A congestion control method based on bandwidth prediction in civil aviation satellite communications

Abstract

The present invention belongs to the technical field of civil aviation mobile communications, and particularly relates to a congestion control method based on bandwidth prediction in civil aviation satellite communications. A bandwidth prediction algorithm is introduced into the three stages of TCP congestion control, namely slow start, congestion avoidance and fast recovery, to adjust the congestion window value. The method comprises the following steps: upon system initialization, the slow start stage of TCP congestion control is entered, and the increase range of the congestion window is adjusted by combining a dichotomy idea and a bandwidth prediction algorithm; when the congestion window value is equal to a slow start threshold, the congestion avoidance stage of TCP congestion control is entered, and an improved congestion avoidance mechanism is adopted to adjust the congestion window; if, in the fast recovery stage, a transmitting end receives three repeated ACKs or a network retransmission timer times out, a first adjustment mechanism or a second adjustment mechanism is adopted to adjust the congestion window and the slow start threshold. The present invention takes into account the remaining bandwidth of a link, reduces the risk of link congestion and packet loss, and improves the throughput of the network.

Description

Technical Field

[0001] The present invention belongs to the technical field of civil aviation mobile communications, and in particular relates to a congestion control method based on bandwidth prediction in civil aviation satellite communications. Background Art

[0002] With the advent of the "mobile phone in the air" era in China's civil aviation, passengers are increasingly demanding in-flight internet access, and this in turn places demands on fast and efficient civil aviation communications technology. Currently, there are two primary technologies for enabling communication between civil aviation users and the ground: satellite-based broadband access and ground-based base station-based broadband access. Satellite-based broadband access, with its wide coverage, long communication distances, and unrestricted geographic presence, is playing an increasingly important role in civil aviation communications.

[0003] As the primary transport layer protocol, the Internet Transmission Control Protocol (TCP) provides users with reliable data flow delivery through flow control, congestion control, and error control. For decades, it has proven highly successful in supporting Internet data communications. Research on congestion control algorithms is crucial to the success of TCP. Similarly, research on TCP congestion control in civil aviation satellite communications plays a crucial role in ensuring high-quality communications for civil aviation passengers. However, the large bandwidth-delay product, high bit error rate, and asymmetric link bandwidth of civil aviation satellite communication links make achieving high-quality communications with TCP difficult.

[0004] After searching, the closest prior art is CN201711033849.9, a satellite network transport layer congestion control method based on forward delay. This invention addresses the asymmetric bandwidth of satellite communication links and proposes a forward congestion control algorithm based on forward delay. This algorithm determines whether the communication link is congested by measuring the forward delay time of data packets and adopts different control strategies based on the different types of congestion in the communication link. In addition, CN201510170924.0, a satellite communication segmented congestion control method based on packet loss rate, addresses the characteristics of high burst bit errors and proposes a segmented congestion control method based on packet loss rate. This method determines the link status in segments based on the packet loss rate, thereby adopting a more accurate window adjustment strategy. Both of these methods only improve one aspect of the link. If applied to civil aviation satellite links, they will not significantly improve performance. Furthermore, neither method considers the link's remaining bandwidth to determine the actual link status, resulting in a blind reduction or increase in the transmission rate. Summary of the Invention

[0005] This paper comprehensively considers the characteristics of asymmetric bandwidth, high bit error rate, and large bandwidth-delay product of civil aviation satellite links, and provides a congestion control method based on bandwidth prediction in civil aviation satellite communications. On the basis of TCP Veno, a bandwidth prediction algorithm is introduced to improve the congestion control algorithm of TCP Veno to achieve higher civil aviation satellite communication quality.

[0006] A congestion control method based on bandwidth prediction in civil aviation satellite communications introduces a bandwidth prediction algorithm to adjust the congestion window value in the three stages of TCP congestion control: slow start, congestion avoidance, and fast recovery. The congestion control method based on bandwidth prediction includes the following steps:

[0007] S1. The system initializes and enters the slow start phase of TCP congestion control. The increase in the congestion window is adjusted by combining the dichotomy method and the bandwidth prediction algorithm.

[0008] S2. When the congestion window value equals the slow start threshold, the TCP congestion control enters the congestion avoidance phase, using the congestion avoidance improvement mechanism to adjust the congestion window.

[0009] S3. If in the fast recovery phase, that is, the sender receives three duplicate ACKs or the network retransmission timer times out, the first adjustment mechanism or the second adjustment mechanism is used to adjust the congestion window and slow start threshold.

[0010] Furthermore, in order to make the data packet sending rate closer to the actual remaining link bandwidth, a bandwidth prediction algorithm is introduced to calculate the remaining link bandwidth. The bandwidth prediction algorithm is expressed as:

[0011]

[0012] BE i represents the predicted value of network bandwidth after receiving the i-th ACK, S data Indicates the total amount of data sent during the smoothed round-trip delay, SRTT i represents the smoothed round trip delay of the i-th node, SRTT i-1 represents the i-1th round-trip delay after smoothing. The smoothing process is expressed as:

[0013]

[0014] k is the weighting factor, T new is the i-th round-trip delay, and T represents the round-trip delay when i=1.

[0015] Furthermore, the network bandwidth prediction value is smoothed, which is expressed as:

[0016]

[0017] Among them, SBE i is the smoothed bandwidth prediction value of the network bandwidth prediction value after receiving the i-th ACK, BE i is the predicted value of network bandwidth after receiving the i-th ACK, SBE i-1 is the smoothed bandwidth prediction value of the network bandwidth prediction value after receiving the i-1th ACK, and α is the weight parameter.

[0018] Furthermore, the congestion window value in the slow start phase is less than the slow start threshold. The formula for adjusting the increase of the congestion window in the slow start phase in step S1 is:

[0019]

[0020] Where S is the slow start threshold, σ is the limiting factor, C(t) is the congestion window value at the current time t, C(t+T) is the congestion window value at the time t+T, and T represents the time of a transmission round.

[0021] Furthermore, the bandwidth prediction algorithm is used to optimize the adjustment formula for the increase of the congestion window in the slow start phase (the congestion window C is less than the slow start threshold S), which is expressed as follows:

[0022]

[0023] Among them, γ is a regulation factor that reflects the degree of link congestion.

[0024] Furthermore, the calculation formula of the adjustment factor is:

[0025]

[0026] a max =2 / (S-2);

[0027] a min =1;

[0028] λ=B t,j / B t,j-1 ;

[0029] B t,j Denotes the predicted bandwidth of the jth transmission round, B t,j-1 represents the predicted bandwidth of the j-1th transmission round, λ is the ratio of the predicted bandwidth of the j-1th transmission round to the predicted bandwidth of the j-1th transmission round, and a max is the maximum adjustment factor value, a min is the minimum adjustment factor value.

[0030] Further, the congestion avoidance improvement mechanism includes:

[0031] S11. After obtaining the current bandwidth prediction value using the bandwidth prediction algorithm, it is determined whether the number of packets accumulated in the link is less than the packet number threshold of 3. If so, step S12 is executed, otherwise, step S13 is executed;

[0032] S12. The network is operating normally, and the congestion window increases by 1 congestion window per transmission round.

[0033] S13. Determine whether the current congestion window is less than the bandwidth prediction value. If so, execute step S14; otherwise, the congestion window value remains unchanged.

[0034] S14. The congestion window increases at a rate of 1 congestion window per two transmission rounds.

[0035] Furthermore, when the sender receives three duplicate ACKs, it uses the first adjustment mechanism to adjust the congestion window and slow start threshold, including:

[0036] S21. After obtaining the current bandwidth prediction value using the bandwidth prediction algorithm, it is determined whether the number of packets accumulated in the link is less than the packet number threshold 3. If so, step S22 is executed, otherwise step S23 is executed;

[0037] S22. Determine whether the congestion window value is less than the current bandwidth prediction value. If so, ssthresh new =ssthresh current , cwnd k+1 =cwnd k If not, cwnd k+1 =ssthresh new ;

[0038] S23. Determine whether the congestion window value is less than the current bandwidth prediction value. If so, cwnd k+1 =ssthresh new If not, cwnd k+1 =ssthresh new ;

[0039] Among them, ssthresh new Indicates the latest calculated slow start threshold, ssthresh current Indicates the current slow start threshold, cwnd k Indicates the congestion window value when receiving the kth ACK, B k+1 Indicates the predicted bandwidth, Seg_size indicates the size of a single segment, RTT ave Indicates the average round-trip delay, RTTmin Indicates the minimum round-trip delay.

[0040] Furthermore, if the network retransmission timer times out, a second adjustment mechanism is used to adjust the congestion window and slow start threshold, including:

[0041] S31 using bandwidth prediction algorithm to predict the bandwidth after determining whether the number of packets accumulated in the link is less than the packet number threshold 3, if so, proceed to step S32, otherwise to step S33;

[0042] S32. cwnd k+1 =ssthresh new ;

[0043] S33. cwnd k+1 =1, restart slow start to relieve congestion;

[0044] Among them, ssthresh new Indicates the updated slow start threshold, cwnd k Indicates the congestion window value when receiving the kth ACK, B k+1 Indicates the predicted bandwidth, Seg_size indicates the size of a single segment, RTT min Indicates the minimum round-trip delay.

[0045] Beneficial effects of the present invention:

[0046] 1. This paper proposes a congestion control method based on bandwidth prediction for civil aviation satellite communications. A bandwidth prediction algorithm is introduced to adjust the congestion window during the slow start, congestion avoidance, and fast recovery phases of TCP congestion control. First, link bandwidth prediction is introduced during the slow start phase of TCP congestion control, making the increase of the congestion window more gradual and more consistent with the network congestion situation. This not only reduces the link packet loss rate, but also shortens the convergence time of slow start.

[0047] 2. The bandwidth prediction-based congestion control method for civil aviation satellite communications described in this invention employs a more conservative congestion window increase method during the congestion avoidance phase. This congestion window increase method not only considers the current link congestion status but also the remaining link bandwidth, reducing the risk of link congestion and packet loss and improving network throughput.

[0048] 3. The congestion control method based on bandwidth prediction in civil aviation satellite communications described in the present invention adopts a more conservative congestion window adjustment method during the fast recovery phase. When the transmitter receives three duplicate ACKs, it distinguishes the cause of packet loss and, based on the relationship between the current congestion window and available bandwidth, adopts a corresponding window adjustment method. This avoids blindly reducing the window and causing bandwidth waste, enabling the window to recover to its optimal value more quickly, while also reducing the risk of link congestion. When the retransmission timer expires, it first determines whether the timeout is due to congestion packet loss or link error packet loss (or ACK loss). Different fast recovery strategies are then adopted based on the different timeout causes, avoiding unnecessary slow starts, reducing bandwidth waste, and improving network performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of the congestion control method based on bandwidth prediction of the present invention;

[0050] Figure 2 This is a flow chart of the bandwidth prediction algorithm of the present invention;

[0051] Figure 3 This is a flow chart of adjusting the congestion window increase range during the slow start phase of the present invention;

[0052] Figure 4 This is a flow chart of the congestion avoidance improvement mechanism of the present invention. DETAILED DESCRIPTION

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0054] This invention provides a TCP congestion control method for civil aviation satellite communications based on bandwidth prediction, supporting high-quality communications for civil aviation passengers. Based on TCP Veno, this method introduces a bandwidth prediction algorithm to improve the slow start mechanism during the slow start phase, reducing the increase in the congestion window in the later stages of slow start and shortening the slow start convergence time. During the congestion avoidance phase, it further differentiates between congestion-induced packet loss. During the fast recovery phase, the TCP Veno mechanism for distinguishing between congestion-induced packet loss and error-induced packet loss is applied to timeout detection, and the situation of receiving three duplicate ACKs is further subdivided.

[0055] In one embodiment, a method for controlling TCP congestion in civil aviation satellite communications based on bandwidth prediction is provided. Figure 1As shown in Figure 1, a bandwidth prediction algorithm is introduced into the three stages of TCP congestion control, namely slow start, congestion avoidance, and fast recovery, to adjust the congestion window value. The congestion control method based on bandwidth prediction includes the following steps:

[0056] S1. The system initializes and enters the slow start phase of TCP congestion control, using the dichotomy approach to adjust the increase in the congestion window.

[0057] S2. When the congestion window value equals the slow start threshold, the TCP congestion control enters the congestion avoidance phase, using the congestion avoidance improvement mechanism to adjust the congestion window.

[0058] S3. If in the fast recovery phase, that is, the sender receives three duplicate ACKs or the network retransmission timer times out, the first adjustment mechanism or the second adjustment mechanism is used to adjust the congestion window and slow start threshold.

[0059] In the present invention, a transmission round refers to the time it takes to receive correct responses corresponding to all data segments after all data segments have been sent out in the current congestion window. j is used to represent the transmission round. Therefore, a total of N data segments are sent out in the j-th transmission round. When the ACK corresponding to the i-th (i≤N)-th data segment is received in the j-th transmission round, a bandwidth prediction algorithm is used to predict the bandwidth. The predicted bandwidth of the j-th transmission round is updated to the predicted network bandwidth value after receiving the i-th ACK calculated by the bandwidth prediction algorithm. In addition, the i-th round-trip delay is the difference between the time the i-th data segment is sent and the time the corresponding ACK is received.

[0060] In one embodiment, during the TCP congestion control process of civil aviation satellite communications, a bandwidth prediction algorithm is used to predict the bandwidth each time a new ACK is received. The bandwidth prediction algorithm formula is expressed as:

[0061]

[0062] Among them, BE i Represents the predicted network bandwidth after receiving the i-th ACK, SRTT i represents the smoothed round trip delay of the i-th node, SRTT i-1 represents the i-1th round trip delay after smoothing, S data Represents the total amount of data sent during the smoothing of round-trip delay variations. Due to the high speed of civil aircraft and the asymmetry of civil aviation satellite links, the network's round-trip delay experiences significant jitter, which introduces significant errors in bandwidth prediction, severely impacting network performance. To mitigate bandwidth prediction errors caused by severe round-trip delay jitter, the bandwidth prediction algorithm smoothes the round-trip delay (RTT) and the predicted network bandwidth value (BE). The smoothing of the round-trip delay (RTT) and the predicted network bandwidth value (BE) is expressed as:

[0063]

[0064]

[0065] Among them, k is the weighting factor, T new is the newly measured round-trip time, T represents the round-trip delay when i=1, SBE i is the smoothed bandwidth prediction value of the network bandwidth prediction value after receiving the i-th ACK, BE i is the predicted value of network bandwidth after receiving the i-th ACK, SBE i-1 is the smoothed bandwidth prediction value of the network bandwidth prediction value after receiving the i-1th ACK, α is a weight parameter, and in this embodiment, k=1 / 4 and α=7 / 8.

[0066] Preferably, the bandwidth prediction algorithm is as follows Figure 2 Shown, including:

[0067] Step A. Initialize the system, set the weighting factor k = 1 / 4, the weight parameter α = 7 / 8, and the number of iterations i, i = 1;

[0068] Step B. Send the i-th segment and record the sending time, then wait for the correct ACK of the i-th segment;

[0069] Step C. After receiving the ACK of the i-th message segment, record the time when the ACK is received, and record the ACK of the i-th message segment as ack i ; Calculate the round trip delay of the i-th segment by the time it is sent and the time it receives its ACK.

[0070] Step D. Determine whether i is 1 (i.e. whether it is the first time to receive ACK). If so, then S data =ack i -1, SRTT = T; otherwise SBE i =α×SBE i-1 +(1-α)×BE i , S data =ack i -ack i-1 , SRTT i =(1-k)×SRTT i-1 +k×T new ;

[0071] Step E: Set the number of iterations i=i+1 and return to step B.

[0072] In one embodiment, in the slow start phase of congestion control (the congestion window C is less than the slow start threshold S), referring to the dichotomy concept, based on TCP Veno, the increase in the congestion window in the late slow start phase is reduced, which can be expressed as:

[0073]

[0074] Where T represents the duration of a transmission round, C(t) is the congestion window value at the current time t, C(t+T) is the congestion window value at time t+T (i.e., the congestion window value for the next transmission round), S is the slow start threshold, and σ is a limiting factor introduced to limit the convergence time of slow start. The value of σ ranges from 2 to S / 2. When σ = 2, the congestion window increases most slowly, but the convergence time is also the longest, the same as the convergence time of TCP Vegas. When σ = S / 2, it is the same as the slow start convergence time of TCP Veno. SC ≥ σ not only limits the convergence time of slow start, but also ensures that the rate of increase of the congestion window in the slow start phase is faster than that in the congestion avoidance phase. Analysis shows that when σ = 2, while ensuring that the slow start rate of increase is faster than that of congestion avoidance, the congestion window increases most slowly and the packet loss rate is lowest. Therefore, in this embodiment, σ is set to 2.

[0075] Preferably, the bandwidth prediction algorithm is used to optimize the adjustment formula of the increase amplitude of the congestion window in the slow start phase (the congestion window C is less than the slow start threshold S) to reduce the slow start convergence time. The optimization formula is expressed as:

[0076]

[0077] Among them, γ is the adjustment factor reflecting the link congestion level (γ≥1), SC>2 and This ensures that the increase in the congestion window in the slow start phase is greater than the increase in the congestion window in the congestion avoidance phase (that is, the slope of the curve of the congestion window increase in the slow start phase is greater than the slope of the curve of the congestion window increase in the congestion avoidance phase), and also ensures that the convergence time of the slow start phase will not be too long.

[0078] Specifically, the calculation formula of the adjustment factor is:

[0079]

[0080] a max =2 / (S-2);

[0081] a min =1;

[0082] λ=B t,j / B t,j-1 ;

[0083] B t,j represents the predicted bandwidth of the jth transmission round, B t,j-1 represents the predicted bandwidth of the j-1th transmission round, λ is the ratio of the predicted bandwidth of the j-1th transmission round to the predicted bandwidth of the j-1th transmission round, and a max is the maximum adjustment factor value, a min is the minimum adjustment factor value.

[0084] Preferably, the process of adjusting the increase of the congestion window in the slow start phase is as follows: Figure 3 Shown, including:

[0085] S101, predicting bandwidth using a bandwidth prediction algorithm and calculating an adjustment factor γ;

[0086] S102: Determine whether the congestion window value C at the current time t is less than half of the slow start threshold S, that is, If yes, the congestion window value at time t+T is calculated based on the adjustment factor γ, and then the process returns to step S101; otherwise, the process goes to step S103;

[0087] S103: Determine whether the difference between the slow start threshold S and the congestion window value C at the current time t is greater than 2, that is, SC>2. If so, execute step S104; otherwise, set the congestion window value at time t+T to be equal to the slow start threshold, and enter the congestion avoidance phase.

[0088] S104, judgment If so, the congestion window value at time t+T is calculated according to the adjustment factor γ, and then the process returns to step S101. Otherwise, the congestion window value at time t+T is set equal to the slow start threshold, and the process enters the congestion avoidance phase.

[0089] In one embodiment, in order to avoid congestion during the congestion avoidance phase, the present invention adopts a more conservative congestion window adjustment method, which is mainly achieved by combining a bandwidth prediction algorithm with the congestion avoidance mechanism of TCP Veno.

[0090] Specifically, congestion avoidance improvement mechanisms such as Figure 4 Shown, including:

[0091] S11. After obtaining the current bandwidth prediction value using the bandwidth prediction algorithm, determine whether the number of packets N accumulated in the link is less than the packet number threshold β (ie, whether N<β, β is 3). If so, execute step S12; otherwise, execute step S13.

[0092] S12. The current network is operating normally, and the congestion window increases at a rate of 1 congestion window per transmission round; that is, Indicates the congestion window increment after receiving an ACK in a transmission round;

[0093] S13. If the link is congested but no packet loss occurs, the current congestion window value is further determined to be less than the current bandwidth prediction value. If so, step S14 is executed. Otherwise, the congestion window value remains unchanged.

[0094] S14. The link still has spare bandwidth. To fully utilize the remaining bandwidth, the congestion window increases at a rate of one congestion window per two transmission rounds; that is, Indicates the congestion window increment after receiving an ACK in a transmission round. In one embodiment, during the fast recovery phase of congestion control, if the sender receives three duplicate ACKs or the network retransmission timer times out, the first adjustment mechanism or the second adjustment mechanism is used to adjust the congestion window and slow start threshold.

[0095] Specifically, when the sender receives three duplicate ACKs, the network capacity may have been reached or a random link error may have caused one or more data packets to be lost. The first adjustment mechanism is used to adjust the congestion window and slow start threshold, including:

[0096] S21. After obtaining the current bandwidth prediction value using the bandwidth prediction algorithm, it is determined whether the number of packets accumulated in the link is less than the packet number threshold 3. If so, step S22 is executed, otherwise step S23 is executed;

[0097] S22. It is assumed that packet loss is caused by link errors, not network congestion. It is determined whether the congestion window value is less than the current bandwidth prediction value. If so, it means that there is enough bandwidth. Then ssthresh new =ssthresh current , cwnd k+1 =cwnd k If not, appropriately reduce the congestion window and threshold, i.e.

[0098] S23. indicates that the network is saturated and packet loss occurs, but no serious congestion occurs. Determine whether the congestion window value is less than the available bandwidth. If so, it means that there is still bandwidth left and it should be fully utilized. Therefore, reduce the reduction of the congestion window and the threshold. cwnd k+1 =ssthresh new If not, cwnd k+1 =ssthresh new ;

[0099] Among them, ssthresh newIndicates the latest calculated threshold, ssthresh current Indicates the current slow start threshold, cwnd k represents the congestion window when receiving the kth ACK, B k+1 Indicates the predicted bandwidth, Seg_size indicates the size of a single segment, RTT ave Indicates the average round-trip delay, RTT min Indicates the minimum round-trip delay. The average round-trip delay is the average of all previous round-trip delays.

[0100] Specifically, if the network retransmission timer times out, the network may be severely congested or the civil aviation satellite link may be in a severe error state, resulting in continuous packet loss or ACK loss or delay. In this case, TCP Veno is introduced to distinguish between congestion and error conditions and use the second adjustment mechanism to adjust the congestion window and slow start threshold, including:

[0101] S31 using bandwidth prediction algorithm to predict the bandwidth after determining whether the number of reports accumulated in the link is less than the maximum packet value, and if so, proceed to step S32, otherwise to step S33;

[0102] S32. Severe bit error causes timer to time out. cwnd k+1 =ssthresh new ;

[0103] S33. cwnd k+1 =1, restart slow start to relieve congestion;

[0104] Among them, ssthresh new Indicates the updated slow start threshold, cwnd k Indicates the congestion window value when receiving the kth ACK, B k+1 Indicates the predicted bandwidth, Seg_size indicates the size of a single segment, RTT min Indicates the minimum round-trip delay.

[0105] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A congestion control method based on bandwidth prediction in civil aviation satellite communications, characterized in that: A bandwidth prediction algorithm is introduced to adjust the congestion window value in the three stages of TCP congestion control: slow start, congestion avoidance, and fast recovery. The congestion control method based on bandwidth prediction includes the following steps: S1. The system initializes and enters the slow start phase of TCP congestion control, using the dichotomy approach to adjust the increase in the congestion window. The congestion window value in the slow start phase is less than the slow start threshold. The formula for adjusting the increase of the congestion window in the slow start phase in step S1 is: Where S is the slow start threshold, σ is the limiting factor, C(t) is the congestion window value at the current time t, C(t+T) is the congestion window value at the time t+T, and T represents the time of a transmission round. The bandwidth prediction algorithm is used to optimize the increase in the congestion window during the slow start phase. The formula is expressed as follows: Among them, γ is the adjustment factor reflecting the link congestion level; The calculation formula of the adjustment factor is: a max =2 / (S-2); a min ＝1； λ=B t,j / B t,j-1 ; B t,j represents the predicted bandwidth of the jth transmission round, B t,j-1 represents the predicted bandwidth of the j-1th transmission round, λ is the ratio of the predicted bandwidth of the j-1th transmission round to the predicted bandwidth of the j-1th transmission round, and a max is the maximum adjustment factor value, a min is the minimum adjustment factor value; S2. When the congestion window value equals the slow start threshold, the TCP congestion control enters the congestion avoidance phase, and the congestion avoidance improvement mechanism is used to adjust the congestion window value. S3. If in the fast recovery phase, that is, the sender receives three duplicate ACKs or the network retransmission timer times out, the first adjustment mechanism or the second adjustment mechanism is used to adjust the congestion window value and the slow start threshold.

2. The congestion control method based on bandwidth prediction in civil aviation satellite communications according to claim 1, characterized in that: The bandwidth prediction algorithm formula is expressed as: BE i It represents the predicted value of network bandwidth after receiving the i-th ACK, S data Indicates the total amount of data sent during the smoothed round-trip time difference, SRTT i represents the smoothed round trip delay of the i-th node, SRTT i-1 represents the i-1th round-trip delay after smoothing. The smoothing process is expressed as: k is the weighting factor, T new is the i-th round-trip delay, and T represents the round-trip delay when i=1.

3. The congestion control method based on bandwidth prediction in civil aviation satellite communications according to claim 2, characterized in that: The network bandwidth prediction value is smoothed and expressed as: Among them, SBE i is the smoothed bandwidth prediction value of the network bandwidth prediction value after receiving the i-th ACK, BE i is the predicted value of network bandwidth after receiving the i-th ACK, SBE i-1 is the smoothed bandwidth prediction value of the network bandwidth prediction value after receiving the i-1th ACK, and α is the weight parameter.

4. The congestion control method based on bandwidth prediction in civil aviation satellite communications according to claim 1, characterized in that: Congestion avoidance improvement mechanisms include: S11. After obtaining the current bandwidth prediction value using the bandwidth prediction algorithm, it is determined whether the number of packets accumulated in the link is less than the packet number threshold of 3. If so, step S12 is executed, otherwise, step S13 is executed; S12. The network is operating normally, and the congestion window increases by 1 congestion window per transmission round. S13 determines whether the current congestion window value is less than the current bandwidth prediction value. If so, execute step S14; otherwise, the congestion window value remains unchanged. S14. The congestion window increases at a rate of 1 congestion window per two transmission rounds.

5. The congestion control method based on bandwidth prediction in civil aviation satellite communications according to claim 1, characterized in that: When three duplicate ACKs are received, the first adjustment mechanism is used to adjust the congestion window and slow start threshold, including: S21. After obtaining the current bandwidth prediction value using the bandwidth prediction algorithm, it is determined whether the number of packets accumulated in the link is less than the packet number threshold 3. If so, step S22 is executed, otherwise step S23 is executed; S22. Determine whether the congestion window value is less than the current bandwidth prediction value. If so, ssthresh new =ssthresh current , cwnd k+1 =cwnd k If not, cwnd k+1 =ssthresh new ; S23. Determine whether the congestion window value is less than the current bandwidth prediction value. If so, cwnd k+1 =ssthresh new If not, cwnd k+1 =ssthresh new ; Among them, ssthresh new Indicates the updated slow start threshold, ssthresh current Indicates the current slow start threshold, cwnd k Indicates the congestion window value when receiving the kth ACK, B k+1 Indicates the predicted bandwidth, Seg_size indicates the size of a single segment, RTT ave Indicates the average round-trip delay, RTT min Indicates the minimum round-trip delay.

6. The congestion control method based on bandwidth prediction in civil aviation satellite communications according to claim 1, characterized in that: If the network retransmission timer times out, the second adjustment mechanism is used to adjust the congestion window and slow start threshold, including: S31 using bandwidth prediction algorithm to predict the bandwidth after determining whether the number of packets accumulated in the link is less than the packet number threshold 3, if so, proceed to step S32, otherwise to step S33; S32. cwnd k+1 =ssthresh new ; S33. cwnd k+1 =1, restart slow start to relieve congestion; Among them, ssthresh new Indicates the updated slow start threshold, cwnd k Indicates the congestion window value when receiving the kth ACK, B k+1 Indicates the predicted bandwidth, Seg_size indicates the size of a single segment, RTT min Indicates the minimum round-trip delay.

Images