A communication method based on the Internet of Things

Abstract

The present application discloses a communication method based on the Internet of Things, which includes: receiving a data set continuously sampled by a terminal within a preset time window, classifying the data set according to data type, and generating a number of encapsulated data packets; based on each encapsulated data packet and the corresponding standard data baseline, using a window data offset algorithm to obtain a window deviation index of the corresponding data type; based on each data type, generating a collaborative evaluation feature vector, inputting it into a collaborative deviation evaluation model, and outputting a prediction deviation index vector of the associated data type; using a preset associated confidence deviation algorithm to obtain the transmission value of the data type; based on the transmission value of each data type, dynamically adjusting the sampling time window size corresponding to the data type, and periodically transmitting the data value sequence of the data type to the cloud platform. In this way, the efficiency and reliability of data transmission in Internet of Things communication are improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a communication method based on the Internet of Things. Background Art

[0002] In today's rapidly developing Internet of Things (IoT) era, IoT systems are widely used in many fields such as industrial production and smart homes due to their powerful connectivity and data interaction capabilities. A typical IoT system mainly consists of IoT terminals, IoT cloud control gateways, and IoT cloud platforms. The three work together to realize data collection, transmission and processing, thereby achieving intelligent management and monitoring of various devices and scenarios.

[0003] IoT terminals, as data sources, come in a wide variety, encompassing various information sensors, smart devices, and more. Deployed in diverse environments and scenarios, they collect real-time data that requires monitoring, connection, and interaction. For example, in industrial production scenarios, sensors monitoring equipment operation within the factory continuously collect key data such as operating temperature, power, and pressure. The IoT cloud control gateway plays a crucial role as a bridge between terminals and the cloud platform. On the one hand, it communicates with numerous IoT terminals, collects data collected by the terminals, and performs preliminary processing and integration on this data. On the other hand, the gateway is responsible for reliably transmitting the processed data to the IoT cloud platform, receiving control commands issued by the cloud platform, and forwarding them to the corresponding terminal devices, enabling remote control of the terminals.

[0004] However, in existing IoT communication processes, gateways typically use a one-size-fits-all approach when transmitting terminal data to cloud platforms, without differentiated processing based on the importance and urgency of the data. This makes it difficult for cloud platforms to quickly filter out key information when receiving large amounts of data, increasing the complexity and time cost of data processing and affecting the communication efficiency of the entire IoT system. Traditional solutions rely primarily on the instantaneous threshold of a single window to determine the transmission value of data, lacking long-term behavioral modeling and unable to achieve efficient and accurate IoT communication monitoring. Summary of the Invention

[0005] The present application improves the efficiency and reliability of data transmission in Internet of Things communications by providing a communication method based on the Internet of Things.

[0006] This application provides a communication method based on the Internet of Things, including:

[0007] S101, receiving a data set continuously sampled by a terminal within a preset time window, classifying the data set according to data type, and generating a number of encapsulated data packets, each of which includes a unique number corresponding to the data type, a data value sequence, and sampling attributes;

[0008] S102, based on each encapsulated data packet and the corresponding standard data baseline, a preset window data offset algorithm is used to obtain a window deviation index of the corresponding data type;

[0009] S103, based on each data type, generating a collaborative evaluation feature vector, inputting it into a pre-trained collaborative deviation evaluation model, and outputting a prediction deviation index vector of the associated data type associated with the data type;

[0010] S104, according to the window deviation index of each data type and the prediction deviation index vector of its associated data type, using a preset associated confidence deviation algorithm, to obtain the transmission value of the data type;

[0011] S105 , dynamically adjusting the sampling time window size corresponding to each data type according to the transmission value of each data type, and periodically transmitting the data value sequence of the data type to the cloud platform.

[0012] Preferably, the encapsulated data packet is represented as , i is the number of the data type, is the encapsulated data packet of the i-th data type, is the data value sequence of the i-th data type, , is the data value of the nth sampling point in the data value sequence of the i-th data type, Sampling attributes of the i-th data type; sampling attributes include time attributes, environment attributes, and sampling time window size.

[0013] Preferably, the preset window data offset algorithm is:

[0014]

[0015] in, is the window deviation index of the corresponding data type, is the mean of the data value sequence, is the standard deviation of the sequence of data values, 、 are the mean and standard deviation of the standard data baseline, is the behavior drift index, 、 are the preset weight factors, , used to indicate the degree of influence of standard deviation and behavior drift on the window deviation index.

[0016] Preferably, the behavior drift index is determined based on a sequence of data values ​​in multiple time windows within a preset time period and is calculated according to the following formula:

[0017]

[0018] is the behavior drift index, is the behavior drift value of data type i within the preset time period, K is the number of time windows within the preset time period, , is the data feature vector of the kth time window, a is the data mean, b is the data variance, c is the data kurtosis difference, is the data feature vector of the k-1th time window, express and The Euclidean distance between The default standard behavior drift value for data type i is determined based on expert experience and historical data.

[0019] Preferably, the collaborative evaluation feature vector is expressed as , i is the number of the data type, is the window deviation index of the data type, is the sampling attribute of the data type, is the number of the associated data type of data type i, and the prediction deviation index vector of the associated data type is expressed as , For associated data types The forecast deviation index.

[0020] Preferably, the preset association confidence deviation algorithm specifically includes:

[0021] S201, calculating a collaborative deviation confidence vector based on the predicted deviation index vector of the associated data type and the actual deviation index vector of the associated data type;

[0022] S202: extracting the correlation factor between the data type and the associated data type from the pre-constructed data correlation factor matrix, and combining it with the collaborative deviation confidence vector to obtain a collaborative deviation confidence index: based on each collaborative deviation confidence in the collaborative deviation confidence vector, using the corresponding correlation factor as the weight value of the collaborative deviation confidence, performing weighted summation to obtain the collaborative deviation confidence index;

[0023] S203: Calculate the transmission value of the data type based on the window deviation index and the collaborative deviation confidence index of the data type.

[0024] Preferably, the transmission value of the data type is calculated according to the following formula:

[0025]

[0026] in, is the transfer value of data type i, The window deviation index of this data type, is the co-deviation confidence index of the data type, It is a pre-set adjustment factor used to adjust the impact of the collaborative deviation confidence index on the exposure deviation index in the transmission value. It is set based on expert experience and historical data and is between 0 and 1.

[0027] Preferably, the S105 specifically includes:

[0028] For each data type, if the transmission value is greater than the preset value threshold, the current time window size is dynamically adjusted using the transmission value, and the collection period for the data type is updated:

[0029]

[0030] in, is the updated sampling time window size, is the original time window size, is the transfer value of data type i, is the preset adjustment coefficient, which is 0.5;

[0031] According to the latest sampling time window size, the data value sequence of the data type is periodically transmitted to the cloud platform until the transmission value of the data type is less than the preset value threshold, and the sampling time window is restored to its initial size.

[0032] Preferably, the method further comprises:

[0033] Based on the updated time window size, steps S101 to S105 are periodically executed to transmit the received data value sequences of multiple data types and their corresponding transmission values ​​and data sources to the IoT cloud platform to realize IoT-based communication.

[0034] Preferably, the collaborative deviation assessment model is obtained in the following manner:

[0035] A1. Collect a large number of historical collaborative evaluation feature vectors of different data types, as well as the actual window deviation index of the corresponding associated data types;

[0036] A2. Label each historical collaborative evaluation feature vector with an actual deviation index vector consisting of the actual window deviation index of the corresponding associated data type.

[0037] A3. Use the labeled historical collaborative evaluation feature vectors as the training sample set, train the pre-selected neural network structure, and continuously optimize the model parameters to obtain the final collaborative deviation evaluation model.

[0038] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0039] The data collected by IoT terminals are classified and packaged by type, and the sampling attributes are recorded in detail to lay the foundation for accurate processing and avoid data confusion; the window deviation index is calculated through the window data offset algorithm, and the behavior drift index is introduced to realize the dual mechanism of real-time deviation detection and long-term behavior drift analysis, effectively capturing chronic attacks, making up for the shortcomings of traditional reliance on instantaneous threshold judgment, and improving the comprehensiveness and reliability of data anomaly detection; using the collaborative deviation evaluation model, the deviation index of the associated data type is predicted based on the collaborative evaluation feature vector, the potential correlation between data is mined, and the value detection dimension is enriched; with the help of the associated confidence deviation algorithm, the collaborative deviation confidence vector and data are combined to form a comprehensive and reliable data anomaly detection model. The correlation factor matrix calculates the transmission value, comprehensively considers the data's own deviations and correlations, accurately evaluates the data value, and provides a scientific basis for dynamic transmission strategies; dynamically adjusts the sampling time window size according to the transmission value, compresses the window for high-value data to strengthen monitoring, and expands the window for low-value data to reduce transmission overhead, thereby achieving reasonable resource allocation; finally, the entire process is executed periodically to continuously process and transmit data to ensure continuous optimized operation of the system, and transmits information such as multiple types of data and transmission value to the cloud platform, providing comprehensive support for cloud platform monitoring and analysis, and overall improving the security of IoT communications, data processing accuracy, transmission efficiency, and system flexibility and real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 The figure is a flow chart of a communication method based on the Internet of Things according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] To facilitate understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0042] It should be noted that the terms “vertical”, “horizontal”, “up”, “down”, “left”, “right” and similar expressions used in this document are for illustrative purposes only and do not represent the only implementation method.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains; the terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0044] Example 1: Figure 1 4 is a flow chart of a communication method based on the Internet of Things according to an embodiment of the present invention.

[0045] Specifically, the IoT terminal can be various devices such as various information sensors, which collect any data information that needs to be monitored, connected, and interacted in real time (for example, operation monitoring of equipment in a factory, operating temperature, power, etc.), communicate through the IoT communication node built into the IoT cloud control gateway, and then upload the collected information to the IoT cloud platform for monitoring and storage through the IoT cloud control gateway, thereby realizing IoT communication.

[0046] like Figure 1 As shown, a communication method based on the Internet of Things is applied to the Internet of Things cloud control gateway, including the following steps:

[0047] S101, receiving a data set continuously sampled by a terminal within a preset time window (which can also be understood as a data collection period), classifying the data set according to data type, and generating a number of encapsulated data packets, each of which includes a unique number of the corresponding data type, a data value sequence, and sampling attributes.

[0048] Among them, the encapsulated data packet is represented as , i is the number of the data type, is the encapsulated data packet of the i-th data type, is the data value sequence of the i-th data type, , is the data value of the nth sampling point in the data value sequence of the i-th data type, The sampling attributes of the i-th data type. For example, the data type includes but is not limited to voltage, power, pressure value, etc.

[0049] Specifically, the sampling attributes include time attributes (set to the time period type to which the current time window belongs, that is, the spatiotemporal characteristics, for encoding representation, for example, the month-date type (working day / rest day)-time period type (peak period / off-peak period / off-peak period), etc. of the terminal operation, which can be set according to the specific situation to reflect the operating characteristics of the terminal in different sampling time windows. Generally, it has regular operating characteristics, and the operating data corresponding to different time attributes are quite different), environmental attributes (can be set to the normalized values ​​of environmental weather parameters such as temperature and humidity, or can be set according to the actual application scenario, as long as it can reflect the scene characteristics, for example, the working scene attributes of a factory) and the sampling time window size.

[0050] It should be noted that the preset time window is the initial value of the time window (such as 10 minutes). As the basic collection period, data classification is automatically completed based on the metadata (sensor ID + data type label) reported by the terminal. The timestamp and environmental sensor reading are automatically attached during packaging to ensure the accuracy of the attributes. This invention will not go into details about this.

[0051] S102 , based on each encapsulated data packet and the standard data baseline, a window deviation index of the corresponding data type is obtained through a preset window data offset algorithm.

[0052] Specifically, the preset window data offset algorithm is:

[0053]

[0054] in, is the window deviation index of the corresponding data type, is the mean of the data value sequence, is the standard deviation of the sequence of data values, 、 are the mean and standard deviation of the standard data baseline, is the behavior drift index, 、 are the preset weight factors, , used to indicate the degree of influence of standard deviation and behavior drift on the window deviation index, and is set according to the actual scenario and requirements, for example, is 0.6, is 0.4.

[0055] It should be noted that the standard data baseline is set up by collecting a large number of historical encapsulated data packets sampled by terminals historically. These historical encapsulated data packets must correspond to standard sampling data under normal operating conditions (data without security risks; for example, abnormal temperatures collected by a factory sensor cannot be determined as standard sampling data). Based on the sampling attributes, all historical encapsulated data packets with the same sampling attributes as the current encapsulated data packet are obtained as standard encapsulated data packets (the similarity between the sampling attributes can be calculated, and historical encapsulated data packets with similarities greater than 0.9 are considered standard encapsulated data packets). The mean of the data value sequence of all standard encapsulated data packets is calculated as the standard data baseline. It is understandable that historical encapsulated data packets with a similarity greater than 0.9 with the current sampling attributes of the data type are filtered from the historical database.

[0056] The behavior drift index is determined based on the data value sequence in multiple time windows within a preset time period (the preset time period is set based on actual conditions and needs to be larger than the time window, that is, the preset time period includes the current time window and multiple historical time windows). It is calculated according to the following formula:

[0057]

[0058] is the behavior drift index, is the behavior drift value of data type i within the preset time period, K is the number of time windows within the preset time period (for example, the past hour contains 12 5-minute time windows), and also represents the number of the current time window. , is the data feature vector of the kth time window, a is the data mean, b is the data variance, c is the data kurtosis difference, is the data feature vector of the k-1th time window, express and The Euclidean distance between The preset standard behavior drift value for data type i is determined based on expert experience and historical data. The historical behavior drift value is calculated based on the standard historical data with the same sampling attributes of the data type (data sampled under normal operation, i.e., normal data, without abnormal risks) as the standard behavior drift value.

[0059] It should be noted that due to the reliance on the instantaneous threshold judgment of the current single window and the lack of long-term behavioral modeling, a drift index is determined by combining behavioral standard drift values ​​to prevent hackers from bypassing threshold detection through long-term fine-tuning of data (such as drifting 0.1°C per day). Traditional solutions cannot capture this type of "boiling frog in warm water" attack. Therefore, the reference to the behavioral drift index is to avoid the hidden dangers of chronic attacks. The drift index captures chronic attacks and can more comprehensively evaluate the value of sampled data.

[0060] In summary, the window data offset algorithm, based on each data type, performs a dual detection mechanism by combining the encapsulated data packets in the current time window with the preset standard data baseline and standard behavior drift value, and performs real-time deviation detection + long-term behavior drift analysis to obtain the window deviation index of the corresponding data type.

[0061] S103: Generate a collaborative evaluation feature vector based on each data type, input it into a pre-trained collaborative deviation evaluation model, and output a prediction deviation index vector of the associated data type associated with the data type.

[0062] Specifically, the collaborative evaluation feature vector , i is the number of the data type, is the window deviation index of the data type, is the sampling attribute of the data type, is the number of the associated data type of data type i (select the data type that has physical association or fluctuation coordination with data type i, determined based on expert experience and historical data). Since the number may not be unique, it is marked with m; the prediction deviation index vector of the associated data type is expressed as , For associated data types The forecast deviation index.

[0063] In some embodiments, the collaborative deviation assessment model is obtained by:

[0064] A1. Collect a large number of historical collaborative evaluation feature vectors of different data types, as well as the actual window deviation index of the corresponding associated data types;

[0065] A2. Label each historical collaborative evaluation feature vector with the actual deviation index vector consisting of the actual window deviation index of the corresponding associated data type.

[0066] A3. Use the labeled historical collaborative evaluation feature vectors as the training sample set, train the pre-selected neural network structure, and continuously optimize the model parameters to obtain the final collaborative deviation evaluation model.

[0067] Therefore, the collaborative deviation evaluation model is used to learn based on the system evaluation feature vector of the input data type, and output the deviation index of other related data types with the maximum probability under the deviation index condition based on the data type.

[0068] S104: Based on the window deviation index of each data type and the prediction deviation index vector of its associated data type, a preset associated confidence deviation algorithm is used to obtain the transmission value of the data type.

[0069] In some embodiments, the preset association confidence deviation algorithm specifically includes:

[0070] S201, based on the predicted deviation index vector of the associated data type and combined with the actual deviation index vector of the associated data type (the window deviation index of each associated data type calculated within the current time window, refer to the calculation method in step S102, and the actual deviation indices of all associated data types are combined into an actual deviation index vector), calculate the collaborative deviation confidence vector.

[0071] Specifically, the collaborative deviation confidence vector is obtained by calculating the similarity value (ranging from 0 to 1) between the predicted deviation index and the actual deviation index of each associated data type, and forming the collaborative deviation confidence vector with the similarity values ​​of all associated data types. , For associated data types The similarity value between the predicted deviation index and the actual deviation index, i.e., the collaborative deviation confidence, reflects the accuracy of the deviation index of the associated data type predicted based on the deviation index of the data type.

[0072] It should be noted that the collaborative deviation confidence vector is used to reflect the collaborative credibility between the data type and all its associated data types.

[0073] S202: extracting the correlation factor between the data type and the associated data type from a pre-constructed data correlation factor matrix, and combining it with the collaborative deviation confidence vector to obtain a collaborative deviation confidence index.

[0074] Among them, the pre-constructed data correlation factor matrix M= , It is the correlation factor between data type i and data type j, that is, the physical correlation strength between the two, which is set according to expert experience. For example, when the voltage value collected by the sensor in a factory fluctuates, the corresponding power value will also fluctuate to a certain extent. For another example, when the temperature of the equipment changes abnormally, the corresponding operating parameters may also change to a certain extent. The corresponding correlation strengths are all set based on experience. The greater the physical correlation strength, the greater the correlation factor. This invention will not elaborate on this.

[0075] Furthermore, the collaborative deviation confidence vector is corrected based on the correlation factor between the associated data type and the data type. That is, the larger the correlation factor is, the more important and credible the deviation assessment between the data type and its associated data type is. It can be understood that the correlation factor can be regarded as a weight value.

[0076] Specifically, step S202 includes: based on each collaborative deviation confidence in the collaborative deviation confidence vector, using the corresponding correlation factor as the weight value of the collaborative deviation confidence, performing weighted summation, and obtaining a collaborative deviation confidence index.

[0077] As an example, the collaborative bias confidence vector is expressed as , Represents data type i and its associated data type The confidence level of collaborative deviation between the two data types is obtained by extracting the correlation factor w between the data type and the related data type from the pre-built data correlation factor matrix. , For data type i and associated data type The correlation factor between the two data types is calculated, and w is used as the weight value to perform a weighted summation of the collaborative deviation confidence vectors to obtain the collaborative deviation confidence index. It can be understood that if the correlation factor is close to 0, it means that there is no significant fluctuation synergy between the two data types. Therefore, multiplying the correlation factor with the corresponding collaborative deviation confidence and thereby reducing the evaluation status of the collaborative deviation confidence in the collaborative deviation confidence index can avoid the accidental high similarity between the deviation index predicted by the model and the actual deviation index.

[0078] Specifically, the collaborative deviation confidence index is calculated according to the following formula:

[0079]

[0080] in, is the co-deviation confidence index of data type i, m is the number of associated data types of data type i, is the correlation factor between data type i and the kth associated data type, The k-th associated data type co-deviation confidence.

[0081] S203: Calculate the transmission value of the data type based on the window deviation index and the collaborative deviation confidence index of the data type.

[0082] Specifically, the transmission value of this data type is calculated according to the following formula:

[0083]

[0084] in, is the transfer value of data type i, The window deviation index of this data type, is the co-deviation confidence index of the data type, It is a pre-set adjustment factor used to adjust the impact of the collaborative deviation confidence index on the exposure deviation index in the transmission value. It is set based on expert experience and historical data, between 0 and 1, and the default is 0.5.

[0085] S105, dynamically adjust the sampling time window size (sampling period) corresponding to the data type according to the transmission value of each data type, and periodically transmit the data value sequence of the data type to the cloud platform according to the latest sampling time window size, until the transmission value of the data type is less than the preset value threshold, and restore the sampling time window to its initial size.

[0086] Specifically, step S105 includes:

[0087] For each data type, if the transmission value is greater than the preset value threshold (set based on actual conditions and historical data to measure whether there are any anomalies in the data value sequence of the data type. If so, the sampling window size needs to be updated to improve the cloud platform's monitoring of the data type), the transmission value is used to dynamically adjust the current time window size and update the collection period for the data type:

[0088]

[0089] in, is the updated sampling time window size, is the original time window size, is the transfer value of data type i, It is a preset adjustment coefficient with a default value of 0.5. It can also be adjusted according to actual dynamic adjustment needs. Therefore, the greater the transmission value, the smaller the corresponding time window, which achieves a sharp reduction in the window size of high-value data, compresses the window of high-value data, and expands the window of low-value data.

[0090] S106, based on the updated time window size, periodically execute steps S101 to S105, and periodically transmit the received data value sequence of multiple data types and their corresponding transmission value, data source (terminal number or sensor ID) to the IoT cloud platform according to the time window size to realize IoT-based communication.

[0091] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0092] Based on each encapsulated data packet and the standard data baseline, a preset window data offset algorithm is used to determine the window deviation index for the corresponding data type. This algorithm combines the mean and standard deviation of the data value sequence, the mean and standard deviation of the standard data baseline, and the behavioral drift index, comprehensively considering the current state and long-term behavior of the data. By introducing the behavioral drift index, the limitation of traditional solutions that rely solely on instantaneous threshold judgment is avoided, and it can effectively capture chronic attacks, such as hackers who fine-tune data over a long period of time. Through the dual detection mechanism (real-time deviation detection + long-term behavioral drift analysis), the value of sampled data can be more comprehensively evaluated.

[0093] Based on each data type, a collaborative evaluation feature vector is generated and input into a pre-trained collaborative deviation evaluation model. The predicted deviation index vector of the associated data types associated with the data type is output. Taking into account the association between different data types, the collaborative deviation evaluation model can predict the deviation index of other associated data types based on the deviation index of one data type, and explore the potential connections between data. This further enriches the dimensions of data value detection, improves the understanding and processing capabilities of complex data relationships, and helps to more comprehensively evaluate the operating status of the entire IoT system.

[0094] Based on the window deviation index of each data type and the prediction deviation index vector of its associated data type, the transmission value of the data type is obtained using the preset associated confidence deviation algorithm. The algorithm finally obtains the transmission value by calculating the collaborative deviation confidence vector and the collaborative deviation confidence index. It comprehensively considers the deviation of the data itself and its relationship with other associated data types. Through the collaborative deviation confidence vector and the association factor matrix, the transmission value of the data is accurately evaluated. The association factor is used as a weight value to reasonably adjust the impact of different associated data types on the transmission value, providing a scientific basis for the dynamic transmission strategy of data. It can reasonably allocate transmission resources according to the actual value of the data, improve the efficiency and pertinence of data transmission, and solve the problem that traditional solutions cannot perform differentiated transmission according to the actual value of data, avoiding resource waste and untimely transmission of important data.

[0095] It realizes dynamic adjustment of transmission strategy according to data value, compresses the window for high-value data to improve the cloud platform's monitoring of this data type; expands the window for low-value data to reduce unnecessary transmission. This dynamic adjustment mechanism can adapt to changes in different data types and improve the flexibility and efficiency of the system.

[0096] In summary, the data collected by IoT terminals are classified and packaged by type, and the sampling attributes are recorded in detail to lay the foundation for accurate processing and avoid data confusion; the window deviation index is calculated through the window data offset algorithm, and the behavior drift index is introduced to realize the dual mechanism of real-time deviation detection and long-term behavior drift analysis, effectively capturing chronic attacks, making up for the shortcomings of traditional reliance on instantaneous threshold judgment, and improving the comprehensiveness and reliability of data anomaly detection; using the collaborative deviation evaluation model, the deviation index of the associated data type is predicted based on the collaborative evaluation feature vector, the potential correlation between data is mined, and the value detection dimension is enriched; with the help of the associated confidence deviation algorithm, combined with the collaborative deviation confidence vector and The data correlation factor matrix calculates the transmission value, comprehensively considers the data's own deviations and correlations, accurately evaluates the data value, and provides a scientific basis for dynamic transmission strategies; dynamically adjusts the sampling time window size according to the transmission value, compresses the window for high-value data to strengthen monitoring, and expands the window for low-value data to reduce transmission overhead, thereby achieving reasonable resource allocation; finally, the entire process is executed periodically to continuously process and transmit data to ensure continuous optimized operation of the system, and transmits information such as multiple types of data and transmission value to the cloud platform, providing comprehensive support for cloud platform monitoring and analysis, and overall improving the security of IoT communications, data processing accuracy, transmission efficiency, and system flexibility and real-time performance.

[0097] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Various modifications and variations are readily apparent to those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A communication method based on the Internet of Things, characterized in that: include: S101, receiving a data set continuously sampled by a terminal within a preset time window, classifying the data set according to data type, and generating a number of encapsulated data packets, each of which includes a unique number corresponding to the data type, a data value sequence, and sampling attributes; S102, based on each encapsulated data packet and the corresponding standard data baseline, a preset window data offset algorithm is used to obtain a window deviation index of the corresponding data type; S103, based on each data type, generating a collaborative evaluation feature vector, inputting it into a pre-trained collaborative deviation evaluation model, and outputting a prediction deviation index vector of the associated data type associated with the data type; S104, according to the window deviation index of each data type and the prediction deviation index vector of its associated data type, using a preset associated confidence deviation algorithm, to obtain the transmission value of the data type; S105 , dynamically adjusting the sampling time window size corresponding to each data type according to the transmission value of each data type, and periodically transmitting the data value sequence of the data type to the cloud platform.

2. The communication method based on the Internet of Things according to claim 1, wherein: The encapsulated data packet is represented as , i is the number of the data type, is the encapsulated data packet of the i-th data type, is the data value sequence of the i-th data type, , is the data value of the nth sampling point in the data value sequence of the i-th data type, Sampling attributes of the i-th data type; sampling attributes include time attributes, environment attributes, and sampling time window size.

3. The communication method based on the Internet of Things according to claim 2, wherein: The preset window data offset algorithm is: in, is the window deviation index of the corresponding data type, is the mean of the data value sequence, is the standard deviation of the sequence of data values, 、 are the mean and standard deviation of the standard data baseline, is the behavior drift index, 、 are the preset weight factors, , used to indicate the degree of influence of standard deviation and behavior drift on the window deviation index.

4. The communication method based on the Internet of Things according to claim 3, wherein: The behavior drift index is determined based on a sequence of data values ​​within multiple time windows within a preset time period and is calculated using the following formula: is the behavior drift index, is the behavior drift value of data type i within the preset time period, K is the number of time windows within the preset time period, , is the data feature vector of the kth time window, a is the data mean, b is the data variance, c is the data kurtosis difference, is the data feature vector of the k-1th time window, express and The Euclidean distance between The default standard behavior drift value for data type i is determined based on expert experience and historical data.

5. The communication method based on the Internet of Things according to claim 4, characterized in that: The collaborative evaluation feature vector is expressed as , i is the number of the data type, is the window deviation index of the data type, is the sampling attribute of the data type, is the number of the associated data type of data type i, and the prediction deviation index vector of the associated data type is expressed as , For associated data types The forecast deviation index.

6. The communication method based on the Internet of Things according to claim 4, characterized in that: The preset association confidence deviation algorithm specifically includes: S201, calculating a collaborative deviation confidence vector based on the predicted deviation index vector of the associated data type and the actual deviation index vector of the associated data type; S202: extracting the correlation factor between the data type and the associated data type from the pre-constructed data correlation factor matrix, and combining it with the collaborative deviation confidence vector to obtain a collaborative deviation confidence index: based on each collaborative deviation confidence in the collaborative deviation confidence vector, using the corresponding correlation factor as the weight value of the collaborative deviation confidence, performing weighted summation to obtain the collaborative deviation confidence index; S203: Calculate the transmission value of the data type based on the window deviation index and the collaborative deviation confidence index of the data type.

7. The communication method based on the Internet of Things according to claim 5, characterized in that: The transmission value of the data type is calculated according to the following formula: in, is the transfer value of data type i, The window deviation index of this data type, is the co-deviation confidence index of the data type, It is a pre-set adjustment factor used to adjust the influence of the collaborative deviation confidence index on the window deviation index in the transmission value. It is set based on expert experience and historical data and is between 0 and 1.

8. The communication method based on the Internet of Things according to claim 1, wherein: The S105 specifically includes: For each data type, if the transmission value is greater than the preset value threshold, the current time window size is dynamically adjusted using its transmission value, and the collection period for that data type is updated: in, is the updated sampling time window size, is the original time window size, is the transfer value of data type i, is the preset adjustment coefficient, which is 0.5; According to the latest sampling time window size, the data value sequence of the data type is periodically transmitted to the cloud platform until the transmission value of the data type is less than the preset value threshold, and the sampling time window is restored to its initial size.

9. The communication method based on the Internet of Things according to claim 8, wherein: The method further comprises: Based on the updated time window size, steps S101 to S105 are periodically executed to transmit the received data value sequences of multiple data types and their corresponding transmission values ​​and data sources to the IoT cloud platform to realize IoT-based communication.

10. The communication method based on the Internet of Things according to claim 5, characterized in that: The collaborative deviation assessment model is obtained in the following manner: A1. Collect a large number of historical collaborative evaluation feature vectors of different data types, as well as the actual window deviation index of the corresponding associated data types; A2. Label each historical collaborative evaluation feature vector with the actual deviation index vector consisting of the actual window deviation index of the corresponding associated data type. A3. Use the labeled historical collaborative evaluation feature vectors as the training sample set, train the pre-selected neural network structure, and continuously optimize the model parameters to obtain the final collaborative deviation evaluation model.

Images