A communication path selection method based on multiple bluetooth networks

Abstract

The application discloses a communication path selection method based on a multi-Bluetooth network, which comprises the following steps: collecting spectrum data, adding time-space labels to the collected spectrum data to generate time-space interference data units, and calculating interference indexes; collecting historical data, identifying interference patterns and extracting interference features based on the collected historical data, and constructing an interference pattern feature table; predicting interference indexes by using a model through the interference pattern feature table, generating interference risk scores, and distributing channels by using the interference risk scores; optimizing path communication quality according to the predicted interference indexes, forming a time-space perception path table, collecting multi-source information such as motion data and spectrum data, combining historical data to mine interference pattern features, optimizing path selection and channel distribution by using the predicted interference indexes, and finally generating a dynamically updated time-space perception path table, so that a more stable, efficient and reliable communication path is provided for Bluetooth ad hoc networks, and the influence of interference on communication quality is reduced.

Description

Technical Field

[0001] This invention relates to the field of data transmission technology, and in particular to a communication path selection method based on multiple Bluetooth networks. Background Technology

[0002] With the rapid development of IoT technology, Bluetooth networks, as a short-range wireless communication technology, have been widely used in smart homes, health monitoring, and industrial control due to their low power consumption, low cost, and ease of deployment. However, Bluetooth networks face many challenges in practical applications, especially in complex environments where multiple Bluetooth devices coexist. The selection and optimization of communication paths have become key factors affecting network performance and user experience.

[0003] Traditional Bluetooth networks often rely on fixed routing protocols or simple Signal Strength Indication (RSSI) for communication path selection. While these methods perform adequately in static or low-dynamic environments, their limitations and shortcomings become increasingly apparent in complex scenarios with high dynamics and heavy interference. For example, fixed routing protocols struggle to adapt to network topology changes caused by device movement, and simple RSSI metrics fail to comprehensively reflect interference and communication quality along the path. Secondly, interference in Bluetooth networks cannot be ignored. With the widespread adoption of Bluetooth devices, multiple Bluetooth networks may operate simultaneously in the same space, leading to spectrum scarcity and severe signal interference. This interference not only reduces data transmission success rate and speed but can also cause communication interruptions, severely impacting user experience. Therefore, effectively identifying and avoiding interference and selecting an optimal communication path has become a crucial issue for improving Bluetooth network performance. Summary of the Invention

[0004] This application provides a communication path selection method based on multiple Bluetooth networks. By collecting multi-source information such as motion data and spectrum data, and combining historical data to mine interference pattern characteristics, the method optimizes path selection and channel allocation using the predicted interference index, and finally generates a dynamically updated spatiotemporally aware path table. This provides a more stable, efficient and reliable communication path for Bluetooth ad hoc networks, reduces the impact of interference on communication quality, and improves the performance of the entire network and the user experience.

[0005] This application provides a communication path selection method based on multiple Bluetooth networks, including:

[0006] S101, Collect spectrum data, add spatiotemporal labels to the collected spectrum data, generate spatiotemporal interference data units based on the collected spectrum data and spatiotemporal labels, and calculate the interference index;

[0007] S102, Collect historical data. The Bluetooth device identifies interference patterns and extracts interference features based on the collected historical data, and constructs an interference pattern feature table based on the interference model and interference features.

[0008] S103, use the model to predict the interference index through the interference mode feature table, generate an interference risk score based on the predicted interference index, and use the interference risk score to allocate the channel.

[0009] S104, optimize path communication quality based on predicted interference index to form a spatiotemporally aware path table.

[0010] Preferably, the interference index is calculated based on the collected spectrum data: in, Interference index For the i-th signal strength, Let be the time occupied by the i-th signal, n be the total number of signals, and Tscan be the total scan time. As the noise baseline, This is the reference noise.

[0011] Preferably, the path communication quality is optimized based on the predicted interference index to form a spatiotemporally aware path table. Specifically, this includes: calculating the interference index by replacing the spectrum data with the predicted interference index; the communication quality parameter = historical transmission efficiency × predicted interference index × device health, where historical transmission efficiency reflects the success rate and speed of data transmission on the path over a past period, and device health measures the operating status and reliability of relevant communication equipment on the path; introducing a path pre-switching mechanism, which, through spatiotemporal pattern prediction, triggers a path pre-switching operation when the interference index of the current path is predicted to increase significantly in the next period; dividing the communication quality parameter by the interference risk score to obtain the path weight, and sorting all candidate paths according to the path weight to generate a spatiotemporally aware path table.

[0012] Preferably, it further includes:

[0013] S201: Collect motion data, calculate motion index using motion data, form mobile-spatiotemporal interference data unit based on interference data and motion data, identify motion trajectory based on mobile-spatiotemporal interference data unit, and construct motion trajectory-interference correlation model based on motion trajectory and interference data;

[0014] S202, adjust the interference risk score according to the movement trajectory, generate different path selection strategies based on devices with different speeds, and calculate the weight of path selection based on communication quality parameters.

[0015] Preferably, the formula for calculating the exercise index using exercise data is: ,in, For the moving index, For the moving speed of the equipment, The rate of change of direction represents the magnitude of change in the direction of movement of the equipment per unit time.

[0016] Preferably, the method for forming a mobile-spatiotemporal interference data unit based on interference data and motion data is as follows: collect the historical movement trajectory of the device and the interference data at the corresponding time points, and combine the motion data and interference data into a mobile-spatiotemporal interference data unit with a certain time interval as the time window.

[0017] Preferably, the step of adjusting the interference risk score according to the movement trajectory is as follows: Real-time monitoring of the device's current movement speed and direction; obtaining the corresponding movement speed-interference fluctuation rate correspondence table and movement direction-regional interference mapping relationship from a pre-built mobile-spatial interference pattern library based on the device's current movement speed and direction; and adjusting the interference risk score according to the movement trajectory. ,in, The adjusted interference risk score, For temporal regularity, it reflects the degree of regularity of the disturbance in the time dimension. Spatial stability refers to the degree of stability of disturbances in the spatial dimension. Equipment density variation represents the change in the number of devices within a given area. The motion index represents the movement trajectory. , and All are weighting coefficients, and .

[0018] Preferably, it further includes:

[0019] S301 collects acceleration change data and real-time user behavior patterns, constructs a behavior-acceleration mapping table, calculates the behavior-acceleration consistency index based on the behavior-acceleration mapping table, and classifies acceleration changes according to the rate of acceleration change.

[0020] S302, predict the movement trajectory based on the current behavior pattern, and correct the predicted movement trajectory based on the rate of change of acceleration;

[0021] S303 calculates the switching timing based on the moving speed and predicted trajectory, and switches the path according to the switching timing.

[0022] Preferably, the characteristics of the real-time collected acceleration change data are compared with the typical acceleration characteristics of the corresponding behavior in the behavior-acceleration mapping table, and the consistency index is obtained by calculating the similarity between them. The formula for calculating the behavior acceleration consistency index is as follows: ,in, The behavioral acceleration consistency index is calculated based on cosine similarity, and its value range is [ 1,1], This is a real-time acquired acceleration feature vector, which contains acceleration feature values ​​in multiple dimensions. This represents the typical acceleration feature vector corresponding to the behavior in the behavior-acceleration mapping table, which also contains typical acceleration feature values ​​in multiple dimensions. Let be the feature value of the i-th dimension of the real-time collected acceleration feature vector A. is the eigenvalue of the i-th dimension of a typical acceleration eigenvector, and n is the number of dimensions of the acceleration eigenvector.

[0023] Preferably, it further includes:

[0024] S401 identifies micro-movement events based on movement trajectory, movement speed, and angle; collects micro-movement event data to construct a micro-movement pattern library; extracts the features of the identified micro-movement events; identifies micro-movement influencing factors based on the micro-movement event features; calculates micro-movement stability coefficients based on the micro-movement influencing factors; and uses the micro-movement stability coefficients to predict the communication quality of the movement.

[0025] S402, construct a multi-level path adjustment mechanism based on micro-movement events.

[0026] One or more technical solutions provided in this application have at least the following technical effects or advantages: comprehensively collecting multi-source information such as motion data and spectrum data, combining historical data to mine interference pattern characteristics, using predicted interference index to optimize path selection and channel allocation, and finally generating a dynamically updated spatiotemporal-aware path table, thereby providing a more stable, efficient and reliable communication path for Bluetooth ad hoc networks, reducing the impact of interference on communication quality, improving the performance and user experience of the entire network, meeting the diverse needs of Bluetooth communication in different scenarios, improving the accuracy of communication quality assessment, optimizing channel allocation, realizing dynamic path pre-switching, ensuring that the spatiotemporal-aware path table is updated in real time, and ultimately improving the overall network performance and user experience;

[0027] By using mobile trajectory-interference correlation models and mobile-aware path lifetime prediction, the impact of device movement on communication paths can be predicted more accurately, allowing for path optimization and switching in advance. Through mobile adaptive path decision-making, interference changes can be addressed in advance, greatly improving communication reliability. By deeply integrating device mobility factors, the original spatiotemporal interference mode is expanded into a mobile-spatiotemporal cooperative model, solving the topology uncertainty problem caused by device movement in Bluetooth networks. This enables the system to predict movement and optimize in advance, significantly improving communication quality in mobile scenarios and providing innovative communication guarantees for mobile IoT applications.

[0028] By combining user behavior recognition with acceleration, a behavior-acceleration mapping table is established, taking into account multi-dimensional information of user behavior and physical motion state. This enables trajectory prediction to more accurately reflect the actual movement of users and improves the accuracy of trajectory prediction. By integrating behavior perception and acceleration analysis, the problem of path switching caused by behavioral uncertainty and sudden motion changes in the mobile environment is solved, realizing an intelligent adaptive path switching mechanism that switches from location-based to behavior-based motion state switching, providing more accurate communication guarantees for dynamic mobile environments.

[0029] The identification and hierarchical adjustment mechanism for micro-motion events avoids macro-path switching due to overreaction, reduces the overhead and risks of switching, and realizes the transformation from macro-path management to micro-parameter adaptation. It identifies and adapts to more subtle motion changes, significantly reduces unnecessary path switching while maintaining path stability, and provides an optimization paradigm for Bluetooth communication in highly dynamic environments. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating a communication path selection method based on multiple Bluetooth networks according to the present invention.

[0031] Figure 2 This is a schematic diagram illustrating the process of constructing the motion trajectory-interference correlation model for this invention.

[0032] Figure 3 This is a schematic diagram of the process for calculating the switching timing in this invention;

[0033] Figure 4 This is a flowchart illustrating the process of constructing a multi-level path adjustment mechanism according to the present invention. Detailed Implementation

[0034] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0035] It should be noted that the terms "vertical," "horizontal," "up," "down," "left," "right," and similar expressions used in this article are for illustrative purposes only and do not represent the only possible implementation.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0037] Example 1: Figure 1 This is a flowchart illustrating a communication path selection method based on multiple Bluetooth networks according to an embodiment of the present invention, including:

[0038] S101, Collect data, add spatiotemporal labels to the collected data, generate spatiotemporal interference data units based on the collected data and spatiotemporal labels, and calculate the interference index based on the collected data;

[0039] Specifically, the collected data is spectrum data. A low-power spectrum sensor is configured on the Bluetooth device to collect spectrum data, including signal strength, channel occupancy, and noise baseline. Signal strength is obtained by collecting the power values ​​emitted by surrounding devices; channel occupancy is obtained by calculating the percentage of active time for each channel; and the noise baseline is obtained by measuring the ambient background noise intensity. The collected spectrum data is time-stamped according to time period encoding (morning / noon / evening), and spatial tags are generated using Bluetooth beacon triangulation. An interference index is calculated based on the collected spectrum data. in, Interference index For the i-th signal strength, Let be the time occupied by the i-th signal, n be the total number of signals, and Tscan be the total scan time. As the noise baseline, For reference noise, The motion state unit and the spatiotemporal interference data unit are obtained by measuring the ambient background noise through calibration equipment; the motion state unit and the spatiotemporal interference data unit are generated by packaging the spectrum data, spatiotemporal tags and calculated interference data; and the motion state unit and the spatiotemporal interference data unit are uploaded to the central node (Bluetooth self-organizing network central device) through Bluetooth ADV frames.

[0040] S102, Collect historical data. The Bluetooth device identifies interference patterns and extracts interference features based on the collected historical data, and constructs an interference pattern feature table based on the interference model and interference features.

[0041] Furthermore, the historical data collection time range covers the past 30 days of spatiotemporal interference data units stored in the Bluetooth ad hoc network central device, and the spatial range covers all Bluetooth device deployment areas. Time tags, spatial tags, interference indices, and motion indices are extracted from the spatiotemporal interference data units. The collected historical data undergoes data cleaning, removing data points with missing time or spatial tags and samples with interference indices exceeding reasonable ranges. Interference pattern identification is performed using both time and spatial modes. For the time mode, the periodicity of the interference index in the time dimension is identified. Historical data is aggregated at a fixed time granularity (e.g., hourly), and the average, peak, and volatility of the interference index for each time period are calculated. The average reflects the overall interference level of the time period, the peak identifies the maximum interference value within the time period, and the volatility measures the stability of the interference within the time period using standard deviation. The periodicity of the interference index is detected using Fast Fourier Transform (FFT), converting the time-domain signal of the interference index to the frequency domain to identify the dominant frequency (e.g., a 24-hour cycle corresponding to a daily cycle). If the FFT result shows a significant 24-hour frequency component, the daily periodicity of the interference is confirmed. For the spatial mode, the spatial attenuation model is implemented using Motley-... The Keenan model is used to construct the spatial attenuation model, which is based on the characteristics of signal propagation in an ideal, unobstructed free space. It indicates that signal strength attenuates inversely with the square of the propagation distance, meaning signal power is inversely proportional to the square of the distance. Simultaneously, it is corrected based on material attenuation coefficients. The inputs to the spatial attenuation model are environmental information and interference data. The environmental information includes wall material data and spatial layout information. The wall material data is obtained from a predefined wall material database containing attenuation coefficients for different materials. The spatial layout information includes the location of the signal source, the location of the receiving point, and the distribution of obstacles such as walls in the space. The interference data includes the interference index and the actual measured interference value. Path loss is calculated based on the spatial attenuation model, and interference propagation hotspots are identified. The formula for calculating path loss is: ,in, This represents the total path loss. Free space path loss, , The propagation distance between the signal source and the receiving point. For signal frequency, For constant terms, This refers to the amount of material degradation. , The material attenuation coefficient, The distance the signal travels through the material is used to identify interference propagation hotspots. Based on the spatial distribution density of the interference index, the kernel density estimation method is used to identify continuous high-value areas. This is achieved by placing a kernel function (Gaussian kernel function) around each data point and then superimposing all kernel functions to obtain the density estimate for the entire space. By setting a threshold, areas with density estimates higher than the threshold are marked as high-interference areas. For example, a corridor may form a continuous high-interference area due to signal reflection. The spatial attenuation model is calculated based on the internal path loss calculation method. Finally, the spatial attenuation model outputs the theoretical attenuation value of the predicted signal from the source to each receiving point and information on interference propagation hotspot areas. The theoretical attenuation value predicted by the model is compared with the actual measured value. The error index is calculated based on the root mean square error, and an error threshold is set. If the obtained error is greater than the preset error threshold, the parameters of the model are adjusted.

[0042] Temporal, spatial, and dynamic features are extracted from the collected historical data. The extracted features are then grouped into time-period interference index and region-interference index pairs. K-means clustering is performed on the time-period interference index and region-interference index pairs. Clustering error curves are plotted for different K values. The inflection point where the error decreases slowly is selected as the K value. The data is divided into K clusters, and a label is assigned to each cluster. An interference pattern feature table is constructed based on the cluster type and semantic label.

[0043] S103, Before the actual path selection, the interference index is predicted through the interference mode feature table, the interference risk score is generated based on the predicted interference index, and the channel is allocated using the interference risk score.

[0044] Specifically, based on the interference index calculated in step S101, a dynamic interference map is constructed using the Louvain algorithm. Subgraphs with dense edge weights in the dynamic interference map are identified as hotspot areas, indicating severe interference between devices within that area. The interference baseline value corresponding to the current time period is loaded from the interference mode feature table. The predicted interference index is obtained based on the spatial attenuation model and the interference baseline value. The predicted interference index is corrected based on the different movement states of the devices. If the device is stationary, the co-channel interference superposition range is less than or equal to -70dBm, meaning the interference between signals will not produce a significant superposition effect, and the impact on communication quality is small. If the device is moving at low speed, the co-channel interference superposition range is between -70dBm and -60dBm. In this state, the signal power will fluctuate to some extent, ranging from 1-3dB. Compared to the stationary state, low-speed movement will slightly change the interference situation. When the device is moving at high speed and the co-channel interference superposition threshold is greater than -60dBm... When interference becomes more severe, not only will the signal power be significantly attenuated, ranging from 5 to 8 dB, but the frequency deviation of the signal will also increase. When the device is in a high-speed moving state, the predicted interference index needs to be multiplied by a correction factor, which is 1.3 here.

[0045] The formula for calculating the interference risk score is: ,in, To interfere with risk scoring, For temporal regularity, it reflects the degree of regularity of the disturbance in the time dimension. Spatial stability refers to the degree of stability of interference in spatial dimensions. Different spatial locations may be affected by different types of interference sources. Spatial stability describes the distribution and variation of interference in different spatial locations. Equipment density variation represents the change in the number of devices within a given area. , and All are weighting coefficients, and .

[0046] Channels are allocated based on the predicted interference index, and the objective function for channel allocation is: Among them, weight , Reflecting the interference relationship between channels, the objective function weights are obtained by multiplying the predicted interference exponent by the geometric mean edge weights. It is a binary decision variable used to represent the decision-making situation of channel allocation. Q-learning reinforcement learning is used to dynamically adjust the channel allocation strategy, with the historical allocation success rate as the reward function. The Q-learning algorithm learns the optimal channel allocation strategy by continuously interacting with the environment. After each channel allocation, the Q value is updated according to the historical allocation success rate. If the historical allocation success rate of the current allocation strategy is high, the Q value corresponding to the strategy is increased; otherwise, the Q value is decreased. Through continuous iteration and learning, the algorithm can gradually find the optimal channel allocation strategy. According to the generated interference risk score, the channel allocation is divided into different priorities. Areas or time periods with high interference risk scores need to be allocated channels with less interference first.

[0047] S104, Optimize path communication quality based on predicted interference index to form a spatiotemporally aware path table;

[0048] Specifically, traditional communication quality parameter calculations often rely on the instantaneous interference index. However, instantaneous data exhibits significant fluctuations and uncertainties, making it difficult to accurately reflect the true communication status of a path over a period of time. Therefore, the communication quality parameter formula is reconstructed, replacing the instantaneous interference index with the predicted interference index. The calculation formula for the communication quality parameter is: Communication Quality Parameter = Historical Transmission Efficiency × Predicted Interference Index × Equipment Health. Here, historical transmission efficiency reflects the success rate and speed of data transmission along the path over a past period, while equipment health measures the operational status and reliability of relevant communication equipment along the path. During the path selection process... In addition to considering communication quality parameters, a path pre-switching mechanism is introduced. Through spatiotemporal pattern prediction, when it is predicted that the interference index of the current path will increase significantly in the next time period, or the interference index of the backup path will decrease significantly, the system will trigger a path pre-switching operation. The communication quality parameters are divided by the interference risk score to obtain the path weight. All candidate paths are sorted according to the path weight to generate a spatiotemporal-aware path table. The spatiotemporal-aware path table contains information such as path identifier, start node, end node, communication quality parameters, risk score, and path weight. The spatiotemporal-aware path table is dynamically updated according to the real-time status of the network and prediction data.

[0049] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: by comprehensively collecting multi-source information such as motion data and spectrum data, combining historical data to mine interference pattern characteristics, and using the predicted interference index to optimize path selection and channel allocation, a dynamically updated spatiotemporal-aware path table is finally generated, thereby providing a more stable, efficient, and reliable communication path for Bluetooth ad hoc networks, reducing the impact of interference on communication quality, improving the performance and user experience of the entire network, meeting the diverse needs of Bluetooth communication in different scenarios, improving the accuracy of communication quality assessment, optimizing channel allocation, realizing dynamic path pre-switching, ensuring that the spatiotemporal-aware path table is updated in real time, and ultimately improving the overall network performance and user experience.

[0050] Example 2: Example 1 assumes the device location is relatively fixed and does not consider dynamic changes caused by device mobility. This example integrates device mobility factors to solve the uncertainty problem in topology prediction caused by device movement. By establishing a movement trajectory-interference correlation model, dynamic parameters such as device movement distance, speed, and direction are deeply bound to spatiotemporal interference patterns, enabling motion-aware path prediction and switching. This allows the communication path to dynamically track device movement, such as... Figure 2 As shown.

[0051] S201: Collect motion data, form a mobile-spatiotemporal interference data unit based on interference data and motion data, identify the movement trajectory based on the mobile-spatiotemporal interference data unit, and construct a movement trajectory-interference correlation model based on the movement trajectory and interference data;

[0052] Specifically, each Bluetooth device integrates a three-axis accelerometer and a gyroscope. The acceleration data collected by the three-axis accelerometer is integrated to calculate the movement speed. Data collected by the gyroscope is used to determine the device's direction of movement. The calculated movement speed is integrated to obtain the device's distance traveled. Tags are then bound based on the movement speed, with a speed range including a minimum threshold and a maximum threshold. When the device speed is below the minimum threshold, the tag is marked as stationary; when the device speed is between the minimum and maximum thresholds, the tag is marked as low-speed movement; and when the device speed is above the maximum threshold, the tag is marked as high-speed movement. A motion index is calculated based on the movement speed. ,in, For the moving index, For the moving speed of the equipment, The direction change rate, calculated from gyroscope data, represents the magnitude of change in the device's movement direction per unit time. The collected motion data, motion tags, and motion indices are packaged to generate motion state units. The historical movement trajectory of the device is collected, including its position information at different time points. Simultaneously, interference data at corresponding time points is collected. Using time windows with specific time intervals (e.g., every 1 minute, every 5 minutes, etc., determined according to actual needs and data characteristics), the preprocessed motion data and interference data are combined. The motion data and corresponding interference data within each time window together constitute a motion-spatiotemporal interference data unit. The K-Means clustering algorithm is used to divide the data units into different clusters based on the similarity between them. Each cluster represents a relatively similar movement pattern. By analyzing these clusters, common movement paths of the device are identified. The results are further analyzed to select representative typical movement corridors from the common movement paths. For example, moving along a fixed direction at a relatively stable speed within a specific area. Based on the identified typical movement corridors, the typical movement corridor models in the motion-spatiotemporal interference pattern library are updated, and the feature information of the newly identified typical movement corridors is stored in the pattern library.

[0053] Based on the identified movement trajectories and interference data, a movement trajectory-interference correlation model is constructed. A certain number of high-speed mobile devices are selected from the device group as samples, covering different device types and usage scenarios. Based on the movement speed of the selected samples, the corresponding interference fluctuation rate is calculated. The interference fluctuation rate is calculated by statistically analyzing the standard deviation of the interference value. The interference fluctuation rate reflects the degree of change of the interference value at that movement speed. Each movement speed and its corresponding interference fluctuation rate are organized into a table to form a movement speed-interference fluctuation rate correspondence table, which is stored in a movement-spatiotemporal interference pattern library. For directional mobile devices, their future positions are predicted based on their historical movement data and the trends of their historical movement paths. For example, if the device has been moving along a straight line in historical data with a relatively stable speed, the position at several future time points can be predicted based on the current position and speed. The average interference value and interference fluctuation rate of directional mobile devices when passing through various areas in different movement directions are statistically analyzed to establish a movement direction-area interference mapping relationship, which is stored in a movement-spatiotemporal interference pattern library. In the spatiotemporal interference pattern library, based on the current moving speed and direction of the device, the corresponding moving speed-interference volatility correspondence table and moving direction-regional interference mapping relationship are obtained from the moving-spatiotemporal interference pattern library. The LSTM neural network in the machine learning prediction model is used to predict the trend of interference value change of the device at several future time points. The collected historical data is divided into training set and validation set to train and validate the LSTM neural network. The obtained moving speed, interference volatility, moving direction and regional interference are input into the LSTM neural network to predict the trend of interference value change of the device at several future time points.

[0054] S202, adjust the interference risk score according to the movement trajectory, generate different path selection strategies based on devices with different speeds, and calculate the weight of path selection based on communication quality parameters;

[0055] Furthermore, the system monitors the device's current speed and direction of movement in real time. Based on the device's current speed and direction, it accurately queries and retrieves the corresponding speed-interference volatility mapping table and direction-regional interference mapping relationship from a pre-built spatiotemporal interference pattern library. The interference risk score is then adjusted according to the device's trajectory (speed and direction). ,in, The adjusted interference risk score, For temporal regularity, it reflects the degree of regularity of the disturbance in the time dimension. Spatial stability refers to the degree of stability of interference in spatial dimensions. Different spatial locations may be affected by different types of interference sources. Spatial stability describes the distribution and variation of interference in different spatial locations. Equipment density variation represents the change in the number of devices within a given area. The motion index represents the movement trajectory. , and All are weighting coefficients, and .

[0056] Different path selection strategies are implemented for devices with different speeds. For high-speed mobile devices, topologies with smooth path switching are selected. Various path topologies in the network are analyzed, and the communication continuity during path switching is evaluated. For example, topologies with large overlap between adjacent paths and minimal signal attenuation during switching are chosen. The communication quality under different topologies is evaluated, and the optimal topology is selected as the path for high-speed mobile devices. For directional mobile devices, relay paths are pre-configured in the direction of movement. Based on the directional movement direction of the device and the movement direction-regional interference mapping relationship, the areas the device may pass through during movement are predicted, and suitable paths are pre-configured in these areas to ensure seamless switching to the next suitable path during movement. For example, if the device moves directionally to the northeast, multiple paths with good communication quality are sequentially configured in the northeast direction based on regional interference conditions to form a relay path.

[0057] Based on the communication quality parameters calculated in step S104, the path selection weight is calculated using the formula: Path selection weight = Communication quality parameter × Mobility adaptability coefficient. The mobility adaptability coefficient is calculated based on the matching degree between path lifetime prediction and device moving speed. The ratio of the predicted path lifetime value to the time required for the device to traverse the path at the current speed is used as the matching degree index. If the matching degree is greater than 80%, it indicates that the matching degree between the path lifetime prediction and the device moving speed is high, and the mobility adaptability coefficient is set to 1.2. If the matching degree is less than 30%, it indicates that the matching degree is low, and the mobility adaptability coefficient is set to 0.8. If the matching degree is between 30% and 80%, the mobility adaptability coefficient can be calculated using the linear interpolation method.

[0058] By comparing the actual movement trajectory with the predicted trajectory, the trajectory prediction error rate is calculated. If the error rate is greater than a preset threshold, the parameters of the machine learning prediction model are updated.

[0059] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: By using the mobile trajectory-interference correlation model and mobile-aware path lifetime prediction, the impact of device movement on the communication path can be predicted more accurately, allowing for path optimization and switching in advance. Through mobile adaptive path decision-making, interference changes can be addressed in advance, greatly improving communication reliability. By deeply integrating device mobility factors, the original spatiotemporal interference mode is expanded into a mobile-spatiotemporal cooperative model, solving the problem of topology uncertainty caused by device movement in Bluetooth networks. This enables the system to predict movement and optimize in advance, significantly improving communication quality in mobile scenarios and providing innovative communication guarantees for mobile IoT applications.

[0060] Example 3: Following Example 2, which considered movement speed and trajectory prediction (the regularity of user behavior (such as walking or driving) determining movement trajectory), this example integrates two dimensions: the rate of change of movement acceleration and user behavior context, to establish a collaborative prediction model of behavior-acceleration-trajectory. By quantitatively analyzing the correlation between acceleration change patterns and behavioral characteristics, it dynamically optimizes path switching timing, solves the path interruption problem caused by sudden movement changes and behavioral uncertainties, and achieves accurate path switching decisions. Figure 3 As shown.

[0061] S301 collects acceleration change data and real-time user behavior patterns, constructs a behavior-acceleration mapping table, calculates the behavior-acceleration consistency index based on the behavior-acceleration mapping table, and classifies acceleration changes according to the rate of acceleration change.

[0062] Specifically, accelerometers and gyroscopes are used to acquire acceleration change data and the user's real-time behavior patterns. Accelerometers measure the acceleration changes of an object along three axes, while gyroscopes detect the object's angular velocity. Combining these two technologies, dynamic information about the user's movement is obtained. The Bluetooth device incorporates a behavior recognition algorithm based on a 3D convolutional neural network (3D CNN). CNN classifies a user's current behavior pattern based on the feature patterns of acceleration and angular velocity. The behavior patterns are divided into stationary, walking, running, and driving. Taking walking behavior as an example, when the acceleration shows periodic changes and its amplitude is within a pre-set range, the algorithm will determine that the user is currently walking. At the same time, the algorithm will calculate the behavior confidence score based on the degree of matching between the real-time collected data and typical walking patterns. The score range is set to 0-100%. The higher the degree of matching, the closer the confidence score is to 100%; conversely, the lower the degree of matching, the closer the score is to 0. A small time interval Δt is selected. Within this time interval, the acceleration values ​​of two adjacent moments are differentially calculated to obtain the change in acceleration Δa. Then, Δa is divided by Δt to obtain the instantaneous rate of change of acceleration. The acceleration trend is judged based on the sign and change of the instantaneous rate of change of acceleration over a period of time. If the instantaneous rate of change of acceleration remains positive during this period, it indicates that the object is in a state of continuous acceleration; if the instantaneous rate of change of acceleration remains negative, it indicates that the object is in a state of continuous deceleration; if the instantaneous rate of change of acceleration changes frequently between positive and negative and the value is small, it indicates that the object's acceleration is in a stable state, that is, the speed remains basically constant. The data obtained above is filtered and normalized to analyze the acceleration characteristics corresponding to different behavior patterns and organize them into a behavior-acceleration mapping table. For example, walking behavior usually corresponds to periodic acceleration changes, with acceleration fluctuations ranging from 0.5 to 2 m / s²; while driving behavior will show sudden acceleration changes during emergency braking, with acceleration values ​​greater than 3 m / s².

[0063] The characteristics of real-time collected acceleration change data are compared with the typical acceleration characteristics of the corresponding behaviors in the behavior-acceleration mapping table. The consistency index is obtained by calculating the similarity between them. The formula for calculating the behavior acceleration consistency index is as follows: ,in, The behavioral acceleration consistency index is calculated based on cosine similarity, and its value range is [ [1,1], the closer the value is to 1, the more similar the real-time acceleration characteristics are to typical acceleration characteristics. This is a real-time acquired acceleration feature vector, which contains acceleration feature values ​​in multiple dimensions. This represents the typical acceleration feature vector corresponding to the behavior in the behavior-acceleration mapping table, which also contains typical acceleration feature values ​​in multiple dimensions. Let be the feature value of the i-th dimension of the real-time collected acceleration feature vector A. is the eigenvalue of the i-th dimension of the typical acceleration eigenvector, and n is the number of dimensions of the acceleration eigenvector, i.e., the number of acceleration features considered.

[0064] Based on the magnitude of the instantaneous acceleration rate of change, acceleration changes are divided into three levels: stable, fluctuating, and drastic. Different ranges of acceleration rate of change are set as grading thresholds. The acceleration rate of change for the stable level is set to be within the range [0, a1], for the fluctuating level within the range (a1, a2], and for the drastic level greater than a2. A path switching sensitivity parameter is defined for each acceleration stability level. The path switching sensitivity parameter reflects the system's sensitivity to path switching under different acceleration stability levels. When the acceleration is stable, it indicates that the object's motion is relatively smooth, so the sensitivity parameter is set to 1. When the acceleration is fluctuating, the object's motion is not very stable, so the sensitivity parameter is set to a value greater than 1, and the specific value can be adjusted according to the degree of fluctuation. When the acceleration changes drastically, the object's motion is very unstable, and the possibility of path switching is high. However, because the acceleration change is too drastic, it will lead to a decrease in the accuracy of trajectory prediction. Therefore, the sensitivity parameter is set to a value less than 1 to reduce the over-response to drastic acceleration changes. The Pearson correlation coefficient between acceleration change and disturbance index is calculated using the following formula: ,in, The Pearson correlation coefficient has a range of values ​​of [ ]. [1,1], positive values ​​indicate a positive correlation, negative values ​​indicate a negative correlation, and the larger the absolute value, the stronger the correlation. Let be the acceleration stability sensitivity parameter for the i-th sample. Let be the interference index of the i-th sample. The mean value of the acceleration stability sensitivity parameter. Let r be the mean of the interference index, and n be the total number of samples. If r > 0.7 (strong positive correlation), it indicates that the more severe the acceleration fluctuation, the higher the interference index (the less reliable the behavior recognition), and the stability control module should be optimized first. If r < 0.7, the interference index is higher. A value of 0.5 (significantly negative correlation) may indicate a reverse correlation (such as increased stability leading to increased interference), suggesting the need to recalibrate parameter thresholds or model logic.

[0065] S302, predict the movement trajectory based on the current behavior pattern, and correct the predicted movement trajectory based on the rate of change of acceleration;

[0066] Furthermore, based on the aforementioned behavior recognition algorithm, the current behavior pattern is identified, and the movement trajectory is predicted according to the identified behavior pattern. The parameters of the movement trajectory are set; if the predicted behavior trajectory is walking, the distance range is set to 0-500m, and the speed range is set to 0.5-1.5m / s. There are no fixed road restrictions, but path constraints from obstacles must be avoided. A Gaussian Mixture Model (GMM) is used to predict the initial trajectory of the walking pattern. The mean vector, covariance matrix, and mixing coefficients of the Gaussian Mixture Model are set, and passable points on the map are randomly selected. The sampling step size is taken from a log-normal distribution to simulate the asymmetry of walking speed. A Gaussian component is randomly selected based on the GMM mixing coefficients, and the directional offset is sampled from the Gaussian distribution of this component. Combining the current position, step size, and direction, the coordinates of the next position are calculated. Based on the probability density function of the GMM, the probability that the trajectory point falls within the 95% confidence interval is calculated, generating a confidence interval. If the predicted trajectory is a driving behavior trajectory, the distance range of the driving trajectory is set to be greater than 1km, the speed range is 5-30m / s, and the movement is along the center line of the road, complying with the path constraints of traffic rules. The algorithm optimizes the route by combining real-time traffic information obtained from a map API, randomly selecting drivable points on the map, and retrieving the road network from OpenStreetMap. The inputs are the start point, end point, and cost function. The algorithm traverses the road network, selects the path with the minimum total cost, and predicts the driving trajectory.

[0067] Based on the calculated rate of change of acceleration, a threshold is set, and the rate of change of acceleration is compared with the preset threshold. If the rate of change of acceleration is greater than the preset threshold, it is marked as a sudden change point, i.e., a trajectory anomaly. After detecting the trajectory anomaly, the trajectory is corrected. If it is a walking trajectory, the endpoint of the walking trajectory is moved forward by 10m. If it is a driving trajectory, the endpoint of the driving trajectory is moved backward by 20m, and re-path planning is triggered. The steps of trajectory re-planning are as follows: discard subsequent predicted points starting from the timestamp of the sudden change point, adjust the speed or distance range according to the type of sudden change, use the Local Random Walk (RRT) algorithm to avoid the area near the sudden change point for walking trajectories, and call the navigation API to re-plan the detour route and use a quadratic Bézier curve to smoothly transition the new trajectory with the original trajectory.

[0068] S303 calculates the switching timing based on the moving speed and predicted trajectory, and switches the path according to the switching timing;

[0069] Specifically, the formula for calculating the switching timing based on the moving speed and predicted trajectory is as follows: ,in, Timing of basic switching To predict the remaining distance of the trajectory, The formula for calculating the adjustment factor based on the standard deviation of acceleration, where the moving speed is the average speed, is: ,in, For acceleration adjustment coefficient, The baseline switching timing is adjusted based on the acceleration adjustment coefficient to determine the acceleration standard deviation. ,in, This is the revised basic switching timing. Timing of basic switching The acceleration adjustment coefficient is set based on the behavior pattern, with a weighting coefficient β=1.5 for driving behavior, β=1.0 for walking behavior, and β=1.2 for running behavior. The final switching timing is then calculated based on these weighting coefficients. ,in, For the final switching time, This is the revised basic switching timing. The behavior weighting coefficient is used; a tiered switching strategy is set according to the switching timing, and a switching threshold range is set, which includes a maximum threshold and a minimum threshold. When the threshold is less than the minimum threshold, the switchover will proceed normally according to the final switchover timing. When the threshold is greater than or equal to the minimum threshold but less than the maximum threshold, extend the switching time until... + ,in For dynamic buffer time, when When the threshold is greater than or equal to the highest threshold, immediately switch to the pre-selected stable path and activate the backup path.

[0070] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: combining user behavior recognition with acceleration to establish a behavior-acceleration mapping table, considering multi-dimensional information of user behavior and physical motion state, enabling trajectory prediction to more accurately reflect the actual movement of the user, improving the accuracy of trajectory prediction, and solving the path switching problem caused by behavioral uncertainty and motion change in the mobile environment by integrating behavior perception and acceleration analysis, realizing an intelligent adaptive path switching mechanism, switching from location-based to behavior-based motion state switching, and providing more accurate communication guarantee for dynamic mobile environments.

[0071] Example 4: Based on the movement speed, acceleration, and trajectory identification of Examples 1 to 3, this example refines the path from macroscopic movement to the microscopic level. By identifying microscopic movement events such as turning, pausing, and small displacements, a microscopic movement-path parameter mapping model is established to achieve precise fine-tuning of communication parameters, avoid excessive path switching, and improve system efficiency and stability. Figure 4 As shown.

[0072] S401 identifies micro-movement events based on movement trajectory, movement speed, and angle; collects micro-movement event data to construct a micro-movement pattern library; extracts the features of the identified micro-movement events; identifies micro-movement influencing factors based on the micro-movement event features; calculates micro-movement stability coefficients based on the micro-movement influencing factors; and uses the micro-movement stability coefficients to predict the communication quality of the movement.

[0073] Furthermore, based on the collected movement trajectory, movement speed, and angle, micro-movement events are identified. These micro-movement events include micro-turning events, micro-displacement events, micro-pause events, and posture fine-tuning events. When the detected angle change is within the range of 1-15 degrees and the duration of this angle change is between 0.5-2 seconds, it is determined to be a micro-turning event; when the detected trajectory displacement is within the range of 0.1-0.5 meters and the calculated average speed is less than 0.3 m / s, it is determined to be a micro-displacement event; when the detected speed is close to zero while the device remains active (e.g., the device's sensors are still working normally and outputting data), it is determined to be a micro-pause event; when a small change in the device angle is detected (1-...), it is determined to be a micro-pause event. (5 degrees) is identified as a posture fine-tuning event; a large amount of micro-movement event data is collected in different environments, and feature extraction and feature quantification are performed on the micro-movement event data. Typical event sequence combinations are sorted out from a large amount of micro-movement event data, such as micro-turning → micro-displacement → micro-pause, micro-pause followed by micro-turning, and other common situations. Then, statistical analysis methods are used to analyze the relationship between the frequency of occurrence and the order of occurrence of different events in various typical event sequences, establish an event chain probability model, associate micro-movement events with the characteristics of the current environment, and construct a micro-movement pattern library. For example, in an indoor environment with many obstacles, micro-displacement events are more frequent and the displacement distance is shorter.

[0074] Different micro-movement impact factors are set according to different types of micro-movement events. For example, for micro-turning events, the impact factor value range is set to 0.1-0.3, and for micro-displacement events, the impact factor value range is set to 0.05-0.15. The micro-movement stability coefficient is calculated based on the micro-movement impact factor. The micro-movement stability coefficient = macro-path stability coefficient × (1 + micro-movement impact factor adjustment value). The micro-movement path quality score is calculated based on the micro-movement stability coefficient. The micro-movement path quality score = macro-path score × micro-movement stability coefficient, where the macro-path score refers to the score of path communication quality in step S104.

[0075] S402, construct a multi-level path adjustment mechanism based on micro-movement events;

[0076] The multi-level path adjustment mechanism refers to adjusting the path in stages, divided into fine-tuning, mid-tuning, and macro-tuning levels. The fine-tuning level optimizes parameters; when a micro-movement event is detected, communication parameters are adjusted first. Transmit power is dynamically adjusted within ±3dB based on changes in signal quality. When signal strength decreases, transmit power is increased appropriately; when signal strength is strong, transmit power is decreased appropriately to save energy. Different signal qualities are suitable for different modulation and coding schemes. Channel selection parameters are fine-tuned; when the current channel is interfered with, the system can automatically switch to other channels with less interference. The mid-tuning level optimizes the path; when a sequence of micro-movement events indicates a trend change, relay node selection is optimized within the current path. For example, if a series of micro-movement events indicates that the device is moving in a certain direction and the signal quality of the current path is gradually decreasing, the system can select a more suitable relay node within the current path to improve communication quality and adjust data. Transmission timing avoids predicted quality low points. Based on communication quality predictions, the system can adjust the timing of data transmission to avoid data transmission during periods of poor signal quality. For example, if it is predicted that the signal quality will deteriorate within a few seconds, the system can schedule the data transmission task for another time period. The macro-tuning level refers to path switching. When micro-mobility analysis indicates that the current path cannot meet the requirements, a complete path switching is performed. If, after fine-tuning and mid-tuning, the communication quality still cannot meet the requirements, and micro-mobility analysis indicates that the signal quality of the current path will continue to deteriorate, the system will perform a complete path switching to another higher-quality path for communication.

[0077] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: the identification and hierarchical adjustment mechanism for micro-movement events avoids macro-path switching due to overreaction, reduces the overhead and risk of switching, realizes the transformation from macro-path management to micro-parameter adaptation, identifies and adapts to more subtle motion changes, and significantly reduces unnecessary path switching while maintaining path stability, providing an optimization paradigm for Bluetooth communication in highly dynamic environments.

[0078] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A communication path selection method based on multiple Bluetooth networks, characterized in that, include: S101: Collect spectrum data, add spatiotemporal labels to the collected spectrum data, generate spatiotemporal interference data units based on the collected spectrum data and spatiotemporal labels, and calculate the interference index at the current moment based on the spectrum data; S102, Collect historical data. The Bluetooth device identifies interference patterns based on the collected historical data through time patterns and spatial patterns. The time pattern includes identifying the periodicity of the interference index in the time dimension and the statistical characteristics of time periods. The spatial pattern includes identifying interference propagation hotspots based on a spatial attenuation model. Time features, spatial features, and dynamic features are extracted as interference features. Cluster the time period-interference index pairs and the region-interference index pairs, and construct an interference pattern feature table based on the cluster type and semantic labels; S103, Based on the interference mode feature table, a prediction model is used to predict the future interference index to obtain the predicted interference index. An interference risk score is generated based on the predicted interference index, and the channel is allocated using the interference risk score. S104. Optimize path communication quality based on predicted interference index. Calculate communication quality parameters by multiplying historical transmission efficiency by predicted interference index and then by device health. Introduce a path pre-switching mechanism. When the predicted increase in interference index of the current path in the next time period exceeds a threshold or the interference index of the backup path decreases, trigger pre-switching. Divide the communication quality parameters by the interference risk score to obtain path weights. Sort all candidate paths according to path weights to generate a spatiotemporally aware path table containing path identifiers, communication quality parameters, risk scores, and path weights. Communication path selection methods also include: S301, collect acceleration change data and real-time user behavior patterns, construct a behavior-acceleration mapping table, compare the features of the real-time collected acceleration change data with the typical acceleration features of the corresponding behaviors in the behavior-acceleration mapping table, and obtain a consistency index by calculating the similarity between them. The formula for calculating the behavior acceleration consistency index is: , where is the behavior acceleration consistency index, calculated based on cosine similarity, and its value range is [ [1,1] represents the real-time acquired acceleration feature vector, which contains acceleration feature values ​​in multiple dimensions. It is the typical acceleration feature vector corresponding to the behavior in the behavior-acceleration mapping table, which also contains typical acceleration feature values ​​in multiple dimensions. It is the feature value of the i-th dimension of the real-time acquired acceleration feature vector A, and the feature value of the i-th dimension of the typical acceleration feature vector. n is the number of dimensions of the acceleration feature vector. The acceleration change is classified according to the acceleration change rate. S302, predict the movement trajectory based on the current behavior pattern, and correct the predicted movement trajectory based on the rate of change of acceleration; S303 calculates the switching timing based on the moving speed and predicted trajectory, and switches the path according to the switching timing.

2. The communication path selection method based on multiple Bluetooth networks as described in claim 1, characterized in that, The interference index is calculated based on the collected spectrum data: in, The interference index, For the i-th signal strength, Let be the time occupied by the i-th signal, n be the total number of signals, and Tscan be the total scan time. As the noise baseline, This is the reference noise.

3. The communication path selection method based on multiple Bluetooth networks as described in claim 2, characterized in that, Optimizing path communication quality based on predicted interference indices and forming a spatiotemporally aware path table involves: 1) Replacing spectrum data with predicted interference indices to calculate the interference index; 2) Calculating the communication quality parameter as: historical transmission efficiency × predicted interference index × device health. Historical transmission efficiency reflects the success rate and speed of data transmission along the path over a past period, while device health measures the operational status and reliability of relevant communication equipment along the path; 3) Introducing a path pre-switching mechanism. Based on spatiotemporal pattern prediction, when a significant increase in the interference index of the current path is predicted in the next time period, the system triggers a path pre-switching operation; 4) Dividing the communication quality parameter by the interference risk score to obtain the path weight, and sorting all candidate paths according to the path weights to generate a spatiotemporally aware path table.

4. The communication path selection method based on multiple Bluetooth networks as described in claim 1, characterized in that, Also includes: S201: Collect motion data, calculate motion index using motion data, form mobile-spatiotemporal interference data unit based on interference data and motion data, identify motion trajectory based on mobile-spatiotemporal interference data unit, and construct motion trajectory-interference correlation model based on motion trajectory and interference data; S202, adjust the interference risk score according to the movement trajectory, generate different path selection strategies based on devices with different speeds, and calculate the weight of path selection based on communication quality parameters.

5. The communication path selection method based on multiple Bluetooth networks as described in claim 4, characterized in that, The formula for calculating the mobility index using motion data is: ,in, For the moving index, For the moving speed of the equipment, The rate of change of direction represents the magnitude of change in the direction of movement of the equipment per unit time.

6. The communication path selection method based on multiple Bluetooth networks as described in claim 5, characterized in that, The method for forming a mobile-spatiotemporal interference data unit based on interference data and motion data is as follows: collect the historical movement trajectory of the device and the interference data at the corresponding time points, and combine the motion data and interference data into a mobile-spatiotemporal interference data unit with a certain time interval as the time window.

7. The communication path selection method based on multiple Bluetooth networks as described in claim 6, characterized in that, The steps for adjusting the interference risk score based on the movement trajectory are as follows: Monitor the device's current speed and direction of movement in real time; obtain the corresponding movement speed-interference volatility correspondence table and movement direction-regional interference mapping relationship from a pre-built mobile-spatial interference pattern library based on the device's current speed and direction; and adjust the interference risk score according to the movement trajectory. ,in, The adjusted interference risk score, To demonstrate temporal regularity, the volatility of historical disturbance indices over time is used as a measure, which reflects the degree of regularity of the disturbances over time. For spatial stability, it is represented by the volatility of historical disturbance indices in the spatial dimension. This volatility is measured by standard deviation and reflects the degree of stability of disturbances in the spatial dimension. Equipment density variation represents the change in the number of devices within a given area. For the moving index, , and All are weighting coefficients, and .

8. The communication path selection method based on multiple Bluetooth networks as described in claim 1, characterized in that, Also includes: S401 identifies micro-movement events based on movement trajectory, movement speed, and angle; collects micro-movement event data to construct a micro-movement pattern library; extracts the features of the identified micro-movement events; identifies micro-movement influencing factors based on the micro-movement event features; calculates micro-movement stability coefficients based on the micro-movement influencing factors; and uses the micro-movement stability coefficients to predict the communication quality of the movement. S402, construct a multi-level path adjustment mechanism based on micro-movement events.

Images