Transmission optimization method and device
By analyzing the adaptive packet loss rate threshold and packet loss time interval trends, combined with the window freezing strategy for satellite network reconfiguration cycles, the performance degradation problem of traditional congestion control algorithms in satellite networks is solved, thereby improving the transmission efficiency and connection stability of satellite networks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- COMP NETWORK INFORMATION CENT CHINESE ACADEMY OF SCI
- Filing Date
- 2026-01-29
- Publication Date
- 2026-05-19
AI Technical Summary
Traditional congestion control algorithms in satellite networks suffer from reduced throughput and bandwidth utilization due to high latency, high bit error rate, and dynamic topology characteristics. Existing optimization schemes are difficult to cover all random packet loss and lack dynamic adaptive capabilities.
By using an adaptive packet loss rate threshold judgment mechanism and packet loss time interval change trend analysis, congestion and non-congestion packet loss can be distinguished, and the congestion control window can be frozen during the satellite network reconfiguration cycle to avoid misjudgment and network interruption.
It achieves end-to-end adaptive capability, accurately distinguishes between congested and non-congested packet loss, improves the transmission efficiency and connection stability of satellite networks, and adapts to the highly dynamic changes of satellite networks.
Smart Images
Figure CN122068945A_ABST
Abstract
Description
Technical Field
[0001] This specification relates to one or more embodiments in the field of network technology, and more particularly to a transmission optimization method and apparatus. Background Technology
[0002] With the growing global demand for seamless connectivity, traditional terrestrial networks are insufficient in coverage, making low-Earth orbit (LEO) satellite networks a crucial component of 6G's integrated space-air-ground system. However, the high latency, high error rate, and dynamic topology of satellite networks cause traditional congestion control algorithms to degrade in this environment, making it difficult to fully utilize bandwidth.
[0003] Satellite network packet loss is mainly divided into congestion-related packet loss and non-congestion-related packet loss (such as channel errors, satellite handover, etc.). Existing packet loss-based algorithms cannot distinguish the cause and are prone to erroneously modifying the parameters of the congestion control window; algorithms based on latency or bandwidth are prone to misjudgment due to the dynamic nature of the network. Both lead to a decrease in throughput and bandwidth utilization.
[0004] To address this issue, various optimization solutions have been proposed, such as sensing handover events, distinguishing packet loss types, and introducing forward error correction. However, most of these solutions are designed for specific scenarios and cannot cover all random packet loss; forward error correction consumes bandwidth and may mask congestion; and they lack dynamic adaptive capabilities, making it difficult to adapt to the highly dynamic changes in satellite networks. Summary of the Invention
[0005] The present invention describes a transmission optimization method and apparatus that can solve the above-mentioned technical problems.
[0006] According to a first aspect, a transmission optimization method is provided, characterized in that it includes:
[0007] Based on the determined packet loss event, update the long-term packet loss monitoring queue and the short-term packet loss monitoring queue for the current period. The long-term packet loss monitoring queue is used to monitor a first number of recently sent data packets that have not successfully reached the other end. The short-term packet loss monitoring queue is used to monitor a second number of recently sent data packets that have not successfully reached the other end, wherein the first number is greater than the second number.
[0008] Based on the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue, a packet loss time interval trend is determined, wherein the packet loss time interval trend indicates the trend of change in the packet loss rate.
[0009] Based on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend, it is determined whether congestion packet loss has occurred. If congestion packet loss is determined to have occurred, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion control window are not changed.
[0010] Based on the above embodiments, it also includes:
[0011] Obtain the RTT of the most recently received data packet and extract the delay information from the fields of the data packet;
[0012] Using the exponentially weighted moving average method, the current delay smoothing value is obtained based on the delay information and historical delay smoothing values;
[0013] The RTT and minimum RTT of the data packet are compared, and the minimum value of the two is taken as the propagation delay;
[0014] If the current delay smoothing value is less than the sum of the propagation delay and the delay threshold, then congestion and packet loss are determined to have occurred.
[0015] If congestion and packet loss are detected, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion window will not be changed.
[0016] Based on the above further embodiments, determining the packet loss time interval trend according to the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue includes:
[0017] Iterate through the timestamps in the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue to obtain the time interval between adjacent packet losses, and store the time interval in the interval array;
[0018] Set a sliding window and slide it step by step on the interval array to calculate the mean of the time interval within each window;
[0019] Compare the average time interval of the current window with the average time interval of the previous window;
[0020] If the mean of the time interval in the current window is less than the mean of the time interval in the previous window, update the downward trend count.
[0021] After comparing the mean of the time intervals of all windows, the downward trend count is compared with a dynamic threshold, which is determined based on the number of windows in the sliding window.
[0022] If the downward trend count is greater than the dynamic threshold, then the packet loss time interval trend is determined to be downward.
[0023] Based on the above embodiments, the step of basing the data on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend includes:
[0024] The long-term packet loss rate is obtained based on the long-term packet loss monitoring queue; the short-term packet loss rate is obtained based on the short-term packet loss monitoring queue.
[0025] Congestion packet loss occurs when the short-term packet loss rate is greater than the sum of the long-term packet loss rate and the dynamic tolerance threshold, and the packet loss time interval trend is decreasing.
[0026] Based on the above-described further embodiments, the method further includes:
[0027] Based on long-term packet loss rate The dynamic tolerance threshold is obtained as follows: ,in, and It is a constant parameter used to make the dynamic tolerance threshold increase as the long-term packet loss rate increases;
[0028] when This confirms that congestion and packet loss have occurred.
[0029] Based on the above-described further embodiments, the method further includes:
[0030] Based on the satellite network reconfiguration cycle, a time period for freezing congestion control is determined, wherein the time period for freezing congestion control covers the satellite network reconfiguration cycle;
[0031] If congestion and packet loss are confirmed, and the current period is within the frozen congestion control window period, the parameters of the congestion control window will not be changed.
[0032] Based on the above-described further embodiments, the method further includes:
[0033] When it is determined that a previously identified packet loss event was a misjudgment, the corresponding record of the packet loss event is removed from both the long-term packet loss monitoring queue and the short-term packet loss monitoring queue.
[0034] According to a second aspect, a transmission optimization apparatus is provided, characterized in that it comprises:
[0035] The first processing module is used to update the long-term packet loss monitoring queue and the short-term packet loss monitoring queue in the current period according to the determined packet loss event. The long-term packet loss monitoring queue is used to monitor a first number of recently sent data packets that have not successfully reached the other end. The short-term packet loss monitoring queue is used to monitor a second number of recently sent data packets that have not successfully reached the other end. The first number is greater than the second number.
[0036] The second processing module is used to determine the packet loss time interval trend based on the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue, wherein the packet loss time interval trend indicates the trend of change in the packet loss rate.
[0037] The third processing module is used to determine whether congestion packet loss has occurred based on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend. If congestion packet loss is determined to have occurred, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion control window are not changed.
[0038] Based on the above embodiments, a fourth processing module is also included, used to obtain the RTT of the most recently received data packet and extract the delay information from the fields of the data packet;
[0039] Using the exponentially weighted moving average method, the current delay smoothing value is obtained based on the delay information and historical delay smoothing values;
[0040] The RTT and minimum RTT of the data packet are compared, and the minimum value of the two is taken as the propagation delay;
[0041] If the current delay smoothing value is less than the sum of the propagation delay and the delay threshold, then congestion and packet loss are determined to have occurred.
[0042] If congestion and packet loss are detected, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion window will not be changed.
[0043] Based on the above further embodiments, the second processing module is used to traverse the timestamps in the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue to obtain the time interval between adjacent packet losses, and store the time interval in the interval array;
[0044] Set a sliding window and slide it step by step on the interval array to calculate the mean of the time interval within each window;
[0045] Compare the average time interval of the current window with the average time interval of the previous window;
[0046] If the mean of the time interval in the current window is less than the mean of the time interval in the previous window, update the downward trend count.
[0047] After comparing the mean of the time intervals of all windows, the downward trend count is compared with a dynamic threshold, which is determined based on the number of windows in the sliding window.
[0048] If the downward trend count is greater than the dynamic threshold, then the packet loss time interval trend is determined to be downward.
[0049] Based on the above embodiments, the third processing module is used to obtain the long-term packet loss rate according to the long-term packet loss monitoring queue; and to obtain the short-term packet loss rate according to the short-term packet loss monitoring queue.
[0050] Congestion packet loss occurs when the short-term packet loss rate is greater than the sum of the long-term packet loss rate and the dynamic tolerance threshold, and the packet loss time interval trend is decreasing.
[0051] Based on the above further embodiments, the third processing module is used to determine the long-term packet loss rate. The dynamic tolerance threshold is obtained as follows: ,in, and It is a constant parameter used to make the dynamic tolerance threshold increase as the long-term packet loss rate increases;
[0052] when This confirms that congestion and packet loss have occurred.
[0053] Based on the above further embodiments, the third processing module is used to determine a frozen congestion control window time period according to the satellite network reconfiguration cycle, wherein the frozen congestion control window time period covers the satellite network reconfiguration cycle;
[0054] If congestion and packet loss are confirmed, and the current period is within the frozen congestion control window period, the parameters of the congestion control window will not be changed.
[0055] Based on the above further embodiments, the first processing module is used to remove the corresponding record of the packet loss event from the long-term packet loss monitoring queue and the short-term packet loss monitoring queue when it is identified that the previously determined packet loss event was a misjudgment.
[0056] According to a third aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the program, implements the transmission optimization method as described in the above technical solutions.
[0057] According to a fourth aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the program, implements the transmission optimization method as described in the above technical solutions.
[0058] In the methods described in the embodiments of this specification, end-to-end adaptive capabilities can be achieved without the need for additional support from the network layer. This allows for accurate differentiation between congested and non-congested packet loss, and effectively addresses periodic reconfiguration and sudden packet loss in satellite networks, thereby comprehensively improving transmission efficiency and connection stability in satellite networks. Attached Figure Description
[0059] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0060] Figure 1 This is an architecture diagram of the transmission optimization method provided by the present invention;
[0061] Figure 2 This is a flowchart illustrating a transmission optimization method provided by the present invention;
[0062] Figure 3 This is a schematic diagram of the long-term packet loss monitoring queue and the short-term packet loss monitoring queue provided by the present invention;
[0063] Figure 4 This is a schematic diagram of the QUIC protocol message provided by the present invention;
[0064] Figure 5 This is a schematic diagram of the packet loss time interval trend algorithm provided by the present invention;
[0065] Figure 6 This is a schematic diagram of the transmission optimization device provided by the present invention. Detailed Implementation
[0066] The solution provided in this specification will now be described with reference to the accompanying drawings.
[0067] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be described below with reference to the accompanying drawings.
[0068] In the description of the embodiments of the present invention, the words "exemplary," "for example," or "for instance" are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design that is described as "exemplary," "for example," or "for instance" in the embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.
[0069] In the description of the embodiments of the present invention, the term "and / or" is merely a description of the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, B existing alone, and A and B existing simultaneously. Furthermore, unless otherwise stated, the term "multiple" means two or more.
[0070] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and their variations all mean "including but not limited to," unless otherwise specifically emphasized.
[0071] With the growing global demand for seamless network connectivity, traditional terrestrial networks, due to their limited coverage and difficulties in deployment in remote areas, are struggling to meet global communication needs. Against this backdrop, satellite networks, particularly low-Earth orbit (LEO) satellite networks, are becoming an important component of the 6G integrated space-air-ground network due to their wide-area coverage and flexible deployment advantages.
[0072] However, the inherent characteristics of satellite networks, such as high latency, high bit error rate, and dynamic topology changes, cause the performance of traditional congestion control algorithms based on terrestrial networks to degrade significantly in this environment, making it impossible to fully utilize available bandwidth.
[0073] Packet loss in satellite networks can generally be categorized into two types: congestion-related packet loss and non-congestion-related packet loss. The latter is primarily caused by factors such as channel errors, weather conditions, and satellite handover, and is unrelated to network congestion. Existing algorithms perform poorly in handling both types of packet loss. One type, packet loss-based algorithms (such as Reno and Cubic), cannot distinguish the cause of packet loss, often misclassifying non-congestion-related packet loss as congestion, leading to unnecessary congestion window reduction. The other type, algorithms based on latency or bandwidth estimation (such as BBR and Copa), are prone to misjudging network conditions due to the highly dynamic nature of satellite networks. The limitations of both types of algorithms severely reduce throughput and bandwidth utilization, hindering the performance of satellite networks.
[0074] To address the failure of traditional congestion control algorithms in satellite networks, various optimization schemes have been proposed. For example, SaTCP and SatGuard detect satellite handover events to avoid misjudging network congestion during such handovers; TP-Satellite uses priority marking to actively distinguish different packet loss types; Polar-QUIC and DRS introduce forward error correction (FEC) technology to directly address packet loss caused by high bit error rates. However, these existing schemes still have significant limitations: most schemes are optimized for specific scenarios such as satellite handover, but non-congestion packet loss scenarios in satellite communication are complex and variable (such as channel errors, weather effects, solar storms, etc.), and a single fixed strategy is difficult to effectively cope with all random packet loss; FEC technology brings new problems: the introduced redundant error correction packets will occupy effective bandwidth and may mask the real congestion signal, thus causing bandwidth fairness issues; lack of dynamic adaptive capability: existing schemes mostly rely on external information input or preset static parameters, making it difficult to autonomously adapt to the highly dynamic changes in satellite network topology and link status.
[0075] To address the aforementioned issues, this invention proposes a transmission optimization method. By analyzing packet loss information and employing an adaptive packet loss rate threshold judgment mechanism, it distinguishes between congestion-related and non-congestion-related packet loss. Furthermore, it introduces a packet loss time interval change trend analysis method to further improve the accuracy of congestion judgment by identifying the temporal characteristics of packet loss events. For latency information, a latency threshold is periodically updated and set to alleviate queue accumulation at intermediate nodes and reduce average transmission latency. Finally, considering the periodic reconfiguration characteristics of satellite networks, a window freezing strategy effectively eliminates the impact of network interruptions and packet loss caused by satellite handover on transmission. This invention can achieve end-to-end adaptive capabilities without additional network layer support, accurately distinguishing between congestion-related and non-congestion-related packet loss, and effectively handling periodic reconfiguration and sudden packet loss in satellite networks, thereby comprehensively improving transmission efficiency and connection stability in satellite networks.
[0076] Figure 1 This is an architecture diagram of the transmission optimization method. (For example...) Figure 1 As shown, the transmission optimization method in this embodiment of the invention optimizes the path from receiving the acknowledgment frame to changing the congestion window. Specifically, it includes a packet loss rate statistics module, a latency statistics module, an adaptive packet loss rate threshold judgment module, a packet loss time interval change trend judgment module, a latency threshold judgment module, and a reconfiguration window freezing module.
[0077] In the packet loss rate statistics module, the packet loss rate is dynamically calculated. This invention employs a dual-window mechanism to maintain a long-term packet loss monitoring queue and a short-term packet loss monitoring queue. In the adaptive packet loss rate threshold judgment module, a packet loss rate threshold is dynamically set, and based on the packet loss rate obtained from the packet loss rate statistics module, packets exceeding the set threshold are judged as congestion-induced packet loss. In the packet loss time interval change trend judgment module, based on the short-term packet loss monitoring queue in the packet loss rate statistics module, the change trend of the packet loss time interval is judged, and packets decreasing beyond the threshold are judged as congestion-induced packet loss.
[0078] In the latency statistics module, designed based on the inherent characteristics of satellite networks, the network employs a periodic reconfiguration strategy. Within each configuration period, latency typically fluctuates within a certain baseline range, with significant changes only occurring during reconfiguration. Therefore, this module re-collects latency-related data at the beginning of each period. In the latency threshold judgment module, based on the results obtained from the latency statistics module, the propagation latency within that period is determined, and a threshold is used to limit the length of the intermediate queue, preventing latency increases.
[0079] In the reconfiguration window freezing module, the congestion window of the congestion control algorithm is frozen during reconfiguration, filtering out a large number of sudden non-congestion packet losses during satellite handover. It also prevents the unnecessary decrease of the congestion window caused by temporary network interruptions during satellite handover, ensuring throughput stability during link handover.
[0080] In summary, an adaptive packet loss rate threshold judgment mechanism and a packet loss time interval change trend analysis method are used to distinguish between congestion-related and non-congestion-related packet loss. For latency information, latency thresholds are periodically updated and set to alleviate queue accumulation at intermediate nodes and reduce average transmission latency. Addressing the periodic reconfiguration characteristics of satellite networks, a window freezing strategy effectively eliminates the impact of network interruptions and packet loss caused by satellite handover on transmission. This achieves end-to-end adaptive capabilities, accurately distinguishing between congestion-related and non-congestion-related packet loss, and effectively responding to periodic reconfiguration and sudden packet loss in satellite networks, thereby comprehensively improving transmission efficiency and connection stability in satellite networks.
[0081] Figure 2 This is a flowchart illustrating a transmission optimization method. (For example...) Figure 2 As shown, the transmission optimization method includes the following steps:
[0082] 210: Based on the identified packet loss events, update the long-term packet loss monitoring queue and the short-term packet loss monitoring queue for the current period.
[0083] In an embodiment of the present invention, Figure 3 This is a diagram illustrating the long-term packet loss monitoring queue and the short-term packet loss monitoring queue, as shown below. Figure 3 As shown, the long-term packet loss monitoring queue Used to monitor the last packet received by the receiving end. Packet loss information in each data packet, including This is the first quantity, set as a configurable parameter. Short-term packet loss monitoring queue. Used to monitor the receiver's confirmation of the last received packet. Packet loss information in each data packet, including The second quantity is set as a configurable parameter, where, > .
[0084] Based on long-term packet loss monitoring To obtain the long-term packet loss rate Based on short-term packet loss monitoring To obtain the long-term packet loss rate Where len is the length of the queue. and This is the number of messages that should be received.
[0085] The period, also known as the update frequency, can be per transmission. Each data packet or each passing time.
[0086] In addition, when the number of packets sent at the beginning is less than At that time, due to insufficient data, an accurate judgment could not be made. and If the value is set to 0, it means that initialization has not yet been completed. If packet loss occurs at this time, it will be directly judged as congestion packet loss.
[0087] In this embodiment of the invention, the QUIC protocol is used as an example for illustration. The QUIC protocol is a novel, UDP-based, secure, reliable, and low-latency Internet transport layer protocol. The QUIC ACK frame is a frame type in the QUIC protocol used to acknowledge that a data packet has been successfully received and to control the data flow. Figure 4 This is a diagram of a QUIC protocol message, as shown below. Figure 4As shown, unlike the traditional NACK scheme, QUIC ACK employs an ACK Range implementation, meaning a single ACK frame can acknowledge multiple discontinuous packet intervals, thus calculating the actual packet loss rate within a specific time period. Furthermore, unlike TCP SACK, QUIC ACK is not limited by the number of acknowledgment blocks, further improving its packet loss detection efficiency. QUIC ACK also includes an ACKDelay field, recording the time interval between receiving a packet and sending the corresponding ACK frame. Subtracting this delay from the measured delay yields an RTT value that more accurately reflects the actual network latency, thereby improving the accuracy of delay-based threshold control and achieving better congestion management and performance optimization.
[0088] Upon receiving an ACK frame, the accurate sequence number of the lost data packet is obtained by parsing the ACK information. The sequence number and timestamp of the lost data packet are simultaneously recorded in the long-term packet loss monitoring queue. and short-term packet loss monitoring queue In the middle, and recalculate when the update conditions are met. and .
[0089] 220: Determine whether congestion and packet loss have occurred based on the long-term packet loss rate and the short-term packet loss rate.
[0090] In specific embodiments of the present invention:
[0091]
[0092] when If the short-term packet loss rate is lower than the long-term packet loss rate, it is determined that no congestion or packet loss has occurred at this time.
[0093] when At that time, and when When the packet loss rate exceeds the packet loss rate threshold in the short term, it is considered that a surge in the observed packet loss rate in the short term has occurred, and it is directly judged as a congestion packet loss.
[0094] Specifically, when setting the packet loss rate threshold, a fixed threshold limit was not used; instead, a function was employed. As a tolerance threshold, the tolerance for changes in packet loss rate is adjusted accordingly. The growth is accompanied by an increase in packet loss rate. This leads to a higher long-term packet loss rate. It exhibits appropriate tolerance for different sizes. The function also limits its maximum tolerance value to... Even below, The value becomes very large, and the tolerance threshold cannot grow indefinitely. If the tolerance threshold is too high, the judgment of non-congestion packet loss is lenient, the window cannot decrease normally during congestion, the application is in an over-processing state, resulting in more packet loss and throughput fluctuations. This approach achieves adaptive packet loss rate threshold judgment.
[0095] when Furthermore, the Downward Trend of the packet loss interval is decreasing, confirming that congestion and packet loss have occurred.
[0096] Specifically, in order to determine whether packet loss events are becoming more frequent, i.e. whether the time interval between packet loss events shows a continuous downward trend, this trend is quantified and detected by calculating and analyzing the moving average of the packet loss interval sequence. Figure 5 This is a schematic diagram of the packet loss time interval trend algorithm, such as... Figure 5 As shown, the specific steps include:
[0097] First, check the length of the input packet loss queue. If the number of packet loss events in the queue is less than the number of packet loss events MA, there are not enough data points to form a meaningful trend judgment, and the process returns directly.
[0098] Iterate through the packet loss monitoring queue and calculate the time interval t between two adjacent packet loss events. i - t i-1 These interval values are then stored in the intervals array. The MAX_INT constant is the maximum number of intervals.
[0099] Next, check the length of the generated time interval array. If the number of intervals is less than 2×MA, there is not enough data to slide the calculation and compare multiple moving averages, so it also returns false, indicating that no congestion or packet loss occurred.
[0100] Next, a sliding window of length MA is used, moving sequentially across the time interval array intervals, to calculate the average packet loss currAvg for the current window at the preset interval. The average packet loss currAvg of the current window is then compared with the average packet loss prevAvg of the previous window.
[0101] If currAvg < prevAvg, it means that the average interval of packet loss in the latest window is shorter than that in the previous window, i.e., the packet loss is more concentrated. In this case, increment decCount by 1.
[0102] Increment the total number of comparisons, totalMA, by 1. Then assign currAvg to prevAvg as the benchmark for the next window comparison. Slide the window back one position and repeat the above process until all possible windows have been traversed.
[0103] Finally, after comparing all possible moving averages, if the total number of comparisons (totalMA) is greater than 0, and the proportion of the number of decreases to the total number of comparisons exceeds the preset threshold (THRESHOLD), i.e., decCount > THRESHOLD × totalMA, then a significant downward trend is determined, and TRUE is returned, indicating that congestion and packet loss have occurred; otherwise, FALSE is returned, indicating that no congestion and packet loss have occurred.
[0104] This allows for the detection of a continuous downward trend in the moving average of packet loss event intervals, thereby determining whether potential congestion is occurring. This trend detection effectively compensates for the shortcomings of dynamic packet loss rate thresholds in dealing with short-term sudden congestion, improving the algorithm's sensitivity.
[0105] Adaptive packet loss rate thresholds allow the transport layer to adapt to a certain degree of continuous non-congestion packet loss and packet jitter. However, this also introduces new problems: the existence of the threshold reduces the algorithm's sensitivity when congestion actually occurs or throughput reaches the link bottleneck, potentially leading to short-term over-transmission. The packet loss time interval trend judgment module is used to compensate for this deficiency. and This can be viewed as an average performance of packet loss over long and short periods, while the changing trend of the packet loss event interval reflects the potential trend of packet loss rate changes within the statistical timeframe of this average performance. A decrease in the packet loss interval often foreshadows an increase in the packet loss rate, i.e., the occurrence of congestion events.
[0106] 230: Based on the latency information, determine whether congestion and packet loss have occurred.
[0107] In this embodiment of the invention, the delay information is also obtained in each cycle based on the received messages.
[0108] The current delay smoothing value, smoothed_rtt, can be obtained based on the following formula.
[0109] min_rtt = min(latest_rtt, min_rtt);
[0110] adjusted_rtt = latest_rtt - ack_delay
[0111] smoothed_rtt = 8 / 7 * smoothed_rtt + 1 / 8 *adjusted_rtt
[0112] Based on the obtained latency information, the minimum RTT value min_rtt in each period can be obtained. Within a single period, min_rtt is used as the propagation latency for judgment. The judgment condition is smoothed_rtt < min_rtt + delay_threshold. If the latency threshold delay_threshold is not met, it indicates that the queue of intermediate nodes is seriously backed up, and it is judged that congestion and packet loss have occurred.
[0113] Specifically, the RTT of the most recently received data packet is obtained as latest_rtt, and the delay information ack_delay is extracted from the data packet's fields.
[0114] Based on the RTT of the data packet and the delay information in the fields of the data packet, the delay adjustment information adjusted_rtt is obtained;
[0115] The current delay smoothing value is obtained by using the exponentially weighted moving average method, based on delay adjustment information and historical delay smoothing values.
[0116] The RTT and min_rtt of the data packet are compared, and the minimum of the two is taken as the new min_rtt, which is the propagation delay;
[0117] If the current delay smoothing value is less than the sum of the propagation delay and the delay threshold, then congestion and packet loss are determined to have occurred.
[0118] 240: If congestion and packet loss are confirmed, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion control window are not changed.
[0119] Specifically, satellite networks undergo periodic reconfigurations, such as satellite handover. This process produces two typical network phenomena: During the handover, the user terminal loses its link with one satellite and establishes a new link with the next, resulting in a brief connectionless state and a brief communication interruption. Data packets being transmitted during the handover are lost. Simultaneously, the establishment of the new link and route updates cause path changes, leading to a step increase in round-trip time (RTT), resulting in a sudden surge of non-congestion packet loss and increased latency.
[0120] Therefore, satellite handover scenarios and sudden network congestion scenarios are highly similar in their characteristics. Both scenarios involve a large number of consecutive packet losses within a short period, with an increase in RTT (Round-Trip Time). Satellite handover scenarios involve changes in network topology. While the packet loss in these scenarios is not due to congestion, the packet loss and increased latency caused by handover can be misinterpreted as sudden network congestion.
[0121] This misjudgment triggers a double negative effect, causing immediate and ongoing damage to transmission performance: unnecessary congestion window reduction. Once congestion is detected, the congestion control algorithm immediately performs window reduction, severely wasting the available bandwidth of the satellite link. After the handover is completed, the link capacity recovers, but the transmitting end slowly increases from a reduced window, causing a sharp drop in throughput and requiring a long recovery time, making it impossible to maintain stable, high-performance transmission. Furthermore, these misjudged congestion packet loss records are entered into a short-term packet loss monitoring queue. And significantly increase the short-term packet loss rate. This directly raises the benchmark for judging the adaptive packet loss rate threshold in the subsequent period, leading to a relaxation of the criteria for judging real congestion. Even if real network congestion occurs after the handover, it cannot be responded to in a timely manner, resulting in continuous queue backlog and performance degradation.
[0122] In this embodiment of the invention, a total duration of 2 * is set before and after the satellite reconfiguration time point. The frozen window, i.e., the window before the reconfiguration event occurs. After the time of occurrence Time. During this freeze window period, adjustments to the congestion window are completely frozen. Regardless of whether a packet loss signal or a packet acknowledgment (ACK) signal is received, the congestion control algorithm does not change the window size.
[0123] By appropriately extending the freezing time This allows the packet loss statistics queue to be updated naturally over time. Packet loss information that is mistakenly recorded during the switchover will be removed due to the queue's rolling update, preventing it from having a lasting negative impact on the judgment after the data is unfrozen.
[0124] In summary, this invention proposes a transmission optimization framework based on packet loss rate and packet loss time interval (Loss-rate & Interval-based Freeze Enhancement, hereinafter referred to as LIFE), which includes five core modules: 1) Packet loss rate statistics module, which adopts a dual-window monitoring mechanism and combines a TCP retransmission timer-based calculation method to process QUIC. 1) ACK, which provides statistics on packet loss rate and latency information; 2) Adaptive packet loss rate threshold judgment module, which, based on the packet loss information provided by the previous module, accurately distinguishes between congestion-related packet loss and non-congestion-related packet loss through an adaptive packet loss rate threshold judgment mechanism, avoiding unnecessary window reduction due to misjudgment; 3) Packet loss time interval change trend judgment module, which introduces a packet loss time interval change trend analysis method, and further improves the accuracy of congestion judgment by identifying the temporal characteristics of packet loss events; 4) Latency threshold judgment module, which periodically updates and sets the latency threshold based on the latency information provided by module 1, in order to alleviate queue accumulation at intermediate nodes and reduce the average transmission latency; 5) Reconfiguration window freezing module, which designs a window freezing strategy to effectively eliminate the impact of network interruption and packet loss caused by satellite switching on transmission, taking into account the periodic reconfiguration characteristics of satellite networks.
[0125] The present invention proposes a transmission optimization method for packet loss-based congestion control algorithms in satellite network environments. By adaptive packet loss rate threshold judgment, packet loss time interval change trend analysis, and delay threshold setting, combined with a window freezing strategy during satellite reconfiguration, it effectively distinguishes between packet loss caused by network congestion and non-congestion packet loss, thereby improving the performance of packet loss-based congestion control algorithms.
[0126] Figure 6 This is a schematic diagram of a transmission optimization device, such as... Figure 6 As shown, it includes:
[0127] The first processing module is used to update the long-term packet loss monitoring queue and the short-term packet loss monitoring queue in the current period according to the determined packet loss event. The long-term packet loss monitoring queue is used to monitor a first number of recently sent data packets that have not successfully reached the other end. The short-term packet loss monitoring queue is used to monitor a second number of recently sent data packets that have not successfully reached the other end. The first number is greater than the second number.
[0128] The second processing module is used to determine the packet loss time interval trend based on the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue, wherein the packet loss time interval trend indicates the trend of change in the packet loss rate.
[0129] The third processing module is used to determine whether congestion packet loss has occurred based on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend. If congestion packet loss is determined to have occurred, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion control window are not changed.
[0130] Based on the above embodiments, a fourth processing module is also included, used to obtain the RTT of the most recently received data packet and extract the delay information from the fields of the data packet;
[0131] Delay adjustment information is obtained based on the RTT of the data packet and the delay information in the fields of the data packet;
[0132] Using the exponentially weighted moving average method, the current delay smoothing value is obtained based on the delay adjustment information and historical delay smoothing values;
[0133] The RTT and minimum RTT of the data packet are compared, and the minimum value of the two is taken as the propagation delay;
[0134] If the current delay smoothing value is less than the sum of the propagation delay and the delay threshold, then congestion and packet loss are determined to have occurred.
[0135] If congestion and packet loss are detected, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion window will not be changed.
[0136] Based on the above further embodiments, the second processing module is used to traverse the timestamps in the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue to obtain the time interval between adjacent packet losses, and store the time interval in the interval array;
[0137] Set a sliding window and slide it step by step on the interval array to calculate the mean of the time interval within each window;
[0138] Compare the average time interval of the current window with the average time interval of the previous window;
[0139] If the mean of the time interval in the current window is less than the mean of the time interval in the previous window, update the downward trend count.
[0140] After comparing the mean of the time intervals of all windows, the downward trend count is compared with a dynamic threshold, which is determined based on the number of windows in the sliding window.
[0141] If the downward trend count is greater than the dynamic threshold, then the packet loss time interval trend is determined to be downward.
[0142] Based on the above embodiments, the third processing module is used to obtain the long-term packet loss rate according to the long-term packet loss monitoring queue; and to obtain the short-term packet loss rate according to the short-term packet loss monitoring queue.
[0143] Congestion packet loss occurs when the short-term packet loss rate is greater than the sum of the long-term packet loss rate and the dynamic tolerance threshold, and the packet loss time interval trend is decreasing.
[0144] Based on the above further embodiments, the third processing module is used to determine the long-term packet loss rate. The dynamic tolerance threshold is obtained as follows: ,in, and It is a constant parameter used to make the dynamic tolerance threshold increase as the long-term packet loss rate increases;
[0145] when This confirms that congestion and packet loss have occurred.
[0146] Based on the above further embodiments, the third processing module is used to determine a frozen congestion control window time period according to the satellite network reconfiguration cycle, wherein the frozen congestion control window time period covers the satellite network reconfiguration cycle;
[0147] If congestion and packet loss are confirmed, and the current period is within the frozen congestion control window period, the parameters of the congestion control window will not be changed.
[0148] Based on the above further embodiments, the first processing module is used to remove the corresponding record of the packet loss event from the long-term packet loss monitoring queue and the short-term packet loss monitoring queue when it is identified that the previously determined packet loss event was a misjudgment.
[0149] According to another embodiment, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the program to implement the transmission optimization method as described in the above technical solution.
[0150] According to another embodiment, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed in a computer, causes the computer to perform a transmission optimization method.
[0151] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using hardware, software, firmware, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium.
[0152] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.
Claims
1. A transmission optimization method, characterized in that, include: Based on the determined packet loss event, update the long-term packet loss monitoring queue and the short-term packet loss monitoring queue for the current period. The long-term packet loss monitoring queue is used to monitor a first number of recently sent data packets that have not successfully reached the other end. The short-term packet loss monitoring queue is used to monitor a second number of recently sent data packets that have not successfully reached the other end, wherein the first number is greater than the second number. Based on the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue, a packet loss time interval trend is determined, wherein the packet loss time interval trend indicates; Based on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend, it is determined whether congestion packet loss has occurred. If congestion packet loss is determined to have occurred, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion control window are not changed.
2. The transmission optimization method according to claim 1, characterized in that, Also includes: Obtain the round-trip time (RTT) of the most recently received data packet and extract the delay information from the fields of the data packet; Using the exponentially weighted moving average method, the current delay smoothing value is obtained based on the delay information and historical delay smoothing values; The RTT and minimum RTT of the data packet are compared, and the minimum value of the two is taken as the propagation delay; If the current delay smoothing value is less than the sum of the propagation delay and the delay threshold, then congestion and packet loss are determined to have occurred. If congestion and packet loss are detected, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion window will not be changed.
3. The transmission optimization method according to claim 1, characterized in that, The step of determining the packet loss time interval trend based on the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue includes: Iterate through the timestamps in the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue to obtain the time interval between adjacent packet losses, and store the time interval in the interval array; Set a sliding window and slide it step by step on the interval array to calculate the mean of the time interval within each window; Compare the average time interval of the current window with the average time interval of the previous window; If the mean of the time interval in the current window is less than the mean of the time interval in the previous window, update the downward trend count. After comparing the mean of the time intervals of all windows, the downward trend count is compared with a dynamic threshold, which is determined based on the number of windows in the sliding window. If the downward trend count is greater than the dynamic threshold, then the packet loss time interval trend is determined to be downward.
4. The transmission optimization method according to claim 3, characterized in that, The step of basing the data on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend includes: The long-term packet loss rate is obtained based on the long-term packet loss monitoring queue; the short-term packet loss rate is obtained based on the short-term packet loss monitoring queue. Congestion packet loss occurs when the short-term packet loss rate is greater than the sum of the long-term packet loss rate and the dynamic tolerance threshold, and the packet loss time interval trend is decreasing.
5. The transmission optimization method according to claim 4, characterized in that, The method further includes: Based on long-term packet loss rate The dynamic tolerance threshold is obtained as follows: ,in, and It is a constant parameter used to make the dynamic tolerance threshold increase as the long-term packet loss rate increases; when This confirms that congestion and packet loss have occurred.
6. The transmission optimization method according to claim 1, characterized in that, The method further includes: Based on the satellite network reconfiguration cycle, a time period for freezing congestion control is determined, wherein the time period for freezing congestion control covers the satellite network reconfiguration cycle; If congestion and packet loss are confirmed, and the current period is within the frozen congestion control window period, the parameters of the congestion control window will not be changed.
7. The transmission optimization method according to claim 1, characterized in that, The method further includes: When it is determined that a previously identified packet loss event was a misjudgment, the corresponding record of the packet loss event is removed from both the long-term packet loss monitoring queue and the short-term packet loss monitoring queue.
8. A transmission optimization device, characterized in that, include: The first processing module is used to update the long-term packet loss monitoring queue and the short-term packet loss monitoring queue in the current period according to the determined packet loss event. The long-term packet loss monitoring queue is used to monitor a first number of recently sent data packets that have not successfully reached the other end. The short-term packet loss monitoring queue is used to monitor a second number of recently sent data packets that have not successfully reached the other end. The first number is greater than the second number. The second processing module is used to determine the packet loss time interval trend based on the long-term packet loss monitoring queue and / or the short-term packet loss monitoring queue, wherein the packet loss time interval trend indicates the trend of change in the packet loss rate. The third processing module is used to determine whether congestion packet loss has occurred based on the long-term packet loss monitoring queue, the short-term packet loss monitoring queue, and the packet loss time interval trend. If congestion packet loss is determined to have occurred, and the current period is within the overlapping period of the satellite network reconfiguration period, the parameters of the congestion control window are not changed.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the transmission optimization method as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the transmission optimization method as described in any one of claims 1 to 7.