Method for testing compatibility of mobile intelligent terminal

Abstract

The invention belongs to the technical field of terminal testing, and particularly relates to a mobile intelligent terminal compatibility testing method which comprises the following steps: S1, aggregating terminal hardware configuration, software environment and network state data through federal learning, and constructing a real-time testing matrix; s2, synthesizing an edge scene test case based on the generative adversarial network; s3, synchronously collecting a touch track, a display frame rate and an audio synchronization deviation, and quantifying user interaction experience in combination with eye tracker data; s4, deploying a lightweight virtualization intermediate layer between the gap / Android / iOS, automatically converting an equipment interconnection instruction, and verifying cross-chain compatibility; and S5, tracking the privacy data flow by adopting dynamic taint analysis, and detecting illegal behaviors in real time. According to the method, through dynamic environment modeling, real-time test matrix construction and hierarchical aggregation architecture, terminal nodes encrypt and upload local data by using differential privacy and homomorphic encryption technologies, so that the security of the data is ensured.

Description

Technical Field

[0001] This invention belongs to the field of terminal testing technology, specifically a testing method for compatibility testing of mobile smart terminals. Background Technology

[0002] With the rapid development of the mobile internet, the types and numbers of mobile smart terminals have increased dramatically. Significant differences exist between these terminals in hardware configurations (such as CPU / GPU performance and screen characteristics), software environments (such as operating system patch levels and API versions), and network conditions (such as 5G slicing types and Wi-Fi 6 channel bandwidth). This has led to increasingly prominent compatibility issues for mobile applications across different terminals, severely impacting user experience and application adoption.

[0003] Traditional mobile smart terminal compatibility testing methods have several shortcomings. Firstly, test data collection is often limited to a single type, failing to comprehensively cover various complex terminal environment combinations, resulting in inaccurate and incomplete test results. Secondly, test case generation lacks intelligence and specificity, making it difficult to effectively cover various edge scenarios, such as multi-task switching under low memory conditions and temperature threshold triggering under high load. Furthermore, in cross-platform testing such as HarmonyOS / Android / iOS, the conversion and compatibility verification of device interconnection commands are complex, and traditional methods are inefficient and prone to errors. Simultaneously, effective means are lacking for protecting privacy data and detecting violations during the testing process.

[0004] Therefore, the present invention provides a testing method for compatibility testing of mobile smart terminals. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve its technical problem is: a mobile smart terminal compatibility testing method according to this invention, comprising the following steps: S1. By aggregating terminal hardware configuration, software environment, and network status data through federated learning, a real-time test matrix is ​​constructed. The hardware configuration includes the CPU / GPU model and screen refresh rate; The software environment includes OS patch levels and API versions; The network status includes 5G slice type and Wi-Fi 6 channel bandwidth; S2. Test cases for synthesizing edge scenarios based on generative adversarial networks, wherein the edge scenarios include multi-task switching under low memory and temperature threshold triggering under high load. The effectiveness of test cases is verified through mutation testing. Mutation strategies include parameter perturbation and state space jump. S3: Synchronously collect touch trajectory, display frame rate and audio synchronization deviation, and combine eye tracker data to quantify user interaction experience; S4. Deploy a lightweight virtualization middleware layer between HarmonyOS / Android / iOS to automatically translate device interconnection commands and verify. Cross-chain compatibility; S5. Employs dynamic taint analysis to track privacy data streams and detect them in real time. Violation of regulations.

[0007] A further improvement of the present invention is that, in step S1: Federated learning employs a hierarchical aggregation architecture, where end nodes upload local data to edge servers after encrypting it with differential privacy. Edge servers dynamically adjust the weighting coefficients of data from each terminal based on an attention mechanism; The central server merges edge results to generate a global test matrix, which contains a three-dimensional correlation model of hardware, software and network. The three-dimensional association model is represented by a graph neural network, where nodes are hardware / software / network parameters and edges are dependencies between parameters.

[0008] A further improvement of the present invention is that, in the hierarchical aggregation architecture: The terminal node uses homomorphic encryption to encrypt local data. The encryption algorithm is as follows: Cryptographic system; The weighting coefficients of edge servers are adjusted based on the entropy and variance of terminal data, with data having higher entropy and lower variance being assigned higher weights. When merging central servers, a hybrid strategy of weighted average and federated average is adopted, with weights dynamically adjusted. When training a 3D correlation model, the loss function includes parameter correlation loss and prediction accuracy loss.

[0009] A further improvement of the present invention is that the mutation test process in step S2 includes: Generative adversarial network models consist of generators and discriminator composition, The input consists of hardware configuration parameters and edge scene labels, and the output is a sequence of test cases. The input is a mixture of real and generated test cases, and the output is the probability of realism. The mutation test employs a multi-objective optimization algorithm to simultaneously optimize test case coverage, execution efficiency, and defect detection rate. Mutation operations include three types: instruction replacement, data injection, and timing disruption. A multi-objective optimization algorithm is adopted, with optimization objectives including use case coverage, execution efficiency, and defect detection rate. Test execution involves running the original test cases and the mutated test cases on the target program, observing the execution results, and comparing the execution results of the original test cases and the mutated test cases. If a certain variant is not detected by any test case, the variant is called a live variant. The effectiveness of the test case set is evaluated by calculating the ratio of the number of detected variants to the total number of variants. At the same time, based on the results of the mutation test, the existing test case set is improved and enhanced to improve its defect detection capability.

[0010] A further improvement of the present invention is that, during the training of the generative adversarial network model: generator use The structure includes encoder-decoder connections and skip connections; Discriminator use The structure outputs the probability of the reality of a local region. The loss function is A combination of distance and gradient penalty term; The multi-objective optimization algorithm for mutation testing adopts The fitness function is: test case coverage × execution efficiency / defect discovery rate. A further improvement of the present invention is that, in step S3: Touch trajectory acquisition is achieved through the fusion of capacitive and piezoelectric sensors, with a sampling rate of 1.2. ; Display frame rate analysis uses a high-speed camera to capture the screen refresh process, combined with... Algorithm extracts frame rate features; Audio synchronization deviation is calculated by comparing the two-channel timestamps, with an error threshold set at 50. ; Eye tracker data includes fixation point coordinates and pupil diameter changes, which are used to quantify visual attention allocation.

[0011] A further improvement of the present invention is that, in step S4: The lightweight virtualization middleware layer uses containerization technology, with each operating system ecosystem corresponding to an independent container; Device interconnection command conversion is achieved through semantic parsing, which will... The encoding and decoding parameters are mapped to a format supported by the target system; Cross-chain compatibility verification includes: cross-chain identity authentication, data signature verification, and transaction consistency check; The verification process uses zero-knowledge proofs. Technology protects user privacy.

[0012] A further improvement of the present invention is that, in the containerization technology: Each container contains an independent operating system kernel and user space; Inter-container communication is achieved through shared memory and sockets; Semantic parsing adopts The model takes raw instructions as input and outputs a structured semantic representation. Cross-chain identity authentication adopts Documents and Verifiable Credentials The signature algorithm is .

[0013] A further improvement of the present invention is that, in step S5: Dynamic taint analysis is achieved through code instrumentation, inserting taint markers at points of access to private data; The taint propagation path is tracked using a data flow graph, where nodes represent data operations and edges represent data flow directions. Violation detection includes: unauthorized clipboard access, location information leakage, and external transmission of contact data; The test results are stored on the blockchain to ensure they cannot be tampered with.

[0014] A further improvement of the present invention is that, in the blockchain evidence storage: It adopts a consortium blockchain architecture, with nodes composed of testing organizations, equipment manufacturers, and regulatory authorities; The evidence includes the type of violation, timestamp, and device identifier; The consensus mechanism is PBFT, ensuring high throughput and low latency; Evidence retrieval is achieved through smart contracts, which input device identifiers and time ranges and output evidence records.

[0015] The beneficial effects of this invention are as follows: 1. This invention aggregates terminal hardware configuration, software environment, and network status data through federated learning to construct a real-time test matrix. It employs a hierarchical aggregation architecture, where terminal nodes use differential privacy and homomorphic encryption to encrypt and upload local data, ensuring data security. Simultaneously, edge servers dynamically adjust data weights based on an attention mechanism, and the central server fuses and generates a global test matrix containing a three-dimensional correlation model of hardware, software, and network. A graph neural network represents the dependencies between parameters, providing a more comprehensive and accurate reflection of the actual terminal environment and a reliable foundation for subsequent testing. This solves the problems of single and inaccurate traditional data collection. The invention verifies the effectiveness of test cases through mutation testing, employing a multi-objective optimization algorithm to simultaneously optimize test case coverage, execution efficiency, and defect detection rate. The mutation operations are rich and diverse. The effectiveness of the test case set is evaluated by calculating the proportion of surviving mutants, and the test case set is improved accordingly, significantly improving the quality and efficiency of test cases and overcoming the shortcomings of traditional test case generation, which lacks intelligence and specificity. 2. This invention employs multiple technical means, such as fusing capacitive and piezoelectric sensors to collect touch trajectories and using high-speed cameras combined with algorithms to extract display frame rates. This enables precise acquisition of various data during user-terminal interaction, providing detailed and accurate data for evaluating user experience on different terminals. Secondly, it utilizes containerization technology, with each operating system ecosystem corresponding to an independent container. Semantic parsing enables device interconnection instruction conversion and cross-chain compatibility verification, while zero-knowledge proof technology protects user privacy. This effectively solves the problems of complex instruction conversion and difficult compatibility verification in cross-platform testing, improving the efficiency and accuracy of cross-platform testing. Dynamic taint analysis tracks privacy data flows, inserting taint markers at privacy data access points and tracing propagation paths through data flow graphs. It can detect unauthorized clipboard reading, location information leakage, and other violations in real time. Detection results are stored on the blockchain, using a consortium blockchain architecture and PBFT consensus mechanism to ensure data immutability and convenient and quick retrieval, providing reliable privacy protection during the testing process. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 , This embodiment provides a testing method for mobile smart terminal compatibility testing, including the following steps: S1. By aggregating terminal hardware configuration, software environment, and network status data through federated learning, a real-time test matrix is ​​constructed. Hardware configuration includes CPU / GPU model and screen refresh rate; The software environment includes OS patch levels and API versions; Network status includes 5G slice type and Wi-Fi 6 channel bandwidth; In step S1: Federated learning employs a hierarchical aggregation architecture, where end nodes upload local data to edge servers after encrypting it with differential privacy. Edge servers dynamically adjust the weighting coefficients of data from each terminal based on an attention mechanism; The central server merges edge results to generate a global test matrix, which contains a three-dimensional correlation model of hardware, software and network. The 3D correlation model is represented by a graph neural network, where nodes represent hardware / software / network parameters and edges represent dependencies between parameters. In a layered aggregation architecture: The terminal node uses homomorphic encryption to encrypt local data. The encryption algorithm is as follows: Cryptographic system; The weighting coefficients of edge servers are adjusted based on the entropy and variance of terminal data, with data having higher entropy and lower variance being assigned higher weights. The entropy value of the terminal data is calculated using the information entropy formula: in, For random variables Information entropy is an indicator that measures the uncertainty of information. It can also be understood as the minimum average number of bits required to encode information. The larger its value, the higher the uncertainty of the information, or the greater the average amount of information contained in the information. As a random variable, it represents various possible events or states, for example, in mobile smart terminal compatibility testing. It can represent a certain characteristic of the terminal, such as different Model number, operating system version, etc.; Represents random variables The number of all possible values, for example, if This indicates the operating system of the terminal, which could be Android, iOS, or HarmonyOS. =3; It is a random variable The There are several possible values. Continuing with the operating system mentioned above as an example, let's consider the following: It can be Android. It can be iOS. It could be HarmonyOS; Represents random variables Values The probability, probability Satisfy 0≤ ≤1, =1; For example, in a terminal testing scenario, if 50% of terminals on the market use the Android system, 30% use the iOS system, and 20% use the HarmonyOS system, then =0.5, =0.3, =0.2; It is a base-2 logarithmic operation, using the base-2 logarithm to ensure that the information unit of the calculation result is bits. If other bases are used, such as the natural logarithm... The unit of information content is the Knight; The variance calculation uses the conventional statistical variance formula. Based on the calculated entropy value and variance, the weight coefficient is determined according to the set rules, such as assigning high weight when the entropy value is higher than a certain threshold and the variance is lower than a certain threshold. When merging central servers, a hybrid strategy of weighted average and federated average is adopted, with weights dynamically adjusted. During the training of the 3D association model, the loss function includes parameter correlation loss and prediction accuracy loss. The parameter correlation loss can be measured by cosine similarity to measure the correlation difference between hardware, software and network parameters. The prediction accuracy loss can be calculated by mean squared error (MSE) to calculate the difference between the model prediction result and the actual value. Then, 1000 sets of terminal data are used as training samples, and the number of training rounds is set to 100 rounds. In each round of training, the model parameters are adjusted according to the value of the loss function. It's important to note that by comprehensively collecting hardware, software, and network status information from the terminals and utilizing a layered aggregation architecture based on federated learning, a global test matrix reflecting the real-world terminal environment is constructed while protecting data privacy. The three-dimensional correlation model within this matrix clearly presents the dependencies between hardware, software, and network parameters, providing a data foundation for generating targeted test cases and ensuring that testing covers various real-world scenario combinations. For example, assuming multiple mobile terminals of different brands and models participate in the test, each terminal node uploads local data such as its CPU model (e.g., a certain brand of mobile phone uses a Snapdragon 888 chip), OS patch level (e.g., updated to the latest patch), and 5G slice type (e.g., industrial internet slice), encrypted using differential privacy and homomorphic encryption techniques, to an edge server. The edge server adjusts the weights based on the data's entropy and variance. For instance, a terminal with high entropy and low variance indicates good data stability and a large amount of information, thus receiving a higher weight. The central server then merges the edge results to generate a global test matrix, using a graph neural network to display the relationships between hardware, software, and network parameters. For example, it can analyze the impact of a specific 5G slice type on application performance under a particular CPU model and OS patch level.

[0020] S2. Test cases for synthesizing edge scenarios based on generative adversarial networks. Edge scenarios include multi-task switching under low memory and temperature threshold triggering under high load. The effectiveness of test cases is verified through mutation testing. Mutation strategies include parameter perturbation and state space jump. The mutation testing process in step S2 includes: Generative adversarial network models consist of generators and discriminator composition, The input consists of hardware configuration parameters and edge scene labels, and the output is a sequence of test cases. The input is a mixture of real and generated test cases, and the output is the probability of realism. The mutation test employs a multi-objective optimization algorithm to simultaneously optimize test case coverage, execution efficiency, and defect detection rate. Mutation operations include three types: instruction replacement, data injection, and timing disruption. A multi-objective optimization algorithm is adopted, with optimization objectives including use case coverage, execution efficiency, and defect detection rate. Test execution involves running the original test cases and the mutated test cases on the target program, observing the execution results, and comparing the execution results of the original test cases and the mutated test cases. If a certain variant is not detected by any test case, the variant is called a live variant. The effectiveness of the test case set is evaluated by calculating the ratio of the number of detected variants to the total number of variants. At the same time, based on the results of the mutation test, the existing test case set is improved and enhanced to improve its defect detection capability; During the training of a generative adversarial network model: generator use The architecture consists of an encoder-decoder and skip connections. The encoder comprises multiple convolutional and pooling layers to extract features from hardware configuration parameters and edge scene labels. The decoder consists of multiple deconvolutional layers to reconstruct the sequence of test cases from the encoded features. Skip connections allow features from certain layers in the encoder to be directly passed to the corresponding layers in the decoder, preserving more detailed information. Discriminator use The structure outputs the probability of authenticity for local regions. The discriminator uses a structure of multiple convolutional layers and fully connected layers. When outputting the probability of authenticity for local regions, the test case sequence can be divided into multiple local regions, the probability of authenticity for each region can be calculated separately, and then a comprehensive evaluation can be performed. The loss function is The combination of distance and gradient penalty term, The distance is calculated using an approximate algorithm, and the weight coefficient of the gradient penalty term is set to 0.1. The multi-objective optimization algorithm for mutation testing adopts The fitness function is calculated as test case coverage × execution efficiency / defect discovery rate. Coverage can be calculated by comparing the number of functionalities covered by the tested test cases to the total number of functionalities. Execution efficiency can be measured by the average execution time of the test cases. The defect discovery rate is the ratio of the number of defects found to the actual number of defects. Secondly, if it is found that the survival rate of mutations of the instruction substitution type is high, more test cases for different instruction combinations can be added, or existing test cases can be modified to improve the detection capability of such mutations. It should be noted that Generative Adversarial Networks (GANs) can synthesize various complex edge scenario test cases, such as low-memory multitasking switching and high-load temperature threshold triggering. These scenarios are difficult to fully cover in conventional testing. Mutation testing verifies the effectiveness and robustness of test cases by performing various types of mutation operations on them, while optimizing the quality of the test case set and improving the defect detection rate. For example, the generator of a GAN generates a corresponding sequence of test cases based on the input hardware configuration parameters, such as a memory capacity of 6GB and the edge scenario label of low-memory multitasking switching. The discriminator judges the real test cases and the generated test cases and outputs the probability of authenticity. In mutation testing, the generated test cases are mutated by instruction substitution, such as replacing the original instruction "Open application A" with "Open application B". The original test cases and the mutated test cases are executed on the target program. If a mutant, such as the modified instruction combination, is not detected by any test cases, it is a live mutant. The effectiveness of the test case set is evaluated by calculating the ratio of the number of detected variants to the total number of variants. If it is found that certain types of variants have a high survival rate, the test case set is improved and enhanced, for example, by adding test cases for specific instruction combinations.

[0021] S3: Synchronously collect touch trajectory, display frame rate and audio synchronization deviation, and combine eye tracker data to quantify user interaction experience; In step S3: Touch trajectory acquisition is achieved through the fusion of capacitive and piezoelectric sensors, with a sampling rate of 1.2. ; Display frame rate analysis uses a high-speed camera to capture the screen refresh process, combined with... Algorithm extracts frame rate features; Audio synchronization deviation is calculated by comparing the two-channel timestamps, with an error threshold set at 50. ; Eye-tracking data includes fixation point coordinates and pupil diameter changes, which are used to quantify visual attention allocation; It's worth noting that the S3 uses multiple technologies to simultaneously collect touch trajectories, display frame rates, audio synchronization deviations, and eye-tracking data, comprehensively and accurately quantifying the user's interaction experience with the mobile smart terminal. This data helps developers and testers understand the actual usage of applications on different devices and identify issues affecting user experience; for example, when testing a game application, the user's touch trajectories are collected through a fusion of capacitive and piezoelectric sensors, achieving a sampling rate of 1.2. It can accurately record the user's touch operation path and force in the game, and a high-speed camera captures the screen refresh process. Combined with algorithms, it extracts the display frame rate characteristics. If the display frame rate is unstable, it may cause the game screen to stutter. Audio synchronization deviation is calculated by comparing the two-channel timestamps, and the error threshold is set at 50. To ensure that the sound effects in the game are synchronized with the visuals, the eye tracker records the coordinates of the user's gaze point and changes in pupil diameter, and analyzes the user's visual attention allocation in the game, such as whether the user pays more attention to a specific area.

[0022] S4. Deploy a lightweight virtualization middleware layer between HarmonyOS / Android / iOS to automatically translate device interconnection commands and verify. Cross-chain compatibility; In step S4: The lightweight virtualization middleware layer uses containerization technology, with each operating system ecosystem corresponding to an independent container; Device interconnection command conversion is achieved through semantic parsing, which will... The encoding and decoding parameters are mapped to a format supported by the target system; Cross-chain compatibility verification includes: cross-chain identity authentication, data signature verification, and transaction consistency check; The verification process uses zero-knowledge proofs. Technology protects user privacy; In containerization technology: Each container contains an independent operating system kernel and user space; Inter-container communication is achieved through shared memory and sockets; Semantic parsing adopts The model takes raw instructions as input and outputs a structured semantic representation. Cross-chain identity authentication adopts Documents and Verifiable Credentials The signature algorithm is ; It's important to note that S4 addresses the issue of device interconnection and compatibility across different operating system ecosystems. The lightweight virtualization middleware layer employs containerization technology, providing an independent environment for each operating system ecosystem. It automatically converts device interconnection commands through semantic parsing and performs cross-chain compatibility verification. Zero-knowledge proof technology protects user privacy, ensuring seamless integration and normal operation of applications across different platforms. For example, when an application needs to implement device interconnection functionality on HarmonyOS, Android, and iOS platforms, such as data synchronization between a phone and a smartwatch, the lightweight virtualization middleware layer creates an independent container for each system. Each container has an independent operating system kernel and user space. Device interconnection commands, through semantic parsing, map the encoding and decoding parameters of the HarmonyOS system to formats supported by Android and iOS systems. In cross-chain compatibility verification, cross-chain identity authentication, data signature verification, and transaction consistency checks are performed, such as verifying the legitimacy of identities between a phone and a watch, and the integrity and consistency of data transmission. Zero-knowledge proof technology is used to complete the verification process without disclosing user privacy information. S5. Employs dynamic taint analysis to track privacy data streams and detect them in real time. Violations; In step S5: Dynamic taint analysis is achieved through code instrumentation, inserting taint markers at points of access to private data; The taint propagation path is tracked using a data flow graph, where nodes represent data operations and edges represent data flow directions. Violation detection includes: unauthorized clipboard access, location information leakage, and external transmission of contact data; The test results are stored on the blockchain to ensure they cannot be tampered with.

[0023] Blockchain-based evidence storage: The consortium blockchain adopts a consortium blockchain architecture, with nodes consisting of testing organizations, equipment manufacturers, and regulatory authorities. When a new node joins, it needs to be verified by the consensus of other nodes. The PBFT consensus mechanism is used, and a new node can join the consortium blockchain when more than two-thirds of the nodes agree. The evidence includes the type of violation, timestamp, and device identifier; The consensus mechanism is PBFT, ensuring high throughput and low latency; When evidence verification is implemented through smart contracts, the specific code logic and input / output formats of the smart contract should be provided. For example, if the smart contract is written in Solidity, the input would be a device identifier and a time range, and the output would be a list of evidence records that meet the specified conditions.

[0024] It's important to note that real-time tracking of the flow of privacy data within applications allows for the timely detection of unauthorized access behaviors, such as clipboard reading and location information leakage. Detection results are stored on the blockchain to ensure data immutability and traceability, providing reliable protection for privacy during testing and serving as a basis for subsequent problem investigation and liability determination. For example, when testing a social application, code instrumentation inserts taint markers at privacy data access points, such as user chat logs and location information. The taint propagation path is tracked using a data flow graph, clearly showing the data flow. If an application is detected reading clipboard content without the user's knowledge, this constitutes a violation. The detection result is uploaded to the blockchain for evidence storage, using a consortium blockchain architecture with nodes composed of testing organizations, equipment manufacturers, and regulatory authorities. The stored data includes the type of violation (e.g., unauthorized clipboard reading), a timestamp (the specific time of occurrence), and a device identifier (which terminal device). When a query is needed, a smart contract is used to input the device identifier and time range to output the stored record.

[0025] The terms "front," "back," "left," "right," "top," and "bottom" all refer to the figures in the accompanying drawings. Figure 1 Based on the perspective of the observer, the side of the device facing the observer is defined as the front, the left side of the observer is defined as the left, and so on.

[0026] In the description of this invention, it should be understood that the terms "center", "longitudinal", "lateral", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the scope of protection of this invention.

[0027] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A test method for compatibility testing of mobile smart terminals, characterized in that, Includes the following steps: S1. By aggregating terminal hardware configuration, software environment, and network status data through federated learning, a real-time test matrix is ​​constructed. The hardware configuration includes the CPU / GPU model and screen refresh rate; The software environment includes OS patch levels and API versions; The network status includes 5G slice type and Wi-Fi 6 channel bandwidth; S2. Test cases for synthesizing edge scenarios based on generative adversarial networks, wherein the edge scenarios include multi-task switching under low memory and temperature threshold triggering under high load. The effectiveness of test cases is verified through mutation testing. Mutation strategies include parameter perturbation and state space jump. S3: Synchronously collect touch trajectory, display frame rate and audio synchronization deviation, and combine eye tracker data to quantify user interaction experience; S4. Deploy a lightweight virtualization middleware layer between HarmonyOS / Android / iOS to automatically translate device interconnection commands and verify. Cross-chain compatibility; S5 employs dynamic taint analysis to track privacy data streams and detect them in real time. Violation of regulations.

2. The testing method for mobile smart terminal compatibility testing according to claim 1, characterized in that: In step S1: Federated learning employs a hierarchical aggregation architecture, where end nodes upload local data to edge servers after encrypting it with differential privacy. Edge servers dynamically adjust the weighting coefficients of data from each terminal based on an attention mechanism; The central server merges edge results to generate a global test matrix, which contains a three-dimensional correlation model of hardware, software and network. The three-dimensional association model is represented by a graph neural network, where nodes are hardware / software / network parameters and edges are dependencies between parameters.

3. The test method for mobile smart terminal compatibility testing according to claim 2, characterized in that: In the hierarchical aggregation architecture: The terminal node uses homomorphic encryption to encrypt local data. The encryption algorithm is as follows: Cryptographic systems; The weighting coefficients of edge servers are adjusted based on the entropy and variance of terminal data, with data having higher entropy and lower variance being assigned higher weights. When merging central servers, a hybrid strategy of weighted average and federated average is adopted, with weights dynamically adjusted. When training a 3D correlation model, the loss function includes parameter correlation loss and prediction accuracy loss.

4. The testing method for mobile smart terminal compatibility testing according to claim 1, characterized in that: The mutation test process in step S2 includes: Generative adversarial network models consist of generators and discriminator composition, The input consists of hardware configuration parameters and edge scene labels, and the output is a sequence of test cases. The input is a mixture of real and generated test cases, and the output is the probability of realism. The mutation test employs a multi-objective optimization algorithm to simultaneously optimize test case coverage, execution efficiency, and defect detection rate. Mutation operations include three types: instruction replacement, data injection, and timing disruption. A multi-objective optimization algorithm is adopted, with optimization objectives including use case coverage, execution efficiency, and defect detection rate. Test execution involves running the original test cases and the mutated test cases on the target program, observing the execution results, and comparing the execution results of the original test cases and the mutated test cases. If a certain variant is not detected by any test case, the variant is called a live variant. The effectiveness of the test case set is evaluated by calculating the ratio of the number of detected variants to the total number of variants. At the same time, based on the results of the mutation test, the existing test case set is improved and enhanced to improve its defect detection capability.

5. The test method for mobile smart terminal compatibility testing according to claim 4, characterized in that: During the training of the generative adversarial network model: generator use The structure includes encoder-decoder connections and skip connections; Discriminator use The structure outputs the probability of the reality of a local region. The loss function is A combination of distance and gradient penalty term; The multi-objective optimization algorithm for mutation testing adopts The fitness function is: test case coverage × execution efficiency / defect discovery rate.

6. The test method for mobile smart terminal compatibility testing according to claim 1, characterized in that: In step S3: Touch trajectory acquisition is achieved through the fusion of capacitive and piezoelectric sensors, with a sampling rate of 1.

2. ; Display frame rate analysis uses a high-speed camera to capture the screen refresh process, combined with... Algorithm extracts frame rate features; Audio synchronization deviation is calculated by comparing the two-channel timestamps, with an error threshold set at 50. ; Eye tracker data includes fixation point coordinates and pupil diameter changes, which are used to quantify visual attention allocation.

7. The test method for mobile smart terminal compatibility testing according to claim 1, characterized in that: In step S4: The lightweight virtualization middleware layer uses containerization technology, with each operating system ecosystem corresponding to an independent container; Device interconnection command conversion is achieved through semantic parsing, which will... The encoding and decoding parameters are mapped to a format supported by the target system; Cross-chain compatibility verification includes: cross-chain identity authentication, data signature verification, and transaction consistency check; The verification process uses zero-knowledge proofs. Technology protects user privacy.

8. The test method for mobile smart terminal compatibility testing according to claim 7, characterized in that: In the containerization technology: Each container contains an independent operating system kernel and user space; Inter-container communication is achieved through shared memory and sockets; Semantic parsing adopts The model takes raw instructions as input and outputs a structured semantic representation. Cross-chain identity authentication adopts Documents and verifiable credentials The signature algorithm is .

9. The test method for mobile smart terminal compatibility testing according to claim 1, characterized in that: In step S5: Dynamic taint analysis is achieved through code instrumentation, inserting taint markers at points of access to private data; The taint propagation path is tracked using a data flow graph, where nodes represent data operations and edges represent data flow directions. Violation detection includes: unauthorized clipboard access, location information leakage, and external transmission of contact data; The test results are stored on the blockchain to ensure they cannot be tampered with.

10. The test method for mobile smart terminal compatibility testing according to claim 9, characterized in that: In the blockchain evidence storage: It adopts a consortium blockchain architecture, with nodes composed of testing organizations, equipment manufacturers, and regulatory authorities; The evidence includes the type of violation, timestamp, and device identifier; The consensus mechanism is PBFT, ensuring high throughput and low latency; Evidence retrieval is achieved through smart contracts, which input device identifiers and time ranges and output evidence records.

Images