Lightweight operating system resource isolation method and system for intelligent terminal

Abstract

The invention discloses a lightweight operating system resource isolation method and system for an intelligent terminal, and relates to the technical field of computers. Analyzing the matching degree between the resource occupation data of the current task process and the resource occupation data of the historical task process; determining the type of the current task process; acquiring historical resource quota data in the execution process of the historical task process, acquiring resources occupied by resource dynamic isolation in the execution process of the historical task process, and floating the historical resource quota data in the execution process of the historical task process; according to the determined type of the current task process, allocating a floating resource quota for the current task process; when the historical task process matched with the current task process is determined, dynamic resource isolation can be stopped, resources can be released, the energy consumption of the intelligent terminal is reduced, the resource consumption of a resource isolation mechanism is reduced, and the cruising ability of the intelligent terminal is improved; the resources of the resource isolation mechanism are used for the task process, and the execution efficiency of the task process is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a lightweight operating system resource isolation method and system for intelligent terminals. Background Art

[0002] With the development of technology, resource isolation technology has been widely used on smart terminals. Developers use microkernel design, hardware abstraction layer design, virtualization technology and other means to achieve resource isolation between different applications or processes. Through resource isolation, resource access rights between different applications or processes are restricted, thereby protecting user data from illegal access or tampering and maintaining user privacy and security. Resource isolation can prevent errors or anomalies in one application or process from affecting other applications or systems, thereby improving system stability and performance.

[0003] Since the resource usage of the terminal may change rapidly, such as network bandwidth, CPU, and load, static resource allocation may not be able to adapt to real-time changes, resulting in resource waste or competition. Therefore, the terminal will adopt dynamic resource isolation. In order to improve the effect of dynamic resource isolation, the terminal will predict resource usage data in advance to reduce resource waste; however, the dynamic isolation mechanism itself may require more system resources, resulting in a reduction in the resources available for the task process and an increase in the energy consumption of the terminal. For battery-dependent terminals, it may shorten the battery life of the smart terminal. At the same time, in order to perform dynamic resource isolation, load prediction will also occupy system resources and shorten battery life. Summary of the Invention

[0004] The purpose of the present invention is to provide a lightweight operating system resource isolation method and system for intelligent terminals to solve the problems raised in the prior art.

[0005] To achieve the above-mentioned objectives, the present invention provides the following technical solutions: a lightweight operating system resource isolation system for intelligent terminals, comprising: a data acquisition module, a data analysis module, a database and a resource isolation module; the output end of the data acquisition module is connected to the input end of the database, the data analysis module and the resource isolation module, for obtaining resource occupancy data during the execution of the task process; the output end of the data analysis module is connected to the input end of the resource isolation module, for obtaining historical resource quota data during the execution of historical task processes, obtaining resources occupied by dynamic resource isolation during the execution of historical task processes, and floating up the historical resource quota data during the execution of historical task processes; according to the determined type of the current task process, the floated resource quota is allocated to the current task process, and the calculation of dynamic resource isolation is stopped; the output end of the database is connected to the input end of the data analysis module, for storing resource occupancy data of historical task processes; the resource isolation module dynamically isolates resources according to the resource occupancy of the task process.

[0006] Specifically, the data analysis module also includes a dependency modeling unit, a resource quota floating unit, a matching analysis unit and a task process determination unit; the dependency modeling unit is used to determine the dependency between task processes; the resource quota floating unit dynamically isolates resources and allocates resource quotas to occupied resources; the matching analysis unit is used to analyze the matching degree and matching degree change rate between the current task process and the historical task process; the task process determination unit is used to determine the type of the current task process and obtain the resource quota of the current task process.

[0007] Specifically, the resource isolation module also includes a dynamic isolation unit and a response unit. The dynamic isolation unit trains a resource occupancy prediction model based on the resource data occupied by the task process and adjusts the resource quota in advance; the response unit stops the dynamic isolation unit and releases the resources occupied by the dynamic isolation unit after receiving the instruction issued by the data analysis module.

[0008] Specifically, the resource quota floating unit obtains the resource data occupied by dynamic isolation of resources during the execution of historical task processes, performs kmeans clustering on historical resource quota data with the same dependency relationship, adds the resource data occupied by dynamic isolation of resources and the historical resource quota data corresponding to the centroid, and obtains the historical resource quota data after floating.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a lightweight operating system resource isolation method for a smart terminal, comprising the following steps:

[0010] S10, the data acquisition module obtains the resource usage data of the current task process and sends the resource usage data of the current task process to the data analysis module;

[0011] S11, the data analysis module analyzes the resource usage data of the current task process one by one, obtains the resource usage data of the historical task process from the database, and analyzes the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process;

[0012] S12, the data analysis module analyzes each resource usage data of the current task process to obtain a matching degree, and further analyzes the obtained matching degree during the process execution to obtain a matching degree change rate;

[0013] S13, the data analysis module analyzes the matching degree and matching degree change rate of the resource usage data of the current task process and the resource usage data of the historical task processes, and determines the type of the current task process based on the analysis results;

[0014] S14, obtaining historical resource quota data during the execution of historical task processes, obtaining resources occupied by dynamic resource isolation during the execution of historical task processes, and floating up the historical resource quota data during the execution of historical task processes; according to the determined type of the current task process, allocating the floated resource quota to the current task process.

[0015] Specifically, in step S14, the floating up of historical resource quota data during the execution of historical task processes further includes the following steps:

[0016] Obtain historical resource quota data and dependencies during the execution of historical task processes, use historical resource quota data with the same dependencies as input features, and perform kmeans clustering on the input features as data points. The specific steps include:

[0017] S20, select num data points as the initial centroid, num is a positive integer;

[0018] S21, assign each data point to the nearest centroid to form a cluster;

[0019] S22, calculate the average value of all data points in each cluster and update the centroid of each cluster;

[0020] S23, return to step S21, and after reaching the preset number of iterations, proceed to step S24;

[0021] S24, obtain the resource data occupied by dynamic isolation of resources during the execution of the historical task process; analyze the maximum absolute error between the centroid and all data points in the same cluster; if there is resource data occupied by dynamic isolation of resources that is less than the maximum absolute error between the centroid and all data points in the same cluster, adjust the number num of initial centroids and return to step S20; if there is no resource data occupied by dynamic isolation of resources that is less than the maximum absolute error between the centroid and all data points in the same cluster, complete kmeans clustering.

[0022] Specifically, the floating up of historical resource quota data during the execution of historical task processes further includes the following steps:

[0023] The resource data occupied by dynamic isolation of resources during the execution of historical task processes is obtained, and the resource data occupied by dynamic isolation of resources is added to the historical resource quota data corresponding to the centroid to obtain the historical resource quota data after floating.

[0024] Specifically, in step S11, the analysis of the matching degree between the resource occupancy data of the current task process and the resource occupancy data of the historical task process further includes the following steps:

[0025] S41, obtain the resource occupancy data of the historical task process in the i-th cluster, the resource occupancy data of each historical task process forms a characteristic curve, the characteristic curves in the cluster are merged into one, and the data analysis module calculates the root mean square error R between the resource occupancy data of the current task process and the characteristic curve ij ; i represents the number of the historical task process, j represents the number of the resource usage data;

[0026] S42, the data analysis module analyzes the difference between the dependency relationship of the current task process obtained from the data acquisition module and the dependency relationship of the historical task process in the i-th cluster obtained from the database, and compares the difference D between each dependency relationship of the current task process and each dependency relationship of the historical task process in the i-th cluster. k , if the dependency of the current task process is the same as the dependency of the historical task process in the i-th cluster, then D k =1, otherwise D k =0, k is a positive integer between [1, b], indicating the number of dependency relationships, and b is the total number of non-repeated dependency relationships of the current task process and the dependency relationships of the historical task processes;

[0027] S43, the data analysis module analyzes the resource occupancy data of the current task process and the dependency relationship to obtain the matching degree F between the resource occupancy data of the current task process and the resource occupancy data of the historical task process in the i-th cluster. ij , Where wj represents the weight of the j-th resource occupancy data, v k represents the weight of the kth dependency, m represents the amount of resource usage data, u ij Represents the average value of the j-th resource occupancy data.

[0028] Specifically, the weight of the resource occupancy data and the weight of the dependency are determined by the following steps:

[0029] The data analysis module calculates the root mean square error (sd) of the j-th resource occupancy data based on the gap between the j-th resource occupancy data of different cluster centers. j , the weight w of the j-th resource occupancy data is obtained according to the root mean square error of the j-th resource occupancy data j , w j =(1 / sd j ) / Σ(1 / sd j ); According to the weight w of resource occupancy data j Get the weight v of the dependency k , v k =(1 / b)×Σ(w j );

[0030] The step of analyzing the obtained matching degree to obtain the matching degree change rate during the process execution further includes the following steps:

[0031] After the current task process executes for a sampling period T, the data analysis module obtains the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process within the sampling period T;

[0032] After waiting for a sampling period T, the data analysis module obtains the matching degree between the resource occupancy data of the current task process and the resource occupancy data of the historical task process within the waiting sampling period T, and then obtains the matching degree change rate within the waiting sampling period T.

[0033] Specifically, in step S13, determining the type of the current task process according to the analysis result further includes the following steps:

[0034] The binary classification neural network model is trained on the resource usage data of the historical task processes in each cluster. The data analysis module inputs the matching degree and the matching degree change rate into the i-th binary classification neural network model, and determines the type of the current task process based on the classification results of the binary classification neural network model:

[0035] If there is only one cluster that matches the current task process, the type of the current task process is determined, and the historical resource quota data after floating is allocated according to the historical task process that matches the task process type, and the calculation of dynamic resource isolation is stopped;

[0036] If there is more than one cluster that matches the current task process, and in the matching clusters, there is historical resource quota data after floating that can meet the requirements of all historical resource quota data in the matching clusters, then the resource limit is allocated according to the historical resource quota data after floating, and the calculation of dynamic resource isolation is stopped;

[0037] Otherwise, the current task process is monitored and resources are dynamically isolated.

[0038] Compared with the existing technology, the beneficial effects of the present invention are: when determining the historical task process that matches the current task process, dynamic resource isolation can be stopped, and there is no need to maintain dynamic resource isolation during the execution of the task process, thereby releasing resources and reducing the energy consumption of the smart terminal, reducing the resource consumption of the resource isolation mechanism, and improving the battery life of the smart terminal; the resources of the resource isolation mechanism are used for the task process, thereby improving the execution efficiency of the task process. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The diagram is a structural diagram of a lightweight operating system resource isolation system for smart terminals according to the present invention. DETAILED DESCRIPTION

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0041] Example: Figure 1 As shown, the present invention provides a technical solution for a lightweight operating system resource isolation system for intelligent terminals, comprising: a data acquisition module, a data analysis module, a database and a resource isolation module; the output end of the data acquisition module is connected to the input end of the database, the data analysis module and the resource isolation module, for obtaining resource occupancy data during the execution of the task process; the output end of the data analysis module is connected to the input end of the resource isolation module, for obtaining historical resource quota data during the execution of historical task processes, obtaining resources occupied by dynamic resource isolation during the execution of historical task processes, and floating up the historical resource quota data during the execution of historical task processes; according to the determined type of the current task process, the floated resource quota is allocated to the current task process, and the calculation of dynamic resource isolation is stopped; the output end of the database is connected to the input end of the data analysis module, for storing resource occupancy data of historical task processes; the resource isolation module dynamically isolates resources according to the resource occupancy of the task process.

[0042] The data analysis module also includes a dependency modeling unit, a resource quota floating unit, a matching analysis unit and a task process determination unit; the dependency modeling unit is used to determine the dependency between task processes; the resource quota floating unit dynamically isolates the occupied resources and performs resource quotas; the matching analysis unit is used to analyze the matching degree and matching degree change rate between the current task process and the historical task process; the task process determination unit is used to determine the type of the current task process and obtain the resource quota of the current task process.

[0043] The resource isolation module also includes a dynamic isolation unit and a response unit. The dynamic isolation unit trains a resource occupancy prediction model based on the resource data occupied by the task process and adjusts the resource quota in advance; the response unit stops the dynamic isolation unit and releases the resources occupied by the dynamic isolation unit after receiving the instruction issued by the data analysis module.

[0044] The resource quota floating unit obtains the resource data occupied by dynamic isolation of resources during the execution of historical task processes, performs kmeans clustering on historical resource quota data with the same dependency relationship, adds the resource data occupied by dynamic isolation of resources and the historical resource quota data corresponding to the centroid, and obtains the historical resource quota data after floating.

[0045] Embodiment: The present invention provides a technical solution for a lightweight operating system resource isolation method for a smart terminal, comprising the following steps:

[0046] S10, the data acquisition module obtains the resource usage data of the current task process and sends the resource usage data of the current task process to the data analysis module;

[0047] S11, the data analysis module analyzes the resource usage data of the current task process one by one, obtains the resource usage data of the historical task process from the database, and analyzes the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process;

[0048] S12, the data analysis module analyzes each resource usage data of the current task process to obtain a matching degree, and further analyzes the obtained matching degree during the process execution to obtain a matching degree change rate;

[0049] S13, the data analysis module analyzes the matching degree and matching degree change rate of the resource usage data of the current task process and the resource usage data of the historical task processes, and determines the type of the current task process based on the analysis results;

[0050] S14, obtaining historical resource quota data during the execution of historical task processes, obtaining resources occupied by dynamic resource isolation during the execution of historical task processes, and floating up the historical resource quota data during the execution of historical task processes; according to the determined type of the current task process, allocating the floated resource quota to the current task process.

[0051] Uploading the historical resource quota data during the execution of historical task processes also includes the following steps:

[0052] Obtain historical resource quota data and dependencies during the execution of historical task processes, use historical resource quota data with the same dependencies as input features, and perform kmeans clustering on the input features as data points. The specific steps include:

[0053] S20, select num data points as the initial centroid, num is a positive integer;

[0054] S21, assign each data point to the nearest centroid to form a cluster;

[0055] S22, calculate the average value of all data points in each cluster and update the centroid of each cluster;

[0056] S23, return to step S21, and after reaching the preset number of iterations, proceed to step S24;

[0057] S24, obtain the resource data occupied by dynamic isolation of resources during the execution of the historical task process; analyze the maximum absolute error between the centroid and all data points in the same cluster; if there is resource data occupied by dynamic isolation of resources that is less than the maximum absolute error between the centroid and all data points in the same cluster, adjust the number num of initial centroids and return to step S20; if there is no resource data occupied by dynamic isolation of resources that is less than the maximum absolute error between the centroid and all data points in the same cluster, complete kmeans clustering.

[0058] Resource usage data includes but is not limited to CPU usage, memory usage, and network bandwidth usage, which are obtained through real-time monitoring tools; resource quota data refers to the CPU, memory, network bandwidth, and other resources allocated to the task process by the resource isolation module based on the resource usage data of the task process. In order to determine the resource quota of the current task process, a cluster analysis is performed on the resource quotas of historical task processes. Since the resource usage of the terminal may change rapidly, such as network bandwidth and CPU load, static resource allocation may not be able to adapt to real-time changes, resulting in resource waste or competition. Therefore, the terminal will adopt dynamic resource isolation. In order to improve the effect of dynamic resource isolation, the terminal will predict resource usage data in advance in order to reduce resource waste. However, the dynamic isolation mechanism itself may require more system resources, resulting in a reduction in the resources available to the task process and an increase in the energy consumption of the terminal. This is a disadvantage for battery-dependent terminals and may shorten the battery life. At the same time, in order to perform dynamic resource isolation, load prediction will also occupy system resources and shorten the battery life.

[0059] To this end, on the basis of clustering, consider allocating the resources occupied by dynamic resource isolation to the task process. First, determine the matching clustering cluster according to the resource occupation data of the task process, and allocate resource quotas for the current task process according to the matching clustering cluster. At the same time, the resource demand of the current task process may be inconsistent with the historical task process. At this time, the resource quota of the historical task process is increased to resist the change of resource needs of the current task process. In this way, there is no need to maintain dynamic resource isolation for the task process. When the clustering cluster matching the current task process can be determined, the dynamic resource isolation can be stopped to release resources and reduce the energy consumption of the smart terminal.

[0060] Uploading the historical resource quota data during the execution of historical task processes also includes the following steps:

[0061] The resource data occupied by dynamic isolation of resources during the execution of historical task processes is obtained, and the resource data occupied by dynamic isolation of resources is added to the historical resource quota data corresponding to the centroid to obtain the historical resource quota data after floating.

[0062] Analyzing the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process also includes the following steps:

[0063] S41, obtain the resource occupancy data of the historical task process in the i-th cluster, the resource occupancy data of each historical task process forms a characteristic curve, the characteristic curves in the cluster are merged into one, and the data analysis module calculates the root mean square error R between the resource occupancy data of the current task process and the characteristic curve ij ; i represents the number of the historical task process, j represents the number of the resource usage data;

[0064] S42, the data analysis module analyzes the difference between the dependency relationship of the current task process obtained from the data acquisition module and the dependency relationship of the historical task process in the i-th cluster obtained from the database, and compares the difference D between each dependency relationship of the current task process and each dependency relationship of the historical task process in the i-th cluster. k , if the dependency of the current task process is the same as the dependency of the historical task process in the i-th cluster, then D k =1, otherwise D k =0, k is a positive integer between [1, b], indicating the number of dependency relationships, and b is the total number of non-repeated dependency relationships of the current task process and the dependency relationships of the historical task processes;

[0065] S43, the data analysis module analyzes the resource occupancy data of the current task process and the dependency relationship to obtain the matching degree F between the resource occupancy data of the current task process and the resource occupancy data of the historical task process in the i-th cluster. ij , Where w j represents the weight of the j-th resource occupancy data, v k represents the weight of the kth dependency, m represents the amount of resource usage data, u ij Represents the average value of the j-th resource occupancy data.

[0066] Some processes may have dependencies, and their resource usage needs to be coordinated. For example, if a process needs to frequently communicate with another background process, a certain resource quota needs to be reserved for the background process. Otherwise, communication delays or data transmission errors may occur. Therefore, consider the dependency relationship to analyze the matching degree between the current task process and the historical task process. Real-time monitoring tools can be used to monitor the resource usage and communication status of the process, and the call links between applications can be analyzed to obtain the dependency relationship.

[0067] The higher the similarity between the resource usage data of the current task process and the characteristic curve, that is, the smaller the root mean square error value, the more likely the resource usage data of the new previous task process is similar to the resource usage data of the historical task process. Therefore, the matching degree is inversely proportional to the root mean square error value. The root mean square error value is compared with the mean of the resource usage data of the historical task process to eliminate the influence of the scale of the resource usage data of different historical task processes. The more the number of the current task process dependency relationships that are the same as the historical task process dependency relationships, the higher the similarity between the resource usage data of the new previous task process and the historical task process. Therefore, the matching degree is proportional to the dependency relationship.

[0068] The weights of resource usage data and dependencies are determined through the following steps:

[0069] The data analysis module calculates the root mean square error (sd) of the j-th resource occupancy data based on the gap between the j-th resource occupancy data of different cluster centers. j , the weight w of the j-th resource occupancy data is obtained according to the root mean square error of the j-th resource occupancy data j , w j =(1 / sd j ) / Σ(1 / sd j ); According to the weight w of resource occupancy data j Get the weight v of the dependency k , v k =(1 / b)×Σ(w j );

[0070] The step of analyzing the obtained matching degree to obtain the matching degree change rate during the process execution further includes the following steps:

[0071] After the current task process executes for a sampling period T, the data analysis module obtains the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process within the sampling period T;

[0072] After waiting for a sampling period T, the data analysis module obtains the matching degree between the resource occupancy data of the current task process and the resource occupancy data of the historical task process within the waiting sampling period T, and then obtains the matching degree change rate within the waiting sampling period T.

[0073] The resource usage data of different task processes are different, so the impact of each resource usage data on the task process is different. Because the larger the root mean square error of the j-th resource usage data, the greater the change in resource usage data required to determine the clustering cluster of the task process. Therefore, the resource usage data and the inverse of the root mean square error are positively correlated. After all the inverses of the root mean square errors are distributed in proportion, the weight is obtained; the dependency relationship needs to be consistent with the scale of the resource usage data, so the dependency weight is determined by the resource usage data weight. Here, the mean of the resource usage data weight is selected.

[0074] Determining the type of the current task process based on the analysis results also includes the following steps:

[0075] The binary classification neural network model is trained on the resource usage data of the historical task processes in each cluster. The data analysis module inputs the matching degree and the matching degree change rate into the i-th binary classification neural network model, and determines the type of the current task process based on the classification results of the binary classification neural network model:

[0076] If there is only one cluster that matches the current task process, the type of the current task process is determined, and the historical resource quota data after floating is allocated according to the historical task process that matches the task process type, and the calculation of dynamic resource isolation is stopped;

[0077] If there is more than one cluster that matches the current task process, and in the matching clusters, there is historical resource quota data after floating that can meet the requirements of all historical resource quota data in the matching clusters, then the resource limit is allocated according to the historical resource quota data after floating, and the calculation of dynamic resource isolation is stopped;

[0078] Otherwise, the current task process is monitored and resources are dynamically isolated.

[0079] The i-th binary classification neural network model is trained through the matching degree and matching degree change rate outside the i-th cluster and the matching degree and matching degree change rate within the i-th cluster. The matching degree and matching degree change rate outside the i-th cluster are assigned a label of 0, and the matching degree and matching degree change rate within the i-th cluster are assigned a label of 1. The label is used as output and the matching degree and matching degree change rate are used as input to train the i-th binary classification neural network model. According to the classification results, it can be determined whether the current task process is consistent with the i-th cluster; the number of binary classification neural network models is the same as the number of clusters.

[0080] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A lightweight operating system resource isolation method for intelligent terminals, characterized in that: The following steps are involved: S10, the data acquisition module obtains the resource usage data of the current task process and sends the resource usage data of the current task process to the data analysis module; S11, the data analysis module analyzes the resource usage data of the current task process one by one, obtains the resource usage data of the historical task process from the database, and analyzes the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process; S12, the data analysis module analyzes each resource usage data of the current task process to obtain a matching degree, and further analyzes the obtained matching degree during the process execution to obtain a matching degree change rate; S13, the data analysis module analyzes the matching degree and matching degree change rate of the resource usage data of the current task process and the resource usage data of the historical task processes, and determines the type of the current task process based on the analysis results; S14, obtaining historical resource quota data during the execution of historical task processes, obtaining resources occupied by dynamic resource isolation during the execution of historical task processes, and floating up the historical resource quota data during the execution of historical task processes; according to the determined type of the current task process, allocating the floated resource quota to the current task process.

2. The lightweight operating system resource isolation method for intelligent terminals according to claim 1, characterized in that: In step S14, the floating up of historical resource quota data during the execution of historical task processes further includes the following steps: Obtain historical resource quota data and dependencies during the execution of historical task processes, use historical resource quota data with the same dependencies as input features, and perform kmeans clustering on the input features as data points. The specific steps include: S20, select num data points as the initial centroid, num is a positive integer; S21, assign each data point to the nearest centroid to form a cluster; S22, calculate the average value of all data points in each cluster and update the centroid of each cluster; S23, return to step S21, and after reaching the preset number of iterations, proceed to step S24; S24, obtain the resource data occupied by dynamic isolation of resources during the execution of the historical task process; analyze the maximum absolute error between the centroid and all data points in the same cluster; if there is resource data occupied by dynamic isolation of resources that is less than the maximum absolute error between the centroid and all data points in the same cluster, adjust the number num of initial centroids and return to step S20; if there is no resource data occupied by dynamic isolation of resources that is less than the maximum absolute error between the centroid and all data points in the same cluster, complete kmeans clustering.

3. The lightweight operating system resource isolation method for intelligent terminals according to claim 2, characterized in that: The floating up of historical resource quota data during the execution of historical task processes further includes the following steps: The resource data occupied by dynamic isolation of resources during the execution of historical task processes is obtained, and the resource data occupied by dynamic isolation of resources is added to the historical resource quota data corresponding to the centroid to obtain the historical resource quota data after floating.

4. The lightweight operating system resource isolation method for intelligent terminals according to claim 3, characterized in that: In step S11, the analysis of the matching degree between the resource occupancy data of the current task process and the resource occupancy data of the historical task process further includes the following steps: S41, obtain the resource occupancy data of the historical task process in the i-th cluster, the resource occupancy data of each historical task process forms a characteristic curve, the characteristic curves in the cluster are merged into one, and the data analysis module calculates the root mean square error R between the resource occupancy data of the current task process and the characteristic curve ij ; i represents the number of the historical task process, j represents the number of the resource usage data; S42, the data analysis module analyzes the difference between the dependency relationship of the current task process obtained from the data acquisition module and the dependency relationship of the historical task process in the i-th cluster obtained from the database, and compares the difference D between each dependency relationship of the current task process and each dependency relationship of the historical task process in the i-th cluster. k , if the dependency of the current task process is the same as the dependency of the historical task process in the i-th cluster, then D k =1, otherwise D k =0, k is a positive integer between [1, b], indicating the number of dependency relationships, and b is the total number of non-repeated dependency relationships of the current task process and the dependency relationships of the historical task processes; S43, the data analysis module analyzes the resource occupancy data of the current task process and the dependency relationship to obtain the matching degree F between the resource occupancy data of the current task process and the resource occupancy data of the historical task process in the i-th cluster. ij , Where w j represents the weight of the j-th resource occupancy data, v k represents the weight of the kth dependency, m represents the amount of resource usage data, u ij Represents the average value of the j-th resource occupancy data.

5. The lightweight operating system resource isolation method for intelligent terminals according to claim 4, characterized in that: The weight of the resource occupancy data and the weight of the dependency are determined by the following steps: The data analysis module calculates the root mean square error (sd) of the j-th resource occupancy data based on the gap between the j-th resource occupancy data of different cluster centers. j , the weight w of the j-th resource occupancy data is obtained according to the root mean square error of the j-th resource occupancy data j , w j =(1 / sd j ) / Σ(1 / sd j ); According to the weight w of resource occupancy data j Get the weight v of the dependency k , v k =(1 / b)×Σ(w j ); The step of analyzing the obtained matching degree to obtain the matching degree change rate during the process execution further includes the following steps: After the current task process executes for a sampling period T, the data analysis module obtains the matching degree between the resource usage data of the current task process and the resource usage data of the historical task process within the sampling period T; After waiting for a sampling period T, the data analysis module obtains the matching degree between the resource occupancy data of the current task process and the resource occupancy data of the historical task process within the waiting sampling period T, and then obtains the matching degree change rate within the waiting sampling period T.

6. The lightweight operating system resource isolation method for intelligent terminals according to claim 5, characterized in that: In step S13, determining the type of the current task process according to the analysis result further includes the following steps: The binary classification neural network model is trained on the resource usage data of the historical task processes in each cluster. The data analysis module inputs the matching degree and the matching degree change rate into the i-th binary classification neural network model, and determines the type of the current task process based on the classification results of the binary classification neural network model: If there is only one cluster that matches the current task process, the type of the current task process is determined, and the historical resource quota data after floating is allocated according to the historical task process that matches the task process type, and the calculation of dynamic resource isolation is stopped; If there is more than one cluster that matches the current task process, and in the matching clusters, there is historical resource quota data after floating that can meet the requirements of all historical resource quota data in the matching clusters, then the resource limit is allocated according to the historical resource quota data after floating, and the calculation of dynamic resource isolation is stopped; Otherwise, the current task process is monitored and resources are dynamically isolated.

7. A lightweight operating system resource isolation system for smart terminals, using the lightweight operating system resource isolation method for smart terminals according to any one of claims 1 to 6, characterized in that: include: Data acquisition module, data analysis module, database and resource isolation module; The output end of the data acquisition module is connected to the input end of the database, data analysis module and resource isolation module, and is used to obtain resource occupancy data during the execution of the task process; the output end of the data analysis module is connected to the input end of the resource isolation module, and is used to obtain historical resource quota data during the execution of historical task processes, obtain the resources occupied by dynamic resource isolation during the execution of historical task processes, and float the historical resource quota data during the execution of historical task processes; according to the determined type of the current task process, the floated resource quota is allocated to the current task process, and the calculation of dynamic resource isolation is stopped; the output end of the database is connected to the input end of the data analysis module, and is used to store resource occupancy data of historical task processes; the resource isolation module dynamically isolates resources according to the resource occupancy of the task process.

8. The lightweight operating system resource isolation system for intelligent terminals according to claim 7, characterized in that: The data analysis module also includes a dependency modeling unit, a resource quota floating unit, a matching analysis unit and a task process determination unit; the dependency modeling unit is used to determine the dependency relationship between task processes; the resource quota floating unit dynamically isolates the occupied resources and performs resource quotas; the matching analysis unit is used to analyze the matching degree and matching degree change rate between the current task process and the historical task process; The task process determination unit is used to determine the type of the current task process and obtain the resource quota of the current task process.

9. The lightweight operating system resource isolation system for intelligent terminals according to claim 8, characterized in that: The resource isolation module also includes a dynamic isolation unit and a response unit. The dynamic isolation unit trains a resource occupancy prediction model based on the resource data occupied by the task process and adjusts the resource quota in advance; the response unit stops the dynamic isolation unit and releases the resources occupied by the dynamic isolation unit after receiving the instruction issued by the data analysis module.

10. The lightweight operating system resource isolation system for intelligent terminals according to claim 9, characterized in that: The resource quota floating unit obtains the resource data occupied by dynamic isolation of resources during the execution of historical task processes, performs kmeans clustering on historical resource quota data with the same dependency relationship, adds the resource data occupied by dynamic isolation of resources and the historical resource quota data corresponding to the centroid, and obtains the historical resource quota data after floating.

Images