Low-power data transmission optimization method and device for Internet of Things devices

Abstract

The present invention discloses a low-power data transmission optimization method and device for Internet of Things devices, which relates to the field of information transmission technology. The method includes: configuring Internet of Things monitoring equipment to collect monitoring data and basic monitoring feature information, analyzing the basic data value to obtain parameters based on the information, classifying and determining the extraction quantity based on the parameters and extracting frame monitoring data at equal intervals, identifying the extracted data to obtain suspicious parameters, optimizing the frame monitoring data compression strategy based on the suspicious parameters, and compressing and transmitting with low power consumption after obtaining the optimal strategy. The optimal strategy includes quality and quantity compression parameters, aiming to reduce power consumption and improve data quality. The method solves the technical problem that the low power consumption requirement and high-quality data transmission cannot be effectively balanced in the existing Internet of Things monitoring data transmission, and achieves the technical effect of ensuring high-quality data transmission and effectively identifying and processing suspicious situations while significantly reducing the power consumption of data transmission.

Description

Technical Field

[0001] The present application relates to the field of information transmission technology, and in particular to a low-power data transmission optimization method and device for Internet of Things devices. Background Art

[0002] In the application scenarios of the Internet of Things, the energy consumption and quality issues of data transmission are particularly prominent, and the contradictions of related requirements are more obvious. How to optimize data transmission to reduce power consumption and ensure data quality has become a crucial link in solving the development of the Internet of Things. Traditional Internet of Things data transmission methods are often simple and extensive, relying only on conventional data processing methods or limited transmission strategies, insufficient mining of data value, lack of comprehensive analysis and utilization of monitoring data, and difficult to accurately assess the importance and risk status of data. There are unreasonable situations in data processing and transmission strategies, resulting in excessive power consumption of some key data transmissions and damaged data quality. The formulation of data transmission plans is relatively simple and fixed, and cannot well cope with the increasingly complex and changeable actual needs of the Internet of Things.

[0003] At present, there is a technical problem in the relevant technologies that the low power consumption requirements and high-quality data transmission cannot be effectively balanced in the IoT monitoring data transmission. Summary of the invention

[0004] The present application provides a low-power data transmission optimization method and device for IoT devices, which adopts the configuration of IoT monitoring devices to collect monitoring data and basic monitoring feature information, analyzes the basic data value to obtain parameters based on the information, classifies the parameters to determine the extraction quantity and extracts frame monitoring data at equal intervals, identifies the extracted data to obtain suspicious parameters, optimizes the frame monitoring data compression strategy based on the suspicious parameters, and compresses and transmits low-power data after obtaining the optimal strategy. The optimal strategy includes quality and quantity compression parameters, which aims to reduce power consumption and improve data quality. The application realizes the intelligent extraction, value analysis, suspicious parameter identification and compression strategy optimization of IoT device monitoring data, and achieves the technical effect of significantly reducing data transmission power consumption while ensuring high-quality data transmission and effectively identifying and handling suspicious situations.

[0005] The present application provides a low-power data transmission optimization method for IoT devices, including:

[0006] By configuring monitoring equipment of the Internet of Things, monitoring data and basic monitoring feature information are collected and obtained, wherein the monitoring data includes multiple frame monitoring data of multiple time frames; based on the basic monitoring feature information, basic data value analysis is performed to obtain basic data value parameters; based on the basic data value parameters, random sampling quantities are obtained by classification, and the multiple frame monitoring data are equidistantly extracted to obtain multiple basic extracted frame monitoring data; the multiple basic extracted frame monitoring data are identified to obtain suspicious image parameters and suspicious action parameters; based on the suspicious image parameters and suspicious action parameters, the compression strategy of the multiple frame monitoring data is optimized to obtain the optimal compression strategy, and the multiple frame monitoring data are compressed and transmitted with low power consumption, wherein the optimal compression strategy includes optimal quality compression parameters and optimal quantity compression parameters, and the optimization includes reducing transmission power consumption and improving transmission data quality.

[0007] The present application also provides a low-power data transmission optimization device for IoT devices, including:

[0008] A data information acquisition module, the data information acquisition module is used to acquire monitoring data and basic monitoring feature information by configuring monitoring equipment of the Internet of Things, wherein the monitoring data includes multiple frame monitoring data of multiple time frames; a value analysis module, the value analysis module is used to perform basic data value analysis based on the basic monitoring feature information to obtain basic data value parameters; a monitoring data extraction module, the monitoring data extraction module is used to classify and obtain a random extraction quantity based on the basic data value parameters, perform equidistant extraction on the multiple frame monitoring data, and obtain multiple basic extracted frame monitoring data; a data identification module, the data identification module is used to identify the multiple basic extracted frame monitoring data and obtain image suspicious parameters and action suspicious parameters; a compression strategy optimization module, the compression strategy optimization module is used to optimize the compression strategy of the multiple frame monitoring data based on the image suspicious parameters and action suspicious parameters, obtain the optimal compression strategy, and compress the multiple frame monitoring data for low-power transmission, wherein the optimal compression strategy includes optimal quality compression parameters and optimal quantity compression parameters, and the optimization includes reducing transmission power consumption and improving transmission data quality.

[0009] The low-power data transmission optimization method and device for IoT devices proposed in this application are first configured to collect monitoring data and basic monitoring feature information of the IoT, analyze the basic data value to obtain parameters based on this, classify the parameters to determine the extraction quantity and extract frame monitoring data at equal intervals, identify the extracted data to obtain suspicious parameters, optimize the frame monitoring data compression strategy based on the suspicious parameters, and compress and transmit at low power after obtaining the optimal strategy. The optimal strategy includes quality and quantity compression parameters, which aims to reduce power consumption and improve data quality. Through intelligent extraction, value analysis, suspicious parameter identification and compression strategy optimization of IoT device monitoring data, the technical effect of significantly reducing data transmission power consumption while ensuring high-quality data transmission and effectively identifying and handling suspicious situations is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the device according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0011] Figure 1 A flowchart of a low-power data transmission optimization method for an Internet of Things device provided in an embodiment of the present application;

[0012] Figure 2 A schematic diagram of the structure of a low-power data transmission optimization device for an Internet of Things device provided in an embodiment of the present application.

[0013] Explanation of the accompanying drawings: data information collection module 10, value analysis module 20, monitoring data extraction module 30, data identification module 40, compression strategy optimization module 50. DETAILED DESCRIPTION

[0014] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0015] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0016] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, device, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0017] The present application embodiment provides a low-power data transmission optimization method for IoT devices, such as Figure 1 As shown, the method includes:

[0018] Step S100, by configuring the monitoring equipment of the Internet of Things, monitoring data and basic monitoring feature information are collected and obtained, wherein the monitoring data includes multiple frame monitoring data of multiple time frames. Specifically, the monitoring data and basic monitoring feature information refer to the frame monitoring data and related time information collected during the Internet of Things monitoring process. Select the type of Internet of Things device to be monitored, prepare the monitoring collection equipment, set the monitoring collection parameters, such as collection frequency, data format, storage path, etc., perform monitoring collection, and obtain frame monitoring data records of multiple time frames. The frame monitoring data record represents the continuous data generated by the device during the monitoring time period, which is characterized by time seriality. At the same time, basic monitoring feature information is collected, including the monitoring start time, end time, etc. Record the relevant basic monitoring features during the entire monitoring process.

[0019] In a possible implementation, monitoring data and basic monitoring feature information are collected by configuring the monitoring device of the Internet of Things, wherein the monitoring data includes multiple frames of monitoring data of multiple time frames. Step S100 further includes step S110, wherein the monitoring data are collected by configuring the monitoring device of the Internet of Things, wherein the monitoring data includes multiple frames of monitoring data of multiple time frames. Specifically, the monitoring device of the Internet of Things is fully configured, including but not limited to determining the network connection settings of the device to ensure that it can communicate stably with the data acquisition system; setting the monitoring range and accuracy of the device to accurately capture the required monitoring object information; adjusting the acquisition frequency of the device, such as collecting several times per second, so as to obtain a sufficient amount of monitoring data per unit time. After the configuration is completed, the monitoring device is started, and the monitoring object is continuously monitored and data collected according to the set frequency and range. For example, if the acquisition frequency of the device is 10 frames per second and the collection is continued for 10 seconds, 100 frames of monitoring data will be obtained, and these data constitute a data set containing time information and specific monitoring content.

[0020] Step S120, collecting the median timestamp information of the multiple time frames as basic monitoring feature information. Specifically, after acquiring the frame monitoring data of multiple time frames, the time frames are arranged in order, for example, arranged in chronological order from small to large, and the time frame in the middle position is determined. If the total number of time frames is an odd number, then the timestamp of the time frame in the middle position is the median timestamp; if the total number is an even number, then the average timestamp of the two middle time frames is used as the median timestamp. For example, if there are 11 time frames in total, then the timestamp of the 6th time frame is the median timestamp; if there are 12 time frames, then the average value of the timestamps of the 6th and 7th time frames is the median timestamp. This median timestamp is extracted as important basic monitoring feature information for subsequent data analysis, processing and decision making.

[0021] Step S200: Perform basic data value analysis based on the basic monitoring feature information to obtain basic data value parameters. Specifically, the basic monitoring feature information collected is obtained. The key part of the information is the time-related data of the monitoring. The potential value of the data is preliminarily judged based on the monitoring time. Taking product security monitoring as an example, if it is at night, which is generally considered to be a sensitive time period with a higher probability of product safety problems, then the monitoring data collected during this period will be given a higher value assessment, because at night, there are relatively few people and security measures may be relatively weak. Once a problem occurs, the impact is greater. If it is during the day, there are many people and security is relatively comprehensive. The possibility of safety problems is low, then the value of the monitoring data collected at this time is relatively low. When conducting value analysis, a set of evaluation standards and weight systems will be established. For example, the night time period is set to a higher weight and the daytime time period is set to a lower weight. Considering other time-related factors, such as holidays, specific activities, etc., the value of the monitoring data in these special time periods will also be adjusted accordingly. By considering the time factor and the corresponding weight, the basic data value parameters that can reflect the value of the data are finally obtained. For example, for the product security monitoring of a factory, it is assumed that 22:00 to 6:00 the next day every day in a week is defined as a high-value time period with a weight of 0.8; 6:00 to 22:00 is a low-value time period with a weight of 0.2. If a large amount of effective monitoring data is collected during high-value time periods within a week, its basic data value parameter will be higher after calculation, otherwise it will be lower.

[0022] In a possible implementation, a basic data value analysis is performed based on the basic monitoring feature information to obtain basic data value parameters, and step S200 further includes step S210, obtaining a sample basic monitoring feature information set based on the historical monitoring data of the monitoring device, and annotating a sample basic data value parameter set based on the probability of occurrence of abnormal monitoring events in each sample basic monitoring feature information within a historical time, wherein the sample basic data value parameters include a basic data value level, and the size of the basic data value level is positively correlated with the size of the probability of occurrence of abnormal monitoring time. Specifically, the historical monitoring data accumulated by the monitoring equipment is analyzed, and representative sample basic monitoring feature information is screened and extracted, and organized into a set. For each sample basic monitoring feature information, the actual occurrence of abnormal monitoring events in the past time period is analyzed. Through statistics and calculations, the probability of occurrence of abnormal monitoring events corresponding to each sample basic monitoring feature information is obtained. Based on the occurrence probability, corresponding basic data value parameters are labeled for each sample. The parameters include the basic data value level. The level setting is directly linked to the probability of occurrence of abnormal monitoring events, that is, the higher the probability, the higher the level. For example, for the historical monitoring data of a warehouse, if an area has frequently experienced abnormal events such as loss of goods in the past, then the probability of occurrence of abnormal monitoring events corresponding to the sample basic monitoring feature information of this area will be higher, and the labeled basic data value level will also be correspondingly higher.

[0023] Step S220, using the sample basic monitoring feature information set and the sample basic data value parameter set, training to obtain a basic value analyzer. Specifically, the sample basic monitoring feature information set and the sample basic data value parameter set previously acquired and annotated are used as training materials, and a neural network is used to allow the model to learn the intrinsic association and pattern between the sample basic monitoring feature information and the basic data value parameters. After multiple training and optimizations, a basic value analyzer that can accurately judge the value of basic monitoring feature information is finally obtained. For example, through a large amount of training, the basic value analyzer can understand the relationship between certain specific monitoring features (such as features of a specific time period or a specific area) and high-value data.

[0024] Step S230, using the basic value analyzer, conduct basic data value analysis on the basic monitoring feature information to obtain basic data value parameters. Specifically, when there is new basic monitoring feature information that needs to be evaluated, it is input into the trained basic value analyzer. The basic value analyzer will use the knowledge and patterns learned from previous training to quickly analyze and calculate the input basic monitoring feature information, and output a parameter that can reflect the value of the basic monitoring feature information, providing an important basis for subsequent data processing and decision-making. For example, when a new day's warehouse monitoring data is generated, the basic monitoring feature information is extracted and input into the basic value analyzer to obtain value assessment parameters for these new data.

[0025] Step S300, according to the basic data value parameter, classify and obtain the random extraction quantity, perform equidistant extraction on the multiple frame monitoring data, and obtain multiple basic extraction frame monitoring data. Specifically, the basic data value parameter calculated previously is obtained, and the parameter reflects the importance or value of the monitoring data. The basic data value parameter is processed according to the pre-set classification rules. For example, the value parameter is divided into several intervals, each interval corresponds to a different extraction quantity range. If the basic data value parameter is large, it means that this section of monitoring data has a higher value. At this time, it will be classified into the corresponding high extraction quantity interval, thereby determining a larger extraction quantity. On the contrary, if the basic data value parameter is small, indicating that its value is relatively low, it will be classified into the low extraction quantity interval, and a smaller extraction quantity will be determined. After the extraction quantity is determined, the equidistant extraction operation of multiple frame monitoring data is started. Assuming that there are a total of 1000 frame monitoring data, the determined extraction quantity is 200, and the extraction interval is first calculated, that is, 1000÷200=5. This means that one frame of monitoring data is extracted every 5 frames, starting from the 1st frame of monitoring data, and then extracted in sequence according to the intervals, to the 1st, 6th, 11th... and so on, until 200 frames of monitoring data are extracted. The extracted frame monitoring data are multiple basic extracted frame monitoring data, and the distribution in the time series is uniform, and the time interval between each data is the same. For example, in a factory production monitoring scenario, if the basic data value parameter of the product quality data for a certain period of time is very high, it may be because the risk of quality problems in the product during this period is greater, then more frame monitoring data will be extracted, such as 300, with an interval of 1000÷300≈3.33, that is, one is extracted every approximately 3 data, in order to more carefully analyze the production situation during this period.

[0026] In a possible implementation, the basic data value parameters are classified to obtain the random extraction quantity, and the multiple frame monitoring data are equidistantly extracted to obtain multiple basic extraction frame monitoring data. Step S300 further includes step S310, collecting a sample basic data value parameter set, and obtaining a sample random extraction quantity set. The size of the sample basic data value parameter is positively correlated with the size of the sample random extraction quantity. Specifically, various basic data value parameters are collected from a large number of monitoring data samples and organized into a set. At the same time, for each sample, the corresponding random extraction quantity is determined, and the extraction quantity is also organized into a set. In the sample, the higher the basic data value parameter, the larger the corresponding random extraction quantity. For example, for a group of video surveillance samples, samples with high value parameters may be set to extract more frames.

[0027] Step S320, using the sample basic data value parameter set and the sample random extraction quantity set, construct an extraction quantity classifier. Specifically, using the collected sample basic data value parameter set and the corresponding sample random extraction quantity set, a decision tree algorithm is used to construct a classifier that can determine the extraction quantity based on the input basic data value parameters, so as to learn the relationship pattern between the value parameters in the sample data and the extraction quantity.

[0028] Step S330, input the basic data value parameter into the extraction quantity classifier, and classify to obtain the random extraction quantity. Specifically, when there is a new basic data value parameter that needs to determine the extraction quantity, it is input into the previously constructed extraction quantity classifier. The classifier will analyze and process the input basic data value parameter according to the previously learned patterns and rules, and output a corresponding random extraction quantity. Assuming that the input basic data value parameter is high, the classifier may give a larger extraction quantity.

[0029] Step S340, according to the random extraction number, the multiple frame monitoring data are equidistantly extracted to obtain multiple basic extracted frame monitoring data. Specifically, after obtaining the extraction number, the multiple frame monitoring data are equidistantly extracted according to this number, and the extraction interval is first calculated, and then starting from the first frame, the frames are sequentially extracted at fixed intervals to finally obtain multiple basic extracted frame monitoring data. For example, there are 100 frame monitoring data, the extraction number is 20, and the interval is 5. The 1st, 6th, 11th...96th frames are sequentially extracted to obtain 20 basic extracted frame monitoring data.

[0030] Step S400, identify the multiple basic extracted frame monitoring data to obtain suspicious image parameters and suspicious action parameters. Specifically, multiple basic extracted frame monitoring data are obtained, and these data are processed using image and action recognition technology. In terms of image recognition, the target's appearance features, such as the person's clothing, the items carried, etc., will be analyzed to evaluate the suspicious level of the image. Since image recognition requires high-quality images, algorithms and models that can ensure image clarity and details will be used to accurately judge the suspicious degree of the image. In the action recognition link, attention will be paid to the target's continuous action changes, such as walking posture, action continuity, etc., to determine the suspicious level of the action. In order to accurately identify the action, as many continuous images as possible are required, so a series of frame monitoring data will be comprehensively analyzed. High-quality images and large amounts of continuous image data will generate greater power consumption during transmission. For example, in a surveillance scenario in a shopping mall, a person with abnormal behavior may be given a higher suspicious image level through a detailed analysis of his image; a higher suspicious action level may be given by observing his continuous actions, such as frequent wandering and avoiding cameras. Based on the obtained suspicious image parameters (i.e., suspicious image level) and suspicious action parameters (i.e., suspicious action level), in order to reduce transmission power consumption while ensuring the monitoring effect, the compression strategy of multiple frames of monitoring data will be optimized to find the best compression method that can reduce the amount of data while retaining key information to the greatest extent.

[0031] In one possible implementation, the multiple basic extracted frame monitoring data are identified to obtain suspicious image parameters and suspicious action parameters, and step S400 further includes step S410, collecting a set of sample frame monitoring data, marking the suspicious image parameters and suspicious action parameters of the visitor in each sample frame monitoring data, and obtaining a sample image suspicious parameter set and a sample action suspicious parameter set. Specifically, a large amount of sample frame monitoring data is collected and organized into a sample frame monitoring data set. For each sample frame monitoring data in the set, professionals or with the help of specific annotation tools carefully observe and analyze the image and action characteristics of the visitor. For the visitor's image, suspicious image parameters are annotated according to the abnormality of his appearance characteristics (such as clothing, items carried, etc.). For actions, suspicious action parameters are annotated according to the abnormality of his actions (such as sudden running, sneaking, etc.). By annotating all samples, a sample image suspicious parameter set and a sample action suspicious parameter set are respectively organized. For example, in a bank's monitoring sample, for a visitor wearing a mask and holding a suspicious item, his image suspicious parameter is marked as high; for a visitor who frequently looks around and moves nervously in front of the counter, his action suspicious parameter is marked as high.

[0032] Step S420, using the sample frame monitoring data set, the sample image suspicious parameter set and the sample action suspicious parameter set, training the extracted monitoring data identifier, identifying the multiple basic extracted frame monitoring data, and obtaining multiple extracted image suspicious parameters and multiple extracted action suspicious parameters. Specifically, the previously collected and labeled sample frame monitoring data set, the sample image suspicious parameter set and the sample action suspicious parameter set are used as training data, and the data is trained using, for example, a convolutional neural network, to construct an extracted monitoring data identifier that can accurately identify the visitor's image and suspicious degree of action in the monitoring data. When multiple new basic extracted frame monitoring data are input, they are processed and analyzed using the trained identifier, and the identifier will output the visitor's extracted image suspicious parameters and extracted action suspicious parameters in each basic extracted frame monitoring data. For example, a new set of basic extracted frames in the shopping mall monitoring data is input into the identifier to obtain the suspicious parameters of the visitor in each frame.

[0033] Step S430, based on the multiple extracted suspicious image parameters and multiple extracted suspicious action parameters, calculate and obtain suspicious image parameters and suspicious action parameters. Specifically, the multiple extracted suspicious image parameters and multiple extracted suspicious action parameters output by the identifier are comprehensively calculated and analyzed, and multiple extracted parameters are integrated into a final suspicious image parameter and suspicious action parameter by using a weighted average, a maximum value, and other calculation methods. For example, different weights are assigned to a group of extracted parameters, and then the weighted average is calculated to obtain the final suspicious image and action parameters, so as to more accurately reflect the suspicious degree of the visitor in the monitoring data.

[0034] Step S500, according to the suspicious image parameters and suspicious action parameters, the compression strategy of the plurality of frame monitoring data is optimized to obtain the optimal compression strategy, and the plurality of frame monitoring data is compressed and transmitted with low power consumption, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter, and the optimization includes reducing the transmission power consumption and improving the transmission data quality. Specifically, the suspicious image parameters and suspicious action parameters calculated previously are obtained, and the compression strategy of the plurality of frame monitoring data is optimized based on the parameters. During the optimization process, multiple factors need to be considered comprehensively. On the one hand, the transmission power consumption should be reduced and the amount of compressed data should be minimized as much as possible, because the less the amount of data, the lower the power consumption required for transmission. On the other hand, the quality of image recognition and action recognition should be guaranteed. By continuously adjusting the quality compression parameters and the quantity compression parameters in the compression strategy, if the data quality is excessively reduced and the amount of data is reduced, although the transmission power consumption can be significantly reduced, the accuracy of image recognition and action recognition will be greatly reduced. On the contrary, if high recognition quality is pursued too much without sufficient compression, the transmission power consumption will be too high. Through repeated experiments and calculations, an optimal balance point is found. For example, starting from an initial compression strategy, the compression degree is gradually increased, and the changes in the quality of image recognition and action recognition and the reduction of transmission power consumption are observed. If the quality drops too fast and the power consumption is not significantly reduced, the compression degree is appropriately adjusted; if the power consumption is not reduced enough and the quality can still be maintained well, the compression is further increased, and finally the optimal compression strategy is determined, which includes the optimal quality compression parameter and the optimal quantity compression parameter. Under the premise of ensuring that the image recognition and action recognition quality meet certain requirements, the transmission power consumption is reduced to the greatest extent and the transmission data quality is improved. For example, in a monitoring scene in a public place, the optimal compression strategy obtained after optimization may be to compress non-critical information such as the background to a greater extent on the basis of maintaining the key features of the character image and action, so as to accurately identify suspicious images and actions while reducing power consumption. After determining the optimal compression strategy, multiple frames of monitoring data are compressed according to the strategy and transmitted in a low-power manner to achieve efficient, energy-saving and monitoring effect data transmission.

[0035] In a possible implementation, according to the suspicious image parameter and the suspicious action parameter, the compression strategy of the plurality of frame monitoring data is optimized to obtain the optimal compression strategy, and the plurality of frame monitoring data is compressed and transmitted with low power consumption, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter, and the optimization includes reducing transmission power consumption and improving the quality of transmission data, and step S500 further includes step S510, obtaining the quality compression parameter space and the quantity compression parameter space, wherein the quality compression parameter includes the data quality compression parameter for compressing each frame monitoring data, and the quantity compression parameter includes the data quantity compression parameter for compressing the plurality of frame monitoring data. Specifically, the quality compression parameter and the quantity compression parameter are defined and the range is determined, for example, the possible range of the resolution of the compressed data is clarified, from various degrees of high resolution to low resolution, and the possible range of the ratio of the number of compressed plurality of frame monitoring data to the original number is determined for the quantity compression parameter, for example, the resolution in the quality compression parameter can be from 1920×1080 to 320×240, and the ratio in the quantity compression parameter can be from 100% to 10%.

[0036] Step S520: construct a compression strategy function according to the suspicious image parameters and the suspicious action parameters, as shown in the following formula: , where CSF is the compression fitness, is the quality weight, is the quantity weight, is the power consumption weight, The image recognition quality parameter for image recognition after the plurality of frames of monitoring data are compressed, Suspicious parameters for the image, The action recognition quality parameter for performing action recognition after compressing the plurality of frame monitoring data, is the suspicious action parameter, and G is the transmission power consumption of the plurality of frames of monitoring data after compression. Specifically, based on the image suspicious parameters and action suspicious parameters obtained previously, a mathematical function for evaluating the effect of the compression strategy is established. In this function, CSF represents compression fitness, which is an indicator used to measure the quality of the compression strategy, and quality weight Indicates the importance of image recognition quality in the overall evaluation, quantitative weight Indicates the importance of action recognition quality and power consumption weight This reflects the importance of transmission power consumption. is the quality parameter for image recognition after compression. It is a suspicious parameter. is the quality parameter for action recognition after compression, is the suspicious action parameter, G is the power consumption of transmission after compression. Through this function, the impact of compression on image and action recognition quality and transmission power consumption can be comprehensively considered.

[0037] Step S530, randomly generating a first quality compression parameter and a first quantity compression parameter in the quality compression parameter space and the quantity compression parameter space as a first compression strategy. Specifically, within the determined quality compression parameter space and the quantity compression parameter space, a set of quality compression parameter and quantity compression parameter values ​​are randomly generated. For example, the randomly generated first quality compression parameter may be to set the resolution to 640×480, and the first quantity compression parameter may be to set the frame quantity ratio to 50%. The randomly generated parameters are combined as a first compression strategy for subsequent testing and evaluation.

[0038] Step S540, analyze and obtain the first image recognition quality parameter and the first action recognition quality parameter of the multiple frame monitoring data after being compressed according to the first compression strategy, and obtain the first transmission power consumption of the multiple frame monitoring data after being compressed according to the first compression strategy, and calculate and obtain the first compression fitness based on the compression strategy function. Specifically, perform actual compression operation on multiple frame monitoring data according to the first compression strategy, perform image recognition and action recognition tests on the compressed monitoring data, obtain the first image recognition quality parameter and the first action recognition quality parameter, obtain the first transmission power consumption of the transmission after being compressed according to the first compression strategy through actual transmission test or calculation, substitute the obtained data into the compression strategy function constructed before, and calculate the first compression fitness to evaluate the effect of this compression strategy.

[0039] Step S550, continue to perform iterative optimization until the optimization converges, output the compression strategy with the largest compression fitness, and obtain the optimal compression strategy, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter. Specifically, after obtaining the first compression fitness, do not stop, but continue to repeat the previous process to iterate, and each iteration re-randomly generate new quality compression parameters and quantity compression parameters to form a new compression strategy, and calculate its corresponding compression fitness. In multiple iterations, continuously adjust and improve the compression strategy until the compression fitness no longer changes significantly, that is, the optimization converges, and finally, output the strategy with the largest compression fitness among all the compression strategies tried. This strategy is the optimal compression strategy, which contains the optimal quality compression parameters (such as the best resolution) and the optimal quantity compression parameters (such as the most appropriate frame quantity ratio).

[0040] In one possible implementation, a first image recognition quality parameter and a first action recognition quality parameter for identifying after compressing the multiple frame monitoring data according to the first compression strategy are analyzed and obtained, and a first transmission power consumption for transmitting after compressing the multiple frame monitoring data according to the first compression strategy is obtained, and a first compression fitness is calculated based on the compression strategy function. Step S540 further includes step S541, collecting a sample quality compression parameter set and a sample quantity compression parameter set according to the recognition record of the data transmitted by the monitoring device, obtaining the image recognition accuracy after compressing the monitoring data with different sample quality compression parameters, marking the obtained sample image recognition quality parameter set, and obtaining the action recognition accuracy after compressing the monitoring data with different sample quantity compression parameters, and marking the obtained sample action recognition quality parameter set. Specifically, data is collected from the recognition records of the monitoring equipment's previous transmitted data, and various sample quality compression parameters and sample quantity compression parameters are collected respectively to form corresponding sets. For each sample quality compression parameter, the image recognition accuracy after compressing the monitoring data is tested, and these accuracy rates are marked as sample image recognition quality parameters to form a sample image recognition quality parameter set. For different sample quantity compression parameters, the action recognition accuracy after compressing the monitoring data is obtained and marked as sample action recognition quality parameters to form a sample action recognition quality parameter set. For example, for sample quality compression parameters, such as compression with a resolution of 1280×720, its image recognition accuracy is 80%, and 80% is marked as the corresponding sample image recognition quality parameter.

[0041] Step S542, using the sample quality compression parameter set and the sample image recognition quality parameter set to train the image recognition quality analysis branch, using the sample quantity compression parameter set and the sample action recognition quality parameter set to train the action recognition quality analysis branch, and combining the image recognition quality analysis branch to obtain a compression recognition quality analyzer. Specifically, using the sample quality compression parameter set and the sample image recognition quality parameter set collected previously, through a suitable machine learning algorithm or model, the image recognition quality analysis branch is trained, the sample quantity compression parameter set and the sample action recognition quality parameter set are used to train the action recognition quality analysis branch, and the trained image recognition quality analysis branch and the action recognition quality analysis branch are combined to form a compression recognition quality analyzer that can simultaneously analyze the image and action recognition quality. For example, a neural network can be used for training to allow the analysis branch to learn the relationship between parameters and recognition accuracy.

[0042] Step S543, using the compression recognition quality analyzer, analyze the first quality compression parameter and the first quantity compression parameter to obtain the first image recognition quality parameter and the first action recognition quality parameter for recognizing after compressing the plurality of frames of monitoring data according to the first compression strategy. Specifically, after the first quality compression parameter and the first quantity compression parameter (i.e., the first compression strategy) are determined, they are input into the trained compression recognition quality analyzer, and the compression recognition quality analyzer will analyze and calculate the input parameters according to the previous learning and training, and finally output the first image recognition quality parameter and the first action recognition quality parameter for recognizing after compressing the plurality of frames of monitoring data according to the first compression strategy, so as to evaluate the effect of the compression strategy in image and action recognition. Assuming that the quality compression parameter in the first compression strategy is a resolution of 800×600, and the quantity compression parameter is a frame compression of 50%, the image recognition quality parameter and the action recognition quality parameter corresponding to the output of the analyzer are 70%, and 65%.

[0043] The embodiment of the present application adopts the configuration of Internet of Things monitoring equipment to collect monitoring data and basic monitoring feature information, analyzes the basic data value based on this to obtain parameters, classifies the parameters to determine the extraction quantity and extracts frame monitoring data at equal intervals, identifies the extracted data to obtain suspicious parameters, optimizes the frame monitoring data compression strategy based on the suspicious parameters, and compresses and transmits low-power data after obtaining the optimal strategy. The optimal strategy includes quality and quantity compression parameters, which aims to reduce power consumption and improve data quality. Through the intelligent extraction, value analysis, suspicious parameter identification and compression strategy optimization of Internet of Things device monitoring data, the technical effect of significantly reducing data transmission power consumption while ensuring high-quality data transmission and effectively identifying and handling suspicious situations is achieved.

[0044] In the above, refer to Figure 1 The low power consumption data transmission optimization method for IoT devices according to an embodiment of the present invention is described in detail. Figure 2 A low power consumption data transmission optimization device for an Internet of Things device according to an embodiment of the present invention is described.

[0045] The low-power data transmission optimization device for IoT devices according to the embodiment of the present invention is used to solve the technical problem that the low-power demand and high-quality data transmission cannot be effectively balanced in the existing IoT monitoring data transmission, so as to achieve the technical effect of significantly reducing the power consumption of data transmission while ensuring high-quality data transmission and effectively identifying and handling suspicious situations. The low-power data transmission optimization device for IoT devices includes: a data information acquisition module 10, a value analysis module 20, a monitoring data extraction module 30, a data identification module 40, and a compression strategy optimization module 50.

[0046] The data information acquisition module 10 is used to acquire monitoring data and basic monitoring feature information by configuring the monitoring equipment of the Internet of Things, wherein the monitoring data includes a plurality of frame monitoring data of a plurality of time frames;

[0047] The value analysis module 20 is used to perform basic data value analysis based on the basic monitoring feature information to obtain basic data value parameters;

[0048] The monitoring data extraction module 30 is used to classify and obtain a random extraction quantity according to the basic data value parameter, and perform equidistant extraction on the multiple frames of monitoring data to obtain multiple basic extracted frames of monitoring data;

[0049] The data identification module 40 is used to identify the plurality of basic extracted frame monitoring data to obtain suspicious image parameters and suspicious action parameters;

[0050] The compression strategy optimization module 50 is used to optimize the compression strategy of the multiple frame monitoring data according to the suspicious image parameters and the suspicious action parameters, obtain the optimal compression strategy, and compress the multiple frame monitoring data for low-power transmission, wherein the optimal compression strategy includes optimal quality compression parameters and optimal quantity compression parameters, and the optimization includes reducing transmission power consumption and improving transmission data quality.

[0051] The specific configuration of the data information acquisition module 10 will be described in detail below. As described above, by configuring the monitoring equipment of the Internet of Things, the monitoring data and basic monitoring feature information are collected and obtained, wherein the monitoring data includes multiple frame monitoring data of multiple time frames, and the data information acquisition module 10 further includes: a monitoring data acquisition unit for collecting and obtaining monitoring data by configuring the monitoring equipment of the Internet of Things, wherein the monitoring data includes multiple frame monitoring data of multiple time frames; and a median timestamp information acquisition unit for collecting median timestamp information of the multiple time frames as basic monitoring feature information.

[0052] The specific configuration of the value analysis module 20 will be described in detail below. As described above, based on the basic monitoring feature information, basic data value analysis is performed to obtain basic data value parameters. The value analysis module 20 further includes: a labeling unit for obtaining a sample basic monitoring feature information set based on the historical monitoring data of the monitoring device, and labeling a sample basic data value parameter set based on the probability of occurrence of abnormal monitoring events in the historical time for each sample basic monitoring feature information, wherein the sample basic data value parameter includes a basic data value level, and the size of the basic data value level is positively correlated with the size of the probability of occurrence of abnormal monitoring time; an analyzer training unit for adopting the sample basic monitoring feature information set and the sample basic data value parameter set to train a basic value analyzer; a value analysis unit for adopting the basic value analyzer to perform basic data value analysis on the basic monitoring feature information to obtain basic data value parameters.

[0053] The specific configuration of the monitoring data extraction module 30 will be described in detail below. As described above, according to the basic data value parameters, the random extraction quantity is obtained by classification, and the multiple frame monitoring data are equidistantly extracted to obtain multiple basic extraction frame monitoring data. The monitoring data extraction module 30 further includes: a sample extraction unit is used to collect a set of sample basic data value parameters and obtain a set of sample random extraction quantities, and the size of the sample basic data value parameters is positively correlated with the size of the sample random extraction quantity; a classifier construction unit is used to use the sample basic data value parameter set and the sample random extraction quantity set to construct an extraction quantity classifier; an extraction quantity acquisition unit is used to input the basic data value parameters into the extraction quantity classifier and classify to obtain a random extraction quantity; an equidistant extraction unit is used to equidistantly extract the multiple frame monitoring data according to the random extraction quantity to obtain multiple basic extraction frame monitoring data.

[0054] The specific configuration of the data identification module 40 will be described in detail below. As described above, the multiple basic extracted frame monitoring data are identified to obtain suspicious image parameters and suspicious action parameters. The data identification module 40 further includes: a suspicious parameter set acquisition unit for collecting a sample frame monitoring data set, marking the visitor's suspicious image parameters and suspicious action parameters in each sample frame monitoring data, and obtaining a sample image suspicious parameter set and a sample action suspicious parameter set; a monitoring data identification unit for using the sample frame monitoring data set, the sample image suspicious parameter set and the sample action suspicious parameter set to train an extracted monitoring data identifier, identify the multiple basic extracted frame monitoring data, and obtain multiple extracted image suspicious parameters and multiple extracted action suspicious parameters; a suspicious parameter calculation unit for calculating and obtaining image suspicious parameters and action suspicious parameters based on the multiple extracted image suspicious parameters and multiple extracted action suspicious parameters.

[0055] The specific configuration of the compression strategy optimization module 50 will be described in detail below. As described above, according to the suspicious image parameters and suspicious action parameters, the compression strategy of the multiple frame monitoring data is optimized to obtain the optimal compression strategy, and the multiple frame monitoring data are compressed and transmitted with low power consumption, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter, and the optimization includes reducing the transmission power consumption and improving the transmission data quality. The compression strategy optimization module 50 further includes: a compression parameter space acquisition unit for acquiring the quality compression parameter space and the quantity compression parameter space, wherein the quality compression parameter includes the data quality compression parameter for compressing each frame monitoring data, and the quantity compression parameter includes the data quantity compression parameter for compressing multiple frame monitoring data; a compression strategy function construction unit, wherein the compression strategy function construction unit is used to construct a compression strategy function according to the suspicious image parameters and the suspicious action parameters, as shown in the following formula: , where CSF is the compression fitness, is the quality weight, is the quantity weight, is the power consumption weight, The image recognition quality parameter for image recognition after the plurality of frames of monitoring data are compressed, Suspicious parameters for the image, The action recognition quality parameter for performing action recognition after compressing the plurality of frame monitoring data, is the action suspicious parameter, G is the transmission power consumption of transmitting the multiple frame monitoring data after compression; the first compression strategy acquisition unit is used to randomly generate a first quality compression parameter and a first quantity compression parameter in the quality compression parameter space and the quantity compression parameter space as the first compression strategy; the first compression fitness calculation unit is used to analyze and obtain the first image recognition quality parameter and the first action recognition quality parameter for identifying the multiple frame monitoring data after compression according to the first compression strategy, and obtain the first transmission power consumption for transmitting the multiple frame monitoring data after compression according to the first compression strategy, and calculate the first compression fitness based on the compression strategy function; the optimal compression strategy acquisition unit is used to continue iterative optimization until the optimization converges, output the compression strategy with the largest compression fitness, and obtain the optimal compression strategy, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter.

[0056] Among them, the first image recognition quality parameter and the first action recognition quality parameter of the plurality of frames of monitoring data compressed according to the first compression strategy are analyzed and obtained, and the first transmission power consumption of the plurality of frames of monitoring data compressed according to the first compression strategy is obtained, and the first compression fitness is calculated based on the compression strategy function. The first compression fitness calculation unit further includes: a marking parameter set subunit is used to collect a sample quality compression parameter set and a sample quantity compression parameter set according to the recognition record of the monitoring device transmission data, obtain the image recognition accuracy after the monitoring data is compressed with different sample quality compression parameters, mark the obtained sample image recognition quality parameter set, and obtain the compressed monitoring data with different sample quantity compression parameters. According to the action recognition accuracy rate after labeling, a sample action recognition quality parameter set is obtained; the quality analysis branch recognition subunit is used to adopt the sample quality compression parameter set and the sample image recognition quality parameter set to train the image recognition quality analysis branch, and adopt the sample quantity compression parameter set and the sample action recognition quality parameter set to train the action recognition quality analysis branch, and combine the image recognition quality analysis branch to obtain a compression recognition quality analyzer; the compression parameter analysis subunit is used to adopt the compression recognition quality analyzer to analyze the first quality compression parameter and the first quantity compression parameter, and obtain the first image recognition quality parameter and the first action recognition quality parameter for identifying the multiple frames of monitoring data after compression according to the first compression strategy.

[0057] The low-power data transmission optimization device for Internet of Things devices provided in an embodiment of the present invention can execute the low-power data transmission optimization method for Internet of Things devices provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0058] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0059] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A low-power data transmission optimization method for IoT devices, characterized in that: The method comprises: By configuring the monitoring equipment of the Internet of Things, monitoring data and basic monitoring feature information are collected and obtained, wherein the monitoring data includes multiple frame monitoring data of multiple time frames; Perform basic data value analysis based on the basic monitoring feature information to obtain basic data value parameters; According to the basic data value parameter, the random extraction quantity is obtained by classification, and the multiple frame monitoring data are extracted at equal intervals to obtain multiple basic extraction frame monitoring data; Identify the plurality of basic extracted frame monitoring data to obtain suspicious image parameters and suspicious action parameters; According to the suspicious image parameters and the suspicious action parameters, the compression strategy of the multiple frame monitoring data is optimized to obtain the optimal compression strategy, and the multiple frame monitoring data are compressed and transmitted with low power consumption, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter, and the optimization includes reducing transmission power consumption and improving transmission data quality; The method of optimizing the compression strategy of the plurality of frame monitoring data according to the suspicious image parameters and the suspicious action parameters to obtain the optimal compression strategy includes: Acquire a quality compression parameter space and a quantity compression parameter space, wherein the quality compression parameter includes a data quality compression parameter for compressing each frame of monitoring data, and the quantity compression parameter includes a data quantity compression parameter for compressing multiple frames of monitoring data; According to the suspicious image parameters and suspicious action parameters, a compression strategy function is constructed as follows: ； Among them, CSF is the compression fitness, is the quality weight, is the quantity weight, is the power consumption weight, The image recognition quality parameter for image recognition after the plurality of frames of monitoring data are compressed, Suspicious parameters for the image, The action recognition quality parameter for performing action recognition after compressing the plurality of frame monitoring data, is the suspicious action parameter, G is the transmission power consumption of transmitting the plurality of frames of monitoring data after compression; Randomly generating a first quality compression parameter and a first quantity compression parameter in the quality compression parameter space and the quantity compression parameter space as a first compression strategy; Analyze and obtain a first image recognition quality parameter and a first action recognition quality parameter for recognizing after compressing the plurality of frames of monitoring data according to the first compression strategy, and obtain a first transmission power consumption for transmitting after compressing the plurality of frames of monitoring data according to the first compression strategy, and calculate and obtain a first compression fitness based on the compression strategy function; Continue to perform iterative optimization until the optimization converges, output the compression strategy with the maximum compression fitness, and obtain the optimal compression strategy, wherein the optimal compression strategy includes the optimal quality compression parameter and the optimal quantity compression parameter.

2. The low-power data transmission optimization method for IoT devices according to claim 1, characterized in that: By configuring IoT monitoring devices, monitoring data and basic monitoring feature information are collected, including: By configuring monitoring equipment of the Internet of Things, monitoring data is collected and obtained, wherein the monitoring data includes multiple frame monitoring data of multiple time frames; The median timestamp information of the multiple time frames is collected as basic monitoring feature information.

3. The low-power data transmission optimization method for IoT devices according to claim 1, characterized in that: According to the basic monitoring feature information, basic data value analysis is performed to obtain basic data value parameters, including: According to the historical monitoring data of the monitoring device, a sample basic monitoring feature information set is obtained, and according to the probability of occurrence of abnormal monitoring events in the historical time of each sample basic monitoring feature information, a sample basic data value parameter set is obtained by marking, wherein the sample basic data value parameter includes a basic data value level, and the size of the basic data value level is positively correlated with the size of the probability of occurrence of abnormal monitoring time; Using the sample basic monitoring feature information set and the sample basic data value parameter set, training a basic value analyzer; The basic value analyzer is used to perform basic data value analysis on the basic monitoring feature information to obtain basic data value parameters.

4. The low-power data transmission optimization method for IoT devices according to claim 3, characterized in that: According to the basic data value parameter, the random extraction quantity is obtained by classification, and the multiple frame monitoring data are extracted at equal intervals to obtain multiple basic extracted frame monitoring data, including: Collect the sample basic data value parameter set and obtain the sample random extraction quantity set. The size of the sample basic data value parameter is positively correlated with the size of the sample random extraction quantity. Using the sample basic data value parameter set and the sample random sampling quantity set, constructing an extraction quantity classifier; Input the basic data value parameter into the extraction quantity classifier, and classify to obtain the random extraction quantity; According to the random sampling quantity, the multiple frame monitoring data are equidistantly sampled to obtain multiple basic sampled frame monitoring data.

5. The low-power data transmission optimization method for IoT devices according to claim 1, characterized in that: Identifying the plurality of basic extracted frame monitoring data to obtain suspicious image parameters and suspicious action parameters includes: Collecting a set of sample frame monitoring data, marking suspicious parameters of visitor image and suspicious parameters of action in each sample frame monitoring data, and obtaining a set of sample image suspicious parameters and a set of sample action suspicious parameters; Using the sample frame monitoring data set, the sample image suspicious parameter set and the sample action suspicious parameter set, training an extracted monitoring data identifier, identifying the multiple basic extracted frame monitoring data, and obtaining multiple extracted image suspicious parameters and multiple extracted action suspicious parameters; The suspicious image parameters and the suspicious action parameters are calculated based on the multiple extracted suspicious image parameters and the multiple extracted suspicious action parameters.

6. The low-power data transmission optimization method for IoT devices according to claim 1, characterized in that: Analyzing and obtaining a first image recognition quality parameter and a first action recognition quality parameter obtained by compressing the plurality of frames of monitoring data according to the first compression strategy and then recognizing the plurality of frames of monitoring data, including: According to the recognition record of the data transmitted by the monitoring device, a sample quality compression parameter set and a sample quantity compression parameter set are collected, the image recognition accuracy rate after the monitoring data is compressed by different sample quality compression parameters is obtained, and the sample image recognition quality parameter set is obtained by marking, and the action recognition accuracy rate after the monitoring data is compressed by different sample quantity compression parameters is obtained, and the sample action recognition quality parameter set is obtained by marking; The sample quality compression parameter set and the sample image recognition quality parameter set are used to train the image recognition quality analysis branch, the sample quantity compression parameter set and the sample action recognition quality parameter set are used to train the action recognition quality analysis branch, and the image recognition quality analysis branch is combined to obtain a compression recognition quality analyzer; The compression recognition quality analyzer is used to analyze the first quality compression parameter and the first quantity compression parameter to obtain the first image recognition quality parameter and the first action recognition quality parameter after compressing the multiple frames of monitoring data according to the first compression strategy.

7. A low-power data transmission optimization device for IoT devices, characterized in that: The device is used to implement the low-power data transmission optimization method for an Internet of Things device according to any one of claims 1 to 6, and the device includes: A data information acquisition module, the data information acquisition module is used to acquire monitoring data and basic monitoring feature information by configuring monitoring equipment of the Internet of Things, wherein the monitoring data includes a plurality of frame monitoring data of a plurality of time frames; A value analysis module, wherein the value analysis module is used to perform basic data value analysis based on the basic monitoring feature information to obtain basic data value parameters; A monitoring data extraction module, wherein the monitoring data extraction module is used to classify and obtain a random extraction quantity according to the basic data value parameter, and to perform equidistant extraction on the multiple frames of monitoring data to obtain multiple basic extracted frames of monitoring data; A data identification module, the data identification module is used to identify the plurality of basic extracted frame monitoring data to obtain suspicious image parameters and suspicious action parameters; A compression strategy optimization module, wherein the compression strategy optimization module is used to optimize the compression strategy of the multiple frame monitoring data according to the suspicious image parameters and the suspicious action parameters, obtain the optimal compression strategy, and compress the multiple frame monitoring data for low-power transmission, wherein the optimal compression strategy includes optimal quality compression parameters and optimal quantity compression parameters, and the optimization includes reducing transmission power consumption and improving transmission data quality.

Images