Round-trip delay correction method and system based on reverse path timestamp

Abstract

The invention provides a round-trip delay correction method and system based on a reverse path timestamp. The round-trip delay correction method comprises the following steps: acquiring a data packet sending timestamp, an acknowledgement packet sending timestamp and an acknowledgement packet receiving timestamp; a reverse path time delay sample is dynamically stored through a sliding window, and reference time delay is calculated by selecting a minimum value method or a least square regression method according to the sample size; removing reverse path congestion interference in the round-trip delay according to a difference value between the current reverse path delay and the reference delay to obtain a corrected round-trip delay; and the reference time delay of the same source-destination pair is shared under the symmetric topology. According to the method, forward path congestion and reverse path congestion can be accurately separated, misjudgment caused by reverse path congestion of traditional RTT is avoided, and the bandwidth utilization rate and the short flow performance of a congestion control algorithm are improved; through hardware timestamp and dynamic baseline estimation, low-cost deployment without hardware transformation is realized, and the method is suitable for high-precision time delay optimization of the data center network.

Description

Technical Field

[0001] The present invention relates to the technical field of congestion control algorithms, and in particular to a round-trip delay correction method and system based on reverse path timestamps. Background Art

[0002] As data center networks evolve, the demand for low-latency, high-throughput network performance is increasing. To effectively support emerging applications such as real-time data analysis, AI training, and high-speed storage access (such as RDMA, Flash, and NVMe technologies), congestion control mechanisms in data center networks have become critical.

[0003] Currently, mainstream congestion control algorithms (such as Timely and Swift) typically use round-trip time (RTT) as a congestion feedback signal. This mechanism determines network congestion by measuring the increase in RTT, dynamically adjusting the sending rate and effectively controlling queue length. This approach is simple and offers good performance. However, it cannot effectively distinguish between congestion in the forward and reverse paths within the RTT. RTT includes both the transmission delay of data packets on the forward path and the return delay of acknowledgment (ACK) packets on the reverse path. When the reverse path is congested, even if the forward path is not congested, the sender may misjudge network congestion and incorrectly reduce the sending rate, resulting in wasted bandwidth and resources and degraded network performance.

[0004] Existing technologies typically use the Network Time Protocol (NTP), Precision Time Protocol (PTP), or Datacenter Time Protocol (DTP) for clock synchronization to separate forward and reverse path delays. Specifically, while the Network Time Protocol is low-cost and easy to implement, its synchronization accuracy cannot meet the microsecond-level latency requirements of data center applications. While the Precision Time Protocol and Datacenter Time Protocol offer high clock synchronization accuracy, they rely on dedicated hardware, resulting in significant upgrade costs for network equipment and limiting their practical application.

[0005] Furthermore, existing technologies (such as DX) require frequent re-measurement of baseline delays. This slows convergence when network conditions change or congestion fluctuates, impacting the performance of short flows. Furthermore, these solutions fail to consider the higher network transmission priority of ACK packets, which compromises the accuracy and real-time nature of baseline delay estimation, making it difficult to meet the stringent requirements of data centers for real-time and accurate congestion control.

[0006] Therefore, there is an urgent need for an RTT correction mechanism that does not require hardware modification, is lightweight, and is easy to deploy on a large scale to address the impact of reverse path congestion on RTT. Summary of the Invention

[0007] In view of this, an embodiment of the present invention provides a round-trip delay correction method and system based on reverse path timestamp to eliminate or improve one or more defects in the prior art.

[0008] In one aspect, the present invention provides a method for round-trip delay correction based on reverse path timestamps, the method comprising the following steps: Collecting a data packet sending timestamp, a confirmation packet sending timestamp, and a confirmation packet receiving timestamp; calculating a reverse path transmission delay based on the confirmation packet sending timestamp and the confirmation packet receiving timestamp; Maintain a fixed-size sliding window, dynamically store a preset number of reverse path transmission delay samples within a preset time period, and construct a reverse path transmission delay sample set; When the data volume of the reverse path transmission delay sample set is less than a preset value, selecting a minimum value from the reverse path transmission delay samples in the sliding window as a reference delay; when the data volume of the reverse path transmission delay sample set is not less than a preset value, using a least squares regression method to fit the reverse path transmission delay samples in the sliding window to obtain a theoretical propagation delay in a no-queue state, and using the theoretical propagation delay as the reference delay; The round-trip delay is calculated based on the data packet sending timestamp and the confirmation packet receiving timestamp; the difference between the current reverse path transmission delay and the benchmark delay is calculated to obtain an additional queuing delay; and the additional queuing delay is eliminated from the round-trip delay to obtain a corrected round-trip delay.

[0009] In some embodiments of the present invention, the method further comprises: The data packet sending timestamp, the confirmation packet sending timestamp and the confirmation packet receiving timestamp are transmitted via a cross-layer timestamp frame structure; The frame structure includes standard protocol fields and extended protocol fields of the link layer, the network layer, and the transport layer, and the extended protocol fields are used to carry various timestamp information.

[0010] In some embodiments of the present invention, the method further comprises: The frame structure reports the timestamp information to the user state through a single-root I / O virtualization technology or a data plane development kit.

[0011] In some embodiments of the present invention, the reverse path transmission delay is calculated based on the confirmation packet sending timestamp and the confirmation packet receiving timestamp, and the calculation formula is: T b =t4-t3; Among them, T brepresents the reverse path transmission delay; t4 represents the timestamp of sending the confirmation packet; t3 represents the timestamp of receiving the confirmation packet.

[0012] In some embodiments of the present invention, a fixed-size sliding window is maintained to dynamically store a preset number of reverse path transmission delay samples within a preset time period to construct a reverse path transmission delay sample set. The reverse path transmission delay sample set is expressed as: Wherein, T represents the reverse path transmission delay sample set; represents the Wth reverse path transmission delay sample in the sliding window; W is the size of the sliding window.

[0013] In some embodiments of the present invention, when the data volume of the reverse path transmission delay sample set is less than a preset value, the minimum value is selected from the reverse path transmission delay samples in the sliding window as the reference delay, and the calculation formula is: in, represents the reference delay; min{·} represents the minimum value.

[0014] In some embodiments of the present invention, when the data volume of the reverse path transmission delay sample set is not less than a preset value, fitting the reverse path transmission delay samples in the sliding window using a least squares regression method to obtain a theoretical propagation delay in a no-queue state, and using the theoretical propagation delay as a reference delay, includes: The confirmation packet sending timestamp is used as the independent variable and the reverse path transmission delay is used as the dependent variable to fit a linear model. The calculation formula of the linear model is: The optimal solution of the linear model formula is solved by the least squares method, and the calculation formula is: in, represents the i-th reverse path transmission delay sample in the sliding window, i∈W; α represents the clock drift rate; represents the timestamp of sending the ith confirmation packet in the sliding window; β represents the fitting intercept; ε i represents the measurement error term; n represents the number of current valid samples in the sliding window; The obtained fitting intercept is used as the reference delay, and the calculation formula is: in, represents the reference delay.

[0015] In some embodiments of the present invention, calculating the round-trip delay based on the data packet sending timestamp and the acknowledgment packet receiving timestamp; calculating the difference between the current reverse path transmission delay and the benchmark delay to obtain an additional queuing delay; and subtracting the additional queuing delay from the round-trip delay to obtain a corrected round-trip delay includes: The round-trip delay is calculated by subtracting the time when the confirmation packet is received from the time when the data packet is sent. The calculation formula is: RTT = t4 - t1; The additional queuing delay is calculated by subtracting the current reverse path transmission delay from the reference delay. The calculation formula is: The additional queuing delay is removed from the round trip delay to obtain the corrected round trip delay, which is calculated as follows: RTT corrected =RTT-ΔT b ; RTT represents the original round trip delay; t4 represents the timestamp of receiving the confirmation packet; t1 represents the timestamp of sending the data packet; ΔT b represents the additional queuing delay; T b represents the reverse path transmission delay; Represents the reference delay; RTT corrected Indicates the corrected round-trip delay.

[0016] In some embodiments of the present invention, the method further comprises: In a symmetric path, connections with the same source and destination addresses share the same baseline delay of the reverse path.

[0017] On the other hand, the present invention also provides a round-trip delay correction system based on reverse path timestamps, which, when executed, implements the steps of any of the methods mentioned above, comprising: The timestamp collection module is used to collect the data packet sending timestamp, confirmation packet sending timestamp and confirmation packet receiving timestamp through the network card hardware; A reference delay estimation module is configured to store reverse path transmission delay samples by maintaining a fixed-size sliding window and dynamically select an estimation strategy based on the number of samples to generate a reference delay; the estimation strategy includes a minimum value estimation method and a least squares regression method; The round-trip delay correction module is used to correct the round-trip delay according to the data packet sending timestamp, the confirmation packet sending timestamp, the confirmation packet receiving timestamp and the reference delay.

[0018] The present invention provides a round-trip delay correction method and system based on reverse path timestamps, comprising: collecting data packet sending timestamps, confirmation packet sending timestamps, and confirmation packet receiving timestamps; dynamically storing reverse path delay samples through a sliding window, and selecting a minimum value method or a least squares regression method to calculate a reference delay based on the sample size; eliminating reverse path congestion interference in the round-trip delay based on the difference between the current reverse path delay and the reference delay to obtain a corrected round-trip delay; and sharing the same source-destination pair reference delay under a symmetric topology. The present invention can accurately separate forward and reverse path congestion, avoiding misjudgments of traditional RTTs due to reverse path congestion, and improving the bandwidth utilization and short-flow performance of congestion control algorithms. Through hardware timestamps and dynamic baseline estimation, low-cost deployment without hardware modification is achieved, making it suitable for high-precision delay optimization in data center networks.

[0019] Additional advantages, objects, and features of the present invention will be set forth in part in the following description and will become apparent to those skilled in the art upon examination of the following or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained by the structures particularly pointed out in the description and drawings.

[0020] Those skilled in the art will understand that the purposes and advantages that can be achieved by the present invention are not limited to the above specific descriptions, and the above and other purposes that can be achieved by the present invention will be more clearly understood based on the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the present invention, constitute a part of this application, and do not constitute a limitation of the present invention. In the drawings: Figure 1 The figure is a schematic diagram of the steps of a round-trip delay correction method based on reverse path timestamps in one embodiment of the present invention.

[0022] Figure 2 Schematic diagram of collecting data packet sending and receiving timestamps and confirmation packet sending and receiving timestamps of any communicating parties in one embodiment of the present invention.

[0023] Figure 3 A typical architecture diagram of a data center network in one embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0025] It should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, the accompanying drawings only show structures and / or processing steps closely related to the solutions according to the present invention, while other details that are not closely related to the present invention are omitted.

[0026] It should be emphasized that the term "include / comprises" when used herein refers to the existence of features, elements, steps or components, but does not exclude the existence or addition of one or more other features, elements, steps or components.

[0027] It should also be noted that, unless otherwise specified, the term "connection" herein may refer not only to a direct connection but also to an indirect connection involving an intermediate.

[0028] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0029] It should be emphasized here that the step marks mentioned below do not limit the order of the steps, but it should be understood that the steps can be executed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be executed simultaneously.

[0030] In order to solve the problems that the existing congestion control technology cannot effectively distinguish the congestion of the forward path and the reverse path in the round-trip delay, resulting in misjudgment of the network status, waste of bandwidth and resources, and degradation of network performance; the existing time synchronization-based methods are high in cost and low in accuracy, and are difficult to adapt to dynamic network environments; and the existing baseline delay update efficiency is low, which cannot meet the high requirements of data centers for real-time and accuracy, the present invention provides a round-trip delay correction method based on the reverse path timestamp, such as Figure 1 As shown, the method includes the following steps S101 to S104: Step S101: collecting the data packet sending timestamp, the confirmation packet sending timestamp and the confirmation packet receiving timestamp; calculating the reverse path transmission delay according to the confirmation packet sending timestamp and the confirmation packet receiving timestamp.

[0031] Step S102: maintaining a sliding window of a fixed size, dynamically storing a preset number of reverse path transmission delay samples within a preset time period, and constructing a reverse path transmission delay sample set.

[0032] Step S103: When the data volume of the reverse path transmission delay sample set is less than a preset value, the minimum value is selected from the reverse path transmission delay samples in the sliding window as the reference delay. When the data volume of the reverse path transmission delay sample set is not less than the preset value, the reverse path transmission delay samples in the sliding window are fitted using the least squares regression method to obtain the theoretical propagation delay in a no-queue state, and the theoretical propagation delay is used as the reference delay. Step S104: Calculate the round-trip delay based on the data packet sending timestamp and the acknowledgment packet receiving timestamp; calculate the difference between the current reverse path transmission delay and the benchmark delay to obtain the additional queuing delay; and remove the additional queuing delay from the round-trip delay to obtain the corrected round-trip delay.

[0033] In step S101, Figure 2 As shown, any two communicating parties, such as host A and host B, pass through multiple forwarding devices (such as switches or routers).

[0034] Collect the timestamp recorded when host A sends the data packet, that is, the data packet sending timestamp; the timestamp recorded when host B returns the confirmation packet (ACK) after receiving the data packet, that is, the confirmation packet sending timestamp; and the timestamp recorded when host A receives the confirmation packet, that is, the confirmation packet receiving timestamp.

[0035] In some embodiments, Rx / Tx timestamps are collected by the network card hardware (such as the PTP function of the Itntel NIC), and the difference between the network card and the host CPU clock is calibrated through linear mapping to ensure time synchronization.

[0036] In some embodiments, the data packet send timestamp, acknowledgment packet send timestamp, and acknowledgment packet receive timestamp are transmitted via a cross-layer timestamp frame structure. As shown in Table 1, the frame structure includes link layer, network layer, and transport layer fields for identifying the standard protocol stack, as well as an extended protocol field for carrying various timestamp information.

[0037] Table 1 In some embodiments, the above-mentioned frame structure can report timestamp information to user state through standard single root I / O virtualization (SR-IOV) technology or data plane development kit (DPDK) without relying on hardware modification, and carry latency analysis data through extended protocol fields to facilitate use by subsequent computing modules.

[0038] In some embodiments, the reverse path transmission delay is calculated based on the confirmation packet sending timestamp and the confirmation packet receiving timestamp, and the calculation formula is shown in formula (1): Tb =t4-t3;(1) Among them, T b Indicates the reverse path transmission delay; t4 indicates the timestamp of sending the confirmation packet; t3 indicates the timestamp of receiving the confirmation packet.

[0039] In step S102, a sliding window of a fixed size is maintained to dynamically store a preset number of reverse path transmission delay samples within a preset time period to construct a reverse path transmission delay sample set.

[0040] In some embodiments, the reverse path transmission delay sample set is as shown in formula (2): Where T represents the reverse path transmission delay sample set; represents the Wth reverse path transmission delay sample in the sliding window; W is the size of the sliding window.

[0041] In step S103, different estimation strategies are selected according to the amount of sample data to obtain a reference delay.

[0042] When the amount of data in the reverse path transmission delay sample set is less than the preset value, the minimum value estimation method is used to select the minimum value from the reverse path transmission delay samples in the sliding window as the reference delay, as shown in formula (3): in, Indicates the benchmark delay; min{·} indicates the minimum value. represents the Wth reverse path transmission delay sample in the sliding window.

[0043] If the current new reverse path transmission delay sample is smaller, the reference delay is updated, as shown in formula (4): in, Indicates the current new reverse path transmission delay sample.

[0044] Otherwise, the current new reverse path transmission delay sample is added to the sliding window, keeping the original reference delay unchanged.

[0045] In some embodiments, in response to estimation deviations caused by expiration of old samples, the time range of the sliding window is set. Preferably, the time range is 2 seconds, and the data in the sliding window is periodically rescanned to ensure the timeliness and robustness of the estimation.

[0046] When the data volume of the reverse path transmission delay sample set is not less than the preset value, the least squares regression method is used to fit the reverse path transmission delay samples in the sliding window to obtain the theoretical propagation delay in the no-queue state, and the theoretical propagation delay is used as the benchmark delay.

[0047] First, we use the confirmation packet sending timestamp as the independent variable and the reverse path transmission delay as the dependent variable to fit a linear model (assuming that the baseline delay changes slowly in a short period of time), as shown in formula (5): Then, the optimal solution of the linear model formula is solved by the least square method, as shown in formulas (6) and (7): in, represents the i-th reverse path transmission delay sample in the sliding window, i∈W; α represents the clock drift rate; represents the timestamp of sending the ith confirmation packet in the sliding window; β represents the fitting intercept; ε i Represents the measurement error term; n represents the number of current valid samples in the sliding window.

[0048] Finally, the obtained fitting intercept β is used as the reference delay, as shown in formula (8): in, Indicates the baseline delay.

[0049] From the above, we can see that depending on the size of the sample data, either the minimum estimation method or the least squares regression method is selected to calculate the benchmark latency. This dynamic selection mechanism can balance accuracy and stability while maintaining low implementation complexity and low resource consumption. It is particularly suitable for deployment in data center environments that do not have hardware clock synchronization conditions.

[0050] Among them, Figure 3 The figure shows a typical architecture diagram of a data center network. A data center network is a communication infrastructure specifically used to connect servers, storage devices, and other computing resources within a data center. It is designed to achieve high-speed, low-latency, and highly reliable data transmission, supporting modern applications such as cloud computing, big data, and AI. As mentioned in the background technology, round-trip delay is a key signal for congestion control in a data center network. However, the existing technology cannot distinguish between forward / reverse path congestion, resulting in bandwidth waste and performance degradation. The reverse path delay correction solution provided by the present invention uses hardware timestamps and dynamic baseline estimation to accurately separate congestion factors, directly optimizing the throughput and latency of the data center network.

[0051] In some embodiments, in a data center network, multiple connections (TCP streams / RDMA QPs, etc.) between the same source-destination host pair (e.g., Host A and Host B) typically follow the same physical path (symmetric topology). If each connection independently measures the reverse path baseline latency, this will result in: multiple calculations of the baseline latency for the same path, wasting CPU and bandwidth resources. Furthermore, due to differences in measurement time, the baseline latency for each connection may differ, affecting the accuracy of round-trip delay correction. Therefore, in this embodiment, connections with the same source and destination addresses share the same reverse path baseline latency.

[0052] Furthermore, a reuse determination mechanism is set, and the reuse benchmark delay must meet the following conditions: Quintuple consistency: The source IP, destination IP, source port, destination port, and transport protocol (such as TCP / UDP) are exactly the same.

[0053] Hop count consistency: Verify that the path hop counts of two connections are the same through TTL (Time-To-Live) or explicit path detection (such as traceroute).

[0054] Time validity: The shared benchmark delay must be within the validity period (e.g. updated within the last 2 minutes). Expiration triggers recalculation of the benchmark delay.

[0055] In step S104, the round trip time is corrected according to the data packet sending timestamp, the confirmation packet sending timestamp, the confirmation packet receiving timestamp and the reference delay.

[0056] First, the original round-trip delay is calculated by subtracting the confirmation packet reception time from the data packet sending timestamp, as shown in formula (9): RTT = t4 - t1; (9) Then, the additional queuing delay is calculated by subtracting the current reverse path transmission delay from the reference delay calculated in step S103, as shown in formula (10): Finally, the additional queuing delay is removed from the original round-trip delay to obtain the corrected round-trip delay, as shown in formula (11): RTT corrected =RTT-ΔT b ;(11) Among them, RTT is the original round-trip delay; t4 is the timestamp of the confirmation packet reception; t1 is the timestamp of the data packet transmission; ΔT b represents the additional queuing delay; T b represents the reverse path transmission delay; Indicates the baseline delay; RTT corrected Indicates the corrected round-trip delay.

[0057] The corrected round-trip time can more accurately reflect the congestion situation on the forward path, thus avoiding misjudgment and unnecessary reduction in transmission rate caused by congestion on the reverse path.

[0058] In some embodiments, the round-trip delay correction method based on reverse path timestamps provided by the present invention can be further applied to the network health assessment system of the financial cloud backbone network to optimize network performance management through accurate delay measurement. Specific implementation methods are as follows: 1. Construction of Network Health Indicators: Reliability Assessment: Uses corrected round-trip delay to identify actual congestion on the forward path, avoiding misjudgments due to reverse path congestion and improving transmission reliability. For example, if the corrected round-trip delay consistently exceeds a threshold, an alarm or path switching is triggered.

[0059] Carrying capacity assessment: Calculates additional queuing delay based on dynamic baseline delay and real-time reverse path delay, quantifies the degree of reverse path congestion, and comprehensively assesses the network's remaining carrying capacity based on bandwidth utilization.

[0060] 2. Synergy between Health and Round Trip Delay Correction: Dynamic policy adjustment: If network health indicates a decrease in reliability (e.g., an abnormal increase in the corrected round-trip delay), congestion control parameters (e.g., the rate reduction gradient of the Swift protocol) will be automatically adjusted. If the carrying capacity is insufficient (e.g., the additional queuing delay continues to increase), traffic scheduling or capacity expansion recommendations will be triggered.

[0061] Data support: The confirmation packet sending timestamp, confirmation packet receiving timestamp and calculated benchmark delay provided by the present invention provide high-precision delay data for the health model, replacing traditional inefficient probe measurements.

[0062] The round-trip delay correction method based on reverse path timestamps provided by the present invention is applied to the network health assessment system of the financial cloud backbone network. The reverse path interference is eliminated by correcting the round-trip delay, making the health assessment more objective. At the same time, sharing the same source-destination pair benchmark delay can reduce the computational overhead of the health assessment.

[0063] Corresponding to the round-trip delay correction method based on reverse path timestamps, the present invention also provides a round-trip delay correction system based on reverse path timestamps. When executed, the system can implement the steps of the round-trip delay correction method based on reverse path timestamps. The system includes: The timestamp collection module is used to collect the data packet sending timestamp, confirmation packet sending timestamp and confirmation packet receiving timestamp through the network card hardware.

[0064] The baseline delay estimation module maintains a fixed-size sliding window to store reverse path transmission delay samples and dynamically selects an estimation strategy based on the number of samples to generate the baseline delay. The estimation strategies include minimum value estimation and least squares regression.

[0065] The round trip delay correction module is used to correct the round trip delay according to the data packet sending timestamp, the confirmation packet sending timestamp, the confirmation packet receiving timestamp and the reference delay.

[0066] Corresponding to the above method, the present invention also provides an electronic device, which includes a computer device, the computer device includes a processor and a memory, the memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the electronic device implements the steps of the method described above.

[0067] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned edge computing server deployment method. The computer-readable storage medium may be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium known in the art.

[0068] It should be understood by those skilled in the art that the various exemplary components, systems and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software or a combination of the two. Whether it is specifically performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present invention are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link via a data signal carried in a carrier.

[0069] It should be understood that the present invention is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted. In the above embodiments, several specific steps are described and illustrated as examples. However, the method of the present invention is not limited to the specific steps described and illustrated. Those skilled in the art may make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present invention.

[0070] In the present invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or replace features of other embodiments.

[0071] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations to the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A round trip delay correction method based on reverse path timestamp, characterized in that: The method comprises the following steps: collecting a data packet sending timestamp, a confirmation packet sending timestamp and a confirmation packet receiving timestamp; calculating a reverse path transmission delay based on the confirmation packet sending timestamp and the confirmation packet receiving timestamp; Maintain a fixed-size sliding window, dynamically store a preset number of reverse path transmission delay samples within a preset time period, and construct a reverse path transmission delay sample set; When the data volume of the reverse path transmission delay sample set is less than a preset value, selecting a minimum value from the reverse path transmission delay samples in the sliding window as a reference delay; when the data volume of the reverse path transmission delay sample set is not less than a preset value, using a least squares regression method to fit the reverse path transmission delay samples in the sliding window to obtain a theoretical propagation delay in a no-queue state, and using the theoretical propagation delay as the reference delay; The round-trip delay is calculated based on the data packet sending timestamp and the confirmation packet receiving timestamp; the difference between the current reverse path transmission delay and the benchmark delay is calculated to obtain an additional queuing delay; and the additional queuing delay is eliminated from the round-trip delay to obtain a corrected round-trip delay.

2. The round trip delay correction method based on reverse path timestamp according to claim 1, characterized in that: The method further comprises: The data packet sending timestamp, the confirmation packet sending timestamp and the confirmation packet receiving timestamp are transmitted via a cross-layer timestamp frame structure; The frame structure includes standard protocol fields and extended protocol fields of the link layer, the network layer, and the transport layer, and the extended protocol fields are used to carry various timestamp information.

3. The round trip delay correction method based on reverse path timestamp according to claim 2, characterized in that: The method further comprises: The frame structure reports the timestamp information to the user state through a single-root I / O virtualization technology or a data plane development kit.

4. The round trip delay correction method based on reverse path timestamp according to claim 1, characterized in that: The reverse path transmission delay is calculated based on the confirmation packet sending timestamp and the confirmation packet receiving timestamp, and the calculation formula is: T b =t4-t3; Among them, T b represents the reverse path transmission delay; t4 represents the timestamp of sending the confirmation packet; t3 represents the timestamp of receiving the confirmation packet.

5. The round trip delay correction method based on reverse path timestamp according to claim 1, characterized in that: Maintain a sliding window of fixed size, dynamically store a preset number of reverse path transmission delay samples within a preset time period, and construct a reverse path transmission delay sample set. The reverse path transmission delay sample set is expressed as: Wherein, T represents the reverse path transmission delay sample set; represents the Wth reverse path transmission delay sample in the sliding window; W is the size of the sliding window.

6. The round trip delay correction method based on reverse path timestamp according to claim 5, characterized in that: When the data volume of the reverse path transmission delay sample set is less than a preset value, the minimum value is selected from the reverse path transmission delay samples in the sliding window as the reference delay, and the calculation formula is: in, represents the reference delay; min{·} represents the minimum value.

7. The round trip delay correction method based on reverse path timestamp according to claim 5, characterized in that: When the data volume of the reverse path transmission delay sample set is not less than a preset value, fitting the reverse path transmission delay samples in the sliding window using a least squares regression method to obtain a theoretical propagation delay in a no-queue state, and using the theoretical propagation delay as a reference delay, including: The confirmation packet sending timestamp is used as the independent variable and the reverse path transmission delay is used as the dependent variable to fit a linear model. The calculation formula of the linear model is: The optimal solution of the linear model formula is solved by the least squares method, and the calculation formula is: in, represents the i-th reverse path transmission delay sample in the sliding window, i∈W; α represents the clock drift rate; represents the timestamp of sending the ith confirmation packet in the sliding window; β represents the fitting intercept; ε i represents the measurement error term; n represents the number of current valid samples in the sliding window; The obtained fitting intercept is used as the reference delay, and the calculation formula is: in, represents the reference delay.

8. The round trip delay correction method based on reverse path timestamp according to claim 1, characterized in that: Calculate the round trip delay based on the data packet sending timestamp and the confirmation packet receiving timestamp; Calculating the difference between the current reverse path transmission delay and the reference delay to obtain an additional queuing delay; Eliminating the additional queuing delay from the round-trip delay to obtain a corrected round-trip delay includes: The round-trip delay is calculated by subtracting the time when the confirmation packet is received from the time when the data packet is sent. The calculation formula is: RTT = t4 - t1; The additional queuing delay is calculated by subtracting the current reverse path transmission delay from the reference delay. The calculation formula is: The additional queuing delay is removed from the round trip delay to obtain the corrected round trip delay, which is calculated as follows: RTT corrected =RTT-ΔT b ; RTT represents the original round trip delay; t4 represents the timestamp of receiving the confirmation packet; t1 represents the timestamp of sending the data packet; ΔT b represents the additional queuing delay; T b represents the reverse path transmission delay; Represents the reference delay; RTT corrected Indicates the corrected round-trip delay.

9. The round trip delay correction method based on reverse path timestamp according to claim 1, characterized in that: The method further comprises: In a symmetric path, connections with the same source and destination addresses share the same baseline delay of the reverse path.

10. A round trip delay correction system based on reverse path timestamp, characterized in that: When the system is executed, the steps of the method according to any one of claims 1 to 9 are implemented, and the system comprises: The timestamp collection module is used to collect the data packet sending timestamp, confirmation packet sending timestamp and confirmation packet receiving timestamp through the network card hardware; A reference delay estimation module is configured to store reverse path transmission delay samples by maintaining a fixed-size sliding window and dynamically select an estimation strategy based on the number of samples to generate a reference delay; the estimation strategy includes a minimum value estimation method and a least squares regression method; The round-trip delay correction module is configured to correct the round-trip delay according to the data packet sending timestamp, the confirmation packet sending timestamp, the confirmation packet receiving timestamp and the reference delay.

Images