System and method for realizing Bluetooth satellite flash dynamic calling based on machine learning

Abstract

The invention relates to the technical field of Internet of Things, in particular to a system and method for achieving Bluetooth satellite flash dynamic calling based on machine learning, and the system comprises a data collection module, a data processing module, a machine learning module, a connection control module and a storage module. The method has the beneficial effects that multiple communication indexes are comprehensively considered by using a machine learning model, a more intelligent and more accurate switching decision can be made, and the limitation of a traditional simple threshold value judgment mode is avoided; a switching strategy can be adaptively adjusted according to a real-time communication environment and an equipment state through the model, so that the stability and the efficiency of communication are improved; and the decision tree which is a simple machine learning model is adopted, so that the training speed is high, the implementation cost is low, and deployment in various gateway devices is easy.

Description

Technical Field

[0001] The present invention relates to the technical field of the Internet of Things, and specifically to a Bluetooth and SparkLink dynamic call system and method based on machine learning. Background Art

[0002] Currently, in the fields of the Internet of Things and short-range communication, both SparkLink and Bluetooth technologies have extensive applications. The Bluetooth technology is mature and has advantages such as low power consumption and low cost, and is widely used in various consumer electronic devices; the SparkLink technology has higher transmission rates, lower latency, and stronger anti-interference capabilities, and is suitable for scenarios with higher requirements for communication quality. If a device can combine Bluetooth and SparkLink and dynamically switch according to environmental perception, it will significantly improve communication performance, effectively reduce power consumption, and bring an excellent user experience.

[0003] However, there are still certain disadvantages and deficiencies in the current gateway system in terms of the dynamic switching between Bluetooth and SparkLink. For example, traditional gateway systems often use a mechanism of scoring key signal parameters for judgment, with insufficient comprehensive consideration of factors and difficult control of switching time. When the communication environment changes, the existing system cannot adaptively adjust the switching strategy, directly affecting communication efficiency. Summary of the Invention

[0004] The purpose of the present invention is to provide a Bluetooth and SparkLink dynamic call system and method based on machine learning to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A Bluetooth and SparkLink dynamic call system based on machine learning, the system includes:

[0006] A data acquisition module, which is used to periodically obtain the signal strength, transmission rate, and packet loss rate data of the target device, and send the data to the data processing module;

[0007] A data processing module, which filters and normalizes the data sent by the data acquisition module, uses the mean filter method to remove the noise in the signal strength data, and uses the Min-Max normalization method to normalize the feature values to the [0,1] interval, and sends the processed data to the machine learning module;

[0008] A machine learning module, which trains and optimizes the decision tree model using historical data, receives the data sent by the data processing module during operation, and outputs the best communication method;

[0009] A connection control module, which dynamically switches the SparkLink or Bluetooth connection according to the result output by the machine learning module to ensure that the communication between the gateway and the device remains in the best state;

[0010] A storage module for storing the collected historical data and the trained machine learning model. The historical data is used for the periodic update and optimization of the model, and the trained model is used for prediction during real-time operation.

[0011] Preferably, the data acquisition module collects the signal strength, transmission rate, and packet loss rate data of StarFlash and Bluetooth at a set time interval, and uses these data as features to input into the data processing module.

[0012] Preferably, the machine learning module adopts the CART decision tree model algorithm. By regularly extracting historical data from the storage module, it divides the data into a training set and a validation set, uses the training set to train the decision tree model, and uses the validation set to evaluate the trained model to improve the accuracy of the model.

[0013] Preferably, when the connection control module switches the communication mode, it is set to execute the switch only when the prediction results are the same for several consecutive times to avoid communication jitter caused by frequent switching.

[0014] Preferably, the system further includes a WEB management interaction interface based on the HTTP protocol, which supports users to manually configure and select Bluetooth or StarFlash connection, and view the information of available devices around. The WEB management interaction interface interacts with the data processing module and the connection control module to achieve the real-time effect of user configuration.

[0015] A method for dynamically calling Bluetooth and StarFlash based on machine learning is implemented using a system for dynamically calling Bluetooth and StarFlash based on machine learning. The method includes the following steps:

[0016] Data acquisition: Periodically obtain the signal strength, transmission rate, and packet loss rate data of the target device, and send the data to data processing.

[0017] Data processing: Filter and normalize the data sent by data acquisition. Use the mean filter method to remove the noise in the signal strength data, and use the Min-Max normalization method to normalize the feature values to the interval [0, 1], and send the processed data to machine learning.

[0018] Machine learning: Use historical data to train and optimize the decision tree model, receive the data sent by data processing during operation, and output the best communication mode.

[0019] Connection control: Dynamically switch the StarFlash or Bluetooth connection according to the result output by machine learning to ensure that the communication between the gateway and the device remains in the best state.

[0020] A storage unit for storing the collected historical data and the trained machine learning model. The historical data is used for the periodic update and optimization of the model, and the trained model is used for prediction during real-time operation.

[0021] Preferably, the data collection module collects the signal strength, transmission rate, and packet loss rate data of SparkLink and Bluetooth at a set time interval, and uses these data as features for data processing.

[0022] Preferably, the machine learning uses the CART decision tree model algorithm. By periodically extracting historical data from the storage module, dividing it into a training set and a validation set, training the decision tree model using the training set, and evaluating the trained model using the validation set, the accuracy of the model is improved.

[0023] Preferably, when the connection control switches the communication mode, it is set to execute the switch only when the prediction results are the same for several consecutive times, so as to avoid communication jitter caused by frequent switching.

[0024] Preferably, the method further includes a WEB management interaction interface based on the HTTP protocol, which supports users to manually configure and select Bluetooth or SparkLink connections, and view the information of available devices around. The WEB management interaction interface interacts with the data processing module and the connection control module to achieve the real-time effectiveness of user configurations.

[0025] Compared with the prior art, the beneficial effects of the present invention are:

[0026] The dynamic call system and method for Bluetooth and SparkLink based on machine learning proposed by the present invention can make more intelligent and accurate switching decisions by using a machine learning model to comprehensively consider various communication metrics, avoiding the limitations of traditional simple threshold judgment methods; the model can adaptively adjust the switching strategy according to the real-time communication environment and device status, improving the stability and efficiency of communication; using a simple machine learning model such as a decision tree, it has a fast training speed, low implementation cost, and is easy to deploy in various gateway devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a system block diagram of the present invention;

[0028] Figure 2 It is a method flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] In order to clearly and completely describe the objectives and technical solutions of the present invention and make its advantages more clearly understood, the following further details the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of them, and are only used to explain the embodiments of the present invention, not to limit the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0030] Embodiment 1. Please refer to Figure 1 , the present invention provides a technical solution: a Bluetooth and XingFlash dynamic call system based on machine learning. The system includes:

[0031] The data acquisition module periodically obtains data such as the signal strength, transmission rate, and packet loss rate of the target device and sends it to the data processing module; the data processing module filters and normalizes the acquired data; the machine learning module trains and optimizes the decision tree model using historical data, and can output the best communication method when inputting data during operation; the connection control module dynamically switches between XingFlash and Bluetooth connections according to the results output by the model; the storage module is used to store the acquired historical data and the trained machine learning model. The historical data can be used for the regular update and optimization of the model, and the trained model is used for prediction during real-time operation.

[0032] The hardware platform is an ARM processor, based on the debian system, and provides a WEB management interaction interface. This interface is developed using the HTTP protocol and is the connection window between users and internal programs. The web interface supports two configuration methods: manual switching and automatic switching; sqlite technology is used to implement data storage.

[0033] The web interface supports two configuration methods: manual selection of Bluetooth / XingFlash and automatic call. In the manual configuration interface, users can select the communication method according to their needs, and the information of available devices scanned around will be displayed on the web interface. Users can select and establish or disconnect a Bluetooth / XingFlash connection with a specific device by themselves. After the automatic call configuration, the XingFlash signal strength, Bluetooth signal strength, and device information within a preset range of the target device will be obtained, and one of XingFlash and Bluetooth will be accurately selected as the connection method through the machine learning model.

[0034] The data acquisition module collects data such as the signal strength, transmission rate, and packet loss rate of XingFlash and Bluetooth at set time intervals. These features will affect the switching decision and send them to the data processing module.

[0035] After the data processing module receives the data, it cleans, selects features, and normalizes the data. The mean filtering method is used to remove the noise in the signal strength data. The signal strength, transmission rate, and packet loss rate are selected as the main features, and the Min-Max normalization method is used to normalize the feature values to the interval [0, 1].

[0036] For the machine learning model module, considering the resource carrying capacity of the embedded device, the CART decision tree model algorithm, which is more suitable for real-time decision-making of embedded devices, is selected. Historical data is regularly extracted from the storage module and divided into a training set and a validation set. The extraction period depends on the application environment and data size. The training set is used to train the decision tree model, and the parameters of the model (such as the depth of the tree, the minimum number of samples in the leaf nodes, etc.) are adjusted to improve the accuracy of the model. The trained model is evaluated using the validation set. If the accuracy does not meet the requirements, the parameters are continued to be adjusted for training.

[0037] The data processing module inputs the real-time processed data into the trained decision tree model. The model outputs the communication method to be used. Considering the switching delay and avoiding the jitter caused by frequent switching, it is set to switch only when the prediction results are the same for several consecutive times. The connection control module controls the connection switching of the gateway according to the output result of the model to ensure that the communication between the gateway and the device always remains in the best state

[0038] Example 2, referring to the appendix Figure 2 On the basis of Example 1, a method for dynamically calling Bluetooth and XingFlash based on machine learning is proposed. The method includes the following steps:

[0039] Data collection: Periodically obtain the signal strength, transmission rate, and packet loss rate data of the target device, and send the data to data processing; the data collection collects the signal strength, transmission rate, and packet loss rate data of XingFlash and Bluetooth at a set time interval, and uses these data as features to input to data processing.

[0040] Data processing: Filter and normalize the data sent by data collection. The mean filtering method is used to remove the noise in the signal strength data, and the Min-Max normalization method is used to normalize the feature values to the interval [0, 1], and the processed data is sent to machine learning.

[0041] Machine learning: Use historical data to train and optimize the decision tree model, receive the data sent by data processing during operation, and output the best communication method; the machine learning adopts the CART decision tree model algorithm. By regularly extracting historical data from the storage module, dividing it into a training set and a validation set, using the training set to train the decision tree model, and using the validation set to evaluate the trained model, the accuracy of the model is improved.

[0042] Connection control, based on the results output by machine learning, dynamically switches between SparkLink or Bluetooth connections to ensure that the communication between the gateway and the device remains in the best state; when the connection control switches the communication method, it is set to execute the switch only when the prediction results are the same for several consecutive times to avoid communication jitter caused by frequent switching.

[0043] Storage, used to store the collected historical data and the trained machine learning model. The historical data is used for the periodic update and optimization of the model, and the trained model is used for prediction during real-time operation.

[0044] The method also includes a WEB management interface based on the HTTP protocol, which supports users to manually configure and select Bluetooth or SparkLink connections, and view the information of available devices around. The WEB management interface interacts with the data processing module and the connection control module to achieve the real-time effectiveness of user configurations.

[0045] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A Bluetooth Star Flash dynamic calling system based on machine learning, characterized by: The system comprises: A data acquisition module is used to periodically obtain the signal strength, transmission rate, and packet loss rate data of the target device and send the data to the data processing module; The data processing module filters and normalizes the data sent by the data acquisition module, uses the mean filtering method to remove the noise in the signal strength data, uses the Min-Max normalization method to normalize the eigenvalues ​​to the [0,1] interval, and sends the processed data to the machine learning module; The machine learning module uses historical data to train and optimize the decision tree model, receives data sent by the data processing module during runtime, and outputs the best communication method; The connection control module dynamically switches between StarFlash and Bluetooth connections based on the results output by the machine learning module to ensure that the communication between the gateway and the device remains in the best state; The storage module is used to store the collected historical data and the trained machine learning models. The historical data is used for regular updating and optimization of the models, and the trained models are used for predictions during real-time runtime.

2. According to claim 1, a system for implementing dynamic calling of Bluetooth Star Flash based on machine learning is characterized in that: The data acquisition module collects data on the signal strength, transmission rate, and packet loss rate of star flash and Bluetooth at set time intervals, and inputs these data as features into the data processing module.

3. According to claim 1, a system for implementing dynamic calling of Bluetooth Star Flash based on machine learning is characterized in that: The machine learning module adopts the CART decision tree model algorithm, which extracts historical data from the storage module regularly, divides it into a training set and a validation set, uses the training set to train the decision tree model, and uses the validation set to evaluate the trained model to improve the accuracy of the model.

4. According to claim 1, a system for implementing dynamic calling of Bluetooth Star Flash based on machine learning is characterized in that: When switching the communication mode, the connection control module sets the switching to be performed only when the prediction results are the same for several consecutive times, so as to avoid communication jitter caused by frequent switching.

5. According to claim 1, a system for implementing dynamic calling of Bluetooth Star Flash based on machine learning is characterized in that: The system also includes a WEB management interaction interface based on the HTTP protocol, which supports users to manually configure and select Bluetooth or StarFlash connection, and view the available device information around. The WEB management interaction interface interacts with the data processing module and the connection control module to realize the real-time effectiveness of user configuration.

6. A method for implementing dynamic calling of Bluetooth Star Flash based on machine learning, which is implemented by using a system for implementing dynamic calling of Bluetooth Star Flash based on machine learning as described in any one of claims 1 to 5, characterized in that: The method comprises the following steps: Data collection, periodically obtaining the signal strength, transmission rate, and packet loss rate data of the target device, and sending the data to data processing; Data processing: filtering and normalizing the data sent by data acquisition, using mean filtering to remove noise from signal strength data, using Min-Max normalization to normalize the eigenvalues ​​to the [0,1] interval, and sending the processed data to machine learning; Machine learning uses historical data to train and optimize the decision tree model, receives data sent by data processing at runtime, and outputs the best communication method; Connection control: dynamically switch between StarFlash and Bluetooth connections based on the results of machine learning output to ensure that the communication between the gateway and the device remains in the best state; Storage is used to store collected historical data and trained machine learning models. Historical data is used for regular updates and optimization of models, and trained models are used for predictions during real-time runtime.

7. The method for implementing dynamic calling of Bluetooth Star Flash based on machine learning according to claim 6, characterized in that: Data collection collects the signal strength, transmission rate, and packet loss rate data of star flash and Bluetooth at set time intervals, and inputs these data into data processing as features.

8. The method for implementing Bluetooth Star Flash dynamic calling based on machine learning according to claim 6, characterized in that: Machine learning adopts the CART decision tree model algorithm, which extracts historical data from the storage module regularly, divides it into training set and validation set, uses the training set to train the decision tree model, and uses the validation set to evaluate the trained model to improve the accuracy of the model.

9. The method for implementing Bluetooth Star Flash dynamic calling based on machine learning according to claim 6, characterized in that: When switching the communication mode, the connection control is set to perform the switch only when the prediction results are the same for several consecutive times to avoid communication jitter caused by frequent switching.

10. The method for implementing Bluetooth Star Flash dynamic calling based on machine learning according to claim 6, characterized in that: The method also includes a WEB management interaction interface based on the HTTP protocol, which supports users to manually configure and select Bluetooth or StarFlash connection, and view the available device information around. The WEB management interaction interface interacts with the data processing module and the connection control module to realize the real-time effectiveness of user configuration.

Images