A communication method and system for intelligent internet of things guarantee low latency

Abstract

The application discloses a communication method and system for guaranteeing low time delay of intelligent Internet of Things, and relates to the technical field of Internet of Things.The method adopts a standard interference feature vector set BFv as input, and outputs a pulse prediction activation value Ap, to realize dynamic prediction of burst communication interference probability in a future scheduling period, which is different from a passive scheduling strategy relying on historical statistics or real-time feedback. The prediction mechanism is used as a soft decision trigger, so that the AGV communication node can predict potential risks before the interference occurs, thereby performing scheduling structure fine tuning or rearrangement operation in advance, effectively avoiding scheduling conflicts, window overlaps and other abnormal behaviors, greatly improving the active adaptation ability of the communication structure to unstable environment, and stably controlling the communication time delay within 5ms, to meet the requirements of high reliability and low delay intelligent Internet of Things scene.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, specifically to a communication method and system for ensuring low latency in smart IoT. Background Technology

[0002] As a crucial direction for the integration of information technology and intelligent sensing systems, IoT communication technology is widely applied in various real-time collaborative scenarios such as smart transportation, intelligent manufacturing, and smart logistics. Especially in the field of smart logistics, ports, as typical high-density node communication environments, typically deploy a large number of Automated Guided Vehicles (AGVs) for material handling and path coordination. These AGVs rely on low-latency, high-stability communication networks to achieve refined scheduling and path synchronization. To enhance the intelligence level of port logistics systems, edge communication modules and intelligent communication scheduling mechanisms have been gradually deployed in AGV onboard terminals and yard edge devices to achieve efficient and reliable communication interaction in port scenarios.

[0003] Currently, existing AGV group communication systems mostly employ fixed-time segmented scheduling or adaptive polling mechanisms for communication scheduling. While these methods offer some delay control under normal interference conditions, in port yard environments, the dense stacking of metal containers, significant multipath diffraction effects, and frequent occurrences of sudden electromagnetic pulse interference prevent traditional scheduling strategies from accurately detecting and predicting potential communication mismatch risks. Especially in dynamic collaborative operations of AGV groups, the uncertainty, suddenness, and local concentration of interference frequently lead to imbalances in node scheduling priorities and overlapping transmission windows, directly impacting the communication stability and overall scheduling efficiency of the scheduling system.

[0004] Because existing communication scheduling strategies lack a mechanism to predict the probability of instantaneous interference pulses, the system can only passively execute window rearrangement or node rate limiting after explicit communication conflicts occur, lacking feedforward dynamic scheduling awareness. This lag in response prevents timely adjustment of scheduling strategies in the initial stage of interference, leading to the following problems: First, the increased packet retransmission rate caused by sudden interference accumulates link load, exacerbating congestion risks; second, time slot conflicts or window overlap directly cause delayed transmission of AGV control commands, disrupting synchronous scheduling between nodes and potentially causing abnormal port operation events such as path conflicts and task failures. Therefore, there is an urgent need to establish an intelligent communication scheduling method with the ability to predict sudden interference and dynamically adjust scheduling priorities and window structures to ensure low latency, high robustness, and high concurrency scheduling capabilities of the communication system in complex interference environments. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a communication method and system for ensuring low latency in the Internet of Things (IoT), solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a communication method for ensuring low latency in the Internet of Things (IoT), comprising the following steps:

[0007] S1. Set up data collection points on the edge devices of the port VAG, and set up collection devices in the collection points to collect interference feature vector set Fv in real time, and transmit the interference feature vector set Fv to the Internet of Things platform for preprocessing to obtain standard interference feature vector set BFv.

[0008] S2. Based on the standard interference feature vector set BFv, construct a neural dynamic burst prediction model, calculate the pulse prediction activation value Ap, and perform a preliminary comparative evaluation based on the output results of the pulse prediction activation value Ap.

[0009] S3. Based on the preliminary comparison and evaluation results, trigger the scheduling priority rearrangement, collect standardized signal data, and calculate and output the signal sparse volatility index SFI.

[0010] S4. Based on the signal sparse volatility index SFI, calculate the scheduling priority factor P of each communication node, and rearrange the scheduling order of the communication nodes according to the size of the scheduling priority factor P to generate an updated communication scheduling table.

[0011] S5. Allocate node sending windows according to the updated communication scheduling table, and calculate the scheduling fitness score Csf by combining the pulse prediction activation value Ap and the signal sparse volatility index SFI of each node; compare and evaluate the scheduling fitness score Csf with the preset fitness interval threshold.

[0012] Preferably, S1 includes S11 and S12;

[0013] S11. By deploying edge communication modules in the smart IoT terminals of AGV vehicles, metal boundaries of storage yards and port communication edge devices in the port, setting up data collection points, and setting up collection devices in the data collection points, the interference feature vector set Fv generated during node communication is collected.

[0014] The data acquisition points include a first acquisition point, a second acquisition point, a third acquisition point, and a fourth acquisition point;

[0015] The interference feature vector set Fv includes the multipath change rate Rm, the subcarrier drift frequency Fs, and the interference pulse frequency Fi;

[0016] The first acquisition point is deployed at the edge node of the AGV of the edge communication module and equipped with an edge communication acquisition device including a spectrum monitoring module and a carrier frequency offset analyzer, which is used to acquire the multipath change rate Rm and the subcarrier drift frequency Fs respectively.

[0017] The second acquisition point is located at the filter of the receiving front end of the edge communication module, and a pulse peak detection module is set at the receiving front end filter to acquire the interference pulse frequency Fi.

[0018] The third acquisition point is located in the signal conditioning and analysis unit at the signal receiver of the edge communication module, and acquires the amplitude change rate ΔS within the sampling period i in real time through the sampling ADC module. i and the standard deviation of amplitude fluctuation within sampling period i i ;

[0019] S12. A multi-protocol transmission adapter configured through the edge communication module, wherein the multi-protocol transmission adapter supports MQTT interface protocol and HTTP interface protocol, encapsulates the interference feature vector set Fv in JSON format, and obtains the data encapsulation packet.

[0020] The edge communication module, through its built-in identity authentication module, establishes a two-way authentication connection with the IoT platform to ensure the security of data transmission by transmitting the data package to the data receiving interface of the IoT platform through an encrypted channel.

[0021] The IoT platform is configured with a data receiving gateway to receive data packages, decapsulate the data packages using JSON, and extract and index and store the interference feature vector set Fv.

[0022] Preferably, S1 further includes S13;

[0023] S13. In the IoT platform, preprocess the unblocked interference feature vector set Fv to obtain the standard interference feature vector set BFv;

[0024] The preprocessing includes outlier removal, normalization, and structural correction.

[0025] The outlier removal process involves performing outlier removal on each dimension parameter of the interference feature vector set Fv, and using the median absolute deviation method to identify and remove boundary noise samples.

[0026] The normalization process normalizes the parameter vector after removing outliers, mapping each parameter value to a preset interval [0,1] to eliminate the interference of dimensional differences on subsequent model calculations.

[0027] The structure correction is performed on the normalized interference feature vector set Fv by constructing a feature mapping template based on historical communication interference datasets, matching the predefined parameter order and format; then the normalized and structure-corrected vector data are combined to form the standard interference feature vector set BFv.

[0028] Preferably, S2 includes S21;

[0029] S21. The multipath change rate Rm, interference pulse frequency Fi and subcarrier drift frequency Fs in the standard interference feature vector set BFv are used as input variables. The input variables are input into the neural activation function model, and nonlinear weighted fusion processing is performed to output the pulse prediction activation value Ap. The predicted probability of sudden interference occurring in the future scheduling period is analyzed.

[0030] The pulse prediction activation value Ap is calculated and output using the following neural activation function model;

[0031] ;

[0032] In the formula, e represents the exponential function, sin represents the cosine function, ln represents the natural logarithm, and N1, N2, and N3 represent the weight values ​​of the multipath change rate Rm, the interference pulse frequency Fi, and the subcarrier drift frequency Fs, respectively.

[0033] Preferably, S2 further includes S22;

[0034] S22. Based on the output results of the pulse prediction activation value Ap, a preliminary comparative evaluation is performed to determine the interference status of the current communication environment. Based on the preliminary comparative evaluation results, the trigger scheduling priority is rearranged. The specific evaluation content is as follows:

[0035] When the pulse prediction activation value Ap < 0.4, the interference state is considered normal, the original communication scheduling structure is maintained, and no window adjustment or node priority change is required.

[0036] When the pulse prediction activation value Ap≥0.4, it is determined to be an abnormal interference state, and a local fine-tuning operation is performed, triggering a reordering of scheduling priorities.

[0037] Preferably, S3 includes S31 and S32;

[0038] S31. Based on the preliminary comparison and evaluation, the scheduling priority is reordered. The scheduling priority is reordered by starting the third acquisition point and setting the sampling ADC module and amplitude analysis module in the signal demodulation analyzer for real-time acquisition of the continuous amplitude sample sequence S of the signal.

[0039] The sampling ADC module acquires multiple amplitude data points within each preset sampling period by setting a sampling rate greater than or equal to 10MHz.

[0040] The amplitude analysis module normalizes the collected continuous amplitude sample sequence S to standardize the range of the continuous amplitude sample sequence S to [0,1], thus forming a standardized amplitude sample sequence.

[0041] Within each sampling period i, standardized signal data is extracted, including the amplitude change rate ΔS within sampling period i. i and the standard deviation of amplitude fluctuation within sampling period i i ;

[0042] The rate of change of amplitude ΔS within the sampling period i i Within the i-th sampling period, the signal amplitude value corresponding to the end time of the continuous amplitude sample sequence S within the i-th sampling period is selected and recorded as the current amplitude value; at the same time, the signal amplitude value corresponding to the end time of the previous sampling period is obtained and recorded as the previous amplitude value; the absolute value of the difference between the current amplitude value and the previous amplitude value is calculated and divided by the time interval between the current sampling period and the previous sampling period.

[0043] The standard deviation of amplitude fluctuation B within the sampling period i i It is obtained by the statistical standard deviation of the continuous amplitude sample sequence S within the i-th sampling period;

[0044] S32. Based on standardized signal data, calculate the sparse volatility index SFI of the output signal to quantify the signal stability and volatility intensity of the communication node within a continuous sampling period.

[0045] The signal sparse volatility index SFI is calculated and output using the following algorithm formula;

[0046] ;

[0047] In the formula, T represents the total sampling period, and Δt i Let represent the time of the i-th sampling period, and log represent the logarithmic function. This indicates the elimination of the zero constant.

[0048] Preferably, S4 includes S41 and S42;

[0049] S41. Extract the pulse prediction activation value Ap and the signal sparse volatility index SFI corresponding to each communication node, and perform fusion calculation to obtain the scheduling priority factor P.

[0050] The scheduling priority factor P is calculated and output using the following algorithm formula;

[0051] ;

[0052] In the formula, P j SFI represents the scheduling priority factor of the j-th communication node. j Ap represents the sparse volatility exponent of the signal at the j-th communication node. j Let represent the pulse prediction activation value of the j-th communication node, and k represent the scheduling coupling adjustment coefficient;

[0053] S42. Based on the scheduling priority factor P of each communication node, sort all participating nodes in descending order according to the scheduling priority factor P value, establish a new scheduling order list, and allocate the available communication time slots within the scheduling period to the corresponding nodes in sequence according to the scheduling order list, thereby generating an updated communication scheduling table.

[0054] Preferably, S5 includes S51;

[0055] S51. Based on the updated communication scheduling table, extract the transmission window disturbance amplitude Aw of each communication node, and combine it with the pulse prediction activation value Ap and the signal sparse volatility index SFI. After normalization, calculate and output the scheduling fitness score Csf to analyze the adaptability score of the current scheduling structure to communication environment disturbances.

[0056] The scheduling fitness score Csf is calculated and output using the following algorithm formula;

[0057] ;

[0058] In the formula, Aw max SFI represents the maximum permissible transmission window perturbation. max This represents the upper limit of the sparse volatility index. This represents the ideal balance value of the pulse prediction activation value Ap.

[0059] Preferably, S5 further includes S52;

[0060] S52. Extract historical scheduling fitness scores Csf samples, analyze the scheduling fitness scores Csf sample values ​​when the scheduling structure is stable and when there is interference risk, and average them respectively. Set fitness interval thresholds, which include a first fitness threshold F1 and a second fitness threshold F2, wherein: the first fitness threshold F1 represents the mean of the scheduling fitness scores Csf sample values ​​when the scheduling structure is stable, and the second fitness threshold F2 represents the mean of the scheduling fitness scores Csf sample values ​​when there is interference risk.

[0061] The scheduling fitness score Csf calculated for each communication node is compared and evaluated twice with the fitness interval threshold to determine the stability of the current scheduling structure. The specific evaluation content is as follows:

[0062] When the scheduling fitness score Csf ≥ the first fitness threshold F1, it means that although interference exists, the communication module has sufficient redundancy and adaptive capability, and the current communication scheduling table is maintained.

[0063] When the second fitness threshold F2 ≤ scheduling fitness score Csf < the first fitness threshold F1, it indicates that there is pressure from the current communication disturbance. At this time, the sending window is reduced by 50%, and S41 is re-executed to rearrange the scheduling queue.

[0064] When the scheduling fitness score Csf < the second fitness threshold F2, it indicates that the system is in an abnormal state of interference and has lost local scheduling stability. At this time, all communication nodes are suspended, the existing communication scheduling table is cleared, the scheduling order table is regenerated, and the distributed congestion control protocol DCC is started.

[0065] The Distributed Congestion Control Protocol (DCC) uses each AGV as a communication node, starts an independent DCC thread in its edge device, runs a real-time load awareness and bandwidth management subroutine, and reduces the data transmission frequency to 70%-50% of the maximum value according to the local status. At the same time, each AGV, as a communication node, periodically broadcasts its local bandwidth usage status to neighboring nodes.

[0066] The broadcast content includes the real-time data packet transmission frequency, the current transmission window position and duration, and the most recent scheduling slot interference collision count.

[0067] A communication system for ensuring low latency in the Internet of Things (IoT) includes an interference feature acquisition module, a burst prediction module, a signal fluctuation identification module, a priority scheduling module, and an adaptive scoring module.

[0068] The interference feature acquisition module sets up data acquisition points on the port VAG edge device and sets up acquisition devices in the acquisition points to acquire interference feature vector set Fv in real time. The interference feature vector set Fv is then transmitted to the Internet of Things platform for preprocessing to obtain standard interference feature vector set BFv.

[0069] The burst prediction module constructs a neural dynamic burst prediction model based on the standard interference feature vector set BFv, calculates the pulse prediction activation value Ap, and performs a preliminary comparative evaluation based on the output results of the pulse prediction activation value Ap.

[0070] The signal fluctuation identification module triggers a reordering of scheduling priorities based on the preliminary comparison and evaluation results, collects standardized signal data, and calculates and outputs the signal sparse volatility index (SFI).

[0071] The priority scheduling module calculates the scheduling priority factor P of each communication node based on the signal sparse volatility index SFI, and rearranges the scheduling order of the communication nodes according to the size of the scheduling priority factor P to generate an updated communication scheduling table.

[0072] The adaptive scoring module allocates node sending windows according to the updated communication scheduling table, and calculates the scheduling fitness score Csf by combining the pulse prediction activation value Ap and the signal sparsity volatility index SFI of each node; the scheduling fitness score Csf is then compared and evaluated with a preset fitness interval threshold.

[0073] This invention provides a communication method and system for ensuring low latency in the Internet of Things (IoT). It offers the following advantages:

[0074] (1) This method constructs a neural dynamic burst prediction model, using the standard interference feature vector set BFv as input and outputting the pulse prediction activation value Ap, to achieve dynamic prediction of the probability of burst communication interference within the future scheduling cycle, which is different from the existing passive scheduling strategy that relies on historical statistics or real-time feedback. By using the prediction mechanism as a soft decision trigger, the system can predict potential risks before interference occurs, thereby performing fine-tuning or rearrangement of the scheduling structure in advance, effectively avoiding abnormal behaviors such as scheduling conflicts and window overlap, and significantly improving the proactive adaptability of the communication structure to unstable environments, so that the communication latency is stably controlled within 5ms, meeting the requirements of high reliability and low latency smart IoT scenarios.

[0075] (2) This method proposes a scheduling priority control mechanism that integrates the Sparse Fluctuation Rate Index (SFI). By collecting the continuous amplitude change rate ΔS and amplitude fluctuation standard deviation B at multiple edge acquisition points, and extracting the signal stability index based on the sparse analysis model, it identifies interference mutation intervals from the signal fluctuation structure, achieving dual-dimensional perception of complex weak interference and high-frequency burst interference. Compared with traditional scheduling evaluation methods based on carrier RSSI or packet error rate, the dual-layer fluctuation modeling mechanism of this scheme can effectively distinguish between interference behavior and normal communication fluctuations, significantly reduce the misjudgment rate, improve the system's sensitivity to real interference, thereby optimizing the accuracy of scheduling priority ranking and improving the overall robustness of the scheduling system.

[0076] (3) This method calculates the scheduling fitness score Csf by fusing the transmission window perturbation amplitude Aw, the signal sparsity volatility index SFI, and the pulse prediction activation value Ap after the scheduling reordering is completed. A secondary comparison with a preset fitness interval threshold is then performed to quantitatively assess the stability and adaptability of the communication scheduling table structure. This scoring system considers signal quality, prediction accuracy, and the impact of structural perturbations. When any dimension shows a significant anomaly, scheduling optimization measures can be triggered promptly, or the Distributed Congestion Control Protocol (DCC) can be activated to prevent the system from falling into a nonlinear degradation state. Compared to traditional static scheduling strategies, this scheme can maintain self-regulation and high availability under dynamic interference scenarios, improving the long-term operational stability and fault tolerance of the scheduling system. Attached Figure Description

[0077] Figure 1 This is a schematic diagram illustrating the steps of a communication method for ensuring low latency in the Internet of Things according to the present invention.

[0078] Figure 2 This is a schematic diagram of a communication system for ensuring low latency in the Internet of Things according to the present invention;

[0079] Figure 3 This is a schematic diagram of the overall architecture of the port's Internet of Things (IoT) communication system. Detailed Implementation

[0080] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0081] Example 1

[0082] Please see Figure 1 and Figure 3 This invention provides a method for ensuring low-latency communication in the Internet of Things (IoT). To achieve the above objective, this invention employs the following technical solution, comprising the following steps:

[0083] S1. Set up data collection points on the edge devices of the port VAG, and set up collection devices in the collection points to collect interference feature vector set Fv in real time, and transmit the interference feature vector set Fv to the Internet of Things platform for preprocessing to obtain standard interference feature vector set BFv.

[0084] S2. Based on the standard interference feature vector set BFv, construct a neural dynamic burst prediction model, calculate the pulse prediction activation value Ap, and conduct a preliminary comparative evaluation based on the output results of the pulse prediction activation value Ap.

[0085] S3. Based on the preliminary comparison and evaluation results, trigger the scheduling priority rearrangement, collect standardized signal data, and calculate and output the signal sparse volatility index SFI.

[0086] S4. Based on the signal sparse volatility index SFI, calculate the scheduling priority factor P of each communication node, and rearrange the scheduling order of the communication nodes according to the size of the scheduling priority factor P to generate an updated communication scheduling table.

[0087] S5. Allocate node sending windows according to the updated communication scheduling table, and calculate the scheduling fitness score Csf by combining the pulse prediction activation value Ap and the signal sparse volatility index SFI of each node; compare and evaluate the scheduling fitness score Csf with the preset fitness interval threshold.

[0088] In this embodiment, the method deploys multi-point acquisition devices on the port VAG edge equipment to collect interference feature vector set Fv, and performs data preprocessing based on the IoT platform to obtain standard interference feature vector set BFv. This is then used to construct a neural dynamic burst prediction model, outputting a pulse prediction activation value Ap to achieve prior prediction of the probability of communication burst interference. Driven by the prediction results, a scheduling priority reordering mechanism is triggered, and combined with standardized signal data sampled continuously over multiple periods, the signal sparse volatility index SFI is calculated to quantify the signal stability of communication nodes. Subsequently, the pulse prediction activation value Ap and the signal sparse volatility index SFI are fused to calculate the scheduling priority factor P, and the scheduling order is reordered to generate an updated communication scheduling table. Based on this, the scheduling fitness score Csf is calculated by further combining the transmission window perturbation amplitude Aw, and a secondary comparative evaluation is performed through a set dual threshold interval to achieve dynamic monitoring and decision control of the scheduling structure stability and interference adaptability. Through the above implementation methods, this invention not only establishes a complete closed-loop mechanism from signal perception, disturbance prediction, scheduling reconfiguration to system stability assessment, but also accurately identifies and responds to high-risk communication behaviors in complex electromagnetic interference environments, achieving adaptive evolution of the scheduling structure and robust interference control. Compared to existing static scheduling mechanisms or simple feedback control methods, this invention, through a combination of neural dynamic prediction and signal sparsity analysis, significantly improves the system's ability to detect sudden interference in advance, its scheduling response flexibility, and its operational stability. It effectively reduces data packet retransmission rate and latency fluctuations, ensuring the stability, real-time performance, and adaptive capabilities of communication systems in key smart IoT scenarios.

[0089] Example 2

[0090] Please see Figure 1 and Figure 3 Specifically: S1 includes S11 and S12;

[0091] S11. By deploying edge communication modules in the smart IoT terminals of AGV vehicles, metal boundaries of storage yards and port communication edge devices in the port, setting up data collection points, and setting up collection devices in the data collection points, the interference feature vector set Fv generated during node communication is collected.

[0092] The data collection points include the first collection point, the second collection point, the third collection point, and the fourth collection point;

[0093] The interference feature vector set Fv includes the multipath change rate Rm, the subcarrier drift frequency Fs, and the interference pulse frequency Fi;

[0094] The first acquisition point is deployed at the edge node of the AGV of the edge communication module, and is equipped with an edge communication acquisition device including a spectrum monitoring module and a carrier frequency offset analyzer, which is used to acquire the multipath change rate Rm and the subcarrier drift frequency Fs respectively.

[0095] The second acquisition point is deployed at the filter of the receiving front end of the edge communication module, and a pulse peak detection module is set at the receiving front end filter to acquire the interference pulse frequency Fi.

[0096] The third acquisition point uses a signal conditioning and analysis unit deployed at the signal receiver end of the edge communication module to collect the amplitude change rate ΔS within the sampling period i in real time through the sampling ADC module. i and the standard deviation of amplitude fluctuation within sampling period i i ;

[0097] S12. A multi-protocol transmission adapter configured through the edge communication module, wherein the multi-protocol transmission adapter supports MQTT interface protocol and HTTP interface protocol, encapsulates the interference feature vector set Fv in JSON format, and obtains the data encapsulation packet.

[0098] The edge communication module, through its built-in identity authentication module, establishes a two-way authentication connection with the IoT platform to ensure the security of data transmission. It transmits the data package to the data receiving interface of the IoT platform through an encrypted channel.

[0099] The IoT platform configures a data receiving gateway to receive data packages, decapsulates the data packages using JSON, extracts the interference feature vector set Fv for indexing and storage, and then performs subsequent interference risk prediction processing.

[0100] S1 also includes S13;

[0101] S13. In the IoT platform, preprocess the unblocked interference feature vector set Fv to obtain the standard interference feature vector set BFv;

[0102] Preprocessing includes outlier removal, normalization, and structural correction;

[0103] Outlier removal is performed on the parameters of each dimension in the interference feature vector set Fv, and the median absolute deviation method is used to identify and remove boundary noise samples.

[0104] Normalization is performed on the parameter vector after removing outliers to map each parameter value to a preset interval [0,1], so as to eliminate the interference of dimensional differences on subsequent model calculations.

[0105] Structure correction is performed on the normalized interference feature vector set Fv by constructing a feature mapping template based on historical communication interference datasets, matching the predefined parameter order and format; then the normalized and structure-corrected vector data are combined to form the standard interference feature vector set BFv, which is used to calculate the pulse prediction activation value Ap in the subsequent neural dynamic burst prediction model.

[0106] In this embodiment, the method deploys edge communication modules in port AGV vehicles, yard metal boundaries, and communication edge devices, establishing multiple heterogeneous acquisition points to collect multipath change rates Rm, subcarrier drift frequencies Fs, and interference pulse frequencies Fi, forming a multidimensional interference feature vector set Fv. This Fv is then encapsulated in JSON format via a multi-protocol transmission adapter and securely transmitted to the IoT platform via authentication and an encrypted channel. The IoT platform decapsulates and indexes the interference feature vector set Fv through a data receiving gateway, and further performs outlier removal, normalization, and structure correction processing, including the median absolute deviation method, to obtain a standard interference feature vector set BFv. This provides a unified and reliable data input foundation for subsequent neural dynamic burst prediction models. Through this integrated acquisition, transmission, and preprocessing mechanism, this invention achieves high-resolution, low-latency, and highly robust acquisition and standardized modeling of interference information during communication, significantly improving the perception capability and data quality assurance capability of communication interference characteristics in the complex electromagnetic environment of ports. Compared with existing methods that rely on a single parameter or passive data acquisition, this method has significant advantages in terms of dimensionality, time accuracy, and parameter standardization. It provides solid data support for subsequent modules such as emergency prediction, scheduling decision-making, and system adaptive control, thereby improving the entire intelligent IoT communication system's early warning sensitivity and adaptive control capability to sudden disturbances from the source.

[0107] Example 3

[0108] Please see Figure 1 Specifically: S2 includes S21;

[0109] S21. The multipath change rate Rm, interference pulse frequency Fi and subcarrier drift frequency Fs in the standard interference feature vector set BFv are used as input variables. The input variables are input into the neural activation function model, and nonlinear weighted fusion processing is performed to output the pulse prediction activation value Ap. The predicted probability of sudden interference occurring in the future scheduling period is analyzed.

[0110] The impulse prediction activation value Ap is calculated and output using the following neural activation function model;

[0111] ;

[0112] In the formula, e represents the exponential function, sin represents the cosine function, ln represents the natural logarithm, and N1, N2 and N3 represent the weight values ​​of the multipath change rate Rm, the interference pulse frequency Fi and the subcarrier drift frequency Fs, respectively, which are set by the user, and N1+N2+N3=1.

[0113] The derivation logic and physical meaning of the formula: This formula is a neuron-like activation function structure, derived from the typical Sigmoid nonlinear activation function in neural networks. It is used to compress complex inputs into smooth output values ​​in the (0,1) interval. It is nonlinear, differentiable, suitable for expressing continuous probability responses, and easy to use as a soft decision trigger, i.e., it is not a hard threshold, but rather the higher the threshold, the more urgent the decision.

[0114] Represents a combination of linear and nonlinear transformations;

[0115] First item: Linear modeling of multipath disturbances represents the direct impact of changes in the multipath environment on sudden disturbances. Rapidly changing multipath structures are often accompanied by sudden disturbances.

[0116] Second item: By using trigonometric functions to simulate the periodic changes in the frequency of interference pulses, periodic features of the interference frequency are extracted to adapt to frequent but low-amplitude interference behavior, thus avoiding the dominance of the entire model when the frequency of interference pulses Fi is too high and improving the sensitivity to intermittent interference.

[0117] Third item: The frequency offset amplitude is represented by logarithmic compression to avoid direct linear superposition of large values. The subcarrier drift frequency Fs is compressed to smooth the fluctuations and avoid a few high drift nodes from excessively pulling the model output.

[0118] The combination of linear and nonlinear transformations is substituted into the sigmoid activation function to output the pulse prediction activation value Ap.

[0119] S2 also includes S22;

[0120] S22. Based on the output results of the pulse prediction activation value Ap, a preliminary comparative evaluation is performed to determine the interference status of the current communication environment. Based on the preliminary comparative evaluation results, the trigger scheduling priority is rearranged. The specific evaluation content is as follows:

[0121] When the pulse prediction activation value Ap < 0.4, the interference state is considered normal, the original communication scheduling structure is maintained, and no window adjustment or node priority change is required.

[0122] When the pulse prediction activation value Ap≥0.4, it is determined to be an abnormal interference state, and a local fine-tuning operation is performed, triggering a reordering of scheduling priorities.

[0123] In this embodiment, the method uses the multipath change rate Rm, interference pulse frequency Fi, and subcarrier drift frequency Fs from the standard interference feature vector set BFv as input variables, inputting them into a constructed neural activation function model. It then uses a fusion of linear modeling and nonlinear function transformation to output the pulse prediction activation value Ap. This neural activation function model introduces a sigmoid structure as an output mapping mechanism, and simultaneously superimposes exponential, trigonometric, and logarithmic function transformations to discriminate between three types of interference behavior, extracting burst, frequent, and steady-state interference features respectively, ensuring the model has both predictive sensitivity and computational stability. After calculating the pulse prediction activation value Ap, a preliminary comparative evaluation is performed based on its value: if the pulse prediction activation value Ap < 0.4, the current environmental interference level is considered normal, and the existing scheduling structure is maintained; if the pulse prediction activation value Ap ≥ 0.4, the risk of burst interference is considered increased, and automatic scheduling priority reordering is triggered to improve robustness. Through the introduction of this neural dynamic burst prediction model, early perception and predictive response to communication interference intensity within future scheduling cycles are achieved. Compared to the delayed response of traditional scheduling mechanisms to interference events, this implementation method can achieve millisecond-level prediction and adaptive triggering, significantly improving the feedforward defense capability and dynamic adjustment efficiency of the scheduling system, and significantly reducing the impact of sudden disturbances on data transmission stability. At the same time, this method combines the advantages of lightweight computing and high sensitivity, adapting to the computing resource limitations of various edge devices, and providing a key foundational module for ensuring low-latency communication in the smart IoT environment.

[0124] Example 4

[0125] Please see Figure 1 Specifically: S3 includes S31 and S32;

[0126] S31. Based on the preliminary comparison and evaluation, the scheduling priority is reordered. The scheduling priority is reordered by starting the third acquisition point and setting the sampling ADC module and amplitude analysis module in the signal demodulation analyzer for real-time acquisition of the continuous amplitude sample sequence S of the signal.

[0127] The sampling ADC module acquires multiple amplitude data points within each preset sampling period by setting a sampling rate greater than or equal to 10MHz.

[0128] The amplitude analysis module normalizes the collected continuous amplitude sample sequence S to standardize the range of the continuous amplitude sample sequence S to [0,1], thus forming a standardized amplitude sample sequence.

[0129] Within each sampling period i, normalized signal data is extracted. The normalized signal data includes the amplitude change rate ΔS within sampling period i. i and the standard deviation of amplitude fluctuation within sampling period i i ;

[0130] The rate of change of amplitude ΔS within sample period i i Within the i-th sampling period, the signal amplitude value corresponding to the end time of the continuous amplitude sample sequence S within the i-th sampling period is selected and recorded as the current amplitude value; at the same time, the signal amplitude value corresponding to the end time of the previous sampling period is obtained and recorded as the previous amplitude value; the absolute value of the difference between the current amplitude value and the previous amplitude value is calculated and divided by the time interval between the current sampling period and the previous sampling period.

[0131] Standard deviation B of amplitude fluctuation within sampling period i i It is obtained by the statistical standard deviation of the continuous amplitude sample sequence S within the i-th sampling period;

[0132] S32. Based on standardized signal data, calculate the sparse volatility index SFI of the output signal to quantify the signal stability and volatility intensity of the communication node within a continuous sampling period.

[0133] The Signal Sparse Volatility Index (SFI) is calculated and output using the following algorithm formula;

[0134] ;

[0135] In the formula, T represents the total sampling period, and Δt i Let represent the time of the i-th sampling period, and log represent the logarithmic function. Indicates the prevention of zero constant;

[0136] This represents the rate component, used to analyze the intensity of signal abrupt changes in the time dimension within each sampling period. If the signal changes significantly over a certain period of time, such as a sudden interference pulse, this value will increase sharply, corresponding to the capture of high-frequency or instantaneous interference behavior.

[0137] This represents the fluctuation weight component. If the standard deviation Bi of the amplitude fluctuation within the sampling period i is small, it indicates that the signal is generally stable; therefore, 1 / B i The larger the value, the less fluctuation there is, and the more sensitive it is to sudden events. Introducing a logarithmic function can prevent small disturbances from being amplified unreasonably and smooth out extreme fluctuations.

[0138] The product of the rate component and the volatility component represents a sudden signal change occurring within a stable region. Since volatility is inherently high, this sudden change may not be considered "abnormal." If B... i Small and △S i Large abrupt changes indicate the presence of real abnormal disturbances, such as narrowband interference or reflection superposition.

[0139] Finally, the entire sampling period is averaged to avoid abnormal fluctuations in individual periods that could lead to uncontrolled indicators, thus obtaining a global signal sparsity fluctuation score.

[0140] In this embodiment, the method activates a third acquisition point in the signal demodulation analyzer, configures a sampling ADC module and an amplitude analysis module, and acquires a continuous amplitude sample sequence S of the communication node in real time. A continuous data stream is obtained at a high-frequency sampling rate of no less than 10MHz, thereby ensuring millisecond-level capture capability for sudden interference signals. Subsequently, the amplitude samples are mapped to the [0,1] standard range through normalization processing. Standardized signal data is extracted within each sampling period i, and the amplitude change rate ΔSi and amplitude fluctuation standard deviation Bi are calculated respectively. Based on this standardized signal data, the signal sparse volatility index SFI is constructed by multiplying the rate change intensity and the fluctuation weight factor, thereby comprehensively quantifying the stability of the communication signal within a continuous period. By introducing the signal sparse volatility index SFI as a dynamic interference evaluation index, this implementation can identify abnormal fluctuations caused by instantaneous interference or frequency offset superposition at the micro-amplitude change level, improving the response sensitivity to high-frequency sudden interference. It is particularly suitable for intelligent IoT communication scenarios with complex electromagnetic environments such as multipath reflection and in-band pulse interference. Compared to traditional methods that use average signal-to-noise ratio or bit error rate as the sole criterion, the Sparse Volatility Index (SFI) has stronger real-time performance and discrimination capabilities, significantly enhancing the scheduling system's ability to predict and judge edge interference risks. It provides accurate volatility criteria support for subsequent scheduling priority reordering and scheduling table generation, thereby improving the stability of communication links and optimizing latency convergence speed under sudden interference conditions.

[0141] Example 5

[0142] Please see Figure 1 Specifically: S4 includes S41 and S42;

[0143] S41. Extract the pulse prediction activation value Ap and the signal sparse volatility index SFI corresponding to each communication node, and perform fusion calculation to obtain the scheduling priority factor P.

[0144] The scheduling priority factor P is calculated and output using the following algorithm formula;

[0145] ;

[0146] In the formula, P j SFI represents the scheduling priority factor of the j-th communication node. j Ap represents the sparse volatility exponent of the signal at the j-th communication node. jrepresents the pulse prediction activation value of the j-th communication node, and k represents the scheduling coupling adjustment coefficient, which is used to adjust the influence of the pulse prediction activation value Ap on the scheduling factor. The value range is 0 to 1.0 and is set by the user to enhance the proportion of sudden interference prediction signals in the scheduling order.

[0147] The calculation logic of the scheduling priority factor P is as follows: when a communication node has a high signal sparse volatility index (SFI) and a high-risk pulse prediction activation value (Ap), the scheduling priority factor P will decrease, thus being delayed in the scheduling order; conversely, stable nodes have lower signal sparse volatility index (SFI) and pulse prediction activation value (Ap), and a larger scheduling priority factor P, thus being scheduled first in the scheduling order.

[0148] This formula constructs an "inverse proportional scheduling rule" that combines the sparse volatility index SFI of communication node signals with the pulse prediction activation value Ap to achieve dynamic interference perception capability for scheduling order, ensuring the real-time performance and robustness of the system in complex interference environments.

[0149] S42. Based on the scheduling priority factor P of each communication node, sort all participating nodes in descending order according to the scheduling priority factor P value, establish a new scheduling order list, and allocate the available communication time slots within the scheduling period to the corresponding nodes in sequence according to the scheduling order list, thereby generating an updated communication scheduling table.

[0150] In this embodiment, the method constructs a fusion calculation model based on the pulse prediction activation value Ap and the signal sparsity volatility index SFI output by each communication node, outputs a scheduling priority factor P, and prioritizes the communication nodes according to the magnitude of the scheduling priority factor P, thereby generating an updated communication scheduling table. By setting a scheduling coupling adjustment coefficient k, the influence ratio of the pulse prediction activation value Ap on the final scheduling factor can be flexibly adjusted according to the system scenario during the calculation. In the formula design, an inverse proportional strategy is adopted to give scheduling delay to communication nodes with high interference levels, and to prioritize the allocation of communication time slots to nodes with stable interference, thereby prioritizing the communication latency and success rate of stable links when the overall system scheduling resources are limited. Through this priority factor-driven dynamic scheduling mechanism, the global scheduling structure is optimized under complex interference backgrounds, solving the problem of delayed response to link health status in traditional static scheduling schemes. Especially in multi-node concurrent, high-load communication environments, this method can dynamically avoid the dominant influence of high-interference, high-fluctuation nodes on the scheduling table, ensuring that the overall scheduling table structure is more stable and interference robust, thereby effectively reducing overall communication latency and improving data transmission success rate. This mechanism is scalable and adaptive, and can be widely adapted to various smart IoT deployment scenarios, especially suitable for complex environments such as ports, factories, and mines where there is spatial electromagnetic coupling and multipath interference.

[0151] Example 6

[0152] Please see Figure 1 Specifically: S5 includes S51;

[0153] S51. Based on the updated communication scheduling table, extract the transmission window disturbance amplitude Aw of each communication node, and combine it with the pulse prediction activation value Ap and the signal sparse volatility index SFI. After normalization, calculate and output the scheduling fitness score Csf to analyze the adaptability score of the current scheduling structure to communication environment disturbances.

[0154] The scheduling fitness score Csf is calculated and output using the following algorithm formula;

[0155] ;

[0156] In the formula, Aw max SFI represents the maximum permissible transmission window perturbation. max This represents the upper limit of the sparse volatility index. The ideal balance value of the pulse prediction activation value Ap is set by the user. The target value of the pulse prediction activation value Ap under ideal conditions is used to quantify the deviation of the pulse prediction activation value Ap. The value is dimensionless.

[0157] The value Aw represents the degree of window perturbation. The smaller the perturbation amplitude Aw of the sending window, the less the system is forced to compress the scheduling, indicating higher stability. When it approaches 1, it indicates that the scheduling structure is stable; when it approaches 0, it indicates that the scheduling structure is chaotic.

[0158] The Sparse Volatility Index (SFI) indicates signal stability. A large SFI indicates that the system signal is unstable or fluctuates violently. The closer the value is to 1, the less the overall signal fluctuation is, and the safer it is to use the current scheduling structure.

[0159] This indicates whether the pulse prediction activation value Ap deviates from the ideal state. When the pulse prediction activation value Ap is close to the ideal equilibrium value of the pulse prediction activation value Ap, it indicates that the pulse prediction activation value Ap deviates from the ideal state. When the pulse prediction activation value Ap deviates too much, it indicates that the prediction environment has undergone a nonlinear change, the system uncertainty has increased, and this value decreases.

[0160] The three factors are combined through multiplication, emphasizing that true stability can only be achieved when all three factors are stable. If any one factor is unstable, the overall value of the scheduling fitness score Csf will decrease, thus playing a strict constraint role. For example, even if the signal sparsity volatility index SFI is low and the pulse prediction activation value Ap is moderate, if the perturbation amplitude Aw of the sending window is large, rearrangement is still required. It is a strong decision-making scoring formula, which is suitable for scheduling scenarios with low fault tolerance.

[0161] S5 also includes S52;

[0162] S52. Extract historical scheduling fitness score Csf samples, analyze the scheduling fitness score Csf sample values ​​when the scheduling structure is stable and when there is interference risk, and perform mean normalization on each. Set fitness interval thresholds, including a first fitness threshold F1 and a second fitness threshold F2, where: the first fitness threshold F1 represents the mean of the scheduling fitness score Csf sample values ​​when the scheduling structure is stable, and the second fitness threshold F2 represents the mean of the scheduling fitness score Csf sample values ​​when there is interference risk.

[0163] The scheduling fitness score Csf calculated for each communication node is compared and evaluated twice with the fitness interval threshold to determine the stability of the current scheduling structure. The specific evaluation content is as follows:

[0164] When the scheduling fitness score Csf ≥ the first fitness threshold F1, it means that although interference exists, the communication module has sufficient redundancy and adaptive capability, and the current communication scheduling table is maintained.

[0165] When the second fitness threshold F2 ≤ scheduling fitness score Csf < the first fitness threshold F1, it indicates that there is pressure from the current communication disturbance. At this time, the sending window is reduced by 50%, for example, from 2ms to 1ms, and S41 is re-executed to rearrange the scheduling queue.

[0166] When the scheduling fitness score Csf < the second fitness threshold F2, it indicates that the system is in an abnormal state of interference and has lost local scheduling stability. At this time, all communication nodes are suspended, the existing communication scheduling table is cleared, the scheduling order table is regenerated, and the distributed congestion control protocol DCC is started.

[0167] The Distributed Congestion Control Protocol (DCC) uses each AGV as a communication node, starts an independent DCC thread in its edge device, runs a real-time load awareness and bandwidth management subroutine, and reduces the data transmission frequency to 70%-50% of the maximum value according to the local status. At the same time, each AGV, as a communication node, periodically broadcasts its local bandwidth usage status to neighboring nodes.

[0168] The broadcast content includes the real-time data packet transmission frequency, the current transmission window position and duration, and the interference conflict count of the most recent scheduling time slot; it is used to dynamically sense the link load pressure and conflict risk for other AGVs as communication nodes based on the broadcast content.

[0169] In this embodiment, the method extracts the transmission window perturbation amplitude Aw, the pulse prediction activation value Ap, and the signal sparsity volatility index SFI of each communication node to construct a multi-factor fusion model, outputting a scheduling fitness score Csf to quantify the adaptability of the current scheduling structure under complex interference environments. The scheduling fitness score Csf is obtained by comprehensively calculating the three normalized indicators, fully reflecting the stability of the scheduling structure, the volatility of the signal environment, and the degree of interference prediction deviation. Furthermore, based on historical samples, a first fitness threshold F1 and a second fitness threshold F2 are dynamically set to conduct a secondary comparative evaluation, enabling the determination and intervention of the health status of the scheduling structure. When the scheduling fitness score Csf is in the high fitness range, the original scheduling structure remains unchanged to minimize resource disturbance. When the scheduling fitness score Csf drops to a medium level, window compression and node scheduling order rearrangement are automatically executed to ensure rapid adaptation under light to moderate interference. When the scheduling fitness score Csf falls below the minimum threshold, the system suspends current communication and initiates the Distributed Congestion Control (DCC) protocol to achieve self-isolation and recovery from high-load interference. The DCC protocol dynamically adjusts the transmission frequency and time slot conflict probability through AGV node local bandwidth status awareness and neighbor node broadcasting, forming a self-organizing, self-aware, and self-adjusting communication recovery mechanism. Through these methods, the system significantly improves communication stability and anti-interference robustness in complex, dynamic, and multi-interference port IoT scenarios, realizing the transformation of communication scheduling from static rule-driven to dynamic intelligent control. This not only reduces the rate of sudden conflicts and retransmission probability but also effectively compresses the overall communication latency, providing a more stable and efficient communication guarantee system for latency-sensitive smart IoT services.

[0170] Example 7

[0171] Please see Figure 1 and Figure 2 A communication system for ensuring low latency in the Internet of Things includes an interference feature acquisition module, a burst prediction module, a signal fluctuation identification module, a priority scheduling module, and an adaptive scoring module.

[0172] The interference feature acquisition module sets up data acquisition points on the port VAG edge device and sets up acquisition devices in the acquisition points to acquire interference feature vector set Fv in real time. The interference feature vector set Fv is then transmitted to the Internet of Things platform for preprocessing to obtain standard interference feature vector set BFv.

[0173] The burst prediction module constructs a neural dynamic burst prediction model based on the standard interference feature vector set BFv, calculates the pulse prediction activation value Ap, and performs a preliminary comparative evaluation based on the output results of the pulse prediction activation value Ap.

[0174] The signal fluctuation identification module triggers a reordering of scheduling priorities based on the preliminary comparison and evaluation results, collects standardized signal data, and calculates and outputs the signal sparse volatility index (SFI).

[0175] The priority scheduling module calculates the scheduling priority factor P of each communication node based on the signal sparse volatility index SFI, and rearranges the scheduling order of the communication nodes according to the size of the scheduling priority factor P to generate an updated communication scheduling table.

[0176] The adaptive scoring module allocates node sending windows according to the updated communication scheduling table and calculates the scheduling fitness score Csf by combining the pulse prediction activation value Ap and the signal sparsity volatility index SFI of each node; the scheduling fitness score Csf is then compared and evaluated with the preset fitness interval threshold.

[0177] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A communication method for ensuring low latency in the Internet of Things (IoT), characterized in that: Includes the following steps: S1. Set up data collection points on the edge devices of the port VAG, and set up collection devices in the collection points to collect interference feature vector set Fv in real time, and transmit the interference feature vector set Fv to the Internet of Things platform for preprocessing to obtain standard interference feature vector set BFv. S2. Based on the standard interference feature vector set BFv, construct a neural dynamic burst prediction model, calculate the pulse prediction activation value Ap, and perform a preliminary comparative evaluation based on the output results of the pulse prediction activation value Ap. S3. Based on the preliminary comparison and evaluation results, trigger the scheduling priority rearrangement, collect standardized signal data, and calculate and output the signal sparse volatility index SFI. S4. Based on the signal sparse volatility index SFI, calculate the scheduling priority factor P of each communication node, and rearrange the scheduling order of the communication nodes according to the size of the scheduling priority factor P to generate an updated communication scheduling table. S4 includes S41 and S42; S41. Extract the pulse prediction activation value Ap and the signal sparse volatility index SFI corresponding to each communication node, and perform fusion calculation to obtain the scheduling priority factor P. The scheduling priority factor P is calculated and output using the following algorithm formula; ； In the formula, P j SFI represents the scheduling priority factor of the j-th communication node. j Ap represents the sparse volatility exponent of the signal at the j-th communication node. j Let represent the pulse prediction activation value of the j-th communication node, and k represent the scheduling coupling adjustment coefficient; S42. Based on the scheduling priority factor P of each communication node, sort all participating nodes in descending order according to the scheduling priority factor P value, establish a new scheduling order list, and allocate the available communication time slots within the scheduling period to the corresponding nodes in sequence according to the scheduling order list, thereby generating an updated communication scheduling table. S5. Allocate node sending windows according to the updated communication scheduling table, and calculate the scheduling fitness score Csf by combining the pulse prediction activation value Ap and the signal sparsity volatility index SFI of each node; compare and evaluate the scheduling fitness score Csf with the preset fitness interval threshold. S5 includes S51; S51. Based on the updated communication scheduling table, extract the transmission window disturbance amplitude Aw of each communication node, and combine it with the pulse prediction activation value Ap and the signal sparse volatility index SFI. After normalization, calculate and output the scheduling fitness score Csf to analyze the adaptability score of the current scheduling structure to communication environment disturbances. The scheduling fitness score Csf is calculated and output using the following algorithm formula; ； In the formula, Aw max SFI represents the maximum permissible transmission window perturbation. max This represents the upper limit of the sparse volatility index. This represents the ideal balance value of the pulse prediction activation value Ap.

2. The communication method for ensuring low latency in the Internet of Things according to claim 1, characterized in that: S1 includes S11 and S12; S11. By deploying edge communication modules in the smart IoT terminals of AGV vehicles, metal boundaries of storage yards and port communication edge devices in the port, setting up data collection points, and setting up collection devices in the data collection points, the interference feature vector set Fv generated during node communication is collected. The data acquisition points include a first acquisition point, a second acquisition point, a third acquisition point, and a fourth acquisition point; The interference feature vector set Fv includes the multipath change rate Rm, the subcarrier drift frequency Fs, and the interference pulse frequency Fi; The first acquisition point is deployed at the edge node of the AGV of the edge communication module and equipped with an edge communication acquisition device including a spectrum monitoring module and a carrier frequency offset analyzer, which is used to acquire the multipath change rate Rm and the subcarrier drift frequency Fs respectively. The second acquisition point is located at the filter of the receiving front end of the edge communication module, and a pulse peak detection module is set at the receiving front end filter to acquire the interference pulse frequency Fi. The third acquisition point is located in the signal conditioning and analysis unit at the signal receiver of the edge communication module, and acquires the amplitude change rate ΔS within the sampling period i in real time through the sampling ADC module. i and the standard deviation of amplitude fluctuation within sampling period i i ; S12. A multi-protocol transmission adapter configured through the edge communication module, wherein the multi-protocol transmission adapter supports MQTT interface protocol and HTTP interface protocol, encapsulates the interference feature vector set Fv in JSON format, and obtains the data encapsulation packet. The edge communication module, through its built-in identity authentication module, establishes a two-way authentication connection with the IoT platform to ensure the security of data transmission by transmitting the data package to the data receiving interface of the IoT platform through an encrypted channel. The IoT platform is configured with a data receiving gateway to receive data packages, decapsulate the data packages using JSON, and extract and index and store the interference feature vector set Fv.

3. The communication method for ensuring low latency in the Internet of Things according to claim 2, characterized in that: S1 also includes S13; S13. In the IoT platform, preprocess the unblocked interference feature vector set Fv to obtain the standard interference feature vector set BFv; The preprocessing includes outlier removal, normalization, and structural correction. The outlier removal process involves performing outlier removal on each dimension parameter of the interference feature vector set Fv, and using the median absolute deviation method to identify and remove boundary noise samples. The normalization process normalizes the parameter vector after removing outliers, mapping each parameter value to a preset interval [0,1] to eliminate the interference of dimensional differences on subsequent model calculations. The structure correction is performed on the normalized interference feature vector set Fv by constructing a feature mapping template based on historical communication interference datasets, matching the predefined parameter order and format; then the normalized and structure-corrected vector data are combined to form the standard interference feature vector set BFv.

4. The communication method for ensuring low latency in the Internet of Things according to claim 3, characterized in that: S2 includes S21; S21. The multipath change rate Rm, interference pulse frequency Fi and subcarrier drift frequency Fs in the standard interference feature vector set BFv are used as input variables. The input variables are input into the neural activation function model, and nonlinear weighted fusion processing is performed to output the pulse prediction activation value Ap. The predicted probability of sudden interference occurring in the future scheduling period is analyzed. The pulse prediction activation value Ap is calculated and output using the following neural activation function model; ； In the formula, e represents the exponential function, sin represents the cosine function, ln represents the natural logarithm, and N1, N2, and N3 represent the weight values ​​of the multipath change rate Rm, the interference pulse frequency Fi, and the subcarrier drift frequency Fs, respectively.

5. A communication method for ensuring low latency in the Internet of Things according to claim 4, characterized in that: S2 further includes S22; S22. Based on the output results of the pulse prediction activation value Ap, a preliminary comparative evaluation is performed to determine the interference status of the current communication environment. Based on the preliminary comparative evaluation results, the trigger scheduling priority is rearranged. The specific evaluation content is as follows: When the pulse prediction activation value Ap < 0.4, the interference state is considered normal, the original communication scheduling structure is maintained, and no window adjustment or node priority change is required. When the pulse prediction activation value Ap≥0.4, it is determined to be an abnormal interference state, and a local fine-tuning operation is performed, triggering a reordering of scheduling priorities.

6. A communication method for ensuring low latency in the Internet of Things according to claim 2, characterized in that: S3 includes S31 and S32; S31. Based on the preliminary comparison and evaluation, the scheduling priority is reordered. The scheduling priority is reordered by starting the third acquisition point and setting the sampling ADC module and amplitude analysis module in the signal demodulation analyzer for real-time acquisition of the continuous amplitude sample sequence S of the signal. The sampling ADC module acquires multiple amplitude data points within each preset sampling period by setting a sampling rate greater than or equal to 10MHz. The amplitude analysis module normalizes the collected continuous amplitude sample sequence S to standardize the range of the continuous amplitude sample sequence S to [0,1], thus forming a standardized amplitude sample sequence. Within each sampling period i, standardized signal data is extracted, including the amplitude change rate ΔS within sampling period i. i and the standard deviation of amplitude fluctuation within sampling period i i ; The rate of change of amplitude ΔS within the sampling period i i Within the i-th sampling period, the signal amplitude value corresponding to the end time of the continuous amplitude sample sequence S within the i-th sampling period is selected and recorded as the current amplitude value; at the same time, the signal amplitude value corresponding to the end time of the previous sampling period is obtained and recorded as the previous amplitude value. Calculate the absolute value of the difference between the current amplitude value and the previous amplitude value, and divide it by the time interval between the current sampling period and the previous sampling period. The standard deviation of amplitude fluctuation B within the sampling period i i It is obtained by the statistical standard deviation of the continuous amplitude sample sequence S within the i-th sampling period; S32. Based on standardized signal data, calculate the sparse volatility index SFI of the output signal to quantify the signal stability and volatility intensity of the communication node within a continuous sampling period. The signal sparse volatility index SFI is calculated and output using the following algorithm formula; In the formula, T represents the total sampling period, and Δt i Let represent the time of the i-th sampling period, and log represent the logarithmic function. This indicates the elimination of the zero constant.

7. A communication method for ensuring low latency in the Internet of Things according to claim 1, characterized in that: S5 further includes S52; S52: Extract historical scheduling fitness score Csf samples, analyze the scheduling fitness score Csf sample values ​​when the scheduling structure is stable and when there is interference risk, and perform mean normalization on each, and set fitness interval thresholds, the fitness interval thresholds include a first fitness threshold F1 and a second fitness threshold F2, wherein: the first fitness threshold F1 represents the mean value of the scheduling fitness score Csf sample values ​​when the scheduling structure is stable, and the second fitness threshold F2 represents the mean value of the scheduling fitness score Csf sample values ​​when there is interference risk; The scheduling fitness score Csf calculated for each communication node is compared and evaluated twice with the fitness interval threshold to determine the stability of the current scheduling structure. The specific evaluation content is as follows: When the scheduling fitness score Csf ≥ the first fitness threshold F1, it means that although interference exists, the communication module has sufficient redundancy and adaptive capability, and the current communication scheduling table is maintained. When the second fitness threshold F2 ≤ scheduling fitness score Csf < the first fitness threshold F1, it indicates that there is pressure from the current communication disturbance. At this time, the sending window is reduced by 50%, and S41 is re-executed to rearrange the scheduling queue. When the scheduling fitness score Csf < the second fitness threshold F2, it indicates that the system is in an abnormal state of interference and has lost local scheduling stability. At this time, all communication nodes are suspended, the existing communication scheduling table is cleared, the scheduling order table is regenerated, and the distributed congestion control protocol DCC is started. The Distributed Congestion Control Protocol (DCC) uses each AGV as a communication node, starts an independent DCC thread in its edge device, runs a real-time load awareness and bandwidth management subroutine, and reduces the data transmission frequency to 70%-50% of the maximum value according to the local status. At the same time, each AGV, as a communication node, periodically broadcasts its local bandwidth usage status to neighboring nodes. The broadcast content includes the real-time data packet transmission frequency, the current transmission window position and duration, and the most recent scheduling slot interference collision count.

8. A communication system for ensuring low latency in a smart Internet of Things (IoT), applied to the communication method for ensuring low latency in a smart IoT as described in any one of claims 1-7, characterized in that: It includes an interference feature acquisition module, a burst prediction module, a signal fluctuation identification module, a priority scheduling module, and an adaptive scoring module; The interference feature acquisition module sets up data acquisition points on the port VAG edge device and sets up acquisition devices in the acquisition points to acquire interference feature vector set Fv in real time. The interference feature vector set Fv is then transmitted to the Internet of Things platform for preprocessing to obtain standard interference feature vector set BFv. The burst prediction module constructs a neural dynamic burst prediction model based on the standard interference feature vector set BFv, calculates the pulse prediction activation value Ap, and performs a preliminary comparative evaluation based on the output results of the pulse prediction activation value Ap. The signal fluctuation identification module triggers a reordering of scheduling priorities based on the preliminary comparison and evaluation results, collects standardized signal data, and calculates and outputs the signal sparse volatility index (SFI). The priority scheduling module calculates the scheduling priority factor P of each communication node based on the signal sparse volatility index SFI, and rearranges the scheduling order of the communication nodes according to the size of the scheduling priority factor P to generate an updated communication scheduling table. The adaptive scoring module allocates node sending windows according to the updated communication scheduling table, and calculates the scheduling fitness score Csf by combining the pulse prediction activation value Ap and the signal sparsity volatility index SFI of each node; the scheduling fitness score Csf is then compared and evaluated with a preset fitness interval threshold.

Images