Industrial Internet of Things (IIoT) based TCP-based low-latency data transmission protocol optimization system and method

Abstract

This application relates to the field of industrial Internet of Things (IIoT) technology, and in particular to an IIoT-based TCP-based low-latency data transmission protocol optimization system and method. It includes a sending-end optimization processing unit and a receiving-end optimization processing unit. The sending end diagnoses the cause of packet loss through a real-time channel state inference module and switches the congestion control mode using a dual-mode congestion control engine module, while adjusting the transmission strategy in conjunction with an adaptive data fragmentation and response request module. The receiving end responds to immediate response requests through an enhanced response generation module. This application can distinguish between "congestion-related packet loss" and "non-congestion-related packet loss," adopting differentiated control strategies to reduce transmission latency and jitter, improve data transmission stability and rapid recovery capabilities, and provide a reliable guarantee for efficient data transmission in the IIoT.

Description

Technical Field

[0001] This invention belongs to the field of industrial Internet of Things (IoT) technology, specifically an optimization system and method for low-latency data transmission protocols based on TCP in industrial IoT. Background Technology

[0002] In the data transmission process of the Industrial Internet of Things (IIoT), the TCP protocol stack serves as the core communication foundation, and its performance directly affects the efficiency and stability of data interaction among massive numbers of devices, sensors, and control systems. Industrial environments place stringent demands on the real-time performance, determinism, and reliability of data transmission. Especially in critical operations such as distributed control, remote operation, and state synchronization, microsecond-level latency fluctuations can have a significant impact on production safety and system stability.

[0003] Currently, data transmission optimization in the Industrial Internet of Things (IIoT) primarily relies on two representative technical solutions. One type of solution constructs a communication transmission model for specific application scenarios, optimizing application-layer or physical-layer parameters such as data packet length and the number of spectrum sensing operations to reduce transmission latency. For example, Chinese patent CN113766536B discloses a method for optimizing wireless image data transmission in ultra-high voltage converter stations. This method uses static modeling to pre-set optimal parameter combinations, avoiding some known performance bottlenecks and achieving good results in relatively stable application scenarios. The other type of solution focuses on network routing and protocol selection. It evaluates end-to-end transmission path performance using shortest path algorithms and dynamically switches communication protocols using Bayesian decision-making. For example, Chinese patent CN117155845B discloses an IoT data interaction method and system. This system selects the optimal path and protocol combination through real-time evaluation and decision-making, providing a flexible adaptation mechanism and improving the overall efficiency of data transmission.

[0004] However, as the complexity and integration of Industrial Internet of Things (IIoT) applications increase, the aforementioned technical solutions are gradually revealing common limitations. These solutions tend to treat the TCP protocol stack as a functionally fixed "black box," with optimization strategies mainly focusing on upper-layer inputs or external selections, failing to delve into the core operating mechanism of the protocol stack for adaptive modification. The IIoT network environment has significantly different characteristics from the traditional Internet: First, business traffic is highly heterogeneous, including both periodically sent small status packets and sudden large data blocks; second, network topology is dynamically changing, with the access and movement of numerous wireless nodes causing rapidly fluctuating channel quality; third, channel interference is complex, with frequent non-congestion packet loss caused by physical factors such as electromagnetic noise and signal obstruction. Against this backdrop, the congestion control mechanisms inherent in standard TCP protocols (such as CUBIC and Reno) become a deep-seated source of performance bottlenecks. This mechanism equates any form of packet loss or timeout with a network congestion signal, triggering a drastic contraction and backoff of the congestion window. When applied to industrial wireless environments, random packet loss caused by channel noise or transient interference can be misjudged as severe congestion, leading to a sharp drop in transmission rate and unnecessary long waiting times, significantly increasing transmission latency and jitter. Existing solutions, whether through static parameter optimization or macroscopic path selection, struggle to fundamentally address this "misjudgment" problem caused by the mismatch between the protocol's inherent logic and the actual physical environment. This disconnect between the optimization level and the root cause of the problem creates a technical paradox: congestion control mechanisms designed to ensure reliability become a major obstacle to low-latency, high-stability transmission in specific industrial scenarios.

[0005] In summary, existing technologies either focus on the static optimization of application-layer parameters or on macroscopic scheduling at the network layer, generally treating the TCP protocol stack as a single functional unit without delving into its internal mechanisms to address the fundamental mismatch between its core congestion control algorithm and the dynamic, disruptive, and heterogeneous channel characteristics of the Industrial Internet of Things (IIoT). This limitation makes it difficult to achieve stable and effective control of transmission delay and jitter in complex industrial scenarios, failing to meet the demands of high-standard industrial applications. Therefore, this invention provides an IIoT-based TCP-based low-latency data transmission protocol optimization system and method. Summary of the Invention

[0006] The purpose of this invention is to overcome the above-mentioned defects of the prior art and provide an optimization method and system for low-latency TCP-based data transmission protocols in the Industrial Internet of Things. It aims to delve into the core of the Transmission Control Protocol and construct an intelligent diagnostic mechanism that can accurately identify network congestion and physical channel disturbances. By linking and implementing differentiated congestion control, data fragmentation, and acknowledgment response strategies, it fundamentally solves the performance problems caused by the "misjudgment" of packet loss in the standard TCP protocol in complex industrial environments, thereby achieving ultra-low latency and high stability in data transmission.

[0007] To achieve the above-mentioned objectives, this invention provides an industrial IoT-based TCP-based low-latency data transmission protocol optimization system. The system is deployed on the sending and receiving devices that constitute the data transmission path. The optimization system includes:

[0008] The sending-end optimization processing unit, located in the sending-end device, is responsible for diagnosing the cause of packet loss based on pre-acquired network transmission dynamic characteristics when a packet loss event is detected, and adaptively switching between at least two different congestion control modes and executing corresponding control strategies according to the diagnosis conclusion.

[0009] The receiver optimization processing unit, located in the receiver device, is used to respond to the coordination instructions sent by the transmitter optimization processing unit in a specific congestion control mode, and adjust the acknowledgment response (ACK) generation strategy to cooperate with the control strategy of the transmitter optimization processing unit.

[0010] Furthermore, the sending end optimization processing unit integrates the following modules:

[0011] The real-time channel state inference module, which is embedded in the kernel protocol stack, continuously monitors and analyzes the dynamic characteristics of the round-trip delay of data packets, and outputs a diagnostic conclusion on whether the packet loss event is "congestion-related packet loss" or "non-congestion-related packet loss" when a packet loss event is detected.

[0012] The dual-mode congestion control engine module has at least two operating states: congestion avoidance state and channel fluctuation adaptation state. It switches between the two states based on the diagnostic conclusions output by the real-time channel state inference module.

[0013] The adaptive data fragmentation and response request module is coupled with the current operating state of the dual-mode congestion control engine module. It is used to perform coordinated operations such as adjusting the data segment size and requesting an immediate response from the receiver when the dual-mode congestion control engine module switches to the channel fluctuation adaptation state.

[0014] Furthermore, the real-time channel state inference module includes the following sub-modules:

[0015] The time delay reference parameter calculation submodule continuously updates a smooth round-trip time delay and a time delay variance based on the real-time received round-trip time delay sample values ​​using an exponentially weighted moving average algorithm.

[0016] The time delay dynamic feature analysis submodule calculates the first derivative of the round-trip time delay sequence in the time series data queue, i.e., the time delay gradient, based on a time series data queue containing N round-trip time delay sample values ​​from the past.

[0017] The packet loss cause diagnosis submodule is activated when a packet loss event is detected. It reads the delay gradient and delay variance before the packet loss event occurs and has a built-in set of diagnostic criteria to determine whether the packet loss is "congestion packet loss" or "non-congestion packet loss" based on the comparison results of the delay gradient and delay variance values ​​with preset thresholds.

[0018] Furthermore, the diagnostic criterion logic built into the packet loss cause diagnosis submodule is as follows:

[0019] If, before a packet loss event occurs, the absolute value of the delay gradient is greater than a preset first gradient threshold, and the delay variance is greater than a preset first variance threshold, then the packet loss event is determined to be a "congestion-related packet loss".

[0020] Conversely, if the absolute value of the delay gradient is less than a preset second gradient threshold, and the delay variance is less than a preset second variance threshold, then the packet loss event is determined to be a "non-congestion packet loss"; wherein, the values ​​of the first gradient threshold and the first variance threshold are dynamically set to the current smooth round-trip time multiplied by a first multiplier factor α; and, the values ​​of the second gradient threshold and the second variance threshold are dynamically set to the current smooth round-trip time multiplied by a second multiplier factor β, wherein the value range of the first multiplier factor α is [0.5, 1.5], the value range of the second multiplier factor β is [0.1, 0.4], and α > β.

[0021] Furthermore, the state transition and control logic of the dual-mode congestion control engine module is as follows:

[0022] When the real-time channel state inference module reports "congestion packet loss", the engine maintains or enters congestion avoidance state, sets the slow start threshold to half of the current congestion window size, and resets the congestion window to a maximum segment length.

[0023] When the real-time channel state inference module reports a "non-congestion packet loss", the engine switches from congestion avoidance state to channel fluctuation adaptation state, while keeping the current slow start threshold unchanged and reducing the congestion window by a preset fixed number.

[0024] Furthermore, the specific implementation logic of the adaptive data sharding and response request module is as follows:

[0025] When the dual-mode congestion control engine module switches to the channel fluctuation adaptation state, its internal dynamic segment length adjustment submodule is activated, and the maximum segment length value used by the sender is temporarily reduced by a specific scaling factor.

[0026] At the same time, its internal customized TCP option generation and parsing submodule is activated, inserting a predefined customized TCP option into the header of all TCP packets sent from the sender, and setting a specific flag bit in the option to request an immediate response from the receiver.

[0027] Furthermore, the receiving end optimization processing unit integrates an enhanced response generation module, the specific implementation logic of which is as follows:

[0028] Under normal circumstances, the response is carried out in accordance with the standard delayed ACK strategy;

[0029] When the received TCP packet contains the customized TCP option and the request immediate acknowledgment flag is set, the acknowledgment behavior mode is changed, and an independent acknowledgment (ACK) is immediately generated and sent for each successfully received packet with the flag set.

[0030] On the other hand, this application also provides an optimization method for TCP-based low-latency data transmission protocols in the Industrial Internet of Things (IIoT), applicable to the aforementioned IIoT TCP-based low-latency data transmission protocol optimization system, specifically including the following steps:

[0031] S1: During data transmission, the sending-end optimization processing unit deployed on the sending-end device continuously monitors the dynamic characteristics of network transmission, and the dynamic characteristics include at least the round-trip delay of data packets and their statistical derived values;

[0032] S2: When the sending end optimization processing unit detects a data packet loss event, it calls its internal real-time channel state inference module to diagnose the cause of the packet loss event based on the network transmission dynamic characteristics recorded before the packet loss event occurred, and outputs the diagnosis conclusion as "congestion packet loss" or "non-congestion packet loss".

[0033] S3: The sending end optimization processing unit performs a differentiated response based on the diagnostic conclusion: if the diagnostic conclusion is "congestion-related packet loss", the slow start threshold is set to half of the current congestion window size, and the congestion window is reset to a maximum segment length; if the diagnostic conclusion is "non-congestion-related packet loss", the unit switches to the channel fluctuation adaptation state, keeps the current slow start threshold unchanged, reduces the congestion window by a preset fixed number, and carries a signaling request for immediate response in subsequent data packets.

[0034] S4: After receiving a data packet with a request for immediate response signaling, the receiver optimization processing unit deployed on the receiver device suspends the standard delayed response strategy and instead sends an immediate acknowledgment response for each successfully received such data packet, so as to coordinate with the sender for rapid recovery.

[0035] Furthermore, the process of diagnosing the cause of packet loss in S2 includes the following steps:

[0036] Before a packet loss event occurs, the smooth round-trip time, delay variance, and delay gradient of the network are continuously calculated and updated.

[0037] When a packet loss event occurs, compare the absolute value of the current delay gradient and the delay variance with a set of dynamic thresholds;

[0038] The dynamic threshold is dynamically set based on the current smooth round-trip time.

[0039] If both the absolute value of the delay gradient and the delay variance are greater than their respective thresholds, the packet loss is determined to be "congestion-related"; otherwise, it is determined to be "non-congestion-related".

[0040] The beneficial effects of this invention are as follows:

[0041] 1. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system and method of the present invention solves the problem of reduced transmission performance caused by misjudgment of packet loss in the industrial IoT environment by introducing a real-time channel state inference module, a dual-mode congestion control engine module, and an adaptive data fragmentation and response request module.

[0042] 2. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system and method of the present invention effectively reduces transmission latency and jitter by distinguishing between "congestion-induced packet loss" and "non-congestion-induced packet loss" and adopting differentiated control strategies for different packet loss types, thereby improving the stability and flexibility of data transmission. In addition, through the synergistic effect of customized TCP options and an immediate response mechanism, the system's rapid recovery capability is further enhanced, providing a reliable guarantee for efficient data transmission in industrial IoT scenarios. Attached Figure Description

[0043] The invention will now be further described with reference to the accompanying drawings.

[0044] Figure 1 This is a structural block diagram of the system of the present invention;

[0045] Figure 2 This is a flowchart illustrating the method of the present invention.

[0046] In the diagram: 1. Sending end optimization processing unit; 2. Receiving end optimization processing unit; 3. Real-time channel state inference module; 4. Dual-mode congestion control engine module; 5. Adaptive data fragmentation and acknowledgment request module; 6. Enhanced acknowledgment generation module; 7. Delay baseline parameter calculation submodule; 8. Delay dynamic feature analysis submodule; 9. Packet loss cause diagnosis submodule; 10. Dynamic segment length adjustment submodule; 11. Customized TCP option generation and parsing submodule. Detailed Implementation

[0047] This invention provides a TCP-based low-latency data transmission protocol optimization system and method for the Industrial Internet of Things (IIoT). Its core lies in the collaborative work of the sending-end optimization processing unit 1 and the receiving-end optimization processing unit 2 to achieve real-time monitoring of dynamic network transmission characteristics, accurate diagnosis of packet loss causes, and execution of differentiated congestion control strategies. The following is combined with… Figure 1 and Figure 2 The specific embodiments of the present invention will be described in detail below.

[0048] exist Figure 1 In this module, the sender optimization processing unit 1 is deployed on the sender device. It contains three main modules: a real-time channel state inference module 3, a dual-mode congestion control engine module 4, and an adaptive data fragmentation and acknowledgment request module 5. These modules are logically connected to form a complete functional chain. The real-time channel state inference module 3 is responsible for continuously monitoring the dynamic changes in the round-trip delay of data packets and activating its internal sub-modules for analysis when a packet loss event is detected. The delay baseline parameter calculation sub-module 7 within this module uses an exponentially weighted moving average algorithm to smooth the received round-trip delay sample values, generating a smoothed round-trip delay and a time-to-smooth delay. The delay variance and delay dynamic feature analysis submodule 8 calculates the first derivative based on the time series data queue of the past N round-trip delay sample values ​​to obtain the delay gradient. After detecting a packet loss event, the packet loss cause diagnosis submodule 9 reads the delay gradient and delay variance, and outputs a diagnosis conclusion according to the preset diagnostic criteria. The core of the diagnostic criteria is to compare whether the absolute value of the delay gradient and the delay variance exceed the dynamically set first gradient threshold and first variance threshold. If both exceed, it is judged as "congestion packet loss"; otherwise, it is judged as "non-congestion packet loss". The cooperation between the above modules ensures the accuracy and real-time performance of the packet loss cause diagnosis process.

[0049] The dual-mode congestion control engine module 4 is directly connected to the real-time channel state inference module 3. It receives the diagnostic conclusions output by the module and switches the operating state accordingly. When the diagnostic conclusion is "congestion-related packet loss", the engine switches to the congestion avoidance state, sets the slow start threshold to half of the current congestion window size, and resets the congestion window to a maximum segment length. When the diagnostic conclusion is "non-congestion-related packet loss", the engine switches to the channel fluctuation adaptation state, keeps the current slow start threshold unchanged, and reduces the congestion window by a preset fixed number. This state switching mechanism enables the congestion control strategy to be dynamically adjusted according to the actual network environment, avoiding the performance degradation problem caused by the traditional TCP protocol when misjudging the cause of packet loss.

[0050] The adaptive data fragmentation and acknowledgment request module 5 operates coupled with the dual-mode congestion control engine module 4. Internally, it includes a dynamic segment length adjustment submodule 10 and a customized TCP option generation and parsing submodule 11. When the dual-mode congestion control engine module 4 switches to the channel fluctuation adaptation state, the dynamic segment length adjustment submodule 10 is activated, temporarily reducing the maximum segment length used by the sender by a specific scaling factor. At the same time, the customized TCP option generation and parsing submodule 11 inserts a predefined customized TCP option into the header of all TCP packets sent from the sender and sets the request for immediate acknowledgment flag in the option. The synergistic effect of these two submodules ensures the flexibility and rapid recovery capability of data transmission in the channel fluctuation adaptation state.

[0051] The receiver optimization processing unit 2 is deployed on the receiver device and integrates an enhanced response generation module 6. Under normal circumstances, this module follows the standard delayed response strategy. However, after receiving a data packet containing customized TCP options and with the request for immediate response flag set, it changes the response behavior mode and immediately generates and sends an independent acknowledgment response (ACK) for each successfully received data packet. This immediate response mechanism forms a closed loop with the adaptive data fragmentation and response request module 5 of the sender optimization processing unit 1, further improving the system's rapid recovery capability.

[0052] Figure 2 The specific flow of the method of the present invention is shown. In step S1, the sending end optimization processing unit 1 continuously monitors the dynamic characteristics of network transmission, including the round-trip time of data packets and their statistical derivatives such as smoothed round-trip time, time delay variance and time delay gradient. These characteristics are calculated and updated by each sub-module in the real-time channel state inference module 3.

[0053] In step S2, when a packet loss event is detected, the packet loss cause diagnosis submodule 9 performs a diagnosis based on the dynamic characteristics of network transmission before the packet loss event occurs. Specifically, it compares the current absolute value of the delay gradient and the delay variance with the dynamically set first gradient threshold and first variance threshold. If both exceed the threshold, it is determined to be "congestion-related packet loss"; otherwise, it is determined to be "non-congestion-related packet loss". The diagnosis conclusion is then transmitted to the dual-mode congestion control engine module 4.

[0054] In step S3, the dual-mode congestion control engine module 4 executes differentiated responses based on the diagnostic conclusion. If the diagnostic conclusion is "congestion-related packet loss", it switches to the congestion avoidance state and adjusts the slow start threshold and congestion window. If the diagnostic conclusion is "non-congestion-related packet loss", it switches to the channel fluctuation adaptation state and reduces the congestion window by a preset fixed number. At the same time, it carries a signaling request for immediate response in the subsequent data packets. This process is completed by the adaptive data fragmentation and response request module 5. Its internal dynamic segment length adjustment submodule 10 and customized TCP option generation and parsing submodule 11 are responsible for adjusting the segment length and inserting customized TCP options, respectively.

[0055] In step S4, after receiving a data packet with a request for immediate response signaling, the enhanced response generation module 6 of the receiving end optimization processing unit 2 suspends the standard delayed response strategy and instead sends an immediate acknowledgment response (ACK) for each successfully received data packet. This immediate response mechanism forms a closed-loop cooperation with the adaptive data fragmentation and response request module 5 of the sending end optimization processing unit 1 to ensure the rapid recovery capability of data transmission under channel fluctuation adaptation conditions.

[0056] In practical applications, such as distributed control systems on industrial automated production lines, the system of this invention can be deployed between multiple sensor nodes and a central controller. The sensor nodes act as transmitting devices, and the central controller acts as receiving devices. During data transmission, due to complex electromagnetic interference and signal obstruction in the industrial environment, random packet loss events occur frequently. Traditional TCP protocols misjudge these random packet losses as congestion signals, triggering a sharp contraction of the congestion window, resulting in a sudden drop in transmission rate and unnecessary long waiting times. However, the system of this invention uses the real-time channel state inference module 3 to accurately diagnose the causes of packet loss, distinguishing between "congestion-related packet loss" and "non-congestion-related packet loss." The dual-mode congestion control engine module 4 executes the corresponding congestion control strategy, effectively reducing transmission delay and jitter. In addition, the adaptive data fragmentation and response request module 5 enhances the system's flexibility and rapid recovery capability by adjusting the segment length and inserting customized TCP options. The enhanced response generation module 6 of the receiver optimization processing unit 2 forms a closed-loop collaboration with the sender through an instant response mechanism, further improving the system's stability and reliability.

[0057] This invention solves the problem of decreased transmission performance caused by misjudgment of packet loss in the industrial Internet of Things environment by the close cooperation between the above modules, and achieves the goal of low latency and high reliability data transmission.

[0058] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principles of this invention are further explained below in conjunction with a specific application scenario.

[0059] In industrial automated production lines, multiple sensor nodes in a distributed control system require high-frequency, low-latency data interaction with the central controller. These sensor nodes, acting as transmitting devices, are equipped with a transmitting-end optimization processing unit 1; the central controller, acting as a receiving device, is equipped with a receiving-end optimization processing unit 2. In this scenario, due to complex electromagnetic interference and signal obstruction in the industrial environment, random packet loss events occur frequently. Traditional TCP protocols misjudge these random packet losses as congestion signals, thereby triggering a sharp contraction of the congestion window, resulting in a sudden drop in transmission rate and unnecessary long waiting times. This invention effectively solves this problem through the collaborative work of a real-time channel state inference module 3, a dual-mode congestion control engine module 4, and an adaptive data fragmentation and response request module 5.

[0060] First, during data transmission, the real-time channel state inference module 3 continuously monitors network dynamic characteristics, including the round-trip time (RTT) of data packets and its statistical derivatives. The delay benchmark parameter calculation submodule 7 uses an exponentially weighted moving average algorithm to smooth the received RTT sample values, generating smoothed RTT and delay variance. Simultaneously, the delay dynamic feature analysis submodule 8 calculates the first derivative based on a time-series data queue of the past N RTT sample values ​​to obtain the delay gradient. When a packet loss event is detected, the packet loss cause diagnosis submodule 9 outputs a diagnostic conclusion based on the comparison between the delay gradient and delay variance and preset thresholds. If both the absolute value of the delay gradient and the delay variance exceed the first gradient threshold and the first variance threshold, it is determined to be "congestion-related packet loss"; otherwise, it is determined to be "non-congestion-related packet loss." This diagnostic mechanism ensures the accuracy of packet loss cause judgment and provides a basis for the subsequent implementation of differentiated congestion control strategies. The core of the diagnostic criterion is comparison... In a preferred embodiment, the first gradient threshold and the first variance threshold used to determine whether the absolute value of the delay gradient and the delay variance exceed dynamically set thresholds are compared. These thresholds are dynamically set by multiplying the current smoothed round-trip time by a first multiplier factor α. The second gradient threshold and the second variance threshold used to determine whether the delay is non-congestion-related are dynamically set by multiplying the current smoothed round-trip time by a second multiplier factor β. The values ​​of factors α and β can be adjusted based on empirical data of the network environment. For example, in wired industrial networks with less signal interference, α can be a small value close to 0.5, and β can be a small value close to 0.1. In complex wireless industrial environments, to avoid misjudgment, α can be a value close to 1.2, and β can be a value close to 0.3. Those skilled in the art can determine the optimal parameters by conducting a limited number of routine experiments in the target network environment to test the diagnostic accuracy under different factor combinations. This is not an excessive burden.

[0061] Secondly, the dual-mode congestion control engine module 4 dynamically switches its operating state based on the diagnosis of packet loss causes. When the diagnosis is "congestion-related packet loss", the engine switches to congestion avoidance state, sets the slow start threshold to half the current congestion window size, and resets the congestion window to a maximum segment length. This operation aims to quickly alleviate network congestion and prevent further deterioration. When the diagnosis is "non-congestion-related packet loss", the engine switches to channel fluctuation adaptation state, keeps the current slow start threshold unchanged, and reduces the congestion window by a preset fixed number. This adjustment method avoids a sudden drop in transmission rate due to misjudgment, while maintaining network stability.

[0062] Subsequently, the adaptive data fragmentation and acknowledgment request module 5 is activated when the dual-mode congestion control engine module 4 switches to the channel fluctuation adaptation state. The dynamic segment length adjustment submodule 10 temporarily reduces the maximum segment length used by the sender by a specific scaling factor to reduce the risk of transmission failure of a single data packet. At the same time, the customized TCP option generation and parsing submodule 11 inserts a predefined customized TCP option into the header of all TCP data packets sent from the sender and sets the request for immediate acknowledgment flag. This operation improves the system's rapid recovery capability by requesting immediate acknowledgment from the receiver.

[0063] At the receiving end, after receiving a data packet with a request for immediate acknowledgment flag, the enhanced acknowledgment generation module 6 suspends the standard delayed acknowledgment strategy and instead immediately generates and sends an independent acknowledgment (ACK) for each successfully received data packet. This immediate acknowledgment mechanism, together with the adaptive data fragmentation and acknowledgment request module 5 of the sending end optimization processing unit 1, forms a closed-loop cooperation to ensure the rapid recovery capability of data transmission under channel fluctuation adaptation. For example, when a data packet is lost due to channel noise, the sending end can quickly detect and resend the lost data packet through the immediate acknowledgment mechanism, thereby avoiding the lengthy waiting time caused by delayed acknowledgment in the traditional TCP protocol.

[0064] Furthermore, the system of this invention further enhances its flexibility through the synergistic effect of dynamic segment length adjustment and an instant response mechanism. In practical applications, when channel quality in industrial environments changes rapidly, the dynamic segment length adjustment submodule 10 can flexibly adjust the segment length according to the channel status, thereby reducing the risk of packet loss while ensuring transmission efficiency. The instant response mechanism, through a rapid feedback mechanism, shortens the recovery time after data packet loss, significantly reducing transmission latency and jitter. In summary, through the close collaboration between the above modules, this invention achieves real-time monitoring of dynamic network transmission characteristics, accurate diagnosis of packet loss causes, and execution of differentiated congestion control strategies. In the actual scenario of industrial automated production lines, the system of this invention can effectively distinguish between "congestion-related packet loss" and "non-congestion-related packet loss," and adopt differentiated control strategies for different packet loss types, thereby significantly reducing transmission latency and jitter and improving the stability and reliability of data transmission. This optimization scheme not only meets the demand for low-latency, high-reliability data transmission in the industrial IoT environment, but also provides a reliable guarantee for efficient communication in complex industrial scenarios.

[0065] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An industrial IoT-based TCP-based low-latency data transmission protocol optimization system, characterized in that, The system is deployed on the sending and receiving devices that constitute the data transmission path, and the optimization system includes: The transmitter optimization processing unit (1) located in the transmitter device is used to diagnose the cause of a data packet loss event based on pre-acquired network transmission dynamic characteristics when a data packet loss event is detected, and adaptively switch between at least two different congestion control modes and execute corresponding control strategies according to the diagnostic conclusions; the transmitter optimization processing unit (1) internally integrates: The real-time channel state inference module (3) is a logical processing unit embedded in the kernel protocol stack. It is used to continuously monitor and analyze the dynamic characteristics of the round-trip delay of data packets, and when a packet loss event is detected, it outputs a diagnostic conclusion on whether the packet loss event is "congestion-induced packet loss" or "non-congestion-induced packet loss". The real-time channel state inference module (3) includes: The time delay reference parameter calculation submodule (7) is used to continuously update a smooth round-trip time delay and a time delay variance based on the real-time received round-trip time delay sample values ​​using an exponential weighted moving average algorithm. The time delay dynamic feature analysis submodule (8) is used to calculate the first derivative of the round-trip time delay sequence in the time series data queue, i.e., the time delay gradient, based on the time series data queue containing the past N round-trip time delay sample values; The packet loss cause diagnosis submodule (9) is activated when a packet loss event is detected, and reads the delay gradient and delay variance before the packet loss event occurs. The packet loss cause diagnosis submodule (9) has a built-in set of diagnostic criteria, which are used to determine whether the packet loss event is "congestion packet loss" or "non-congestion packet loss" based on the comparison results of the values ​​of the delay gradient and delay variance with preset thresholds. The dual-mode congestion control engine module (4), as a congestion controller, has at least two operating states, namely congestion avoidance state and channel fluctuation adaptation state. The dual-mode congestion control engine module (4) switches between the congestion avoidance state and the channel fluctuation adaptation state according to the diagnostic conclusion output by the real-time channel state inference module (3). The adaptive data fragmentation and response request module (5) is coupled with the current operating state of the dual-mode congestion control engine module (4) and is used to perform coordinated operations of adjusting the data segment size and requesting an immediate response from the receiver when the dual-mode congestion control engine module (4) switches to the channel fluctuation adaptation state. The receiver optimization processing unit (2) is set in the receiver device. It is used to respond to the coordination instructions sent by the transmitter optimization processing unit (1) in a specific congestion control mode, and adjust its acknowledgment response generation strategy accordingly to cooperate with the control strategy of the transmitter optimization processing unit (1).

2. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system according to claim 1, characterized in that, The logic of the diagnostic criteria built into the packet loss cause diagnosis submodule (9) is as follows: If, before a packet loss event occurs, the absolute value of the delay gradient is greater than a preset first gradient threshold, and the delay variance is greater than a preset first variance threshold, then the packet loss event is determined to be "congestion-related packet loss". Conversely, if the absolute value of the delay gradient is less than a preset second gradient threshold and the delay variance is less than a preset second variance threshold, then the packet loss event is determined to be "non-congestion packet loss". The values ​​of the first gradient threshold and the first variance threshold are dynamically set to the current smooth round-trip time multiplied by a first multiplier factor α; and the values ​​of the second gradient threshold and the second variance threshold are dynamically set to the current smooth round-trip time multiplied by a second multiplier factor β, wherein the value range of the first multiplier factor α is [0.5, 1.5], the value range of the second multiplier factor β is [0.1, 0.4], and α > β.

3. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system according to claim 1, characterized in that, The state transition and control logic of the dual-mode congestion control engine module (4) is as follows: When the real-time channel state inference module (3) reports "congestion packet loss", the engine maintains or enters a congestion avoidance state and executes a standard congestion control response, which includes setting the slow start threshold to half of the current congestion window size and resetting the congestion window to a maximum segment length. When the real-time channel state inference module (3) reports a "non-congestion packet loss", the engine switches from congestion avoidance state to channel fluctuation adaptation state and executes a non-punitive window adjustment strategy.

4. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system according to claim 3, characterized in that, The non-penalty window adjustment strategy executed by the engine in the channel fluctuation adaptation state is specifically as follows: Keep the current slow start threshold unchanged; Reduce the congestion window by a preset fixed amount, less than its current value; After a successful retransmission and confirmation, the congestion window is restored at a controlled rate. Furthermore, when the preset exit conditions are met, the engine automatically returns from the channel fluctuation adaptation state to the congestion avoidance state. The exit conditions include: the number of data packets that are successfully retransmitted and acknowledged in the channel fluctuation adaptation state reaches a preset value, or the time spent in the channel fluctuation adaptation state exceeds a maximum duration.

5. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system according to claim 1, characterized in that, The specific functions of the adaptive data sharding and response request module (5) are as follows: When the dual-mode congestion control engine module (4) switches to the channel fluctuation adaptation state, the dynamic segment length adjustment submodule (10) inside the adaptive data fragmentation and response request module (5) is activated, and the maximum segment length value used by the sender is temporarily reduced by a specific scaling factor. The customized TCP option generation and parsing submodule (11) inside the adaptive data fragmentation and response request module (5) is activated. A predefined customized TCP option is inserted into the header of all TCP data packets sent from the sender, and a specific flag bit in the option is set to request an immediate response from the receiver.

6. The industrial IoT-based TCP-based low-latency data transmission protocol optimization system according to claim 5, characterized in that, The receiver optimization processing unit (2) integrates an enhanced response generation module (6); the specific function of the enhanced response generation module (6) is as follows: Under normal circumstances, a standard delayed response strategy should be followed when responding; When it detects that the received TCP packet contains the customized TCP option and that the request for immediate acknowledgment flag is set, it changes its acknowledgment behavior mode and immediately generates and sends an independent acknowledgment response for each successfully received packet with the flag set.

7. A method for optimizing a low-latency TCP-based data transmission protocol for the Industrial Internet of Things (IIoT), applicable to the IIoT-based TCP-based low-latency data transmission protocol optimization system as described in any one of claims 1-6; the optimization method includes the following steps: S1: During data transmission, the sending end optimization processing unit (1) deployed on the sending end device continuously monitors the dynamic characteristics of network transmission, the dynamic characteristics including at least the round-trip delay of data packets and their statistical derived values; S2: When the sending end optimization processing unit (1) detects a data packet loss event, it calls its internal real-time channel state inference module (3) to diagnose the cause of the packet loss event based on the network transmission dynamic characteristics recorded before the packet loss event occurred, and outputs the diagnosis conclusion as "congestion packet loss" or "non-congestion packet loss". S3: The sending end optimization processing unit (1) performs a differentiated response based on the diagnostic conclusion: if the diagnostic conclusion is "congestion-related packet loss", then it performs strict congestion control measures, including significantly reducing the congestion window; If the diagnosis is "non-congestion packet loss", a series of coordinated optimization actions are performed, including: switching to a control state designed to adapt to channel fluctuations, performing a gentle, non-punitive reduction of the congestion window, temporarily reducing the maximum segment length of the transmitted data packets, and carrying a signaling request for immediate response in subsequent data packets. S4: After receiving a data packet with a request for immediate response signaling, the receiver optimization processing unit (2) deployed on the receiver device suspends the standard delayed response strategy and instead sends an immediate acknowledgment response for each such data packet that is successfully received, so as to coordinate with the sender for rapid recovery.

8. The method for optimizing low-latency TCP-based data transmission protocols for the Industrial Internet of Things according to claim 7, characterized in that, The process of diagnosing the cause of packet loss in S2 further includes: Before a packet loss event occurs, the smooth round-trip time, delay variance, and delay gradient of the network are continuously calculated and updated. When a packet loss event occurs, compare the absolute value of the current delay gradient and the delay variance with a set of dynamic thresholds; The dynamic threshold is dynamically set based on the current smooth round-trip time. If both the absolute value of the delay gradient and the delay variance are greater than their respective thresholds, it is determined to be "congestion-related packet loss" and the stringent congestion control measures are triggered; otherwise, it is determined to be "non-congestion-related packet loss" and the series of coordinated optimization actions are triggered.

Images