _i-1 ) is the bit rate at the current moment, i.e. time t _i-1 ; incoming_bitrate is the received data calculated according to the statistics of the received video data packets Bit rate; according to the rate of receiving video data packets, calculate whether the difference between the distance detection bandwidth is greater than the set threshold.

2) When the current state of the state machine is the Increase state, and the distance from the detection bandwidth is not greater than the set threshold, the code rate at the next moment, that is, moment t _i , is:

A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate)

Among them, averOfPacket is the average data volume of video data packets, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets;

averOfPacket=bits_per_frame/packets_per_frame

Where bits_per_frame is the amount of data per frame, and packets_per_frame is the number of packets contained in a frame. According to the settings, the maximum packet size is 1200 bytes.

packets_per_frame=bits_per_frame/(8×1200)

3) When the current state of the state machine is the Decrease state, the code rate at the next moment, i.e. at moment t _i , is:

_Ar (t _i )=min(0.75×incoming_bitrate, Ar ( _{t i-1} ))

The incoming_bitrate is the received data bitrate calculated according to the statistics of the received video data packets.

4) When the current state signal of the state machine is Hold, the code rate remains unchanged, that is, the code rate at the next time t _i is equal to the code rate at the current time t _i-1 .

The bit rate calculator calculates the real-time adjusted sending bit rate and sends it to the video sending end.

In the embodiment of the present invention, the video sending end adjusts the code rate according to the real-time adjusted sending code rate according to the set adjustment time interval.

FIG. 5 shows a bandwidth tracking effect diagram in which the bandwidth is changed at an interval of 4 seconds. It can be seen that the congestion control scheme for ensuring video quality in real-time video provided by the embodiment of the present invention can meet the requirements.

FIG. 6 shows a bandwidth tracking effect diagram in which the bandwidth is changed at an interval of 6 seconds. It can be seen that the congestion control solution for ensuring video quality in real-time video provided by the embodiment of the present invention can meet the requirements.

To sum up, the above are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (8)

1. the congestion control method of guaranteeing video quality in a real-time video, is characterized in that, comprises the steps: Step 1. Receive video data packets in real time, and calculate the delay gradient; Step 2: Perform overload detection on the current network state according to the delay gradient to obtain a current network state signal; the current network state signal is an overload state signal Overuse, a normal state signal Normal, or a low load state signal Underuse; Step 3: Constructing a finite state machine: including the code rate reduction state Decrease, the code rate maintaining state Hold and the code rate increasing state Increase, the initial state of the finite state machine is preset, and the current network state signal is input to the finite state machine. In the state machine, if the input is Overuse, the finite state machine is directly converted to the Decrease state; if the input is Normal, the finite state machine is directly converted to the Hold state; if the input is Underuse, and the current state of the finite state machine is Decrease, then the finite state machine is converted to the Hold state; if the input is Underuse, and the current state of the finite state machine is not Decrease, then the finite state machine is converted to the Incease state; Step 4. According to the current state of the state machine, calculate the code rate as follows: 1) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is greater than the set threshold, then the code rate at the next moment, i.e. at moment t _i , is: A _r (t _i )=max(min(A _r (t _i-1 )+max(0.2×A _r (t _i-1 ), 1000), 1.4×incoming_bitrate), incoming_bitrate) where Ar (t _i ) is the bit rate at time t _{i ; A r (} t _i-1 ) is the bit rate at the current moment, i.e. time t _i-1 ; incoming_bitrate is the received data calculated according to the statistics of the received video data packets bit rate; according to the rate at which the video data packet is received, calculate whether the distance detection bandwidth difference is greater than the set threshold; 2) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is not greater than the set threshold, then the code rate at the next moment, i.e. at moment t _i , is: A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate) Among them, averOfPacket is the average data volume of video data packets, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets; averOfPacket=bits_per_frame/packets_per_frame Where bits_per_frame is the amount of data per frame, packets_per_frame is the number of packets contained in a frame; 3) When the current state of the state machine is the Decrease state, the code rate at the next moment, i.e. at moment t _i , is: _Ar (t _i )=min(0.75×incoming_bitrate, Ar ( _{t i-1} )) Among them, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets; 4) When the current state signal of the state machine is Hold, the code rate remains unchanged, that is, the code rate at the next moment t _i is equal to the code rate at the current moment t _i-1 ; Step 5: Set the sending code rate of the video data packet according to the calculated code rate at the next moment, that is, time t _i . 2. method as claimed in claim 1 is characterized in that, described step 1, receives video data packet in real time, and calculates delay gradient, is specially: The time interval of the data packet group is preset, and all video data packets received within the time interval of a data packet group are grouped into one group; For the current group of video data packets, the time delay is the arrival time of the last video data packet minus the sending time of the first video data packet; The difference between the time delays of the current group of video data packets and the previous group of video data packets is used as the delay gradient of the current group of video data packets. 3. The method according to claim 1 or 2, wherein the step 2 is specifically: Preset overload threshold Th ₁ and low load threshold Th ₀ , if the delay gradient is greater than the overload threshold Th ₁ , the current network status signal is Overuse; If the delay gradient is less than the low load threshold Th ₀ , the current network status signal is Underuse; If the delay gradient is in the range of [Th ₀ , Th ₁ ], the current network state signal is Normal. 4. The method of claim 3, wherein in the step 4, according to the rate of receiving the video data packet, calculate whether the distance detection bandwidth difference is greater than a set threshold, specifically: The reception rate of the video data packet is incoming_bitrate, and the initial value of the set average maximum code rate avg_max_bitrate is the reception rate of the first group of video data packets, and the average maximum code rate avg_max_bitrate is updated every time a group of video data packets is received: avg_max_bitrate=(1-α)×avg_max_bitrate_last+α×incoming_bitrate where α is the set coefficient constant; Update the maximum bit rate variance var_max_bitrate to: var_max_bitrate=(1-α)×var_max_bitrate_last+α×(avg_max_bitrate-incoming_bitrate) ² Update the maximum bit rate standard deviation std_max_bitrate to:

In the Increase state, if the incoming_bitrate of the received bit rate is greater than the sum of avg_max_bitrate and std_max_bitrate, it is considered that the distance detection bandwidth difference is greater than the set threshold. 5. The method according to claim 1, 2 or 4, wherein the bits_per_frame is an amount of data per frame, and the bits_per_frame is updated after each bit rate calculation is completed: bits_per_frame=current_bitrate/frame_per_second Where frame_per_second is the number of frames per second; current_bitrate is the current bit rate. 6 . The method of claim 1 , wherein the code rate adjustment time interval is set to be within 4s, and steps 1 to 5 are performed to adjust the code rate within each code rate adjustment time interval. 7 . 7. A congestion control system that guarantees video quality in real-time video, is characterized in that, this system comprises video sending end, video receiving end, time filter, overload detector, state converter and code rate calculator; The video sending terminal is used for receiving video data packets according to the sending code rate adjusted in real time; The video receiving end is used to receive video data packets in real time; The time filter is used to calculate the delay gradient according to the sending time and the receiving time of the video data packet; The overload detector is configured to perform overload detection on the current network state according to the delay gradient to obtain a current network state signal; the current network state signal is an overload state signal Overuse, a normal state signal Normal or a low load state signal Underuse; The state transition machine is used to construct a finite state machine: including a code rate reduction state Decrease, a code rate maintaining state Hold and a code rate increasing state Increase, the initial state of the finite state machine is preset, and the current network state signal Input into the finite state machine, if the input is Overuse, the finite state machine is directly converted to the Decrease state; if the input is Normal, the finite state machine is directly converted to the Hold state; if the input is Underuse, and the finite state machine The current state of the state machine is Decrease, then the finite state machine is converted to the Hold state; if the input is Underuse, and the current state of the finite state machine is not Decrease, then the finite state machine is converted to the Incease state; The code rate calculator is used to calculate the code rate according to the current state of the state machine as follows: 1) When the current state of the state machine is the Increase state, and the difference from the detection bandwidth is greater than the set threshold, the code rate at the next moment, i.e. at moment t _i , is: A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate) where Ar (t _i ) is the bit rate at time t _{i ; A r (} t _i-1 ) is the bit rate at the current moment, i.e. time t _i-1 ; incoming_bitrate is the received data calculated according to the statistics of the received video data packets bit rate; according to the rate at which the video data packet is received, calculate whether the difference from the detection bandwidth is greater than the set threshold; 2) When the current state of the state machine is the Increase state, and the distance from the detection bandwidth is not greater than the set threshold, the code rate at the next moment, that is, moment t _i , is: A _r (t _i )=max(min(A _r (t _i-1 )+max(averOfPacket, 1000), 1.4×incoming_bitrate), incoming_bitrate) Among them, averOfPacket is the average data volume of video data packets, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets; averOfPacket=bits_per_frame/packets_per_frame Where bits_per_frame is the amount of data per frame, packets_per_frame is the number of packets contained in a frame; 3) When the current state of the state machine is the Decrease state, the code rate at the next moment, i.e. at moment t _i , is: _Ar (t _i )=min(0.75×incoming_bitrate, Ar ( _{t i-1} )) Among them, incoming_bitrate is the received data bit rate calculated according to the statistics of received video data packets; 4) When the current state signal of the state machine is Hold, the code rate remains unchanged, that is, the code rate at the next moment t _i is equal to the code rate at the current moment t _i-1 ; The sending code rate adjusted in real time calculated by the code rate calculator is sent to the video sending end. 8 . The system of claim 7 , wherein the video sending end adjusts the code rate according to the real-time adjusted sending code rate according to the set adjustment time interval. 9 .

Images