Multi-hop pipeline communication method based on TDMA time slot synchronization

Abstract

The present invention discloses a multi-hop pipeline propagation method based on TDMA time slot synchronization, comprising the following steps: S1, constructing a TDMA frame format; S2, constructing a local pseudo-transition time mapping table in the uplink receive 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 based on the model and adjusting the local clock; S5, completing data packet processing operations in the local processing time slot; S6, in the downlink transmit time slot, the node completes service data transmission and appends local clock drift trend model parameters; S7, entering a low-power listening 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; S8, the node injects Gaussian perturbations into the local clock within a set frame period and regularly updates the model. The present invention uses the bat resonance fitting algorithm and time slot modeling method to achieve multi-hop node autonomous synchronization.

Description

Technical Field

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

[0002] In multi-hop wireless communication networks, clock synchronization between nodes remains a key technology for efficient and reliable data transmission. Existing technologies generally employ time-slot scheduling mechanisms based on TDMA (time division multiple access) to avoid conflicts and improve channel utilization. In such systems, to ensure smooth data relay between nodes on multi-hop links, precise time synchronization is essential to ensure that each node receives, processes, and transmits data in the correct time slot. However, as the network scale expands and the number of node hops increases, clocks between nodes gradually drift due to inconsistent crystal oscillators, uncertain propagation delays, and interference from hardware noise. This significantly increases the probability of time slot misalignment and relay failure.

[0003] The current mainstream synchronization methods are mainly based on the following two ideas: the first is centralized time reference broadcast, 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 collaborative synchronization mechanism, that is, each node establishes a relative clock relationship and adjusts it by exchanging timestamps with adjacent nodes and calculating offsets. Although these two methods can improve synchronization accuracy to a certain extent, they both have obvious limitations. The centralized method suffers from serious synchronization error accumulation in dynamic topologies and multi-hop chain structures. Once the master reference node fails, the entire system may lose its synchronization capability; the distributed method faces the problems of high communication overhead, slow convergence speed, and lack of stable trend modeling capabilities. It is especially difficult to deploy in scenarios with extremely low node duty cycles and limited energy consumption.

[0004] Furthermore, existing TDMA multi-hop synchronization mechanisms mostly use simple linear fitting, least squares methods, or clock filters to estimate inter-node time deviations. While computationally complex, these algorithms lack the ability to model the nonlinear characteristics of clock offsets, making them difficult to adapt to dynamic clock offset fluctuations caused by factors such as power supply fluctuations, temperature variations, and RF disturbances in real wireless environments. This, in turn, makes synchronization prediction errors difficult to control. Furthermore, current synchronization mechanisms generally use static configuration parameters and are unable to adaptively adjust the clock model structure and update rhythm based on feedback during operation. Consequently, they lack flexibility and robustness, and are unable to meet the synchronization accuracy requirements in long-period, large-span, and highly dynamic scenarios.

[0005] Therefore, how to provide a multi-hop pipeline transmission method based on TDMA time slot level synchronization is a problem that those skilled in the art urgently need to solve. Summary of the Invention

[0006] One object 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 rapid adaptation 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 includes the following steps:

[0008] S1. Construct a TDMA frame format with a four-time slot structure, where each frame includes an uplink receive time slot, a local processing time slot, a downlink transmit time slot, and a protection sleep time slot.

[0009] S2. In the uplink receive 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 based on the local receive timestamp, and constructs a local pseudo-transition time mapping table;

[0010] 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 hop node based on a bat resonance fitting algorithm.

[0011] S4. The node predicts the expected arrival time of the next uplink synchronization based on 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;

[0012] S5. In the local processing time slot, the node completes the decoding, verification, address resolution and queue management operations of the data packet;

[0013] 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;

[0014] S7. In the protection sleep time slot, the node enters a low-power listening state, receives 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, and updates the local clock drift trend model if the difference exceeds a preset tolerance threshold;

[0015] 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.

[0016] Optionally, 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; and the fourth time slot is a protection sleep time slot, and the duration is set to 20% to 30% of the frame period.

[0017] Optionally, the S2 specifically includes:

[0018] S21: After entering the uplink receive timeslot, the node starts the RF receiving circuit, performs continuous detection within the set channel energy detection threshold, and completes synchronization lock after detecting the preamble of the data packet sent by the previous hop node;

[0019] S22. When the node receives the starting boundary of the synchronization field of the data packet, it records the current count value of the local clock as the receiving timestamp ;

[0020] S23. Parse 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 sending operation and the end of this time slot calculated by the previous hop node based on the local clock;

[0021] S24. Calculate the estimated value of the end of the time slot :

[0022] ;

[0023] The timeslot end estimated value represents the end time of the current uplink receiving timeslot in the local clock domain estimated by the node;

[0024] S25, the node sets the network identifier of the previous hop node , current frame number , receiving timestamp , remaining time slot duration and the time slot end estimate Composing a structured record :

[0025] ;

[0026] 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.

[0027] Optionally, the S3 specifically includes:

[0028] 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. The local drift error sequence is used to represent the trend of the clock offset of the local node relative to the previous hop node over time.

[0029] 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 connecting edges between adjacent state nodes to form a non-Euclidean search space for path search;

[0030] S33, introduce the bat resonance fitting algorithm, the node initializes the bat fitting set, the bat fitting set includes Each simulated bat individual carries a set of fitting function parameters to construct a potential drift trend path and guide the path direction through the acoustic resonance structure function:

[0031] ;

[0032] in, Indicates the The acoustic resonance structure function corresponding to the simulated bat individual, represents the amplitude factor, Indicates the search frequency, represents the initial phase, represents the exponential decay coefficient;

[0033] S34. The node constructs a frequency hopping self-tuning modulation kernel function for each simulated bat individual:

[0034] ;

[0035] in, Indicates the The frequency hopping self-tuning modulation kernel function corresponding to the simulated bat individual, represents the gradient of the acoustic resonance structure function, Indicates the modulation sensitivity factor, which is used to dynamically adjust the step size and frequency search range;

[0036] S35, the node in each round of fitting iteration is based on the acoustic resonance structure function Frequency Hopping Modulation Kernel Function Together they form a search guidance mechanism until the drift error value fitting residual meets the preset convergence threshold or reaches the maximum number of iterations;

[0037] S36, nodes are selected from all simulated bat individuals The simulated bat individuals with the smallest fitting residuals are , build a local clock drift trend model of the node relative to the previous hop node :

[0038] ;

[0039] S37. The node stores the local clock drift trend model in a fitting buffer.

[0040] Optionally, the S4 specifically includes:

[0041] S41. The node calls a local clock drift trend model from a fitting buffer, where the local clock drift trend model records the offset change trend of the node's local clock relative to the previous node's clock over multiple frame periods.

[0042] S42: The node obtains the current frame period number and uses the next frame period number as a prediction input. The node uses the local clock drift trend model to calculate the local clock drift deviation expected to occur in the next frame period. The node then adds the local clock drift deviation to the start time of the standard uplink receive timeslot to obtain the expected arrival time of the event sent by the preceding node in the local clock.

[0043] 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 synchronization event of the previous hop node;

[0044] S44. The node adjusts the local clock according to the calculated time deviation so that the local uplink receiving window is opened in advance in the next frame period, ensuring that the local operation is aligned with the synchronization event of the previous hop node.

[0045] Optionally, the S6 specifically includes:

[0046] S61. The node assembles a data packet in a downlink transmission time slot and encapsulates the service data into a data payload area.

[0047] S62. The node extracts currently effective local clock drift trend model parameters from the fitting buffer, where the local clock drift trend model parameters include acoustic resonance structure function parameters and frequency hopping self-tuning modulation kernel function parameters.

[0048] S63. The node packages the local clock drift trend model parameters into a synchronization information field, embeds the field into the additional control field of the data packet header, and sends the field to the next hop node together with the service data.

[0049] 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 basis for the previous-hop node to predict the end time of the uplink transmission time slot, and combines the received timestamp to reversely infer the expected transmission time of the previous-hop node under the local clock, calculates the corresponding local time slot end estimate, and jointly constructs a new record in the local pseudo-transition time mapping table with the received timestamp.

[0050] Optionally, the S7 specifically includes:

[0051] S71. The node switches to a low-power listening mode in the protection sleep time slot, leaving only the RF receiving path and the synchronization control module in working state to receive synchronization feedback data from the next-hop node;

[0052] S72. The node parses the feedback field in the received data packet during the monitoring process and extracts the local clock drift trend model parameters returned by the next hop node;

[0053] 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 a field level to obtain differences.

[0054] S74. The node determines whether the difference exceeds an acceptable range based on a preset tolerance threshold. If so, a refitting mechanism is triggered to update the local clock drift trend model from new historical records in the local pseudo-transition time mapping table.

[0055] Optionally, the S8 specifically includes:

[0056] S81. At each set frame period boundary, the node reads the local clock drift trend model stored in the current fitting buffer;

[0057] S82. Injecting a Gaussian perturbation into the local clock drift trend model parameters through local control logic, where the Gaussian perturbation is a pseudo-random perturbation with zero mean and set variance;

[0058] S83: The node uses the local clock drift trend model injected with the Gaussian disturbance as the input for a new round of prediction calculation, and rewrites the updated local clock drift trend model into the fitting buffer.

[0059] The beneficial effects of the present invention are:

[0060] (1) Realize autonomous inter-hop fitting synchronization without the need for centralized control nodes: By constructing a local pseudo-transition time mapping table and introducing a bat resonance fitting algorithm, nodes can autonomously model the clock drift trend relative to the previous hop node in a multi-hop link, achieving decentralized and distributed slot-level time synchronization.

[0061] (2) Improve synchronization accuracy and robustness to adapt to dynamic environmental changes: Use acoustic resonance structure function and frequency hopping self-tuning modulation kernel function to perform nonlinear modeling of the drift trend between nodes, effectively suppress the fitting error of traditional linear fitting in complex wireless interference environment, and enhance synchronization accuracy and system stability.

[0062] (3) Supporting cross-node transmission of synchronization parameters to achieve pipeline link collaboration: By appending local clock drift trend model parameters to downlink data, it supports the orderly transmission of synchronization trend information between the previous hop and the next hop nodes, enabling the nodes to establish their own local pseudo-transition time mapping table, forming a cascade synchronization mechanism for clock modeling.

[0063] (4) It has the ability of model self-evolution and is suitable for low communication density scenarios: by introducing Gaussian perturbations within the set frame period and triggering the model update mechanism, the node can still maintain the freshness and adaptability of the model even in the period without external interaction, thereby improving the system's self-healing ability in scenarios such as link disconnection and interference.

[0064] (5) Support low-power monitoring feedback mechanism to ensure closed-loop update of synchronization model accuracy: During the protection sleep time slot, the node receives the model parameters fed back from the downstream with a power consumption of no more than 2.5 mW, and judges the accuracy of the local model based on this, constructs a perturbation correction closed loop of the synchronization link, and further improves the long-term reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The accompanying 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 of the present invention. In the accompanying drawings:

[0066] Figure 1 This is an overall flow chart of the multi-hop pipeline transmission method based on TDMA time slot level synchronization proposed by the present invention;

[0067] Figure 2 A processing flow chart of generating a local pseudo-transition time mapping table in a multi-hop pipeline propagation method based on TDMA time slot-level synchronization proposed by the present invention;

[0068] Figure 3 This is a flowchart of the steps of constructing a local clock drift trend model based on a bat resonance fitting algorithm in a multi-hop pipeline propagation method based on TDMA time slot-level synchronization proposed by the present invention. DETAILED DESCRIPTION

[0069] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0070] refer to Figure 1-Figure 3 The multi-hop pipeline propagation method based on TDMA time slot level synchronization includes the following steps:

[0071] S1. Construct a TDMA frame format with a four-time slot structure, where each frame includes an uplink receive time slot, a local processing time slot, a downlink transmit time slot, and a protection sleep time slot.

[0072] S2. In the uplink receive 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 based on the local receive timestamp, and constructs a local pseudo-transition time mapping table;

[0073] 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 hop node based on a bat resonance fitting algorithm.

[0074] S4. The node predicts the expected arrival time of the next uplink synchronization based on 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;

[0075] S5. In the local processing time slot, the node completes the decoding, verification, address resolution and queue management operations of the data packet;

[0076] 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;

[0077] S7. In the protection sleep time slot, the node enters a low-power listening state, receives 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, and updates the local clock drift trend model if the difference exceeds a preset tolerance threshold;

[0078] 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.

[0079] By constructing a four-time-slot TDMA frame format and clearly dividing the operational processes of uplink reception, local processing, downlink transmission, and protection sleep, the transmission behavior between nodes is highly deterministic in time, avoiding conflicts and competition. At the same time, the introduction of a local pseudo-transition time mapping table combined with a local clock drift trend model enables hop-by-hop modeling and continuous prediction of clock offsets in multi-hop networks. Compared to traditional solutions based on broadcast timestamps or centralized synchronization, this method does not require central node control and has fully distributed synchronization capabilities with controllable synchronization accuracy and strong link transferability. Combining the downlink trend model transmission with the protection time slot feedback mechanism, a bidirectional closed-loop system of forward prediction and reverse calibration can be formed, giving the entire synchronization network the comprehensive advantages of strong robustness, dynamic adaptability, and low-power operation.

[0080] In this embodiment, in the TDMA frame format of the four-slot structure, the first time slot is the uplink reception time slot, and the duration is set to 20% to 25% of the frame period; the second time slot is the local processing time slot, and the duration is set to 30% to 35% of the frame period; the third time slot is the downlink transmission time slot, and the duration is set to 20% to 25% of the frame period; the fourth time slot is the protection sleep time slot, and the duration is set to 20% to 30% of the frame period.

[0081] 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 sleep can be finely divided. Compared with the traditional structure with fixed ratio or equal time slot configuration, this method optimizes the layout according to the multi-hop pipeline forwarding characteristics, so that the processing delay and transmission delay are balanced, effectively improving the frame-level throughput. In particular, a relatively higher proportion of time is reserved for local processing, leaving sufficient computing time for model fitting, data queue scheduling and error checking, and ensuring the accuracy of trend modeling. The flexible configuration of the protection sleep time slot length also allows the system to dynamically adjust the monitoring time according to the energy budget and feedback density, thereby taking into account the goals of clock accuracy maintenance and energy consumption minimization.

[0082] In this embodiment, S2 specifically includes:

[0083] S21: After entering the uplink receive timeslot, the node starts the RF receiving circuit, performs continuous detection within the set channel energy detection threshold, and completes synchronization lock after detecting the preamble of the data packet sent by the previous hop node;

[0084] S22. When the node receives the starting boundary of the synchronization field of the data packet, it records the current count value of the local clock as the receiving timestamp ;

[0085] S23. Parse 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 sending operation and the end of this time slot calculated by the previous hop node based on the local clock;

[0086] S24. Calculate the estimated value of the end of the time slot :

[0087] ;

[0088] The timeslot end estimated value represents the end time of the current uplink receiving timeslot in the local clock domain estimated by the node;

[0089] S25, the node sets the network identifier of the previous hop node , current frame number , receiving timestamp , remaining time slot duration and the time slot end estimate Composing a structured record :

[0090] ;

[0091] 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.

[0092] During uplink reception, nodes incorporate energy-aware triggering of the preamble and precise recording of the reception timestamp, enabling a highly robust reception boundary capture mechanism. By extracting the remaining time slot duration embedded by the preceding node from the received data packet and combining it with the local reception time to calculate the estimated end of the current time slot, a precise inference of the inter-hop time slot boundary is formed. This estimated value is written into the local pseudo-transition time mapping table as a structured five-tuple and dynamically maintained using a historical sliding window mechanism. This allows nodes to continuously obtain high-quality raw data sets of time drift, supporting subsequent trend modeling and improving the stability of drift estimation and its ability to withstand sudden errors.

[0093] In this embodiment, S3 specifically includes:

[0094] 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. The local drift error sequence is used to represent the trend of the clock offset of the local node relative to the previous hop node over time.

[0095] 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 connecting edges between adjacent state nodes to form a non-Euclidean search space for path search;

[0096] S33, introduce the bat resonance fitting algorithm, the node initializes the bat fitting set, the bat fitting set includes Each simulated bat individual carries a set of fitting function parameters to construct a potential drift trend path and guide the path direction through the acoustic resonance structure function:

[0097] ;

[0098] in, Indicates the The acoustic resonance structure function corresponding to the simulated bat individual, represents the amplitude factor, Indicates the search frequency, represents the initial phase, represents the exponential decay coefficient;

[0099] S34. The node constructs a frequency hopping self-tuning modulation kernel function for each simulated bat individual:

[0100] ;

[0101] in, Indicates the The frequency hopping self-tuning modulation kernel function corresponding to the simulated bat individual, represents the gradient of the acoustic resonance structure function, Indicates the modulation sensitivity factor, which is used to dynamically adjust the step size and frequency search range;

[0102] S35, the node in each round of fitting iteration is based on the acoustic resonance structure function Frequency Hopping Modulation Kernel Function Together they form a search guidance mechanism until the drift error value fitting residual meets the preset convergence threshold or reaches the maximum number of iterations;

[0103] S36, nodes are selected from all simulated bat individuals The simulated bat individuals with the smallest fitting residuals are , build a local clock drift trend model of the node relative to the previous hop node :

[0104] ;

[0105] S37. The node stores the local clock drift trend model in a fitting buffer.

[0106] The bat resonance fitting algorithm is used to model the local drift error sequence, overcoming the problem that traditional linear fitting methods are unable to express non-stationary offset trends. By introducing the acoustic resonance structure function and the frequency hopping self-tuning modulation kernel function to construct a multi-dimensional search space, the path search behavior of multiple bat individuals is simulated, making the model sensitive and adaptable to dynamic drift trends. The algorithm can self-adjust the step size and frequency range during the search process to avoid falling into local optimality, and select the optimal individual to construct the final trend model when the fitting residual reaches the convergence condition. The 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 capability of multi-hop synchronization systems.

[0107] In this embodiment, the S4 specifically includes:

[0108] S41. The node calls a local clock drift trend model from a fitting buffer, where the local clock drift trend model records the offset change trend of the node's local clock relative to the previous node's clock over multiple frame periods.

[0109] S42: The node obtains the current frame period number and uses the next frame period number as a prediction input. The node uses the local clock drift trend model to calculate the local clock drift deviation expected to occur in the next frame period. The node then adds the local clock drift deviation to the start time of the standard uplink receive timeslot to obtain the expected arrival time of the event sent by the preceding node in the local clock.

[0110] 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 synchronization event of the previous hop node;

[0111] S44. The node adjusts the local clock according to the calculated time deviation so that the local uplink receiving window is opened in advance in the next frame period, ensuring that the local operation is aligned with the synchronization event of the previous hop node.

[0112] Nodes use local clock drift trend models to accurately predict the expected arrival time of the next received event, effectively adjusting the opening time of the receive window in advance, thereby ensuring that local operations are aligned at the time slot level with the transmission behavior of the previous hop node. Compared with a single synchronization reference method, it can continuously track the offset trend and perform dynamic compensation, significantly improving synchronization consistency and reducing the probability of receiving jitter and misaligned 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, nodes can complete adaptive synchronization without external control. This is particularly suitable for multi-hop network environments without master nodes and independently operating clocks.

[0113] During local processing slots, nodes perform multi-level processing from the physical layer to the network layer to ensure the reliability and forwarding efficiency of received data. During physical layer processing, nodes first perform error correction decoding on received packets, using lightweight low-density parity-check (LDPC) codes to minimize computational complexity while maintaining error correction capabilities, adapting to resource-constrained device environments. Furthermore, nodes use pilot symbols in the packets to perform channel estimation, obtaining information about the current link quality and providing a basis for transmit power control in downlink transmission slots.

[0114] After completing physical layer processing, the node enters the network layer processing phase. It first parses the destination address field of the data packet and performs time slot mapping using a preset hash function. For example, a modulo operation can be performed on the lower two bits of the destination address to map the data to a designated downlink transmit time slot, enabling forwarding scheduling without central control. Subsequently, the node performs queue management based on data type and service priority, prioritizing urgent data (such as alarm information) to a high-priority transmit queue to ensure the real-time performance of critical services. Routine services are queued to a low-priority queue on a first-in, first-out basis, achieving a balance between latency control and resource scheduling. The design of local processing time slots ensures rapid decoding and proper queuing of data upon receipt, supporting efficient forwarding by the next-hop node.

[0115] In this embodiment, S6 specifically includes:

[0116] S61. The node assembles a data packet in a downlink transmission time slot and encapsulates the service data into a data payload area.

[0117] S62. The node extracts currently effective local clock drift trend model parameters from the fitting buffer, where the local clock drift trend model parameters include acoustic resonance structure function parameters and frequency hopping self-tuning modulation kernel function parameters.

[0118] S63. The node packages the local clock drift trend model parameters into a synchronization information field, embeds the field into the additional control field of the data packet header, and sends the field to the next hop node together with the service data.

[0119] 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 basis for the previous-hop node to predict the end time of the uplink transmission time slot, and combines the received timestamp to reversely infer the expected transmission time of the previous-hop node under the local clock, calculates the corresponding local time slot end estimate, and jointly constructs a new record in the local pseudo-transition time mapping table with the received timestamp.

[0120] During downlink transmission slots, nodes embed the current local clock drift trend model parameters into the control field of the data packet. This allows synchronization information to propagate naturally to the next-hop node along with the data transmission, achieving chained inheritance of synchronization parameters. This approach requires no dedicated synchronization packets or signaling payload, offering the advantage of zero additional communication overhead. The next-hop node then infers the reception time and the model parameters it carries, constructing its own pseudo-transition time mapping table, thereby forming a continuous and transferable trend modeling chain. This structured propagation and update mechanism for the synchronization model improves the consistency and efficiency of drift modeling across the entire network, laying the foundation for prediction and correction.

[0121] In this embodiment, the S7 specifically includes:

[0122] S71. The node switches to a low-power listening mode in the protection sleep time slot, leaving only the RF receiving path and the synchronization control module in working state to receive synchronization feedback data from the next-hop node;

[0123] S72. The node parses the feedback field in the received data packet during the monitoring process and extracts the local clock drift trend model parameters returned by the next hop node;

[0124] 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 a field level to obtain differences.

[0125] S74. The node determines whether the difference exceeds an acceptable range based on a preset tolerance threshold. If so, a refitting mechanism is triggered to update the local clock drift trend model from new historical records in the local pseudo-transition time mapping table.

[0126] During protection sleep time slots, nodes enter a low-power listening state and receive drift trend model parameter feedback from downstream without activating the main processing module, achieving closed-loop verification and accuracy maintenance of the synchronization model. Nodes compare field-level differences to determine whether the current model deviates from the actual system trend. If the tolerance is exceeded, a refit process is triggered to ensure that the model always reflects the latest drift characteristics. This feedback path forms a calibration mechanism for drift modeling, providing robustness against long-term error accumulation and the ability to correct for drift errors. Furthermore, its low-power listening strategy ensures the system maintains a high level of energy efficiency.

[0127] In this embodiment, the S8 specifically includes:

[0128] S81. At each set frame period boundary, the node reads the local clock drift trend model stored in the current fitting buffer;

[0129] S82. Injecting a Gaussian perturbation into the local clock drift trend model parameters through local control logic, where the Gaussian perturbation is a pseudo-random perturbation with zero mean and set variance;

[0130] S83: The node uses the local clock drift trend model injected with the Gaussian disturbance as the input for a new round of prediction calculation, and rewrites the updated local clock drift trend model into the fitting buffer.

[0131] By injecting Gaussian perturbations into the local drift model at set frame intervals, the model maintains its ability to evolve even without external data input, preventing the accumulation of synchronization prediction errors due to model overfitting or rigidity. The perturbed model performs a self-prediction and updates the fitting parameters, forming a self-supervised evolutionary correction path. This approach is suitable for maintaining node synchronization strategies during discontinuous communication or link interruptions, significantly enhancing model robustness and self-healing capabilities, and supporting network synchronization stability even under abnormal conditions.

[0132] Example 1:

[0133] To verify the feasibility of the present invention in practice, the present invention was applied to a typical low-power multi-hop wireless communication system to simulate continuous industrial Internet of Things scenarios, including tunnel structural health monitoring, petrochemical pipeline leak detection, or forest boundary environmental monitoring tasks. In this system, multiple sensor nodes report information at fixed time intervals, and the network topology presents a linear chain structure. The network supports data transmission links of up to 20 hops. Each node adopts a four-slot frame structure based on TDMA, with each frame period fixed at 400ms, which is divided into uplink reception, local processing, downlink transmission, and protection sleep time slots, with the ratio range set to: 25%, 30%, 20%, and 25%.

[0134] In terms of synchronization, each node in this system constructs a local pseudo-transition time mapping table and uses a bat resonance fitting algorithm to generate a local clock drift trend model to achieve prediction and alignment of uplink synchronization events. The node appends the model parameters to the downlink data packet, allowing the next-hop node to construct its own mapping table, thus ensuring link transmission of the synchronization trend. Simultaneously, during protection sleep time slots, the node monitors the drift model parameter feedback from the successor node in low-power mode. Based on this information, the node determines whether the model deviation exceeds the tolerance, triggering a refitting mechanism. Every 10 frames, the node actively injects Gaussian perturbations into the fitted model, enabling the model to evolve and maintaining the timeliness and robustness of drift prediction.

[0135] 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 was constructed to test indicators such as synchronization accuracy, control overhead, maximum supported hops, node unit energy consumption, and response time to topology changes. The results are summarized in the following table.

[0136] Table 1 Comparison of the core performance indicators of the present invention and existing typical protocols

[0137]

[0138] The test results in Table 1 above show that the synchronization accuracy achieved by the present invention is consistently maintained within ±50 microseconds, far exceeding LoRaWAN Class B (±1ms) and approaching the TSCH standard (±10μs), meeting the time-slot-level synchronization accuracy requirements of industrial-grade scenarios. In terms of energy efficiency, due to the use of predictive trend modeling and an extremely low-duty-cycle RF activation mechanism, the unit energy consumption of nodes is significantly lower than that of other protocols, at only 0.25 mJ / node, approximately 41.7% of IEEE802.15.4 TSCH and approximately 16.7% of LoRaWAN Class B. Regarding topology change responsiveness, thanks to the link transfer of model parameters and a rapid feedback mechanism, the present invention can complete synchronization updates under the new topology within a single frame period, significantly surpassing the minute-level resynchronization overhead of existing solutions.

[0139] At the same time, under multi-hop conditions, the present invention successfully supports 20-hop stable communication without increasing control signaling, while TSCH synchronization errors begin to accumulate significantly at 10 hops, and LoRaWAN does not have effective synchronization guarantee after 2 hops due to its non-time slot scheduling structure.

[0140] This embodiment fully demonstrates the technical advantages of the present invention in terms of time slot-level synchronization accuracy, hop count support, energy consumption control, and topology adaptability by applying it to a typical multi-hop wireless communication system. Compared to existing synchronization mechanisms, this solution can stably control inter-node synchronization accuracy to within ±50 microseconds without the need for central coordination and additional signaling overhead, far exceeding the millisecond-level synchronization accuracy provided by traditional protocols. Furthermore, the system can maintain an end-to-end latency of less than 50 milliseconds over a 20-hop link, demonstrating the present invention's excellent latency controllability and scalability. Node unit energy consumption is reduced to 0.25 millijoules, significantly lower than mainstream solutions, ensuring energy sustainability during long-term operation. Furthermore, when the topology changes, synchronization can be reestablished in just one frame period, demonstrating strong dynamic response and robustness. This fully demonstrates the present invention's broad adaptability and engineering practical value in complex industrial, environmental monitoring, and low-power ad hoc networking scenarios.

[0141] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection 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, where each frame includes an uplink receive time slot, a local processing time slot, a downlink transmit time slot, and a protection sleep time slot. S2. In the uplink receive 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 based on the local receive 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 hop node based on a bat resonance fitting algorithm. S4. The node predicts the expected arrival time of the next uplink synchronization based on 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 by 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 perturbations into the local clock within the set frame period and automatically updates the local clock drift trend model at regular intervals. 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. The local drift error sequence is used to represent the trend of the clock offset of the local 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 connecting edges 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, the bat fitting set includes Each simulated bat individual carries a set of fitting function parameters to construct a potential drift trend path and guide the path direction through the acoustic resonance structure function: ； in, Indicates the The acoustic resonance structure function corresponding to the simulated bat individual, represents the amplitude factor, Indicates 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 The frequency hopping self-tuning modulation kernel function corresponding to the simulated bat individual, represents the gradient of the acoustic resonance structure function, Indicates the modulation sensitivity factor, which is used to dynamically adjust the step size and frequency search range; S35, the node in each round of fitting iteration is based on the acoustic resonance structure function Frequency Hopping Modulation Kernel Function Together they form a search guidance mechanism until the drift error value fitting residual meets the preset convergence threshold or reaches the maximum number of iterations; S36, nodes are selected from all simulated bat individuals The simulated bat individuals with the smallest fitting residuals are , build 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 a fitting buffer.

2. The multi-hop pipeline propagation method based on TDMA time slot level synchronization according to claim 1, characterized in that: In the TDMA frame format of the four-slot structure, the first time slot is the uplink reception time slot, and the duration is set to 20% to 25% of the frame period; the second time slot is the local processing time slot, and the duration is set to 30% to 35% of the frame period; the third time slot is the downlink transmission time slot, and the duration is set to 20% to 25% of the frame period; the fourth time slot is the 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 receive timeslot, the node starts the RF receiving circuit, performs continuous detection within the set channel energy detection threshold, and completes synchronization lock after detecting the preamble of the data packet sent by the previous hop node; S22. When the node receives the starting boundary of the synchronization field of the data packet, it records the current count value of the local clock as the receiving timestamp ; S23. Parse 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 sending operation and the end of this time slot calculated by the previous hop node based on the local clock; S24. Calculate the estimated value of the end of the time slot : ； The timeslot end estimated value represents the end time of the current uplink receiving timeslot in the local clock domain estimated by the node; S25, the node sets the network identifier of the previous hop node , current frame number , receiving timestamp , remaining time slot duration and the time slot end estimate Composing 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, characterized in that: The S4 specifically includes: S41. The node calls a local clock drift trend model from a fitting buffer, where the local clock drift trend model records the offset change trend of the node's local clock relative to the previous node's clock over multiple frame periods. S42: The node obtains the current frame period number and uses the next frame period number as a prediction input. The node uses the local clock drift trend model to calculate the local clock drift deviation expected to occur in the next frame period. The node then adds the local clock drift deviation to the start time of the standard uplink receive timeslot to obtain the expected arrival time of the event sent by the preceding node 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 synchronization event of the previous hop node; S44. The node adjusts the local clock according to the calculated time deviation so that the local uplink receiving window is opened in advance in the next frame period, ensuring that the local operation is aligned with the synchronization event of the previous hop node.

5. The multi-hop pipeline propagation method based on TDMA time slot level synchronization according to claim 1, characterized in that: The S6 specifically includes: S61. The node assembles a data packet in a downlink transmission time slot and encapsulates the service data into a data payload area. S62. The node extracts currently effective local clock drift trend model parameters from the fitting buffer, where 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 packages the local clock drift trend model parameters into a synchronization information field, embeds the field into the additional control field of the data packet header, and sends the field 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 basis for the previous-hop node to predict the end time of the uplink transmission time slot, and combines the received timestamp to reversely infer the expected transmission time of the previous-hop node under the local clock, calculates the corresponding local time slot end estimate, and jointly constructs a new record in the local pseudo-transition time mapping table with the received timestamp.

6. The multi-hop pipeline propagation method based on TDMA time slot level synchronization according to claim 1, characterized in that: The S7 specifically includes: S71. The node switches to a low-power listening mode in the protection sleep time slot, leaving only the RF receiving path and the synchronization control module in working state to receive synchronization feedback data from the next-hop node; S72. The node parses the feedback field in the received data packet during the monitoring 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 a field level to obtain differences. S74. The node determines whether the difference exceeds an acceptable range based on a preset tolerance threshold. If so, a refitting mechanism is triggered to update the local clock drift trend model from new historical records in the local pseudo-transition time mapping table.

7. The multi-hop pipeline propagation method based on TDMA time slot level synchronization according to claim 1, characterized in that: The S8 specifically includes: S81. At each set frame period boundary, the node reads the local clock drift trend model stored in the current fitting buffer; S82. Injecting a Gaussian perturbation into the local clock drift trend model parameters through local control logic, where the Gaussian perturbation is a pseudo-random perturbation with zero mean and set variance; S83: The node uses the local clock drift trend model injected with the Gaussian disturbance as the input for a new round of prediction calculation, and rewrites the updated local clock drift trend model into the fitting buffer.

Images