Testing method, system and storage medium for mobile communication indoor signal monitor

Abstract

The present invention discloses a testing method, system, and storage medium for a mobile communication indoor signal monitor. The method includes collecting multiple RSSI values ​​and current environmental test parameters of a test area, calculating an initial RSSI test value based on the multiple RSSI values, performing cross-modal aggregation analysis based on the current environmental test parameters to obtain an aggregated coding feature, and optimizing the initial RSSI test value based on the aggregated coding feature to obtain an optimized RSSI test value. By comprehensively considering the impact of multiple factors on the RSSI test value, including ambient temperature, humidity, wall material, and electromagnetic environment, the present invention can more accurately reflect the true quality of mobile communication indoor signals and improve the accuracy and reliability of signal monitoring.

Description

Technical Field

[0001] The present invention belongs to the technical field of indoor signal monitoring, and in particular relates to a testing method, system and storage medium of a mobile communication indoor signal monitor. Background Art

[0002] With the continuous advancement of mobile communication technology, indoor signal quality has become particularly important to ensure a smooth and reliable user experience. With the popularization of fifth-generation mobile communication (5G) networks and the surge in the number of Internet of Things (IoT) devices, higher requirements are being placed on communication quality in indoor environments. Therefore, accurately monitoring the quality of mobile communication indoor signals is crucial for optimizing network coverage and service quality. Traditional RSSI (Received Signal Strength Indicator) measurement methods typically only provide a rough estimate of signal strength. In real-world applications, indoor scenes are complex and varied. Numerous factors, including building structure, temperature and humidity fluctuations, and electromagnetic interference, can significantly affect the propagation of mobile communication signals. This can lead to deviations between RSSI value fluctuations and actual signal quality, making it impossible to accurately monitor indoor signal quality. Summary of the Invention

[0003] To solve the above problems, the present invention provides a testing method, system and storage medium for a mobile communication indoor signal monitor to solve the problem that traditional signal measurement methods are interfered with by various factors in actual applications, resulting in a deviation between the fluctuation of the received signal strength indicator value and the actual signal quality, and are unable to meet the requirements of accurate monitoring of indoor signal quality.

[0004] A method for testing a mobile communication indoor signal monitor, comprising:

[0005] Collect multiple RSSI values ​​of the area to be tested and the current environmental test parameters;

[0006] Calculate an RSSI initial test value based on multiple RSSI values;

[0007] Perform cross-modal aggregation analysis based on the current environment test parameters to obtain aggregated coding features;

[0008] The RSSI initial test value is optimized based on the aggregated coding feature to obtain the RSSI optimized test value.

[0009] According to a specific embodiment of the present invention, collecting multiple RSSI values ​​of the area to be tested and current environmental test parameters includes:

[0010] Select multiple reference points in the test area to arrange signal detectors, and collect multiple RSSI values ​​through the signal detectors;

[0011] Collect the current environmental test parameters of the test area, including ambient temperature, ambient humidity, wall material, and electromagnetic environment scanning spectrum image.

[0012] According to a specific embodiment of the present invention, calculating the RSSI initial test value based on multiple RSSI values ​​includes:

[0013] Calculate the average of multiple RSSI values ​​to obtain the initial RSSI test value.

[0014] According to a specific embodiment of the present invention, cross-modal aggregation analysis is performed based on the current environment test parameters to obtain the aggregated coding features including:

[0015] Taking the electromagnetic environment scanning spectrum image as the main variable and the ambient temperature, ambient humidity and wall material as covariates, a main-covariate cross-modal aggregation analysis is performed on the current environmental test parameters to obtain the aggregated coding features of the current environmental test parameters.

[0016] According to a specific embodiment of the present invention, the electromagnetic environment scanning spectrum image is used as the main variable, and the ambient temperature, ambient humidity, and wall material are used as covariates. The main-covariate cross-modal aggregation analysis of the current environmental test parameters is performed to obtain the aggregated coding features of the current environmental test parameters, which further includes:

[0017] Extract electromagnetic distribution features based on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map;

[0018] Perform feature fusion encoding on ambient temperature, ambient humidity and wall material to obtain the spliced ​​encoding vector of each environmental parameter;

[0019] The electromagnetic distribution feature map and the spliced ​​coding vector are input into the cross-modal analysis network to obtain the aggregated coding vector of the current environmental test parameters. The aggregated coding vector is the aggregated coding feature.

[0020] According to a specific embodiment of the present invention, the feature fusion coding of the ambient temperature, ambient humidity and wall material is performed to obtain the spliced ​​coding vector of each environmental parameter, further comprising:

[0021] Based on deep learning technology, low-dimensional embedding coding is performed on the ambient temperature, ambient humidity and wall material respectively, and the corresponding low-dimensional embedding coding vector of the ambient temperature, the low-dimensional embedding coding vector of the ambient humidity and the low-dimensional embedding coding vector of the wall material are obtained;

[0022] The low-dimensional embedded coding vectors of each environmental parameter are concatenated to obtain a concatenated coding vector of each environmental parameter.

[0023] According to a specific embodiment of the present invention, the electromagnetic distribution feature map and the spliced ​​coding vector are input into the cross-modal analysis network to obtain the aggregated coding vector of the current environmental test parameters. The aggregated coding vector, i.e., the aggregated coding feature, includes:

[0024] Perform feature decoupling and feature flattening on the electromagnetic distribution feature map to obtain a set of local feature vectors of the electromagnetic distribution;

[0025] Calculate the cluster center of the local feature vector set and the concatenated encoding vector to obtain the cluster center encoding vector;

[0026] The cluster center encoding vector and the set of local feature vectors are input into the cross-modal aggregation descriptor based on the cluster center, and the aggregated encoding vector of the current environment test parameters is output.

[0027] According to a specific embodiment of the present invention, calculating the cluster center of the local feature vector set and the concatenated code vector to obtain the cluster center code vector further includes:

[0028] The local feature vector set is input into the modal kernel feature extraction network to obtain the kernel semantic feature encoding vector of the electromagnetic distribution;

[0029] The cluster center encoding vector is determined based on the concatenated encoding vector and the kernel semantic feature encoding vector of the electromagnetic distribution.

[0030] According to a specific embodiment of the present invention, optimizing the RSSI initial test value based on the aggregated coding feature to obtain the RSSI optimized test value further includes:

[0031] calculating a correction coefficient based on the aggregated coding features;

[0032] Multiply the correction coefficient by the initial RSSI test value to obtain the optimized RSSI test value.

[0033] According to a specific embodiment of the present invention, calculating the correction coefficient based on the aggregated coding feature further includes:

[0034] The aggregated coding vector is input into a correction coefficient estimation module based on a feedforward neural network model to obtain the correction coefficient.

[0035] A test system for a mobile communication indoor signal monitor, comprising:

[0036] The data acquisition module is used to collect multiple RSSI values ​​and current environmental test parameters of the test area, including ambient temperature, ambient humidity, wall material, and electromagnetic environment scanning spectrum images;

[0037] A calculation module, configured to calculate an RSSI initial test value based on multiple RSSI values;

[0038] The analysis module is used to perform cross-modal aggregation analysis based on the current environment test parameters to obtain aggregate coding features;

[0039] The optimization module is used to optimize the RSSI initial test value based on the aggregated coding feature to obtain the RSSI optimized test value.

[0040] According to a specific embodiment of the present invention, the analysis module further includes:

[0041] A feature extraction module is used to extract electromagnetic distribution features based on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map;

[0042] The feature fusion module is used to perform feature fusion encoding on the ambient temperature, ambient humidity and wall material to obtain the spliced ​​coding vector of each environmental parameter;

[0043] The aggregation coding module is used to input the electromagnetic distribution feature map and the splicing coding vector into the cross-modal analysis network to obtain the aggregation coding vector of the current environmental test parameters. The aggregation coding vector is the aggregation coding feature.

[0044] According to a specific embodiment of the present invention, the feature fusion module further includes:

[0045] A low-dimensional embedding coding module is used to perform low-dimensional embedding coding on the ambient temperature, ambient humidity, and wall material based on deep learning technology, and obtain a low-dimensional embedding coding vector for the ambient temperature, a low-dimensional embedding coding vector for the ambient humidity, and a low-dimensional embedding coding vector for the wall material;

[0046] The vector splicing module is used to splice the low-dimensional embedded coding vectors of each environmental parameter to obtain the spliced ​​coding vectors of each environmental parameter.

[0047] According to a specific embodiment of the present invention, the aggregation coding module includes:

[0048] A feature decoupling module is used to perform feature decoupling and feature flattening processing on the electromagnetic distribution feature map to obtain a set of local feature vectors of the electromagnetic distribution;

[0049] The cluster center module is used to calculate the cluster center of the local feature vector set and the spliced ​​coding vector to obtain the cluster center coding vector;

[0050] The aggregation coding generation module is used to input the cluster center coding vector and the local feature vector set into the cluster center-based cross-modal aggregation descriptor, and output the aggregation coding vector of the current environment test parameters.

[0051] According to a specific embodiment of the present invention, the optimization module further includes:

[0052] A correction coefficient calculation module, configured to calculate a correction coefficient based on aggregated coding features;

[0053] The RSSI optimization module is used to multiply the correction coefficient by the RSSI initial test value to obtain the RSSI optimized test value.

[0054] An electronic device includes: a processor and a memory, wherein a computer program is stored in the memory and loaded and executed by the processor to implement the above-mentioned test method for the mobile communication indoor signal monitor.

[0055] A computer-readable storage medium stores a computer program, which is loaded and executed by a processor to implement the above-mentioned testing method for a mobile communication indoor signal monitor.

[0056] Compared with the prior art, the present invention provides a testing method, system, and storage medium for a mobile communication indoor signal monitor, which has the following advantages:

[0057] The present invention first arranges multiple signal detectors in the indoor test area to collect RSSI values, and uses the average of the multiple RSSI values ​​collected as the RSSI initial test value. By obtaining the current ambient temperature, ambient humidity, wall material and electromagnetic environment scanning spectrum image, and introducing deep learning-based data processing technology to perform cross-modal correlation analysis on multi-source environmental data, the degree of influence of the current environment on the RSSI test value is quantified, and a correction coefficient is generated to compensate and optimize the RSSI initial test value. By comprehensively considering the influence of multiple factors such as ambient temperature, humidity, wall material and electromagnetic environment on the RSSI test value, this method can more accurately reflect the true quality of mobile communication indoor signals and improve the accuracy and reliability of signal monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 The figure is a flow chart of a testing method for a mobile communication indoor signal monitor according to an embodiment of the present invention.

[0060] Figure 2 This is a flow chart of a method for collecting multiple RSSI values ​​of a test area and current environmental test parameters according to an embodiment of the present invention.

[0061] Figure 3This is a flow chart of a method for performing cross-modal aggregation analysis based on current environment test parameters provided according to an embodiment of the present invention.

[0062] Figure 4 This is a flow chart of a method for feature fusion coding of ambient temperature, humidity and wall material provided according to an embodiment of the present invention.

[0063] Figure 5 This is a flow chart of a method for calculating an aggregated coding vector of current environment test parameters provided according to an embodiment of the present invention.

[0064] Figure 6 4 is a flow chart of a method for calculating cluster centers of a set of local feature vectors and a concatenated coding vector according to an embodiment of the present invention.

[0065] Figure 7 4 is a flow chart of a method for optimizing an initial RSSI test value based on aggregated coding features according to an embodiment of the present invention.

[0066] Figure 8 1 is a structural diagram of a test system for a mobile communication indoor signal monitor according to an embodiment of the present invention.

[0067] Figure 9 is a structural diagram of an analysis module provided according to an embodiment of the present invention.

[0068] Figure 10 2 is a structural diagram of a feature fusion module provided according to an embodiment of the present invention.

[0069] Figure 11 2 is a structural diagram of an aggregate coding module provided according to an embodiment of the present invention.

[0070] Figure 12 is a structural diagram of an optimization module provided according to an embodiment of the present invention.

[0071] Figure 13 It is a schematic diagram of the structure of a computer device provided according to an embodiment of the present invention.

[0072] Reference numerals:

[0073] 01-Data acquisition module; 02-Calculation module; 03-Analysis module; 04-Optimization module;

[0074] 031-feature extraction module; 032-feature fusion module; 033-aggregation coding module;

[0075] 0321-Low-dimensional embedding coding module; 0322-Vector splicing module;

[0076] 0331-Feature decoupling module; 0332-Clustering center module; 0333-Aggregation code generation module;

[0077] 041-Correction coefficient calculation module; 042-RSSI optimization module. DETAILED DESCRIPTION

[0078] In order to make those skilled in the art understand the concept and thought of the present invention more clearly, the present invention is described in detail below in conjunction with specific embodiment.It should be understood that the embodiment provided herein is only a part of all possible embodiments of the present invention.After reading the specification of the application, those skilled in the art have the ability to make improvements, transformations, or replacements to part or all of the following embodiments, and these improvements, transformations, or replacements are also included in the scope of protection claimed in the present invention.

[0079] In this document, the terms "advance", "entry" and other similar words are not intended to imply any order, quantity and importance, but are merely used to distinguish different elements. In this document, the terms "one", "an" and other similar words are not intended to indicate that there is only one thing, but rather that the relevant description is only for one of the things, and the thing may have one or more. In this document, the terms "comprise", "include" and other similar words are intended to indicate logical relationships, and cannot be regarded as indicating relationships in spatial structure. For example, "A includes B" is intended to indicate that B logically belongs to A, and does not mean that B is spatially located inside A. In addition, the meanings of the terms "comprise", "include" and other similar words should be regarded as open, not closed. For example, "A includes B" is intended to indicate that B belongs to A, but B does not necessarily constitute the whole of A, and A may also include other elements such as C, D, and E.

[0080] In this document, the terms "embodiment," "this embodiment," "one embodiment," and "an embodiment" do not indicate that the description applies only to a specific embodiment, but rather indicate that the description may also apply to one or more other embodiments. Those skilled in the art should understand that any description of a particular embodiment herein may be substituted, combined, or otherwise combined with the description of one or more other embodiments. New embodiments resulting from such substitution, combination, or other combination are readily conceivable by those skilled in the art and fall within the scope of protection of this invention.

[0081] Example 1

[0082] Additional aspects and advantages of embodiments of the present invention will be given in part in the following description and will become apparent from the following description or learned through practice of embodiments of the present invention. Figure 1-Figure 7, an embodiment of the present invention provides a testing method for a mobile communication indoor signal monitor, comprising:

[0083] S1: Collect multiple RSSI values ​​of the area to be tested and the current environmental test parameters.

[0084] S2: Calculate an RSSI initial test value based on multiple RSSI values.

[0085] S3: Perform cross-modal aggregation analysis based on the current environment test parameters to obtain aggregated coding features.

[0086] S4: Optimize the RSSI initial test value based on the aggregated coding feature to obtain an RSSI optimized test value.

[0087] The present invention first arranges multiple signal detectors in the indoor test area to collect RSSI values, and uses the average of the multiple RSSI values ​​collected as the RSSI initial test value. By obtaining the current ambient temperature, ambient humidity, wall material and electromagnetic environment scanning spectrum image, and introducing deep learning-based data processing technology to perform cross-modal correlation analysis on multi-source environmental data, the degree of influence of the current environment on the RSSI test value is quantified, and a correction coefficient is generated to compensate and optimize the RSSI initial test value. By comprehensively considering the influence of multiple factors such as ambient temperature, humidity, wall material and electromagnetic environment on the RSSI test value, the present invention can more accurately reflect the true quality of mobile communication indoor signals and improve the accuracy and reliability of signal monitoring.

[0088] Specifically, step S1 collects multiple RSSI values ​​of the area to be tested and current environmental test parameters including:

[0089] S11: Select multiple reference points in the area to be tested and arrange signal detectors, and collect multiple RSSI values ​​through the signal detectors.

[0090] A signal detector senses the wireless signal strength at its location and converts it into an RSSI (Received Signal Strength Indicator) value. Considering that signal strength can vary in different indoor areas due to factors such as distance from the signal source and obstruction, to reduce measurement deviations caused by local signal fluctuations or special environmental factors, the present invention implements a comprehensive indoor coverage system by placing signal detectors at multiple reference points within the test area. This provides a more accurate and comprehensive RSSI data foundation for subsequent data analysis and processing.

[0091] In a specific embodiment of the present invention, in order to build a monitoring network that can fully cover the target space and accurately reflect the distribution of mobile communication signal strength in the area, it is necessary to comprehensively consider various factors when selecting and arranging these signal detectors. First of all, the selection of the test area is crucial. The ideal test area should be able to represent different types of indoor environments, such as open office areas, conference rooms, corridors, and closed rooms. By covering as wide a range of environment types as possible, more comprehensive data can be obtained for subsequent analysis. At the same time, considering the complex and diverse internal structures of buildings, including but not limited to differences in floor heights, differences in wall thickness and materials, changes in the number and position of windows, etc., all of which will affect the propagation of wireless signals, it is necessary to try to select locations with representative characteristics as reference points to ensure that the collected data can truly reflect the signal changes in the actual environment.

[0092] In a specific embodiment of the present invention, after the test area is determined, the next step is to specifically arrange the positions of the signal detectors to achieve the optimal spatial layout. Ideally, the spacing between signal detectors should be determined based on the expected measurement accuracy. Generally speaking, in relatively open areas with fewer obstacles, the distance between detectors can be appropriately increased. In areas with more obstacles or obvious signal attenuation, the distance between detectors needs to be shortened to ensure that the data acquisition density is high enough to capture subtle signal fluctuations. In addition, care should be taken to avoid placing the detector near devices that may generate strong electromagnetic interference, such as microwave ovens, wireless routers, etc., so as to prevent the interference signals emitted by them from having a negative impact on the measurement results.

[0093] In one embodiment of the present invention, due to the propagation characteristics of radio waves, signal strength varies with height, particularly in the presence of floor partitions. Therefore, properly distributing the detector heights vertically helps to obtain a more three-dimensional signal distribution map, providing more detailed information support for subsequent analysis. For example, in a multi-story building, detectors can be installed at strategic locations on each floor, ensuring that they are located at similar height levels, thereby forming a data collection grid within a three-dimensional coordinate system.

[0094] In a specific embodiment of the present invention, in order to ensure that the signal detector can work stably and continuously provide accurate and reliable RSSI values, the power supply method and data transmission mechanism also need to be considered. For fixed-installed detectors, they can be directly connected to the mains or use a battery-powered solution with long-term battery life. For portable or temporarily deployed detectors, rechargeable batteries are used for power supply and are equipped with a low-power mode to extend working time. In terms of data transmission, wired connections (such as Ethernet) or wireless connections (such as Wi-Fi, Bluetooth, Zigbee, etc.) are selected according to the requirements and constraints of the application scenario. Since wireless transmission may be affected by the surrounding environment and may cause delays or packet loss, it is also necessary to fully evaluate various possible risks and take corresponding measures to avoid them.

[0095] S12: Collect current environmental test parameters of the area to be tested, including ambient temperature, ambient humidity, wall material, and electromagnetic environment scanning spectrum image.

[0096] Because indoor environments are subject to signal reflection, attenuation, and shielding, signal acquisition accuracy can be insufficient. Therefore, the present invention comprehensively considers the impact of environmental factors on RSSI test values. By acquiring current environmental test parameters, the impact of these parameters on signal propagation characteristics is analyzed and used to calibrate the initial RSSI test value. The current environmental test parameters include ambient temperature, ambient humidity, wall material, and an electromagnetic environment scanning spectrum image.

[0097] In a specific embodiment of the present invention, when considering the ambient temperature, a high-precision temperature sensor network needs to be deployed to cover the entire test area. These temperature sensors have a fast response capability and can accurately reflect changes in the ambient air temperature in a short period of time. Taking into account that there may be local temperature differences in different locations, for example, places near windows or vents are usually colder or hotter than the center of the room, the layout of temperature sensors needs to take these special areas into account to ensure data representativeness. In addition, in order to improve the temporal resolution of the measurement results, the embodiment of the present invention adopts a continuous monitoring mode, that is, the temperature readings are automatically recorded at fixed time intervals (such as every minute), and the data are transmitted to the central processing unit via wireless or wired connection for storage and preliminary analysis.

[0098] In a specific embodiment of the present invention, humidity levels also significantly affect radio wave propagation, especially at high frequencies. The present invention utilizes products that can simultaneously measure both relative and absolute humidity. Relative humidity reflects the ratio of the water vapor content in the air relative to the saturation state, while absolute humidity directly indicates the actual mass of water vapor contained per unit volume of air. By combining these two indicators, more complete information can be obtained. The humidity sensor should be installed in the same location as the temperature sensor to ensure that both collect data synchronously for later correlation analysis.

[0099] In a specific embodiment of the present invention, different types of building materials, such as concrete, masonry, wood, glass, etc., each have unique electrical properties that will hinder or reflect radio signals to varying degrees. In order to accurately obtain wall information, the present invention adopts two complementary methods: one is to consult the architectural design drawings to extract detailed descriptions of the wall structure and the materials used; the other is to conduct on-site surveys and use portable testing equipment to directly measure the thickness and surface characteristics of the wall. For the latter, there are many instruments on the market specifically designed for building structure assessment, such as non-destructive testers, which can complete the task without damaging the wall. In addition, considering the existence of multi-layer composite wall structures, it is also necessary to use advanced tools such as ultrasonic detectors to deeply explore the internal layer distribution.

[0100] In a specific embodiment of the present invention, the acquisition of the electromagnetic environment scanning spectrum image mainly relies on a spectrum analyzer, which is a device that can accurately identify and quantify the presence of various radio frequency signals within a wide frequency range. During operation, the appropriate scanning range must first be set, usually covering from a few hundred megahertz to a few gigahertz, and then the spectrum analyzer is moved point by point along a predetermined path to ensure that every corner of the test area is covered. After each movement, the device automatically performs a full-band scan and generates a corresponding spectrum diagram. These images intuitively show the distribution of signal strength at each frequency, making it easier to identify interference sources and evaluate signal purity. Due to the complexity and variability of the electromagnetic environment, instantaneous strong interference may occur in certain periods of time. Therefore, the average value can be taken through repeated scanning to reduce the impact of accidental errors. In addition, directional antennas can be used in conjunction to enhance the receiving sensitivity in a specific direction and further refine the details of the spectrum image.

[0101] Specifically, step S2 of calculating the RSSI initial test value based on multiple RSSI values ​​includes:

[0102] Calculate the average of multiple RSSI values ​​to obtain the initial RSSI test value.

[0103] The present invention takes into account that the RSSI value of a single reference point may be affected by factors such as instantaneous interference and signal reflection, resulting in large fluctuations. Therefore, in order to eliminate the influence of some random noise and local interference on signal strength measurement, the present invention calculates the average of multiple RSSI values, thereby smoothing out noise fluctuations to a certain extent, thereby obtaining a relatively stable RSSI initial test value that can represent the overall signal strength in the room.

[0104] In a specific embodiment of the present invention, before calculation begins, all RSSI values ​​used for calculation must be initially screened to remove obviously erroneous or invalid data points, such as those with extreme values ​​far below or above the normal range. Furthermore, given the potential for dramatic signal fluctuations in real-world environments, this embodiment of the present invention utilizes a sliding window mechanism to collect continuous RSSI readings over a period of time, rather than relying solely on a single snapshot. This mechanism better reflects the temporal trend of signal strength and reduces the impact of transient interference. Next, based on the prepared data, the mean RSSI value is calculated. Specifically, assuming there are N valid RSSI measurements, denoted as RSSI_1, RSSI_2, ..., RSSI_N, the initial RSSI test value (i.e., the average RSSI value) can be expressed as the sum of these values ​​divided by the total number N. This allows for the rapid and efficient aggregation of dispersed independent observations to generate a single, generalized metric describing the average signal strength across the entire monitoring area.

[0105] Specifically, step S3 performs cross-modal aggregation analysis based on the current environment test parameters, and obtains the aggregated coding features including:

[0106] Taking the electromagnetic environment scanning spectrum image as the main variable and the ambient temperature, ambient humidity and wall material as covariates, a main-covariate cross-modal aggregation analysis is performed on the current environmental test parameters to obtain the aggregated coding features of the current environmental test parameters.

[0107] The electromagnetic environment scanning spectrum image can reflect the frequency distribution of other electromagnetic devices in the current environment that may interfere with mobile communication signals. It is a key factor affecting the RSSI test value. Therefore, the present invention focuses on it as the main variable for analysis. Changes in ambient temperature and humidity can alter the dielectric constant of the air, thereby affecting the signal propagation speed and attenuation. Furthermore, differences in wall materials can lead to differences in signal reflection, refraction, and attenuation during penetration. Although the impact of ambient temperature, humidity, and wall material on the RSSI test value is relatively small compared to electromagnetic interference, these three factors do indirectly affect signal propagation in different ways and also have a certain modulating effect on the propagation of electromagnetic interference signals. Therefore, the present invention analyzes ambient temperature, humidity, and wall material as covariates and comprehensively considers their interactions with the main variable, the electromagnetic environment scanning spectrum image, to more accurately quantify the impact of environmental factors on the RSSI test value.

[0108] Specifically, step S3 performs cross-modal aggregation analysis based on the current environment test parameters, and obtains the aggregated coding features including:

[0109] S31: Extracting electromagnetic distribution features based on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map.

[0110] In one embodiment of the present invention, considering that electromagnetic environment scan spectrum images contain rich information about the frequency distribution of electromagnetic devices, a neural network model is employed to extract key electromagnetic distribution features from the electromagnetic environment scan spectrum images, thereby generating an electromagnetic distribution feature map. For example, a suitable pre-trained CNN model, such as VGGNet or ResNet, is selected as a feature extractor. Using the concept of transfer learning, the pre-trained model is fine-tuned on the electromagnetic environment scan spectrum image dataset, enabling it to accurately identify key features of the electromagnetic environment, such as electromagnetic frequency distribution patterns and peak changes in electromagnetic intensity. This generates an electromagnetic distribution feature map, providing an important basis for subsequent RSSI value correction.

[0111] S32: Perform feature fusion coding on the ambient temperature, ambient humidity, and wall material to obtain a concatenated coding vector of each environmental parameter, further comprising:

[0112] S321: Based on deep learning technology, low-dimensional embedding coding is performed on the ambient temperature, ambient humidity and wall material respectively, and a low-dimensional embedding coding vector of the ambient temperature, a low-dimensional embedding coding vector of the ambient humidity and a low-dimensional embedding coding vector of the wall material are obtained accordingly.

[0113] In a specific embodiment of the present invention, considering that ambient temperature, humidity, and wall material have different physical properties and data types, in order to incorporate them into a unified analysis framework, the present invention first performs low-dimensional embedding coding on the ambient temperature, ambient humidity, and wall material in the current environmental test parameters, converting them into numerical vectors to facilitate comprehensive analysis with electromagnetic distribution characteristics. Low-dimensional embedding coding is an effective data dimensionality reduction technique that can transform originally complex and heterogeneous environmental test parameters into feature representations with a unified format and dimension while preserving the information meaning of the original data. In a specific example of the present invention, linear transformation or neural network-based embedding methods can be used for numerical ambient temperature and humidity data. For example, through a fully connected neural network layer, the one-dimensional temperature and humidity data are mapped into a low-dimensional vector space to obtain low-dimensional embedding coding vectors for the ambient temperature and humidity. However, as wall material is non-numeric data, the information it contains is relatively complex and difficult to directly quantify. Therefore, the present invention uses embedding techniques based on natural language processing, such as Word2Vec and the BERT model, to convert the descriptive text of the wall material into numerical vectors to obtain low-dimensional embedding coding vectors for the wall material. The BERT model is a pre-trained language representation model that converts text into numerical vectors. When applied to text describing wall materials, the BERT model understands the semantic information in the text and generates a low-dimensional embedding encoding vector that reflects the characteristics of the wall material. Leveraging the self-attention mechanism in the Transformer architecture, it can simultaneously consider the relationship between each word and all other words in the text, rather than just the dependencies between preceding and following words, when processing sequential data. When encoding the wall material description, the BERT model not only focuses on the meaning of individual words but also captures the meaning of words in the context of the entire sentence and their interactions, thereby more accurately expressing the specific characteristics of the wall material. The Word2Vec model analyzes the co-occurrence of words in the wall material description text and converts these descriptions into numerical vectors to generate a low-dimensional embedding encoding vector for the wall material. This model, based on neural network technology, primarily employs two training architectures: Continuous Bag-of-Words (CBOW) and Skip-gram. The core idea of ​​these two methods is to predict a word based on its surrounding context (CBOW) or, conversely, to predict the surrounding context words based on a word (Skip-gram). When applied to wall material description, the Word2Vec model traverses the description text, learns the probability distribution of each word appearing in different contexts, and encodes this information into a fixed-length vector representation.

[0114] S322: Concatenate the low-dimensional embedded coding vectors of the environmental parameters to obtain concatenated coding vectors of the environmental parameters.

[0115] In a specific embodiment of the present invention, through the above-mentioned low-dimensional embedded coding processing, the ambient temperature, ambient humidity and wall material are converted into low-dimensional vector representations with unified dimension and format. In order to comprehensively consider the feature information of multiple covariates, the present invention further combines the ambient temperature low-dimensional embedded coding vector, the ambient humidity low-dimensional embedded coding vector and the wall material low-dimensional embedded coding vector through feature splicing operation to form a low-dimensional splicing coding vector of the environmental parameter covariate, thereby integrating multiple covariate information into a unified feature space, providing comprehensive and accurate environmental parameter information for the subsequent calculation of correction coefficients.

[0116] S33: Input the electromagnetic distribution feature map and the spliced ​​coding vector into the cross-modal analysis network to obtain the aggregated coding vector of the current environmental test parameters, which is the aggregated coding feature.

[0117] Because ambient temperature, humidity, and wall material not only affect the propagation of mobile communication signals but also modulate the propagation of electromagnetic interference signals, the present invention further captures the potential correlation between the low-dimensional concatenated coding vectors of environmental parameter covariates and the electromagnetic distribution characteristic maps by performing cross-modal interaction analysis. This study studies the modulation effect of the covariates on the primary variables, thereby more accurately revealing the impact pattern of environmental factors on RSSI test values.

[0118] Furthermore, in step S33, the electromagnetic distribution feature map and the concatenated coding vector are input into the cross-modal analysis network to obtain an aggregated coding vector of the current environmental test parameters. The aggregated coding vector, i.e., the aggregated coding feature, further includes:

[0119] S331: Perform feature decoupling and feature flattening processing on the electromagnetic distribution feature map to obtain a set of local feature vectors of the electromagnetic distribution.

[0120] In a specific embodiment of the present invention, step S331 is expressed as follows:

[0121] (1)

[0122] in, Represents the electromagnetic distribution characteristic diagram, Indicates feature decoupling processing, represents the set of local eigenvectors of electromagnetic distribution, 、 、 and They represent the first, second, and third local eigenvectors of the electromagnetic distribution. and local eigenvectors of the electromagnetic distribution, is the number of local eigenvectors of the electromagnetic distribution.

[0123] In order to achieve effective cross-domain aggregation interaction between the low-dimensional splicing encoding vector of the environmental parameter covariate and the electromagnetic distribution feature map, it is necessary to ensure that the data dimensions of the two are consistent. Traditional pooling methods often lead to indiscriminate loss of information during the dimensionality reduction process, which may lead to the loss of key feature information, affecting the model's accurate assessment of the electromagnetic distribution characteristics. Based on this, the present invention first performs feature decoupling and feature flattening on the electromagnetic distribution feature map to subdivide it into a series of smaller-grained sets of electromagnetic distribution local feature vectors. This not only retains more detailed information, but also provides a more delicate foundation for subsequent cross-domain interaction.

[0124] S332: Calculating the cluster center of the local feature vector set and the concatenated code vector to obtain the cluster center code vector, further comprising:

[0125] S3321: Input the local feature vector set into the modal kernel feature extraction network to obtain the kernel semantic feature encoding vector of the electromagnetic distribution.

[0126] In a specific embodiment of the present invention, step S3321 is expressed as follows:

[0127] (2)

[0128] in, The first in the set of local eigenvectors of electromagnetic distribution local eigenvectors of the electromagnetic distribution, The first in the set of local eigenvectors of electromagnetic distribution local eigenvectors of the electromagnetic distribution, represents the modal kernel feature extraction network, express The semantic difference factor of the set of local eigenvectors of the electromagnetic distribution, represents the one-norm of a vector, represents the exponential function with base e, represents the electromagnetic distribution kernel semantic feature encoding vector, and n represents the total number of electromagnetic distribution local feature vectors in the set of electromagnetic distribution local feature vectors.

[0129] S3322: Determine a cluster center encoding vector based on the concatenated encoding vector and the kernel semantic feature encoding vector of the electromagnetic distribution.

[0130] In a specific embodiment of the present invention, step S3322 is expressed as follows:

[0131] (3)

[0132] in, represents the low-dimensional concatenated encoding vector of the environmental parameter covariates, represents the weight matrix, Indicates cascade, represents the bias vector, Represents the center encoding vector of the fine-grained clustering of environmental parameter principal-covariate.

[0133] The present embodiment first uses a modal kernel feature extraction network to process a collection of local electromagnetic distribution feature vectors. During the dimensionality reduction process, it simultaneously selects more core and representative electromagnetic distribution features. This approach preserves key information in the electromagnetic distribution feature map and avoids the loss of important details, thereby generating an electromagnetic distribution kernel semantic feature encoding vector. Next, a neural network model is used to perform cross-domain association learning on the low-dimensional concatenated encoding vectors of the environmental parameter covariates after dimensionality unification and the electromagnetic distribution kernel semantic feature encoding vectors. This captures the potential connections between the electromagnetic distribution features and the covariate environmental parameter features. A cluster center is then identified in the high-dimensional feature space, focusing on the core correlation information between the two, to generate a fine-grained cluster center encoding vector for the environmental parameter principal and covariate. This cluster center encoding vector reflects the core correlation characteristics between the electromagnetic distribution and the covariate environmental parameters, providing an important reference for subsequent feature interaction analysis. This ensures that the core correlation between the electromagnetic distribution information and the environmental parameter information is accurately captured. It also guides the model to more closely focus on the potential correlations and interaction patterns between the two, thereby improving the accuracy of RSSI test value correction.

[0134] S333: Input the cluster center encoding vector and the local feature vector set into the cluster center-based cross-modal aggregation descriptor, and output the aggregated encoding vector of the current environment test parameters.

[0135] In a specific embodiment of the present invention, step S333 is expressed as follows:

[0136] (4)

[0137] (5)

[0138] (6)

[0139] in, express and The Poincare distance between represents the inverse hyperbolic cosine function, represents the gating threshold, express The corresponding gated clustering weight coefficient, Indicates the The local eigenvector of the electromagnetic distribution The eigenvalues ​​at the positions, Represents the first The eigenvalues ​​at the positions, Represents the fine-grained aggregate encoding vector of the principal and covariates of environmental parameters.

[0140] The present invention uses cluster center coding vectors as a benchmark to guide the collection of local electromagnetic distribution feature vectors for fine-grained aggregation analysis, generating fine-grained aggregation coding vectors of environmental parameter principal and covariate variables. The cluster center coding vectors are constructed in a cross-domain feature space based on the feature association between electromagnetic distribution information and covariate environmental parameter information. Using them as a benchmark for aggregation analysis can guide each local electromagnetic distribution feature to perform more refined aggregation based on the strength of its association with the covariate environmental parameter feature, thereby more accurately characterizing the interactive relationship between the electromagnetic distribution features and the covariate environmental parameter features. At the same time, it avoids feature confusion and information loss that may result from directly fusing data from different modalities, providing more accurate data support for subsequent RSSI test value correction.

[0141] Specifically, step S4 optimizes the RSSI initial test value based on the aggregated coding feature to obtain the RSSI optimized test value, including:

[0142] S41: Calculating a correction coefficient based on the aggregated coding features, specifically including:

[0143] The aggregated coding vector is input into a correction coefficient estimation module based on a feedforward neural network model to obtain the correction coefficient.

[0144] The feedforward neural network has powerful nonlinear mapping and self-learning capabilities. Through the layer-by-layer transmission of the internal hidden layer and the action of the nonlinear activation function, it can gradually extract and abstract the deep correlation features between multi-source environmental information from the input aggregated coding vector, and estimate the correction coefficient based on this to quantify the impact of electromagnetic interference on the RSSI test value under the current environmental conditions, thereby achieving accurate correction of the RSSI test value.

[0145] In a specific embodiment of the present invention, first, based on the eigenvalues ​​of the aggregated coding vector Distance and distance matrix and aggregated coded distance matrix ;

[0146] Secondly, the weighted sum of the aggregate coding distance matrix 1 and the aggregate coding distance matrix 2 is calculated to obtain the aggregate coding joint distance matrix, which is expressed as:

[0147] (7)

[0148] in, represents the aggregated coding-distance matrix, represents the aggregated encoding two-distance matrix, and Represents different weight parameters, represents dot product, Indicates point addition, represents the aggregated coded joint distance matrix.

[0149] Then, determine the eigenvalues ​​of the aggregated coding joint distance matrix , and form the aggregated coding joint distance eigenvector ;

[0150] Next, the aggregated coding vector as a row vector is matrix-multiplied with the aggregated coding-distance matrix to obtain the aggregated coding-distance query vector, which is expressed as:

[0151] (8)

[0152] in, represents the aggregated encoding vector, represents the aggregated encoding-distance query vector;

[0153] Then, the aggregated coding two-distance matrix is ​​matrix-multiplied with the autocorrelation matrix of the aggregated coding vector to obtain the aggregated coding two-distance correlation matrix, which is expressed as:

[0154] (9)

[0155] in, represents the transpose of a vector, represents the aggregated coding two-distance association matrix;

[0156] Next, after matrix multiplication of the aggregated coding distance query vector and the aggregated coding distance association matrix, the optimized aggregated coding vector is obtained by further dot multiplication with the aggregated coding joint distance eigenvector, which can be expressed as:

[0157] (10)

[0158] in, represents the aggregated coding joint distance eigenvector, Represents the optimized aggregate encoding vector.

[0159] Finally, the aggregated coding vector is input into the correction coefficient estimation module based on the feedforward neural network model to obtain the correction coefficient.

[0160] When performing cross-domain fine-grained aggregation analysis, where the low-dimensional splicing coding vector of environmental parameter covariates and the electromagnetic distribution feature map represent the low-dimensional embedded splicing features of environmental parameter covariates and the electromagnetic distribution semantic coding features of the electromagnetic environment scanning spectrum image, respectively, insufficient prior cross-domain correlation correspondence of different modal data will cause sparse fine-grained correlation aggregation of the aggregated coding vector, which will reduce the accuracy of the correction coefficient obtained by the correction coefficient estimation module based on the feedforward neural network model due to the lack of feedforward reasoning.

[0161] Therefore, the one-distance matrix and the two-distance matrix of the aggregate coding vector are used as the fine-grained metric association cluster representation of the aggregate coding vector, and dynamic programming of the relationship between association clusters of different association clusters is performed on the aggregate coding vector and the auto-association representation of the aggregate coding vector respectively to simulate the sparse activation based on neuron clusters of the association system, and the intrinsic representation of the metric association cluster of the one-distance matrix and the two-distance matrix of the aggregate coding vector is used to coordinate the fine-grained predictable sparsity of the aggregate coding vector, so as to avoid the lack of association caused by sparsity affecting the loss of feedforward reasoning, and improve the accuracy of the correction coefficient obtained by the correction coefficient estimation module based on the feedforward neural network model of the aggregate coding vector input.

[0162] S42: Multiply the correction coefficient by the RSSI initial test value to obtain the RSSI optimized test value.

[0163] The correction coefficient is multiplied by the initial RSSI test value to achieve linear adjustment of the initial RSSI test value through multiplication, eliminating or reducing the impact of environmental factors on signal strength measurement, thereby obtaining an optimized RSSI test value that is closer to the actual signal strength.

[0164] In a specific embodiment of the present invention, after obtaining the RSSI optimization test value, adjustments related to the existing network configuration are immediately carried out. The current network layout is examined to see if it is reasonable based on the optimized RSSI value, areas with insufficient or excessive signal strength are identified, and necessary improvement measures are planned accordingly. For areas with weak signal coverage, consider adding additional base stations or micro-cellular devices to enhance the coverage range, and for locations where interference problems are caused by excessively strong signals, the situation is improved by adjusting the transmission power or other parameters. At the same time, the optimized RSSI value is used to evaluate the rationality of the currently used frequency band, and frequency planning adjustments are made when necessary to ensure optimal resource utilization. In addition, for specific environmental characteristics, such as the complexity of building structures, directional antennas or other special technical means are used for targeted deployment to ensure that every corner can obtain stable and high-quality signal services.

[0165] Conduct in-depth data analysis based on optimized RSSI values ​​to uncover the underlying insights. By comparing and analyzing large amounts of historical data with real-time data, we can identify regular trends and predict future changes, allowing us to proactively prepare for them. Combined with other relevant metrics (such as bit error rate (BER) and throughput), we can comprehensively assess network performance from multiple perspectives.

[0166] To ensure that the optimized RSSI value produces the desired effect in a real-world environment, a detailed long-term monitoring and maintenance strategy is also necessary. Given the dynamic and complex nature of wireless communication environments, routine inspections and maintenance of hardware facilities are essential. A detailed inspection plan should be developed, clearly defining the responsible individuals and deadlines for each task to ensure that every maintenance activity is completed on time and to quality standards. Furthermore, for various possible emergencies, a detailed contingency plan should be prepared in advance, including troubleshooting procedures and the establishment of an emergency repair team, to ensure that any problems can be quickly and effectively addressed. Long-term monitoring is not only about keeping an eye on the status of hardware and software, but also about continuously accumulating experience and developing a set of effective response plans to keep the entire system in optimal operating condition.

[0167] Example 2

[0168] Based on the above method, the embodiment of the present invention also provides a test system for a mobile communication indoor signal monitor, such as Figures 8-12 Shown, including:

[0169] The data acquisition module 01 is used to collect multiple RSSI values ​​and current environmental test parameters of the area to be tested.

[0170] The calculation module 02 is configured to calculate an RSSI initial test value based on multiple RSSI values.

[0171] The analysis module 03 is used to perform cross-modal aggregation analysis based on the current environment test parameters to obtain aggregated coding features.

[0172] The optimization module 04 is used to optimize the RSSI initial test value based on the aggregated coding feature to obtain the RSSI optimized test value.

[0173] The present invention first uses the data acquisition module 01 to collect multiple RSSI values ​​and current environmental test parameters of the area to be tested, including the current ambient temperature, ambient humidity, wall material and electromagnetic environment scanning spectrum image, and uses the calculation module 02 to calculate the average of the multiple RSSI values ​​as the RSSI initial test value. The analysis module 03 performs cross-modal correlation analysis on the multi-source environmental data to quantify the degree of influence of the current environment on the RSSI test value, generates a correction coefficient, and finally compensates and optimizes the RSSI initial test value through the optimization module 04. By comprehensively considering the influence of multiple factors such as ambient temperature, humidity, wall material and electromagnetic environment on the RSSI test value, the present invention can more accurately reflect the true quality of the mobile communication indoor signal and improve the accuracy and reliability of signal monitoring.

[0174] Specifically, the analysis module 03 also includes:

[0175] The feature extraction module 031 is used to extract electromagnetic distribution features based on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map.

[0176] In a specific embodiment of the present invention, considering that the electromagnetic environment scanning spectrum image contains rich information about the frequency distribution of electromagnetic devices, in order to extract key electromagnetic distribution features from it, the present invention employs a feature extraction module 031 to perform feature extraction on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map. For example, a suitable pre-trained CNN model, such as VGGNet or ResNet, is selected as a feature extractor. Utilizing the concept of transfer learning, the pre-trained model is fine-tuned on the electromagnetic environment scanning spectrum image dataset, enabling it to accurately identify key features of the electromagnetic environment, such as electromagnetic frequency distribution patterns and peak changes in electromagnetic intensity. This generates an electromagnetic distribution feature map, providing an important basis for subsequent RSSI value correction.

[0177] The feature fusion module 032 is used to perform feature fusion coding on the ambient temperature, ambient humidity and wall material to obtain a spliced ​​coding vector of each environmental parameter.

[0178] Specifically, the feature fusion module 032 further includes:

[0179] The low-dimensional embedding coding module 0321 is used to perform low-dimensional embedding coding on the ambient temperature, ambient humidity and wall material based on deep learning technology, and obtain a low-dimensional embedding coding vector for the ambient temperature, a low-dimensional embedding coding vector for the ambient humidity and a low-dimensional embedding coding vector for the wall material.

[0180] In a specific embodiment of the present invention, considering that the ambient temperature, humidity and wall material have different physical properties and data types, in order to incorporate them into a unified analysis framework, the present invention uses a low-dimensional embedding coding module 0321 to perform low-dimensional embedding coding on the ambient temperature, ambient humidity and wall material in the current environmental test parameters, and converts them into a numerical vector form to facilitate comprehensive analysis with electromagnetic distribution characteristics.

[0181] The vector splicing module 0322 is used to splice the low-dimensional embedded coding vectors of each environmental parameter to obtain a spliced ​​coding vector of each environmental parameter.

[0182] In a specific embodiment of the present invention, through the above-mentioned low-dimensional embedded coding processing, the ambient temperature, ambient humidity and wall material are converted into low-dimensional vector representations with unified dimensions and formats. In order to comprehensively consider the characteristic information of multiple covariates, the present invention performs a feature splicing operation through the vector splicing module 0322, and combines the ambient temperature low-dimensional embedded coding vector, the ambient humidity low-dimensional embedded coding vector and the wall material low-dimensional embedded coding vector to form a low-dimensional splicing coding vector of the environmental parameter covariate, thereby integrating multiple covariate information into a unified feature space, providing comprehensive and accurate environmental parameter information for the subsequent calculation of the correction coefficient.

[0183] The aggregate coding module 033 is used to input the electromagnetic distribution feature map and the spliced ​​coding vector into the cross-modal analysis network to obtain the aggregate coding vector of the current environmental test parameters. The aggregate coding vector is the aggregate coding feature.

[0184] Because ambient temperature, humidity, and wall material not only affect the propagation of mobile communication signals but also modulate the propagation of electromagnetic interference signals, the present invention further utilizes the aggregation coding module 033 to perform cross-modal interactive analysis on the low-dimensional concatenated coding vector of the environmental parameter covariates and the electromagnetic distribution characteristic map to capture the potential correlation between the two and learn the modulation effect of the covariates on the main variables, thereby more accurately revealing the impact pattern of environmental factors on RSSI test values.

[0185] Specifically, the aggregation coding module 033 includes:

[0186] The feature decoupling module 0331 is used to perform feature decoupling and feature flattening processing on the electromagnetic distribution feature map to obtain a set of local feature vectors of the electromagnetic distribution.

[0187] In order to achieve effective cross-domain aggregation interaction between the low-dimensional spliced ​​encoding vector of the environmental parameter covariate and the electromagnetic distribution feature map, it is necessary to ensure that the data dimensions of the two are consistent. Traditional pooling methods often lead to indiscriminate loss of information during the dimensionality reduction process, which may in turn lead to the loss of key feature information, affecting the model's accurate assessment of the electromagnetic distribution characteristics. Based on this, the present invention uses the feature decoupling module 0331 to perform feature decoupling and feature flattening processing on the electromagnetic distribution feature map to subdivide it into a series of smaller-grained sets of electromagnetic distribution local feature vectors. This not only retains more detailed information, but also provides a more delicate foundation for subsequent cross-domain interaction.

[0188] The cluster center module 0332 is used to calculate the cluster center of the local feature vector set and the concatenated code vector to obtain the cluster center code vector.

[0189] In a specific embodiment of the present invention, the set of local feature vectors is input into the modal kernel feature extraction network via the cluster center module 0332 to obtain the kernel semantic feature encoding vector of the electromagnetic distribution. The cluster center encoding vector is then determined based on the concatenated encoding vector and the kernel semantic feature encoding vector of the electromagnetic distribution.

[0190] The present embodiment first uses a modal kernel feature extraction network to process a collection of local electromagnetic distribution feature vectors. During the dimensionality reduction process, it simultaneously selects more core and representative electromagnetic distribution features. This approach preserves key information in the electromagnetic distribution feature map and avoids the loss of important details, thereby generating an electromagnetic distribution kernel semantic feature encoding vector. Next, a neural network model is used to perform cross-domain association learning on the low-dimensional concatenated encoding vectors of the environmental parameter covariates after dimensionality unification and the electromagnetic distribution kernel semantic feature encoding vectors. This captures the potential connections between the electromagnetic distribution features and the covariate environmental parameter features. A cluster center is then identified in the high-dimensional feature space, focusing on the core correlation information between the two, to generate a fine-grained cluster center encoding vector for the environmental parameter principal and covariate. This cluster center encoding vector reflects the core correlation characteristics between the electromagnetic distribution and the covariate environmental parameters, providing an important reference for subsequent feature interaction analysis. This ensures that the core correlation between the electromagnetic distribution information and the environmental parameter information is accurately captured. It also guides the model to more closely focus on the potential correlations and interaction patterns between the two, thereby improving the accuracy of RSSI test value correction.

[0191] The aggregated coding generation module 0333 is used to input the cluster center coding vector and the local feature vector set into the cluster center-based cross-modal aggregation descriptor, and output the aggregated coding vector of the current environment test parameters.

[0192] In a specific embodiment of the present invention, the cluster center encoding vector and the set of local feature vectors are input into a cluster center-based cross-modal aggregation descriptor via the aggregation encoding generation module 0333, which outputs an aggregate encoding vector for the current environmental test parameters. The present invention uses the cluster center encoding vector as a benchmark to guide the set of electromagnetic distribution local feature vectors for fine-grained aggregation analysis, generating a fine-grained aggregate encoding vector for the principal and covariate environmental parameters. The cluster center encoding vector is constructed in a cross-domain feature space based on the feature correlation between electromagnetic distribution information and covariate environmental parameter information. Using it as a benchmark for aggregation analysis can guide the more refined aggregation of individual electromagnetic distribution local features based on their correlation strength with the covariate environmental parameter features, thereby more accurately characterizing the interactive relationship between the electromagnetic distribution features and the covariate environmental parameter features. This avoids feature confusion and information loss that can result from directly fusing data from different modalities, providing more accurate data support for subsequent RSSI test value correction.

[0193] Specifically, the optimization module 04 also includes:

[0194] The correction coefficient calculation module 041 is used to calculate the correction coefficient based on the aggregated coding features.

[0195] In a specific embodiment of the present invention, the correction coefficient is obtained by inputting the aggregated code vector into the correction coefficient calculation module 041 for calculation. This is used to quantitatively reflect the degree of impact of electromagnetic interference on the RSSI test value under current environmental conditions, thereby achieving accurate correction of the RSSI test value.

[0196] The RSSI optimization module 042 is configured to multiply the correction coefficient by the RSSI initial test value to obtain an RSSI optimized test value.

[0197] In a specific embodiment of the present invention, the correction coefficient is multiplied by the RSSI initial test value through the RSSI optimization module 042 to achieve linear adjustment of the RSSI initial test value, eliminate or reduce the impact of environmental factors on the signal strength measurement, and thus obtain an RSSI optimized test value that is closer to the actual signal strength.

[0198] Example 3

[0199] like Figure 13 As shown, an embodiment of the present invention further provides an electronic device comprising: a processor and a memory, wherein the memory stores a computer program, which is loaded and executed by the processor to implement the above-mentioned testing method for a mobile communication indoor signal monitor. The device in the present invention can be a server, a PC, a PAD, a mobile phone, etc.

[0200] Furthermore, an embodiment of the present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program is loaded and executed by a processor to implement the above-mentioned testing method of the mobile communication indoor signal monitor.

[0201] In summary, the testing method, system, and storage medium for a mobile communication indoor signal monitor provided by the present invention have the following advantages:

[0202] The present invention first arranges multiple signal detectors in the indoor test area to collect RSSI values, and uses the average of the multiple RSSI values ​​collected as the RSSI initial test value. By obtaining the current ambient temperature, ambient humidity, wall material and electromagnetic environment scanning spectrum image, and introducing deep learning-based data processing technology to perform cross-modal correlation analysis on multi-source environmental data, the degree of influence of the current environment on the RSSI test value is quantified, and a correction coefficient is generated to compensate and optimize the RSSI initial test value. By comprehensively considering the influence of multiple factors such as ambient temperature, humidity, wall material and electromagnetic environment on the RSSI test value, this method can more accurately reflect the true quality of mobile communication indoor signals and improve the accuracy and reliability of signal monitoring.

[0203] The concepts, principles, and concepts of the present invention have been described in detail above with reference to specific implementation methods (including embodiments and examples). Those skilled in the art should understand that the present invention may be implemented in more than just the forms described above. After reading this application document, those skilled in the art may make any possible improvements, substitutions, and equivalent forms to the steps, methods, systems, and components in the above-described implementation methods. Such improvements, substitutions, and equivalent forms should be deemed to fall within the scope of the present invention. The scope of protection of the present invention shall be determined solely by the claims.

Claims

1. A testing method for a mobile communication indoor signal monitor, characterized in that: include: Collect multiple RSSI values ​​and current environmental test parameters of the area to be tested, wherein the current environmental test parameters include ambient temperature, ambient humidity, wall material, and electromagnetic environment scanning spectrum image; Calculating an RSSI initial test value based on the multiple RSSI values; Based on the current environment test parameters, cross-modal aggregation analysis is performed to obtain aggregated coding features, specifically including: Extracting electromagnetic distribution features based on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map; Performing feature fusion coding on the ambient temperature, the ambient humidity, and the wall material to obtain a spliced ​​coding vector of each environmental parameter further includes: Based on deep learning technology, low-dimensional embedding coding is performed on the ambient temperature, the ambient humidity, and the wall material, respectively, to obtain a low-dimensional embedding coding vector of the ambient temperature, a low-dimensional embedding coding vector of the ambient humidity, and a low-dimensional embedding coding vector of the wall material; The low-dimensional embedded coding vectors of the environmental parameters are concatenated to obtain concatenated coding vectors of the environmental parameters; Inputting the electromagnetic distribution characteristic map and the spliced ​​coding vector into a cross-modal analysis network to obtain an aggregated coding vector of the current environmental test parameter, the aggregated coding vector being an aggregated coding feature, further comprising: Performing feature decoupling and feature flattening processing on the electromagnetic distribution feature map to obtain a set of local feature vectors of the electromagnetic distribution; Calculating the cluster center of the local feature vector set and the concatenated code vector to obtain a cluster center code vector; Inputting the cluster center encoding vector and the local feature vector set into a cluster center-based cross-modal aggregation descriptor, and outputting an aggregated encoding vector of the current environment test parameter; The RSSI initial test value is optimized based on the aggregate coding feature to obtain an RSSI optimized test value.

2. The testing method of the mobile communication indoor signal monitor according to claim 1, characterized in that: The collecting of multiple RSSI values ​​of the area to be tested and current environmental test parameters includes: Select multiple reference points in the test area to arrange signal detectors, and collect multiple RSSI values ​​through the signal detectors; Collect the current environmental test parameters of the test area, including ambient temperature, ambient humidity, wall material, and electromagnetic environment scanning spectrum image.

3. The testing method of the mobile communication indoor signal monitor according to claim 2, characterized in that: Calculating the RSSI initial test value based on the multiple RSSI values ​​includes: An average value of the multiple RSSI values ​​is calculated to obtain an initial RSSI test value.

4. The testing method of the mobile communication indoor signal monitor according to claim 3, characterized in that: The calculating the cluster center of the local feature vector set and the concatenated code vector to obtain the cluster center code vector further includes: Inputting the local feature vector set into a modal kernel feature extraction network to obtain a kernel semantic feature encoding vector of the electromagnetic distribution; A cluster center coding vector is determined based on the concatenated coding vector and the kernel semantic feature coding vector of the electromagnetic distribution.

5. The testing method of the mobile communication indoor signal monitor according to claim 4, characterized in that: Optimizing the RSSI initial test value based on the aggregated coding feature to obtain the RSSI optimized test value further includes: calculating a correction coefficient based on the aggregated coding feature; The correction coefficient is multiplied by the RSSI initial test value to obtain the RSSI optimized test value.

6. The testing method of the mobile communication indoor signal monitor according to claim 5, characterized in that: Calculating the correction coefficient based on the aggregated coding feature further includes: The aggregated coding vector is input into a correction coefficient estimation module based on a feedforward neural network model to obtain a correction coefficient.

7. A test system for a mobile communication indoor signal monitor, characterized in that: include: A data acquisition module is used to collect multiple RSSI values ​​and current environmental test parameters of the area to be tested, wherein the current environmental test parameters include ambient temperature, ambient humidity, wall material, and electromagnetic environment scanning spectrum image; a calculation module, configured to calculate an RSSI initial test value based on the multiple RSSI values; An analysis module, configured to perform cross-modal aggregation analysis based on the current environment test parameters to obtain aggregated coding features; An optimization module, configured to optimize the RSSI initial test value based on the aggregated coding feature to obtain an RSSI optimized test value; The analysis module also includes: A feature extraction module is used to extract electromagnetic distribution features based on the electromagnetic environment scanning spectrum image to obtain an electromagnetic distribution feature map; A feature fusion module is used to perform feature fusion coding on the ambient temperature, the ambient humidity and the wall material to obtain a spliced ​​coding vector of each environmental parameter; an aggregate coding module, configured to input the electromagnetic distribution feature map and the spliced ​​coding vector into a cross-modal analysis network to obtain an aggregate coding vector of the current environmental test parameter, wherein the aggregate coding vector is an aggregate coding feature; The feature fusion module also includes: A low-dimensional embedding coding module is used to perform low-dimensional embedding coding on the ambient temperature, the ambient humidity, and the wall material based on deep learning technology, and obtain a low-dimensional embedding coding vector for the ambient temperature, a low-dimensional embedding coding vector for the ambient humidity, and a low-dimensional embedding coding vector for the wall material; A vector splicing module is used to splice the low-dimensional embedded coding vectors of each environmental parameter to obtain a spliced ​​coding vector of each environmental parameter; The aggregate coding module further includes: a feature decoupling module, configured to perform feature decoupling and feature flattening processing on the electromagnetic distribution feature map to obtain a set of local feature vectors of the electromagnetic distribution; A cluster center module, configured to calculate the cluster center of the local feature vector set and the concatenated code vector to obtain a cluster center code vector; The aggregate coding generation module is used to input the cluster center coding vector and the local feature vector set into a cluster center-based cross-modal aggregation descriptor, and output the aggregate coding vector of the current environment test parameter.

8. The test system for mobile communication indoor signal monitor according to claim 7, characterized in that: The optimization module also includes: a correction coefficient calculation module, configured to calculate a correction coefficient based on the aggregated coding feature; The RSSI optimization module is used to multiply the correction coefficient by the RSSI initial test value to obtain an RSSI optimized test value.

9. An electronic device, characterized in that: include: A processor and a memory, wherein a computer program is stored in the memory, and the computer program is loaded and executed by the processor to implement the testing method of the mobile communication indoor signal monitor according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which is loaded and executed by a processor to implement the testing method of the mobile communication indoor signal monitor according to any one of claims 1 to 6.

Images