Multi-hop pipeline propagation method based on time division multiple access (TDMA) time slot level synchronization

Abstract

The invention discloses a multi-hop pipeline propagation method based on TDMA time slot level synchronization. The method comprises the following steps: S1, constructing a TDMA frame format; s2, constructing a local pseudo transition time mapping table in an uplink receiving time slot; s3, constructing a local clock drift trend model based on a bat resonance fitting algorithm; s4, predicting the expected arrival time of the next uplink synchronization according to the model, and adjusting a local clock; s5, completing data packet processing operation in the local processing time slot; s6, in a downlink sending time slot, the node completes service data sending and adds local clock drift trend model parameters; s7, entering a low-power-consumption monitoring state in the protection sleep time slot, comparing the difference between the feedback model parameters and the model parameters, and updating the model if the difference exceeds a preset tolerance threshold; and S8, the node injects Gaussian disturbance into a local clock in a set frame period, and the model is updated regularly. According to the method, a bat resonance fitting algorithm and a time slot modeling method are adopted, and multi-hop node autonomous synchronization is achieved.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and particularly to a multi-hop pipeline propagation method based on TDMA time slot-level synchronization. Background Art

[0002] In a multi-hop wireless communication network, clock synchronization between nodes has always been one of the key technologies for achieving efficient and reliable data transmission. Existing technologies generally adopt a time slot scheduling mechanism based on TDMA (Time Division Multiple Access) to avoid collisions and improve channel utilization. In such systems, in order to ensure that data on a multi-hop link can be smoothly relayed between different nodes, an accurate time synchronization mechanism must be relied on to ensure that each node completes reception, processing, and transmission in the correct time slot. However, as the network scale expands and the number of node hops increases, due to inconsistent crystal oscillators, uncertain propagation delays, and interference from hardware noise, the clocks between nodes will gradually drift, resulting in a significant increase in the probability of time slot misalignment and relay failure.

[0003] Currently, the mainstream synchronization methods are mainly based on the following two ideas: the first is centralized time reference broadcasting. For example, the master node periodically sends a global time beacon, and all nodes align their local clocks according to the reception time; the second is a distributed cooperative synchronization mechanism, that is, each node constructs a relative clock relationship and adjusts it by exchanging timestamps with adjacent nodes and calculating offsets. Although these two methods can improve the synchronization accuracy to a certain extent, they both have obvious limitations. The centralized method has serious cumulative synchronization errors in dynamic topologies and multi-hop chain structures. Once the main reference node fails, the entire system may lose its synchronization ability; the distributed method faces problems such as high communication overhead, slow convergence speed, and lack of stable trend modeling ability, especially in scenarios where the node duty cycle is extremely low and energy consumption is limited, it is difficult to be actually deployed.

[0004] In addition, in existing TDMA multi-hop synchronization mechanisms, most use simple linear fitting, least squares method, or clock filters to estimate the time deviation between nodes. Although these algorithms have low computational complexity, they lack the ability to model the non-linear characteristics of clock offsets and are difficult to adapt to the dynamic fluctuations of clock offsets caused by factors such as power fluctuations, temperature changes, and radio frequency disturbances in the actual wireless environment, resulting in difficult-to-control synchronization prediction errors. At the same time, current synchronization mechanisms generally use static configuration parameters and cannot adaptively adjust the structure and update rhythm of the clock model according to feedback during operation. They lack flexibility and robustness and cannot meet the synchronization accuracy requirements in long-period, large-span, and strongly dynamic scenarios.

[0005] Therefore, how to provide a multi-hop pipeline propagation method based on TDMA time slot-level synchronization is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0006] One purpose of the present invention is to propose a multi-hop pipeline propagation method based on TDMA time slot level synchronization. The present invention adopts a bat resonance fitting algorithm and a pseudo-transition time mapping mechanism, combined with a TDMA four-time slot structure, to achieve autonomous time synchronization and clock drift modeling between multi-hop nodes without centralized control. Through model parameter transmission and feedback comparison, a multi-hop pipeline propagation system with TDMA time slot level synchronization is constructed. The system has the advantages of high precision, low power consumption, strong anti-interference and fast adaptability to topology changes, and is suitable for industrial Internet of Things and self-organizing network scenarios in large-scale complex environments.

[0007] The multi-hop pipeline propagation method based on TDMA time slot level synchronization according to an embodiment of the present invention comprises the following steps: S1, construct a TDMA frame format with a four-time slot structure, each frame includes an uplink receiving time slot, a local processing time slot, a downlink transmitting time slot and a protection sleep time slot; S2. In the uplink receiving time slot, the node receives the data packet from the previous hop node, extracts the remaining time slot duration carried in the data packet, calculates the time slot end estimate value in combination with the local receiving timestamp, and constructs a local pseudo-transition time mapping table; S3, the node extracts several cycles of historical records from the local pseudo-transition time mapping table, and constructs a local clock drift trend model of the node relative to the previous jump node based on the bat resonance fitting algorithm; S4. The node predicts the expected arrival time of the next uplink synchronization according to the local clock drift trend model, and adjusts the local clock in advance to align the local operation with the synchronization event of the previous hop node; S5. In the local processing time slot, the node completes the decoding, verification, address resolution and queue management operations of the data packet; S6. In the downlink transmission time slot, the node completes the service data transmission and adds the local clock drift trend model parameters as synchronization auxiliary information for the next hop node to establish its own local pseudo transition time mapping table; S7. In the protection sleep time slot, the node enters a low-power listening state, receives feedback clock drift trend model parameters sent from the next hop node, compares the difference between the feedback clock drift trend model parameters and the local clock drift trend model parameters, and updates the local clock drift trend model if the difference exceeds a preset tolerance threshold; S8. The node injects Gaussian disturbance into the local clock within the set frame period and automatically updates the local clock drift trend model at regular intervals.

[0008] Optionally, in the TDMA frame format of the four-slot structure, the first slot is an uplink receiving slot, and its duration is set to 20% to 25% of the frame period; the second slot is a local processing slot, and its duration is set to 30% to 35% of the frame period; the third slot is a downlink transmitting slot, and its duration is set to 20% to 25% of the frame period; the fourth slot is a protection and sleep slot, and its duration is set to 20% to 30% of the frame period.

[0009] Optionally, the S2 specifically includes: S21. After the node enters the uplink receiving slot, it activates the radio frequency receiving circuit, performs continuous detection within the set channel energy detection threshold range, and completes synchronization locking after detecting the packet preamble sent by the previous hop node; S22. When the node receives the start boundary of the packet synchronization field, it records the current count value of the local clock as the reception timestamp ; S23. Analyze the control field in the packet, and extract the remaining slot duration inserted by the previous hop node during transmission data field, and the remaining slot duration represents the time length between the current transmission operation of the previous hop node calculated under the local clock and the end of this slot; S24. Calculate the slot end estimate : ; The slot end estimate represents the end time of the current uplink receiving slot speculated by the node within the local clock domain; S25. The node combines the network identifier , the current frame number , the reception timestamp , the remaining slot duration and the slot end estimate to form a structured record : ; S26. The node writes the structured record into the local pseudo-transition time mapping table in the form of a five-tuple. The local pseudo-transition time mapping table is a local data structure with a historical sliding window mechanism, supporting retrieval, clearing, and updating sorted by frame number or timestamp.

[0010] Optionally, the S3 specifically includes: S31. The node extracts the historical records within several consecutive frame periods from the local pseudo-transition time mapping table. The node obtains the drift error value based on the difference in the slot end estimation values between adjacent historical records, and constructs a local drift error sequence, which is used to represent the trend of the clock offset of this node relative to the previous hop node changing over time; S32. The node constructs a drift state diagram according to the local drift error sequence, maps the drift error value corresponding to each historical frame period to a state node, and establishes a connection edge between adjacent state nodes to form a non-Euclidean search space for path search; S33. Introduce the bat resonance fitting algorithm. The node initializes the bat fitting set, and the bat fitting set includes simulated bat individuals. Each simulated bat individual carries a set of fitting function parameters, which are used to construct a potential drift trend path and guide the path direction through the acoustic resonance structure function: ; Among them, represents the acoustic resonance structure function corresponding to the th simulated bat individual, represents the amplitude factor, represents the search frequency, represents the initial phase, represents the exponential decay coefficient; S34. The node constructs a frequency hopping self-tuning modulation kernel function for each simulated bat individual: ; Among them, represents the frequency hopping self-tuning modulation kernel function corresponding to the th simulated bat individual, represents the gradient of the acoustic resonance structure function, represents the modulation sensitivity factor, which is used to dynamically adjust the step size and the frequency search range; S35. In each round of fitting iteration, the node is based on the search guidance mechanism jointly composed of the acoustic resonance structure function and the frequency hopping self-tuning modulation kernel function until the fitting residual of the drift error value meets the preset convergence threshold or reaches the maximum number of iterations; S36. The node selects simulated bat individuals with the smallest fitting residuals from all simulated bat individuals, where , and constructs a local clock drift trend model of this node relative to the previous hop node: ; S37. The node stores the local clock drift trend model in the fitting buffer.

[0011] Optionally, S4 specifically includes: S41. The node calls the local clock drift trend model from the fitting buffer area. The local clock drift trend model records the offset change trend of the node's local clock relative to the previous-hop node's clock within multiple frame periods; S42. The node obtains the current frame period number, uses the next frame period number as the prediction input, calculates the local clock drift deviation expected to occur in the next frame period using the local clock drift trend model, and then superimposes the local clock drift deviation on the starting time of the standard uplink reception time slot, so as to obtain the expected arrival time of the previous-hop node's transmission event in the local clock; S43. The node compares the expected arrival time with the current local clock value and calculates the time deviation of the local operation relative to the previous-hop node's synchronization event; S44. The node adjusts the local clock according to the calculated time deviation to advance the opening of the local uplink reception window in the next frame period, ensuring that the local operation is aligned with the previous-hop node's synchronization event.

[0012] Optionally, S6 specifically includes: S61. The node assembles a data packet in the downlink transmission time slot and encapsulates the service data into the data payload area; S62. The node extracts the currently effective local clock drift trend model parameters from the fitting buffer area. The local clock drift trend model parameters include acoustic resonance structure function parameters and frequency hopping self-adjusting modulation kernel function parameters; S63. The node packs the local clock drift trend model parameters into a synchronization information field and embeds them into the additional control field of the data packet header, and sends them to the next-hop node together with the service data; S64. After receiving the data packet containing the local clock drift trend model parameters, the next-hop node regards the local clock drift trend model parameters as the prediction basis for the end time of the uplink transmission time slot of the previous-hop node, combines the received timestamp to reverse the expected transmission time of the previous-hop node under the local clock, calculates the corresponding local time slot end estimate value, and constructs a new record in the local pseudo-transition time mapping table together with the received timestamp.

[0013] Optionally, S7 specifically includes: S71. The node switches to the low-power listening mode in the protection sleep time slot, and only keeps the radio frequency reception path and the synchronization control module in the working state to receive the synchronization feedback data from the next-hop node; S72. The node analyzes the feedback field in the received data packet during the listening process and extracts the local clock drift trend model parameters returned by the next-hop node; S73. The node performs a field-level comparison between the received feedback local clock drift trend model parameters and the local clock drift trend model parameters in the current local fitting buffer to obtain the difference. S74. The node determines whether the difference exceeds the acceptance range according to a preset tolerance threshold. If it exceeds, it triggers a refitting mechanism, re-obtains new historical records from the local pseudo-transition time mapping table, and updates the local clock drift trend model.

[0014] Optionally, the S8 specifically includes: S81. When the node reaches the boundary of each set frame period, it reads the local clock drift trend model saved in the current fitting buffer. S82. Inject Gaussian perturbations into the local clock drift trend model parameters through local control logic. The Gaussian perturbations are pseudo-random perturbations with zero mean and a set variance. S83. The node uses the local clock drift trend model after injecting the Gaussian perturbations as the input for the new round of prediction calculation, and rewrites the updated local clock drift trend model into the fitting buffer.

[0015] The beneficial effects of the present invention are as follows: (1) Achieve inter-hop autonomous fitting synchronization without a centralized control node: By constructing a local pseudo-transition time mapping table and introducing a bat resonance fitting algorithm, the node can autonomously model the clock drift trend relative to the previous-hop node in a multi-hop link, realizing non-centralized and distributed slot-level time synchronization.

[0016] (2) Improve synchronization accuracy and robustness and adapt to dynamic environmental changes: Use the acoustic resonance structure function and the frequency hopping self-adjusting modulation kernel function to perform non-linear modeling on the drift trend between nodes, effectively suppressing the fitting error of traditional linear fitting in a complex wireless interference environment, and enhancing synchronization accuracy and system stability.

[0017] (3) Support the cross-node transfer of synchronization parameters and achieve pipelined link cooperation: By attaching the local clock drift trend model parameters to the downlink data, support the orderly transfer of synchronization trend information between the previous-hop and the next-hop nodes, enabling the node to establish its own local pseudo-transition time mapping table and form a cascaded synchronization mechanism for clock modeling.

[0018] (4) Have the ability of model self-evolution and be applicable to low communication density scenarios: By introducing Gaussian perturbations and triggering the model update mechanism within a set frame period, even during the period without external interaction, the node can still maintain the freshness and adaptability of the model, improving the self-healing ability of the system in scenarios such as link breakage and interference.

[0019] (5) Support the low-power listening feedback mechanism to ensure the closed-loop update of the synchronization model accuracy: The node receives the model parameters fed back by the downstream in the protected sleep time slot with a power consumption not higher than 2.5 milliwatts, and judges the local model accuracy based on this, constructs a perturbation correction closed-loop of the synchronization link, and further improves the reliability of the system during long-term operation. Description of the Drawings

[0020] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. In the drawings: Figure 1 It is the overall flowchart of the multi-hop pipeline propagation method based on TDMA time slot-level synchronization proposed by the present invention; Figure 2 It is the processing flowchart of generating the local pseudo-transition time mapping table of the multi-hop pipeline propagation method based on TDMA time slot-level synchronization proposed by the present invention; Figure 3 It is the step flowchart of constructing the local clock drift trend model based on the bat resonance fitting algorithm of the multi-hop pipeline propagation method based on TDMA time slot-level synchronization proposed by the present invention. Detailed Embodiments

[0021] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only showing the basic structure of the present invention in a schematic way, so they only show the components related to the present invention.

[0022] Refer to Figures 1 - 3 , the multi-hop pipeline propagation method based on TDMA time slot-level synchronization includes the following steps: S1. Construct a TDMA frame format with a four-time-slot structure. Each frame sequentially includes an uplink reception time slot, a local processing time slot, a downlink transmission time slot, and a protected sleep time slot; S2. In the uplink reception time slot, the node receives the data packet from the previous-hop node, extracts the remaining time slot duration carried in the data packet, calculates the time slot end estimate value in combination with the local reception timestamp, and constructs a local pseudo-transition time mapping table; S3. The node extracts the historical records of several cycles from the local pseudo-transition time mapping table, and constructs a local clock drift trend model of the present node relative to the previous-hop node based on the bat resonance fitting algorithm; S4. The node predicts the expected arrival time of the next uplink synchronization according to the local clock drift trend model, and adjusts the local clock in advance to align the local operation with the synchronization event of the previous-hop node; S5. In the local processing time slot, the node completes the operations of decoding, verifying, address resolution, and queue management of the data packet; S6. In the downlink transmission time slot, the node completes the transmission of service data and attaches the local clock drift trend model parameters as synchronization auxiliary information for the next-hop node to establish its own local pseudo-transition time mapping table. S7. In the protection and sleep time slot, the node enters the low-power listening state, receives the feedback clock drift trend model parameters sent by the next-hop node, compares the difference between the feedback clock drift trend model parameters and the local clock drift trend model parameters. If the difference exceeds the preset tolerance threshold, the local clock drift trend model is updated. S8. The node injects Gaussian perturbation into the local clock within the set frame period to automatically update the local clock drift trend model at regular intervals.

[0023] By constructing a TDMA frame format with a four-time-slot structure, the operation processes of each stage, namely uplink reception, local processing, downlink transmission, and protection and sleep, are clearly defined, making the transmission behavior between nodes highly deterministic in time, avoiding conflicts and competitions. Meanwhile, by introducing a method that combines the local pseudo-transition time mapping table with the local clock drift trend model, hop-by-hop modeling and continuous prediction of clock offsets in a multi-hop network are realized. Compared with traditional schemes based on broadcast timestamps or centralized synchronization, this method does not require the control of a central node, has a fully distributed synchronization ability, and has controllable synchronization accuracy and strong link transitivity. Combining the downlink trend model transmission with the protection time slot feedback mechanism can form a two-way closed-loop system of forward prediction and reverse calibration, enabling the entire synchronization network to have comprehensive advantages such as strong robustness, dynamic adaptability, and low-power operation.

[0024] In this embodiment, in the TDMA frame format with a four-time-slot structure, the first time slot is the uplink reception time slot, and its duration is set to 20% to 25% of the frame period; the second time slot is the local processing time slot, and its duration is set to 30% to 35% of the frame period; the third time slot is the downlink transmission time slot, and its duration is set to 20% to 25% of the frame period; the fourth time slot is the protection and sleep time slot, and its duration is set to 20% to 30% of the frame period.

[0025] By setting the proportion range of each time slot in the TDMA frame period, the time resources of each link of uplink reception, local processing, downlink transmission, and protection and sleep are finely divided. Compared with the traditional structure with a fixed proportion or equal time slot configuration, this method is optimally arranged according to the multi-hop pipeline forwarding characteristics, enabling the processing delay and transmission delay to reach a balance and effectively improving the frame-level throughput capacity. Especially, a relatively higher proportion of the duration is reserved for local processing, leaving sufficient computing time for model fitting, data queue scheduling, and error checking to ensure the accuracy of trend modeling. The flexible configuration of the protection and sleep time slot duration also enables the system to dynamically adjust the listening timing according to the energy budget and feedback density, thus taking into account both the maintenance of clock accuracy and the goal of minimizing energy consumption.

[0026] In this embodiment, S2 specifically includes: S21. After the node enters the uplink receiving time slot, it activates the radio frequency receiving circuit, performs continuous detection within the set channel energy detection threshold range, and completes synchronization locking after detecting the packet preamble sent by the previous-hop node; S22. When the node receives the start boundary of the packet synchronization field, it records the current count value of the local clock as the reception timestamp ; S23. Analyze the control field in the packet, and extract the remaining time slot duration data field inserted by the previous-hop node during transmission. The remaining time slot duration represents the time length between the current transmission operation calculated by the previous-hop node under the local clock and the end of this time slot; S24. Calculate the estimated value of the end of the time slot : ; The estimated value of the end of the time slot represents the end time of the current uplink receiving time slot speculated by the node within the local clock domain; S25. The node combines the network identifier , current frame number , reception timestamp , remaining time slot duration and the estimated value of the end of the time slot to form a structured record : ; S26. The node writes the structured record into the local pseudo-transition time mapping table in the form of a five-tuple. The local pseudo-transition time mapping table is a local data structure with a historical sliding window mechanism, supporting retrieval, clearing, and updating sorted by frame number or timestamp.

[0027] During the uplink reception process, the node introduces energy awareness triggering for the preamble and precise recording of the reception timestamp, realizing a highly robust reception boundary capture mechanism. By extracting the remaining time slot duration embedded by the previous-hop node from the received packet and combining the local reception time to calculate the estimated value of the end of the current time slot, an accurate speculation of the inter-hop time slot boundary is formed. This estimated value record is written into the local pseudo-transition time mapping table in the form of a structured five-tuple and dynamically maintained using the historical sliding window mechanism, enabling the node to continuously obtain a high-quality original dataset of time drift, supporting subsequent trend modeling requirements, and improving the stability of drift estimation and the ability to resist burst errors.

[0028] In this embodiment, S3 specifically includes: S31. The node extracts the historical records within several consecutive frame periods from the local pseudo-transition time mapping table. The node obtains the drift error value based on the difference in the slot end estimation values between adjacent historical records, and constructs a local drift error sequence, which is used to represent the trend of the clock offset of this node relative to the previous hop node changing over time; S32. The node constructs a drift state diagram according to the local drift error sequence, maps the drift error value corresponding to each historical frame period to a state node, and establishes a connection edge between adjacent state nodes to form a non-Euclidean search space for path search; S33. Introduce the bat resonance fitting algorithm. The node initializes the bat fitting set, and the bat fitting set includes simulated bat individuals. Each simulated bat individual carries a set of fitting function parameters, which are used to construct a potential drift trend path and guide the path direction through the acoustic resonance structure function: ; Among them, represents the acoustic resonance structure function corresponding to the th simulated bat individual, represents the amplitude factor, represents the search frequency, represents the initial phase, represents the exponential decay coefficient; S34. The node constructs a frequency hopping self-adjusting modulation kernel function for each simulated bat individual: ; Among them, represents the frequency hopping self-adjusting modulation kernel function corresponding to the th simulated bat individual, represents the gradient of the acoustic resonance structure function, represents the modulation sensitivity factor, which is used to dynamically adjust the step size and the frequency search range; S35. In each round of fitting iteration, the node is based on the search guidance mechanism jointly composed of the acoustic resonance structure function and the frequency hopping self-adjusting modulation kernel function until the fitting residual of the drift error value meets the preset convergence threshold or reaches the maximum number of iterations; S36. The node selects simulated bat individuals with the smallest fitting residuals from all simulated bat individuals, where , and constructs a local clock drift trend model of this node relative to the previous hop node: ; S37. The node stores the local clock drift trend model in the fitting buffer.

[0029] The bat resonance fitting algorithm is used to model the local drift error sequence, breaking through the problem that traditional linear fitting methods cannot express non-stationary offset trends. By introducing the acoustic resonance structure function and the frequency hopping self-adjusting modulation kernel function to construct a multi-dimensional search space, the path search behavior of multiple bat individuals is simulated, enabling the model to have sensitivity and adaptability to dynamic drift trends. This algorithm can self-adjust the step size and frequency range during the search process to avoid falling into local optima, and select the optimal individual to construct the final trend model when the fitting residuals reach the convergence condition. This model is stored in the fitting buffer in the form of structured parameters and can be used for prediction and downlink synchronization parameter embedding. It has the advantages of high accuracy, sustainable evolution, and strong algorithm stability, significantly improving the clock offset modeling ability of multi-hop synchronization systems.

[0030] In this embodiment, step S4 specifically includes: S41. The node calls the local clock drift trend model from the fitting buffer. The local clock drift trend model records the offset change trend of the node's local clock relative to the previous hop node's clock within multiple frame periods. S42. The node obtains the current frame period number and uses the next frame period number as the prediction input. It calculates the local clock drift deviation expected to occur in the next frame period using the local clock drift trend model, and then adds the local clock drift deviation to the starting time of the standard uplink reception time slot to obtain the expected arrival time of the previous hop node's transmission event in the local clock. S43. The node compares the expected arrival time with the current local clock value to calculate the time deviation of the local operation relative to the previous hop node's synchronization event. S44. The node adjusts the local clock according to the calculated time deviation to advance the opening of the local uplink reception window in the next frame period, ensuring that the local operation is aligned with the previous hop node's synchronization event.

[0031] The node uses the local clock drift trend model to accurately predict the expected arrival time of the next reception event, effectively advancing the opening time of the reception window, so that the local operation is aligned with the previous hop node's transmission behavior at the time slot level. Compared with the single synchronization reference method, it can continuously track the offset trend and perform dynamic compensation, significantly improving synchronization consistency and reducing the probability of reception jitter and misalignment packet loss. By comparing the current clock value with the predicted time to calculate the time deviation and performing local timer fine-tuning based on this deviation, the node can complete adaptive synchronization without external control, which is especially suitable for multi-hop network environments without a master node and with independent clock operation.

[0032] In the local processing time slot, the node performs multi-level processing operations from the physical layer to the network layer to ensure the reliability and forwarding efficiency of the received data. In the physical layer processing, the node first performs error correction decoding on the received data packet, using lightweight low-density parity-check (LDPC) codes to control the computational complexity while ensuring the error correction ability and adapting to the operating environment of resource-constrained devices. At the same time, the node uses the pilot symbols in the data packet for channel estimation to obtain the current link quality information, providing a basis for the transmit power control in the downlink transmission time slot.

[0033] After completing the physical layer processing, the node enters the network layer processing stage. First, by parsing the destination address field of the data packet and combining with a preset hash function, it performs time slot mapping operations. For example, the modulo operation on the lower 2 bits of the destination address can be used to map the data to the specified downlink transmission time slot, realizing a centerless control forwarding scheduling. Subsequently, the node performs queue management according to the data type and service priority, arranging the emergency data (such as alarm information) in the high-priority transmission queue preferentially to ensure the real-time performance of critical services; the regular services are queued in the low-priority queue according to the first-in-first-out principle to achieve a balance between delay control and resource scheduling. The design of the local processing time slot ensures the rapid decoding and reasonable queuing of the data after reception, providing support for the effective forwarding of the next-hop node.

[0034] In this embodiment, the specific steps of S6 include: S61. The node assembles the data packet in the downlink transmission time slot and encapsulates the service data into the data payload area; S62. The node extracts the currently effective local clock drift trend model parameters from the fitting buffer area. The local clock drift trend model parameters include the acoustic resonance structure function parameters and the frequency hopping self-adjusting modulation kernel function parameters; S63. The node packs the local clock drift trend model parameters into a synchronization information field and embeds it into the additional control field of the data packet header, and sends it to the next-hop node together with the service data; S64. After receiving the data packet containing the local clock drift trend model parameters, the next-hop node regards the local clock drift trend model parameters as the prediction basis for the end time of the uplink transmission time slot of the previous-hop node, and combines the received timestamp to reverse the expected transmission moment of the previous-hop node under the local clock, calculates the corresponding local time slot end estimate value, and constructs a new record in the local pseudo-transition time mapping table together with the received timestamp.

[0035] During the downlink transmission time slot, the node embeds the current local clock drift trend model parameters into the packet control field, enabling the synchronization information to naturally spread to the next-hop node along with the data transmission and realizing the chain inheritance of synchronization parameters. This method does not require dedicated synchronization packets or signaling loads and has the advantage of zero additional communication overhead. The next-hop node performs backward inference based on the reception time and the carried model parameters to construct its own pseudo-transition time mapping table, thereby forming a continuous and transferable trend modeling link. The structured propagation and update mechanism of this synchronization model improves the consistency and propagation efficiency of the network-wide drift modeling, laying a foundation for prediction and correction.

[0036] In this embodiment, the S7 specifically includes: S71. The node switches to the low-power listening mode during the protection sleep time slot, and only keeps the radio frequency reception path and the synchronization control module in the working state to receive the synchronization feedback data from the next-hop node; S72. The node analyzes the feedback field in the received packet during the listening process and extracts the local clock drift trend model parameters returned by the next-hop node; S73. The node performs a field-level comparison between the received feedback local clock drift trend model parameters and the local clock drift trend model parameters in the current local fitting buffer to obtain the difference; S74. The node determines whether the difference exceeds the acceptable range according to the preset tolerance threshold. If it exceeds, it triggers the re-fitting mechanism, re-obtains the new historical records from the local pseudo-transition time mapping table, and updates the local clock drift trend model.

[0037] The node enters the low-power listening state during the protection sleep time slot and receives the downstream feedback drift trend model parameters without activating the main processing module, realizing the closed-loop verification and accuracy maintenance of the synchronization model. The node judges whether the current model deviates from the actual system trend through field-level difference comparison and triggers the re-fitting process when the tolerance is exceeded to ensure that the model always reflects the latest offset characteristics. The feedback path forms a calibration mechanism for drift modeling, which has the ability to resist long-term error accumulation and correct mis-drift. At the same time, its low-power listening strategy gives the system a great advantage in energy consumption control.

[0038] In this embodiment, the S8 specifically includes: S81. When the node reaches the boundary of each set frame period, it reads the local clock drift trend model saved in the current fitting buffer; S82. Inject Gaussian perturbations into the local clock drift trend model parameters through the local control logic. The Gaussian perturbations are pseudo-random perturbations with zero mean and a set variance; S83. The node uses the local clock drift trend model after injecting the Gaussian perturbation as the input for the new round of prediction calculation, and rewrites the updated local clock drift trend model into the fitting buffer.

[0039] By regularly injecting Gaussian perturbations into the local drift model within the set frame period, the model can still maintain its evolution ability without external data input, preventing the accumulation of synchronization prediction errors caused by model overfitting or rigidity. Use the perturbed model to perform a self-prediction and update the fitting parameters to form a self-supervised evolution correction path. It is applicable to the synchronization strategy maintenance when the node's communication is discontinuous or the link is interrupted, significantly enhancing the model's robustness and self-healing ability, and providing support for the network to maintain synchronization stability in abnormal states.

[0040] Embodiment 1: To verify the feasibility of the present invention in implementation, the present invention is applied to a typical low-power multi-hop wireless communication system to simulate continuous industrial Internet of Things scenarios, including tunnel structure health monitoring, petrochemical pipeline leakage detection, or forest boundary environment monitoring tasks. In this system, multiple sensing nodes report information at fixed time intervals, and the networking topology is a linear chain structure. The network supports data transmission links of more than 20 hops at most. Each node adopts a four-time-slot frame structure based on TDMA. Each frame period is fixed at 400 ms and is divided into uplink reception, local processing, downlink transmission, and protection sleep time slots, and the proportion range is set as: 25%, 30%, 20%, 25%.

[0041] In the synchronization mechanism, each node in this system constructs a local pseudo-transition time mapping table internally and uses the bat resonance fitting algorithm to generate a local clock drift trend model to achieve the prediction alignment of uplink synchronization events. The node attaches the model parameters to the downlink data packet for the next-hop node to construct its own mapping table to achieve the link transmission of the synchronization trend. At the same time, the node listens for the feedback drift model parameters from the successor node in the protection sleep time slot in a low-power mode, and judges whether the model deviation exceeds the tolerance based on this, triggering the refitting mechanism. Within every 10 frame periods, the node actively injects Gaussian perturbations into the fitting model to make the model have the evolution ability and maintain the timeliness and robustness of drift prediction.

[0042] To comprehensively compare the performance differences between the present invention and mainstream synchronization mechanisms in a multi-hop wireless communication environment, the following control test system is constructed to test indicators such as synchronization accuracy, control overhead, maximum supported hop count, node unit energy consumption, and response time to topology changes. The results are summarized in the following table.

[0043] Table 1 Comparison table of the present invention and existing typical protocols in core performance indicators

[0044] The test results in Table 1 above show that the synchronization accuracy achieved by the present invention is stably maintained within ±50 microseconds, far superior to LoRaWAN Class B (±1 ms) and approaching the TSCH standard (±10 μs), meeting the accuracy requirements for slot-level synchronization in industrial scenarios. In terms of energy efficiency, due to the adoption of prediction-driven trend modeling and a radio frequency activation mechanism with an extremely low duty cycle, the unit energy consumption of nodes is significantly lower than that of other protocols, only 0.25 millijoules / node, which is approximately 41.7% of IEEE 802.15.4 TSCH and approximately 16.7% of LoRaWAN Class B. In terms of the ability to respond to topological changes, with the help of the link transfer and fast feedback mechanism of the model parameters, the present invention can complete the synchronization update under the new topology within 1 frame period, far leading the minute-level resynchronization overhead in existing solutions.

[0045] Meanwhile, under multi-hop conditions, the present invention successfully supports stable communication of 20 hops without increasing control signaling, while the synchronization error of TSCH begins to accumulate significantly at 10 hops, and LoRaWAN does not have effective synchronization guarantee after 2 hops due to its non-slot scheduling structure.

[0046] In this embodiment, by applying the present invention to a typical multi-hop wireless communication system, the technical advantages of the present invention in terms of slot-level synchronization accuracy, hop count support ability, energy consumption control, and topological adaptability are fully verified. Compared with the existing synchronization mechanism, without the need for central coordination and additional signaling overhead, the synchronization accuracy between nodes can be stably controlled within ±50 microseconds, far superior to the millisecond-level synchronization accuracy provided by traditional protocols. At the same time, the system can still maintain an end-to-end delay of less than 50 milliseconds under a 20-hop link, indicating that the present invention has good delay controllability and scalability. The unit energy consumption of nodes is reduced to 0.25 millijoules, significantly lower than the mainstream solutions, ensuring the energy sustainability during long-term operation. In addition, when the topology changes, the synchronization reconstruction can be completed in only 1 frame period, demonstrating extremely strong dynamic response ability and robustness, fully illustrating the wide adaptability and engineering practical value of the present invention in complex industrial, environmental monitoring, and low-power self-organizing network scenarios.

[0047] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, should be covered by the protection scope of the present invention.

Claims

1. A multi-hop pipeline propagation method based on TDMA time slot-level synchronization, characterized in that The steps include: S1, construct a TDMA frame format with a four-time slot structure, each frame includes an uplink receiving time slot, a local processing time slot, a downlink transmitting time slot and a protection sleep time slot; S2. In the uplink receiving time slot, the node receives the data packet from the previous hop node, extracts the remaining time slot duration carried in the data packet, calculates the time slot end estimate value in combination with the local receiving timestamp, and constructs a local pseudo-transition time mapping table; S3, the node extracts several cycles of historical records from the local pseudo-transition time mapping table, and constructs a local clock drift trend model of the node relative to the previous jump node based on the bat resonance fitting algorithm; S4. The node predicts the expected arrival time of the next uplink synchronization according to the local clock drift trend model, and adjusts the local clock in advance to align the local operation with the synchronization event of the previous hop node; S5. In the local processing time slot, the node completes the decoding, verification, address resolution and queue management operations of the data packet; S6. In the downlink transmission time slot, the node completes the service data transmission and adds the local clock drift trend model parameters as synchronization auxiliary information for the next hop node to establish its own local pseudo transition time mapping table; S7. In the protection sleep time slot, the node enters a low-power listening state, receives feedback clock drift trend model parameters sent from the next hop node, compares the difference between the feedback clock drift trend model parameters and the local clock drift trend model parameters, and updates the local clock drift trend model if the difference exceeds a preset tolerance threshold; S8. The node injects Gaussian disturbance into the local clock within the set frame period and automatically updates the local clock drift trend model at regular intervals.

2. The multi-hop pipeline propagation method based on TDMA time slot-level synchronization according to claim 1, wherein In the TDMA frame format of the four-slot structure, the first time slot is an uplink reception time slot, and the duration is set to 20% to 25% of the frame period; the second time slot is a local processing time slot, and the duration is set to 30% to 35% of the frame period; the third time slot is a downlink transmission time slot, and the duration is set to 20% to 25% of the frame period; the fourth time slot is a protection sleep time slot, and the duration is set to 20% to 30% of the frame period.

3. The multi-hop pipeline propagation method based on TDMA time slot-level synchronization according to claim 1, characterized in that The S2 specifically includes: S21, after entering the uplink receiving time slot, the node starts the RF receiving circuit, performs continuous detection within the set channel energy detection threshold range, and completes synchronization locking after detecting the preamble code of the data packet sent by the previous hop node; S22. When the node receives the start boundary of the packet synchronization field, record the current count value of the local clock as the reception timestamp ; S23. Analyze the control field in the data packet and extract the remaining time slot duration inserted by the previous hop node during transmission. Data field, the remaining time slot duration Indicates the time length between the current transmission operation calculated by the previous hop node under the local clock and the end of this time slot. S24. Calculate the estimated value of the end of the time slot : ； The time slot end estimate value represents the end time of the current uplink receiving time slot in the local clock domain estimated by the node; S25. The node takes the network identifier of the previous hop node , the current frame number , the reception timestamp , the remaining slot duration and the slot end estimate to form a structured record : ； S26. The node writes the structured record into a local pseudo-transition time mapping table in the form of a five-tuple. The local pseudo-transition time mapping table is a local data structure with a historical sliding window mechanism, and supports retrieval, clearing and updating sorted by frame number or timestamp.

4. The multi-hop pipeline propagation method based on TDMA time slot-level synchronization according to claim 1, wherein The S3 specifically includes: S31, the node extracts historical records within several consecutive frame periods from the local pseudo-transition time mapping table, the node obtains a drift error value based on the difference between the time slot end estimation values between adjacent historical records, and constructs a local drift error sequence, wherein the local drift error sequence is used to represent the trend of the clock offset of the node relative to the previous hop node over time; S32. The node constructs a drift state graph based on the local drift error sequence, maps the drift error value corresponding to each historical frame period to a state node, and establishes connection edges between adjacent state nodes to form a non-Euclidean search space for path search; S33. Introduce the bat resonance fitting algorithm, and initialize the bat fitting set at the node. The bat fitting set includes simulated bat individuals. Each simulated bat individual carries a set of fitting function parameters, which are used to construct the potential drift trend path and guide the path direction through the acoustic resonance structure function: ； Among them, represents the acoustic resonance structure function corresponding to the th simulated bat individual, represents the amplitude factor, represents the search frequency, represents the initial phase, represents the exponential decay coefficient; S34. The node constructs a frequency-hopping self-tuning modulation kernel function for each simulated bat individual; ； in, Indicates The frequency hopping self-tuning modulation kernel function corresponding to the simulated bat individual, represents the acoustic resonance structure function gradient, Represents the modulation sensitivity factor, which is used to dynamically adjust the step size and frequency search range; S35. In each round of fitting iteration, the node is based on the acoustic resonance structure function and the frequency hopping self - tuning modulation kernel function to jointly form a search guidance mechanism until the fitting residual of the drift error value meets the preset convergence threshold or reaches the maximum number of iterations; S36. The node selects simulated bat individuals with the smallest fitting residuals from all simulated bat individuals, where , and constructs a local clock drift trend model of the node relative to the previous hop node : ​​​ ； S37. The node stores the local clock drift trend model in the fitting buffer.

5. The multi-hop pipeline propagation method based on TDMA time slot-level synchronization according to claim 1, wherein The specific steps of S4 are as follows: S41. The node calls the local clock drift trend model from the fitting buffer. The local clock drift trend model records the offset change trend of the node's local clock relative to the clock of the previous-hop node over multiple frame periods; S42. The node obtains the current frame period number, uses the next frame period number as the prediction input, calculates the local clock drift deviation expected to occur in the next frame period using the local clock drift trend model, and then superimposes the local clock drift deviation on the start time of the standard uplink reception time slot to obtain the expected arrival time of the previous-hop node's transmission event in the local clock; S43. The node compares the expected arrival time with the current local clock value to calculate the time deviation of the local operation relative to the synchronization event of the previous-hop node; S44. The node adjusts the local clock according to the calculated time deviation to advance the opening of the local uplink reception window in the next frame period, ensuring the alignment of the local operation with the synchronization event of the previous-hop node.

6. The multi-hop pipeline propagation method based on TDMA time slot-level synchronization according to claim 1, wherein The specific steps of S6 are as follows: S61. The node assembles a data packet in the downlink transmission time slot and encapsulates the service data into the data payload area; S62. The node extracts the currently effective local clock drift trend model parameters from the fitting buffer. The local clock drift trend model parameters include acoustic resonance structure function parameters and frequency-hopping self-tuning modulation kernel function parameters; S63. The node packs the local clock drift trend model parameters into a synchronization information field and embeds it into the additional control field of the data packet header, and sends it to the next-hop node together with the service data; S64. After receiving the data packet containing the local clock drift trend model parameters, the next-hop node regards the local clock drift trend model parameters as the prediction basis for the end time of the uplink transmission time slot of the previous-hop node, and combines the reception timestamp to back-calculate the expected transmission time of the previous-hop node in the local clock, calculates the corresponding local time slot end estimate value, and constructs a new record in the local pseudo-transition time mapping table together with the reception timestamp.

7. The multi-hop pipeline propagation method based on TDMA time slot level synchronization according to claim 1, characterized in that The specific steps of S7 are as follows: S71. The node switches to the low-power listening mode in the guard sleep time slot, and only keeps the radio frequency reception path and the synchronization control module in the working state to receive the synchronization feedback data from the next-hop node; S72. The node parses the feedback field in the received data packet during the listening process and extracts the local clock drift trend model parameters returned by the next-hop node; S73. The node compares the received feedback local clock drift trend model parameters with the local clock drift trend model parameters in the current local fitting buffer at the field level to obtain the differences; S74. The node determines whether the difference exceeds the acceptable range according to the preset tolerance threshold. If it exceeds, the re-fitting mechanism is triggered, and the node re-obtains the new historical records from the local pseudo-transition time mapping table and updates the local clock drift trend model.

8. The multi-hop pipeline propagation method based on TDMA time slot-level synchronization according to claim 1, wherein The specific steps of S8 are as follows: S81. When the boundary of each set frame period arrives, the node reads the local clock drift trend model saved in the current fitting buffer. S82. Gaussian perturbation is injected into the parameters of the local clock drift trend model through the local control logic. The Gaussian perturbation is a pseudo-random perturbation with zero mean and a set variance. S83. The node uses the local clock drift trend model after injecting the Gaussian perturbation as the input for the new round of prediction calculation, and rewrites the updated local clock drift trend model into the fitting buffer.

Images