Embedded Internet of Things communication bus application deployment method based on swan gap system

Abstract

The invention discloses an embedded Internet of Things communication bus application deployment method based on a swan-gap system, and relates to the field of Internet of Things and embedded technologies, and the method comprises the following steps: S1, collecting a sleep wake-up trigger sequence of embedded equipment under a unified time baseline, constructing an instantaneous topology difference map, and carrying out the analysis of the instantaneous topology difference map; determining the amplitude, the duration and the influence range of the topology sudden change; s2, constructing a causal residual matrix based on the topological difference atlas, training a loop precursor recognition model for positioning an initial boundary of a topological closed loop, and extracting time delay features of key nodes; according to the method, a topological difference map is constructed based on a unified time base line, a closed loop precursor is identified, a shadow routing table is generated, path feasibility is corrected, stable forwarding of data is ensured by adopting segmented injection and amplitude limiting migration, path weight is dynamically adjusted in combination with an error frequency and mirror image traction mechanism, closed loop formation is effectively inhibited, and the reliability of the system is improved. And the stability and the safety of the communication bus in a dynamic environment are improved.

Description

Technical Field

[0001] This invention relates to the fields of Internet of Things (IoT) and embedded technology, and specifically to a method for deploying embedded IoT communication bus applications based on the HarmonyOS system. Background Technology

[0002] The deployment of embedded IoT communication bus applications based on the HarmonyOS system refers to using the HarmonyOS operating system as the underlying runtime environment in embedded devices and the IoT communication bus as the core data interaction channel. Through driver adaptation, protocol parsing, and task scheduling, various applications distributed across different sensors, actuators, and edge nodes are efficiently integrated into a unified system framework, enabling stable data transmission, message publishing, and command issuance between devices. This process not only solves the problems of application fragmentation and inconsistent interfaces in traditional embedded IoT systems, but also leverages the distributed characteristics and lightweight kernel design of the HarmonyOS system to achieve rapid application deployment, dynamic updates, and cross-device collaboration, thereby improving the real-time performance, scalability, and system security of IoT communication.

[0003] The existing technology has the following shortcomings: In existing technologies, during the deployment of embedded IoT communication buses, when some nodes enter a sleep state due to power management strategies and are then awakened, the topology of the communication bus can change drastically within a very short time. However, existing routing table update mechanisms generally rely on periodic scanning or delayed reconstruction. When the topology change exceeds its response time, the routing table is prone to failure to complete real-time reconstruction. In such cases, some data packets will continue to be forwarded in incorrectly identified loop paths, causing prolonged blockage of the global communication link and even triggering systemic deadlock, directly threatening the reliability and security of IoT applications in critical scenarios.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method for deploying embedded IoT communication bus applications based on the HarmonyOS system, so as to solve the problems in the background technology mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for deploying an embedded IoT communication bus application based on the HarmonyOS system, comprising the following steps: S1. Acquire the sleep-wake trigger sequence of the embedded device under a unified time baseline, construct an instantaneous topological difference map, and determine the amplitude, duration and scope of the topological mutation based on it; S2, based on the topological differential map, constructs a causal residual matrix, trains a loop precursor identification model, is used to locate the initial boundary of the topological closed loop, and extracts the time delay features of key nodes; S3. Construct a topology-consistent routing shadow table based on the initial boundary, and use the time-inversion path correction method to generate a zero-loop handover path sequence and a path protection threshold set. S4 dynamically injects the shadow table into the online forwarding table according to the segmented time stamp, and forwards the data packets in micro-batch mode through pulse-level amplitude limiting write-back, ensuring the continuity and loop-free nature of the path switching process; After injection, S5 combines the golden ratio frequency shifting traction and dual time-scale mirroring traction to periodically adjust the path weights using a breathing-like oscillation, thereby dynamically suppressing the path loop and maintaining the stable operation of the communication bus.

[0007] Preferably, step S1 includes: Under a unified time baseline, the sleep-wake trigger sequence of embedded IoT devices is continuously collected, and a complete state evolution sequence is constructed by combining the physical connection status of the nodes, signal strength level and link delay status. Based on time alignment, the communication relationships between nodes within consecutive time windows before and after wake-up are extracted, and superimposed to generate a topological differential graph. The connection change rate, communication delay change magnitude, and link stability offset gradient are introduced as edge weight attributes of the differential graph. Identify regions of edge weight variation in the difference graph, statistically analyze the perturbation expansion trend, delineate propagation boundaries, and classify events. The hierarchical results are mapped to the temporal coordinates of the nodes, and a multidimensional correlation matrix with the wake-up timestamp as the main axis is constructed to form a topological evolution data structure for path prediction and anomaly identification.

[0008] Preferably, in the topological differential map, the expansion rate of the disturbance propagation boundary is combined with the topological depth of the central node to determine the impact level of the mutation event, and the topological disturbance trajectory of the node is extracted based on the impact level to construct a topological behavior matrix with retrospective and predictive capabilities, which serves as the basis for judging the path evolution trend.

[0009] Preferably, step S2 includes: After completing the construction of the topological differential map, the structural state and performance parameters of the communication path in continuous time segments are extracted based on the unified time baseline, a set of topological evolution trajectories is constructed, and a causal residual matrix is ​​constructed. Identify concentrated perturbation paths in the causal residual matrix, and determine the closed-loop evolution region by combining path location and topological boundary features; A loop precursor recognition model is trained based on evolutionary trajectory, and a path-by-path scan is performed on the topological map to be detected to identify the closed loop germination boundary. Extract the time delay evolution trajectory of nodes associated with the closed-loop path and construct a time delay feature set of key communication nodes.

[0010] Preferably, when constructing the causal residual matrix, the path state before wake-up is used as a benchmark, and the path state offset is formed by comparing the path state after wake-up. The number of times the path state offset exceeds a preset threshold is used as a disturbance concentration index to screen high-risk communication paths within the closed-loop evolution area.

[0011] Preferably, S3 includes: After completing the initial boundary identification of the closed loop and the extraction of time delay features of key communication nodes, the communication paths in the closed loop boundary are selected as candidate targets for path replacement, and a dynamic candidate path set is constructed by combining the edge weight change trend and node load. The candidate path set is corrected by time inversion, the path stability is judged based on the historical topological evolution trajectory, high-risk paths are eliminated and delay offset compensation is performed on the retained paths to generate a corrected path structure with topological consistency. Based on the path score, a switching path sequence is generated, and a set of protection thresholds including the hop count variation range and data load limits are set to achieve smooth path migration and closed-loop avoidance.

[0012] Preferably, time-reversal path correction includes retrospective analysis of the connection status, link delay fluctuations and node interaction strength of candidate paths within historical topology cycles, and screening out high-risk paths based on disturbance frequency, while retaining paths achieves time alignment through delay compensation.

[0013] Preferably, S4 includes: Based on the path protection threshold set, a time-segmented injection plan is set, and the path structure is decomposed into the starting jump point, relay node segment and the final convergence segment and then activated sequentially. Before each time period injection, a path segment mapping conflict retrieval is performed. If a conflict exists, the injection is delayed, and the path availability is confirmed by the target node response delay. During the path segmentation injection process, a pulse time window is set to limit the number of micro-batch data packets forwarded and attach a path activation identifier; By analyzing message timestamps and priorities, micro-batch cross-segment migrations are implemented between path switching cycles, and data granularity is adjusted to ensure forwarding continuity. After completing path injection, monitor the path's running status, determine its closure capability, release old path entries, and complete path switching convergence.

[0014] Preferably, when setting the pulse time window, the pulse period is dynamically adjusted according to the network stability during the path switching phase. When the path is detected to be in a highly unstable state, the pulse interval is shortened to reduce the data packet injection rate, so as to avoid communication blockage or loop backflow caused by sudden load increase during the path switching process.

[0015] Preferably, S5 includes: Set asynchronous weight update cycles for all injected paths, and use a frequency shifting method that combines global control frequency and golden ratio constant to periodically shift the path weights. Two sets of mirror time-stamped sequences are established to synchronously constrain the current performance and historical risk status of the path, respectively, so as to achieve positive incentives and reverse constraints during the weight adjustment process; The path weights are periodically adjusted by a breathing-type oscillation curve, which increases, stabilizes, and decreases to maintain a slow and stable scheduling rhythm. The monitoring path changes in connection density and backflow probability fluctuations within the oscillation cycle. If the dynamic threshold is exceeded, the path is triggered to enter a suppression protection state. The overall network closed-loop risk index is assessed based on the dispersion of network weight distribution and the overlap of active paths, and the error frequency ratio and mirror baseline parameters are reset to complete the global weight scheduling rebalancing.

[0016] The beneficial effects provided by the present invention in the above technical solution are as follows: This invention constructs a time-series map of device wake-up behavior using a unified time baseline, accurately capturing the boundaries and fluctuation ranges of topology mutations. Then, based on differential topology, it establishes a causal residual matrix to effectively identify early signs of communication loop formation. Furthermore, by combining a time-reversal path correction strategy, it constructs a routing shadow table that satisfies topology consistency and generates switching paths and protection thresholds with risk awareness capabilities, fundamentally preventing loop paths from entering the forwarding table. Simultaneously, through segmented timescale injection and pulse-level amplitude-limiting write-back mechanisms, data packets are smoothly migrated to the new path in micro-batch form, achieving seamless connection and continuous transmission during path switching. Finally, through a coordinated control method of golden ratio frequency misalignment traction and dual-timescale mirror traction, it implements periodic breathing-style oscillation adjustments to the path weights, dynamically suppressing the continuous strengthening trend of potential closed-loop paths without interfering with the main path scheduling. In summary, this method not only solves the problems of delayed routing table updates and difficulty in real-time identification and prevention of communication loops in existing technologies, but also achieves adaptive control of topology disturbances in embedded IoT environments while maintaining system communication stability and real-time performance. It has structural innovation, precise control and deployment feasibility, and improves the reliability and security of IoT communication buses in dynamic scenarios. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart of a method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to the present invention. Detailed Implementation

[0019] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0020] This invention provides, for example Figure 1 The method for deploying an embedded IoT communication bus application based on the HarmonyOS system, as shown, includes the following steps: S1. Collect the sleep and wake-up trigger sequences of embedded IoT devices under a unified time baseline, construct an instantaneous topological differential map based on the state changes before and after wake-up, and determine the amplitude, duration and scope of the topological structural change based on the map. To address the issue of unrecognized topology shifts caused by node sleep and wake-up processes in embedded IoT communication buses, a method for acquiring sleep and wake-up behavior and performing topology differential analysis based on a unified time baseline is proposed. Through a high-precision time synchronization mechanism, multi-level node state sampling, and graph-based structure reconstruction, this method enables quantifiable modeling and impact assessment of dynamic changes in the network structure. The specific steps are as follows: By embedding a high-precision time base in the embedded IoT device operating environment, using the system startup time as the reference zero point, a unified time tick generator is embedded within all devices. This generator achieves nanosecond-level synchronization accuracy based on the kernel timing resources of the HarmonyOS operating system, thus ensuring that all devices have alignable timing coordinates during communication. Within this unified time framework, the sleep entry time and wake-up trigger time of each embedded device are periodically collected, recording its activation boundary changes in the network. This collection process includes not only the timestamps of the device's power-on and power-off states, but also the micro-evolution sequence of its physical connection status with other neighboring nodes, signal strength levels, and link delay states. Unlike existing technologies that only obtain limited state changes through event triggering, this approach combines active probing with continuous sampling, ensuring the continuity and traceability of details of each device state change, providing a solid dynamic foundation for subsequent topology reconstruction.

[0021] Based on time alignment and state sequence acquisition, the communication relationship states between all nodes in the network are extracted in two consecutive time windows before and after each wake-up. A time-related table is constructed to capture changes in the connectivity edges between devices. Then, the two topological states before and after wake-up are differentially overlaid in a unified coordinate system to generate a time-sensitive topological differential graph. This graph uses nodes as vertices, communication relationships as edges, and edge state changes as weights. Color coding and weighted line segments represent the reconstruction trajectory of the network structure on the time axis. Notably, this topological differential graph introduces three-dimensional correlation information: node connection change rate, communication latency change magnitude, and link stability offset gradient. This not only provides information on topological changes but also reflects the dynamic drift of network performance, significantly different from the static view of traditional graph structures that only reflects connectivity.

[0022] After the topological differential graph is constructed, regions with edge weight changes are analyzed in a focused manner. The state drift amplitude of each changing edge is statistically analyzed per unit time, forming a set of edge perturbation sequences. By comparing the expansion trend of the perturbation boundary over time, the propagation boundary of the perturbation is extracted and its propagation rate is measured. Furthermore, by combining the topological depth of the perturbation region center point with the edge weight fluctuation frequency, mutation events are classified and graded. Based on this grading result, the impact range of each type of mutation event in the network is quantified, specifically including parameters such as the number of affected nodes, the maximum path length change ratio, and the local path replacement rate, thus forming an impact assessment report for a single sleep-wake event. This report has a clear spatial boundary definition and temporal extensibility characterization, providing structural input for subsequent path prediction and anomaly detection. However, existing technologies lack a dynamic structural description method similar to topological differential graphs, and usually can only rely on qualitative judgment or coarse path failure statistics, failing to achieve refined mutation source tracing and impact analysis.

[0023] By mapping the marginal parameters of mutation events and the distribution of topological perturbations to node time-series coordinates, a complete data structure reflecting the impact of sleep-wake behavior is formed. This structure uses the wake-up timestamp as the main axis, combined with its corresponding topological perturbation level and propagation speed, to construct a multi-dimensional correlation matrix. In this matrix, all nodes have a traceable historical topological perturbation trajectory and possess behavioral trend criteria that can be extrapolated forward. This matrix not only allows for locating the network mutation segment triggered by a specific wake-up behavior but also enables predictive modeling of potential topological changes caused by similar future behaviors. Finally, this series of processes is summarized into a set of standard topological evolution description data that can be directly referenced by subsequent path analysis mechanisms, providing structured input conditions for causal residual modeling and precursor identification.

[0024] S2, a causal residual matrix is ​​constructed based on the instantaneous topological differential map, and a loop precursor identification model is trained based on this matrix. The initial boundary of the topological closed loop is accurately identified through the identification model, and the time delay features of the key communication nodes involved are extracted. To identify potential communication loop risks in advance when network topology changes abruptly, and to quantitatively assess the possible location and boundaries of loops, a method based on constructing a causal residual matrix using topological difference maps is proposed. Based on this, a precursor identification model is established to extract the delay variation characteristics of key communication paths, which is then used to construct a precise loop early warning mechanism. The specific steps are as follows: After constructing the instantaneous topological differential map, the structural state, connection weights, and transmission performance parameters of the same communication path in multiple consecutive time segments are selected, and these temporal evolution sequences are aligned according to a unified time scale to form a set of topological evolution trajectories. This set uses the communication path as the basic analysis unit, tracking the behavior of each path before and after wake-up, and during the topological stability and perturbation phases. To ensure the continuity and diversity of the analysis, all communication edges within the perturbation region of the topological structure are selected as high-priority sampling targets, and their connection status (whether connected), bandwidth utilization, latency changes, and the number of connections of adjacent nodes are organized in vector form to construct a temporal structural state comparison set with the path as the main line. After this set is formed, based on the pre-wake-up state as the prediction baseline state, the post-wake-up path state is subtracted from it to construct a causal residual matrix. This matrix uses the time slice as the horizontal axis and the communication path as the vertical axis, with each cell value representing the state offset of the path at a specific time point. Unlike existing technologies that only collect facts about topological changes, this method establishes a causal relationship between the behavior of previous and subsequent paths by using a unified time scale, so that every communication disturbance has a mathematical basis in which the source can be traced and the change can be quantified.

[0025] After constructing the causal residual matrix, cluster analysis is performed on all non-zero units in the matrix to identify path sets exhibiting highly anomalous characteristics during the topological mutation phase. A perturbation concentration index is introduced into the analysis, which measures the number of times a path state offset exceeds a set threshold within a unit time interval. This index is used to screen communication paths exhibiting rapid mutation trends during topological perturbations. For these paths, their positions in the topological difference map are further observed, and combined with the topological depth and boundary characteristics of the nodes they connect, the potential regions for closed-loop evolution are preliminarily identified. This process emphasizes the temporal synchronicity and spatial overlap of path perturbations to ensure that the extracted path clusters possess high local convergence and a potential tendency for loop propagation. Unlike traditional topological scanning methods that cannot proactively identify closed-loop risks, this region-focusing method based on the causal residual matrix can detect early signs of topological loop evolution before the loop is fully formed.

[0026] After identifying the suspected closed-loop propagation areas, all communication paths within these areas are selected as training data samples. A loop precursor identification model is established through supervised learning. During training, the path state evolution trajectory is used as input features, and whether a path ultimately participates in the formation of a closed-loop structure is used as the output label. A time-sensitive classification mechanism is employed to enhance the model's ability to capture early disturbances. After training, the model is used to perform path-by-path scanning analysis of the topology to be detected, determining whether each path belongs to the closed-loop germination boundary and marking potential inward or outward expansion paths. Compared with traditional simple loop detection algorithms that rely on graph structures, this method introduces historical evolution dimensions and disturbance trend judgments, evolving from a static judgment of "whether a loop exists" to a dynamic early warning judgment of "whether a loop is forming," thus possessing higher practicality and proactive response.

[0027] After identifying the communication paths constituting the closed-loop boundary, the focus is on the key communication nodes connected to these paths, extracting the time delay evolution trajectories of these nodes during topology changes. This trajectory reflects the changing trend of a node's signal forwarding response time after a topology change, including the average delay to enter the response state, the time window required for path stabilization, and the number of reconnection failures. By standardizing these delay features on a unified time baseline, a set of time delay features for key nodes is formed. This set can not only characterize the stability level of a single node during topology disturbances but also be used in subsequent path planning to determine whether a node is suitable to act as a communication relay.

[0028] S3. Based on the identified initial boundary of the topology closed loop, construct a routing shadow table that satisfies topology consistency. Perform feasibility calculation on the candidate paths in the shadow table using the time inversion path correction method to generate a switching path sequence and path protection threshold set that do not contain communication loops. To ensure that new communication loops are not triggered during communication path switching after a sudden change in topology, and to guarantee a smooth transition of path replacement before the topology stabilizes, a method is proposed to construct a routing shadow table based on the initial boundary of the closed loop. Path feasibility is then corrected through time inversion, thereby generating a path replacement sequence and a set of protection thresholds with topological consistency and loop avoidance capabilities. The specific steps are as follows: Based on the identification of the initial closed-loop boundary and the extraction of latency features of key communication nodes, all communication paths within the closed-loop evolution boundary are selected as candidate paths for replacement. Combining the edge weight change trend in the causal residual matrix, the path's persistent perturbation characteristics are further assessed, and priorities are assigned accordingly, with paths exhibiting higher perturbation frequencies and more core topological positions prioritized for inclusion in the path reconstruction list. Furthermore, for each path to be replaced, alternative path detection requests are initiated to its upstream and downstream communication nodes. These requests are responded to under a unified time baseline, providing information on currently available alternative paths, including hop count, average transmission delay, historical stability coefficient, and cross-domain transmission capability. By integrating the topological attributes and performance parameters of multiple candidate paths, several sets of candidate path structures are preliminarily selected. Unlike existing technologies that use static path replacement, this implementation combines topological perturbation trends and key node load evolution to construct a dynamic set of candidate paths that is highly adaptable to the current network state, laying the foundation for subsequent path feasibility correction.

[0029] For the candidate path set, a time-reversal path correction method is used for feasibility calculation. Time-reversal path correction refers to, while keeping the current topology snapshot unchanged, reversing the actual trajectory of network topology evolution over a past period to simulate potential local topology reconstruction events after path switching. Specifically, for each candidate path, its connection status, link delay fluctuations, and node interaction strength in the previous period are retrospectively reconstructed to determine its probability of re-entering the topology disturbance center in the future. If a path has a significant fluctuation trend in the past or frequent intersections with closed-loop boundaries, it is identified as a potentially high-risk path and eliminated. Based on this, low-risk paths that pass the retrospective verification are retained, and delay offset compensation is applied to these paths using the delay fingerprints of key nodes, generating a time-aligned path structure with transmission continuity and topological stability. This method significantly differs from traditional path availability judgments based on the current network state; it combines topology evolution trends with path behavior prediction, achieving temporal consistency and preventative measures for path replacement strategies.

[0030] After obtaining the time-reversed path set, a set of switching path sequences without communication loop structures is generated based on their feasibility scores and path replacement levels, and a set of path protection thresholds is constructed. The switching path sequence is ranked according to path stability and load balancing of key nodes, guiding the original communication links to migrate to new paths in stages. During this process, to prevent sudden load increases or communication delays caused by path mutations, a dynamic protection threshold is set on each candidate path. This threshold includes three sub-parameters: maximum allowable data load, maximum hop count variation, and minimum remaining available bandwidth, serving as a control threshold for switching execution. If any indicator exceeds the preset protection threshold during path switching execution, the activation process of that path is immediately suspended, and the next level path in the alternative path sequence is re-evaluated. Simultaneously, this switching sequence also has adaptive adjustment capabilities, automatically adjusting priority weights and switching to the next optimal path as the overall system topology evolves further. This approach fundamentally avoids secondary loops or communication interruptions caused by delayed path replacement responses or uncontrollable switching processes in existing technologies, greatly improving the stability and responsiveness of the embedded IoT communication bus in dynamic environments.

[0031] S4 dynamically injects the routing shadow table that meets the consistency condition into the current online forwarding table according to the preset segmented time scale, and uses a pulse-level amplitude-limited write-back strategy to enable data packets to be forwarded in the new path through micro-batch cross-segment migration, so as to realize continuous transmission and loop-free flow of communication path during the switching process. After completing path feasibility correction and shadow routing table generation, to ensure a smooth transition of communication links to new paths during topology changes and to avoid communication interruptions or loop backflows caused by uneven path injection or switching synchronization misalignment, a dynamic migration mechanism combining segmented time-stamp injection and pulse-level amplitude-limited write-back is proposed. This mechanism enables batch-based transition forwarding of data packets and continuous path switching. The specific steps are as follows: Based on the estimated carrying capacity and topology sensitivity of each candidate path recorded in the path protection threshold set, a corresponding time-segmented injection plan is set. This injection plan uses a unified system time baseline as a reference, dividing the path switching time into multiple incremental time windows. Each window only allows the injection of a portion of the path structure, specifically including the path's starting hop, relay node segments, and the connection relationships, delay scalars, and forwarding rules of the endpoint convergence segment. By decomposing the entire path into time-series segments and activating them sequentially, the network synchronization pressure and abrupt state transitions caused by a one-time path injection can be effectively avoided. This segmented injection method differs from the static update strategy of replacing the entire path in traditional schemes. It achieves gradual evolutionary control of route activation through time segmentation and structural decoupling, ensuring that the shadow path gradually "integrates" into the entire topology rather than being "embedded" instantaneously.

[0032] Before each injection period begins, a mapping conflict retrieval is performed on the path segment to be injected based on the current usage status and remaining entry space of the online forwarding table. This ensures that the new path does not overwrite critical communication paths in use or cause existing path entries to become invalid. If a conflict area is found, the injection of that segment is suspended and postponed to the next time window, waiting for the original path to complete its forwarding cycle and automatically exit. This process not only ensures the decoupled execution of the old and new paths in the time dimension but also avoids routing oscillations caused by path overlap during path switching at critical nodes. Furthermore, to ensure that each injected segment can be accurately propagated to the target node, a node-level confirmation feedback mechanism is introduced during the path segment injection process. By comparing the delayed response time returned by the target node with a preset standard delay range, it is confirmed that the path has been correctly identified and is available before activating the next injection segment.

[0033] Building upon segmented injection into the routing shadow table, a pulse-level limiting write-back strategy is employed to prevent forwarding failures or duplicate transmissions during path switching due to incomplete activation of the target path. Specifically, a pulse time window is set for each hop path in the forwarding table, allowing only one micro-batch of data packets to enter the forwarding process within this window. A maximum allowed number of packets per batch and a delay drift range are also set. When the system detects that the path switching phase is in a highly unstable region (e.g., multiple paths activating simultaneously, severe disturbances at the network topology edge), it automatically compresses the pulse period, reducing the forwarding frequency to ensure data is injected into the network in a gradual manner, minimizing the instantaneous impact load during path switching. Simultaneously, a path activation identifier is appended to the end of each pulse batch to identify the new path the data packet should take, rather than the historical path, thus preventing loop phenomena where data repeatedly tumbles between old and new paths.

[0034] To improve forwarding efficiency and transmission continuity during path data migration, a micro-batch cross-segment migration mechanism is introduced. This mechanism analyzes the timestamps and priorities of the data to be forwarded, creating multiple fine-grained data transition windows between path switching cycles in the forwarding table. This allows high-priority packets to be preemptively transferred to activated paths, while low-priority packets gradually enter paths injected in subsequent time periods. Through dynamic granular adjustment of data batches, path switching achieves a smooth transition not only at the structural level but also at the content forwarding level. Unlike the full-forwarding strategy in existing technologies, this approach, through packet granular control and staggered scheduling of path activity windows, enables high-smoothness data migration during path replacement, avoiding communication jitter or congestion caused by repeated forwarding in loop areas.

[0035] After injecting all path segments and confirming that the connection status of all hops is consistent with the forwarding table update status, the path switching convergence process is initiated. This process continuously monitors the operational status of all injected path segments based on the maximum write-back period set according to the path protection threshold, and makes continuity judgments based on data throughput, packet retransmission counts, and forwarding delay fluctuations. Once it is determined that the path has full-segment closure capability and no abnormal return links, the existing forwarding table entries in the original path are released, and the old path structure is completely removed in the next cycle. During this process, if some path segments are still in an unstable state, the dual-path coexistence mechanism of the original path and the new path is maintained until a stable transition occurs. This method, through a strict five-step chain of injection—confirmation—limiting—migration—convergence, enables the path switching process to logically possess complete closed-loop control capabilities, effectively solving the key problems of path updates easily causing loop errors, switching link breaks, and data loss in existing technologies.

[0036] S5, after completing the injection of all shadow tables, uses a combination of golden ratio frequency shifting and dual time-scale mirroring to periodically adjust the path weights of the routing shadow tables in a breathing-like oscillation manner, dynamically suppressing the possible path loop phenomenon and maintaining the transmission stability of the communication bus under network topology changes. After all segments of the routing shadow table are injected and the forwarding table is fully activated, the network communication path has been dynamically replaced. However, due to the structural variability and rapid state drift of the embedded IoT communication bus in a high-density node topology environment, relying solely on static forwarding table weights for path selection will be insufficient to effectively adapt to connectivity disturbances and link reconstruction pressures caused by topological changes. Therefore, to maintain communication stability under dynamic changes and suppress the potential generation trend of path loops, a weight control strategy based on a combination of golden ratio frequency misalignment traction and dual-timescale mirror traction is proposed. This strategy periodically adjusts the shadow table path weights through a breathing oscillation method, constructing a dynamic balance mechanism with adaptive loop resistance. The specific steps are as follows: Based on the injected routing shadow table path set, the current transmission performance of all active paths is sampled, including path hop count, average node latency, packet success rate, and path boundary interference factor. To avoid centralized skew of path priorities during topology disturbances, which could lead to traffic concentrating on certain short-term high-efficiency paths and induce loop enhancement, a "golden ratio frequency shifting" strategy is introduced at this stage. Specifically, each path is assigned an independent update cycle, which is shifted by a golden ratio constant multiplied by a non-integer multiple of the global control frequency. This results in overlapping, asynchronous, and staggered arrival of weight updates for each path, forming a natural disturbance time distribution and preventing synchronous increases in path weights during adjustment that could lead to global network weight focusing. This method differs from existing techniques that adjust path weights in batches at fixed periods. It constructs a non-resonant adjustment order at the timing control level, significantly reducing the probability of loop formation.

[0037] After configuring the misalignment update cycle, two sets of mirrored timescale sequences are constructed based on the historical performance fluctuation data of all paths: one set is a time baseline describing the current transmission state of the path, and the other set is used to trace the performance degradation trend of the path in past abrupt change cycles. These two sets of mirrored timescales assign each path a set of positive adjustment weights and negative correction weights, ensuring that it can respond to the current performance enhancement trend while also being constrained by its historical risk characteristics within the same time period. For example, if a path currently has excellent performance but experienced data congestion or loop induction in the previous cycle, its positive weight growth will be restrained by the mirrored historical weights, keeping the weight growth rate within a reasonable threshold. This "timescale mirroring traction" strategy effectively balances the immediacy and security of path selection, preventing over-scheduling due to the "short-term excellence" of a path, significantly different from traditional path scoring methods based on moving averages.

[0038] Based on a dual mechanism of frequency misalignment control and mirror constraint, the system enters the dynamic adjustment phase of path weights. This phase employs a "breathing oscillation" mode, where within a complete weight adjustment cycle, the weight of each path is set to a continuously changing curve that first increases gradually, then stabilizes, and then decreases gradually, simulating the rhythm of physiological breathing and providing a degree of flexibility and periodic buffer in the path adjustment process. During the weight increase phase, adjustments are made gradually based on the actual transmission performance and historical reliability of the path; during the stabilization phase, the current weight is maintained to observe network convergence; and during the decrease phase, the priority of some high-weight but potentially risky paths is gradually reduced to free up scheduling space for other paths. This oscillation process creates a dynamic balance mechanism for path weights, ensuring that the network retains self-healing and load-distribution capabilities even during high-frequency fluctuations.

[0039] Within the oscillation cycle, boundary monitoring is performed on the weight adjustment curve of each path, and a corresponding dynamic loop early warning threshold is established. Specifically, by continuously tracking the changes in connection overlap density and data backflow probability caused by the weight increase phase of a path, if a path exhibits characteristics such as a decrease in hop count but an increase in backflow probability, and a bimodal shift in forwarding delay over multiple cycles, it is triggered to enter a "suppression protection state." In the next weight adjustment cycle, its maximum weight is forcibly reduced, and the duration of its decline phase is extended. This loop protection mechanism does not depend on whether a closed loop has actually been formed in the topology structure, but rather reduces the risk of its induction through dynamic weighting before the structure is formed. It is a proactive defense measure in the path-level loop prevention strategy.

[0040] After the path weight adjustment mechanism has been running stably for several cycles, the topological closed-loop risk index of the overall network is periodically assessed. This assessment is based on a comprehensive calculation of three types of data: the dispersion of path weight distribution, path activity, and loop potential score, forming a set of comprehensive indicators reflecting the dynamic stability of the network. If an excessive concentration of path weights is detected in the network or overlapping paths occur in a specific area, a global adjustment mechanism is activated to rebalance the path scheduling pattern by shuffling the weight update order, resetting the error frequency ratio, or switching the mirror reference baseline. This closed-loop mechanism works together from five dimensions: periodic patterns, historical constraints, dynamic disturbances, early warning response, and global control, constructing a technical guarantee framework for the embedded IoT communication bus to maintain loop-free paths and stable scheduling in extreme dynamic environments.

[0041] This invention constructs a time-series map of device wake-up behavior using a unified time baseline, accurately capturing the boundaries and fluctuation ranges of topology mutations. Then, based on differential topology, it establishes a causal residual matrix to effectively identify early signs of communication loop formation. Furthermore, by combining a time-reversal path correction strategy, it constructs a routing shadow table that satisfies topology consistency and generates switching paths and protection thresholds with risk awareness capabilities, fundamentally preventing loop paths from entering the forwarding table. Simultaneously, through segmented timescale injection and pulse-level amplitude-limiting write-back mechanisms, data packets are smoothly migrated to the new path in micro-batch form, achieving seamless connection and continuous transmission during path switching. Finally, through a coordinated control method of golden ratio frequency misalignment traction and dual-timescale mirror traction, it implements periodic breathing-style oscillation adjustments to the path weights, dynamically suppressing the continuous strengthening trend of potential closed-loop paths without interfering with the main path scheduling. In summary, this method not only solves the problems of delayed routing table updates and difficulty in real-time identification and prevention of communication loops in existing technologies, but also achieves adaptive control of topology disturbances in embedded IoT environments while maintaining system communication stability and real-time performance. It has structural innovation, precise control and deployment feasibility, and improves the reliability and security of IoT communication buses in dynamic scenarios.

[0042] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for deploying an embedded IoT communication bus application based on the HarmonyOS system, characterized in that, Includes the following steps: S1. Acquire the sleep-wake trigger sequence of the embedded device under a unified time baseline, construct an instantaneous topological difference map, and determine the amplitude, duration and scope of the topological mutation based on it; S2, based on the topological differential map, constructs a causal residual matrix, trains a loop precursor identification model, is used to locate the initial boundary of the topological closed loop, and extracts the time delay features of key nodes; S3. Construct a topology-consistent routing shadow table based on the initial boundary, and use the time-inversion path correction method to generate a zero-loop handover path sequence and a path protection threshold set. S4 dynamically injects the shadow table into the online forwarding table according to the segmented time stamp, and forwards the data packets in micro-batch mode through pulse-level amplitude limiting write-back, ensuring the continuity and loop-free nature of the path switching process; S5, after injection, combines the golden ratio frequency shifting traction and dual time-scale mirroring traction to periodically adjust the path weights using a breathing-like oscillation.

2. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 1, characterized in that, Step S1 includes: Under a unified time baseline, the sleep-wake trigger sequence of embedded IoT devices is continuously collected, and a complete state evolution sequence is constructed by combining the physical connection status of the nodes, signal strength level and link delay status. Based on time alignment, the communication relationships between nodes within consecutive time windows before and after wake-up are extracted, and superimposed to generate a topological differential graph. The connection change rate, communication delay change magnitude, and link stability offset gradient are introduced as edge weight attributes of the differential graph. Identify regions of edge weight variation in the difference graph, statistically analyze the perturbation expansion trend, delineate propagation boundaries, and classify events. The hierarchical results are mapped to the temporal coordinates of the nodes, and a multidimensional correlation matrix with the wake-up timestamp as the main axis is constructed to form a topological evolution data structure for path prediction and anomaly identification.

3. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 2, characterized in that, In the topological differential map, the expansion rate of the disturbance propagation boundary is combined with the topological depth of the central node to determine the impact level of the mutation event. Based on the impact level, the topological disturbance trajectory of the node is extracted to construct a topological behavior matrix with retrospective and predictive capabilities, which serves as the basis for judging the path evolution trend.

4. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 1, characterized in that, Step S2 includes: After completing the construction of the topological differential map, the structural state and performance parameters of the communication path in continuous time segments are extracted based on the unified time baseline, a set of topological evolution trajectories is constructed, and a causal residual matrix is ​​constructed. Identify concentrated perturbation paths in the causal residual matrix, and determine the closed-loop evolution region by combining path location and topological boundary features; A loop precursor recognition model is trained based on evolutionary trajectory, and a path-by-path scan is performed on the topological map to be detected to identify the closed loop germination boundary. Extract the time delay evolution trajectory of nodes associated with the closed-loop path and construct a time delay feature set of key communication nodes.

5. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 4, characterized in that, When constructing the causal residual matrix, the path state before wake-up is used as the benchmark. The path state offset is formed by comparing the path state after wake-up. The number of times the path state offset exceeds the preset threshold is used as the disturbance concentration index to screen high-risk communication paths within the closed-loop evolution area.

6. The embedded IoT communication bus application deployment method based on HarmonyOS system according to claim 1, characterized in that, S3 include: After completing the initial boundary identification of the closed loop and the extraction of time delay features of key communication nodes, the communication paths in the closed loop boundary are selected as candidate targets for path replacement, and a dynamic candidate path set is constructed by combining the edge weight change trend and node load. The candidate path set is corrected by time inversion, the path stability is judged based on the historical topological evolution trajectory, high-risk paths are eliminated and delay offset compensation is performed on the retained paths to generate a corrected path structure with topological consistency. Based on the path score, a switching path sequence is generated, and a set of protection thresholds including the hop count variation range and data load limits are set to achieve smooth path migration and closed-loop avoidance.

7. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 6, characterized in that, Time-reversal path correction involves retrospectively analyzing the connection status, link delay fluctuations, and node interaction strength of candidate paths within historical topology cycles, and filtering out high-risk paths based on disturbance frequency, while retaining paths achieves time alignment through delay compensation.

8. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 1, characterized in that, S4 includes: Based on the path protection threshold set, a time-segmented injection plan is set, and the path structure is decomposed into the starting jump point, relay node segment and the final convergence segment and then activated sequentially. Before each time period injection, a path segment mapping conflict retrieval is performed. If a conflict exists, the injection is delayed, and the path availability is confirmed by the target node response delay. During the path segmentation injection process, a pulse time window is set to limit the number of micro-batch data packets forwarded and attach a path activation identifier; By analyzing message timestamps and priorities, micro-batch cross-segment migrations are implemented between path switching cycles, and data granularity is adjusted to ensure forwarding continuity. After completing path injection, monitor the path's running status, determine its closure capability, release old path entries, and complete path switching convergence.

9. The method for deploying an embedded IoT communication bus application based on the HarmonyOS system according to claim 8, characterized in that, When setting the pulse time window, the pulse period is dynamically adjusted according to the network stability during the path switching phase. When the path is detected to be in a highly unstable state, the pulse interval is shortened to reduce the data packet injection rate, so as to avoid communication blockage or loop backflow caused by sudden load increases during the path switching process.

10. The method for deploying an embedded IoT communication bus application based on HarmonyOS according to claim 1, characterized in that, S5 include: Set asynchronous weight update cycles for all injected paths, and use a frequency shifting method that combines global control frequency and golden ratio constant to periodically shift the path weights. Two sets of mirror time-stamped sequences are established to synchronously constrain the current performance and historical risk status of the path, respectively, so as to achieve positive incentives and reverse constraints during the weight adjustment process; The path weights are periodically adjusted by a breathing-type oscillation curve, which increases, stabilizes, and decreases to maintain a slow and stable scheduling rhythm. The monitoring path changes in connection density and backflow probability fluctuations within the oscillation cycle. If the dynamic threshold is exceeded, the path is triggered to enter a suppression protection state. The overall network closed-loop risk index is assessed based on the dispersion of network weight distribution and the overlap of active paths, and the error frequency ratio and mirror baseline parameters are reset to complete the global weight scheduling rebalancing.

Images