Self-aware resource allocation method and system for cloud-edge collaboration

Abstract

The present invention discloses a cloud-edge collaborative self-perceiving resource allocation method, including: the edge end receives basic records regularly issued by various platforms in the central cloud and stores them after overwriting and updating, and saves the device activity records of the device end; the edge end performs header enhancement processing on the request header of the data reporting request initiated by each currently connected device end and forwards it to the central cloud end; the central cloud end decrypts the encrypted data in the received request header and calculates the cloud resource pre-allocation rule; the central cloud end performs resource scheduling and allocation according to the cloud resource pre-allocation rule calculated for each device end. The present invention realizes reasonable resource allocation in an automated manner, effectively reduces labor costs, and realizes plug-in transformation of resource allocation and use.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things and cloud computing, and in particular to a self-perceiving resource allocation method and system for a cloud-edge system. Background Art

[0002] In recent years, applications in various industries have developed towards scale, and the demand for the Internet of Things and cloud computing has increased. Most vendors in the cloud service market do not have the ability to operate the Internet of Things network, and most vendors with the ability to operate the Internet of Things network have weak cloud service capabilities. The relatively independent operation modes of the Internet of Things and cloud computing in the past have difficulty meeting the needs of comprehensive scenarios in terms of technology and methods. For example, in scenarios such as industrial Internet and video surveillance that require both Internet of Things and cloud computing resources, the number of device terminals is large, the user levels are multiple, and the use of resources such as cloud storage is scattered. It is difficult to achieve reasonable allocation of Internet of Things and cloud computing resources, and users need to repeatedly connect to multiple resource systems, which is very troublesome. For example, in the video surveillance scenario, it is necessary to customize and develop business management platforms according to different monitoring business needs, and it is necessary to manually associate Internet of Things traffic for each user point or camera, purchase storage resources, and set storage policies. It is difficult to automatically and reasonably allocate and control Internet of Things and cloud computing resources, and there is a waste of resource utilization. In addition, during the control process, technical personnel who understand business, cloud computing, and Internet of Things are required to operate on various business management platforms, which consumes a lot of manpower and learning costs. Summary of the invention

[0003] The present invention provides a cloud-edge collaborative self-perceiving resource allocation method and system, which helps users to flexibly, safely and reasonably allocate IoT and cloud computing resources to diversified IoT devices without intruding into actual business in terms of external supply and integrated operation of IoT and cloud computing resources, and realizes self-perceiving allocation and scheduling of the device's cloud storage, cloud computing power, IoT traffic and other decomposable resources.

[0004] To achieve this object, the present invention adopts the following technical solutions:

[0005] A cloud-edge collaborative self-aware resource allocation method is provided, comprising the steps of:

[0006] S1, the edge receives the basic records regularly issued by various platforms in the center cloud and stores them after overwriting and updating, while saving the device activity records on the device side;

[0007] S2, the edge end performs header enhancement processing on the request header of the data reporting request initiated by each of the currently connected device ends and forwards it to the central cloud;

[0008] S3, the central cloud decrypts the encrypted data in the received request header and calculates the cloud resource pre-allocation rule;

[0009] S4, the central cloud performs resource scheduling and allocation according to the cloud resource pre-allocation rules calculated for each of the device ends.

[0010] Preferably, the basic records stored in the first recording module at the edge end include any one or more of a user unique ID, a device type, a device unique identification number, an IoT card unique identification number, and a device initial resource allocation level.

[0011] Preferably, the device activity record stored in the first recording module at the edge end includes any one or more of a device data reporting log, a device control instruction reception log, a device online and offline record log, a device server resource occupation status record, and a ratio of actual device resource usage to the allocated size.

[0012] Preferably, in step S2, the method steps of the second processing module at the edge end performing header enhancement processing on the request header include:

[0013] S21, obtaining the basic records of all the device ends that initiate data reporting requests from the first recording module according to a preset fixed interval period M, and obtaining the device activity records of all the device ends that initiate data reporting requests from the first recording module according to a fixed interval period N;

[0014] S22, calculating an indicator value for each of the device ends according to the acquired basic records and device activity records of each of the device ends;

[0015] S23, extract the maximum value Max of each type of indicator i 、Minimum value i ;

[0016] S24, normalize each indicator value of each device end to obtain the indicator relative value G corresponding to each indicator value i ;

[0017] S25, according to the relative value G corresponding to each indicator i and the corresponding preset weight ω i , calculate the activity of the device end of the feature value to be calculated

[0018] S26, the activity corresponding to each of the device terminals connected to the edge terminal Perform logistic regression operations;

[0019] S27, inserting the ciphertext encrypted by the first processing module at the edge into the request header.

[0020] Preferably, N=2M hours.

[0021] Preferably, in step S22, the indicator values ​​to be calculated include the number of bytes uploaded by the device in the past N hours X1, the number of control instructions received by the device X2, the online time of the device in the past N hours X3, the proportion of the edge server CPU occupied by the device requests X4, the proportion of the edge server memory occupied by the device requests X5, the proportion of the edge server bandwidth occupied by the device requests X6, the proportion of the edge server storage occupied by the device requests X7, the initial resource allocation level of the device X8, and the proportion of the actual resource usage of the device in the past N hours to the allocation size X9, any one or more thereof.

[0022] Preferably, in step S24, the relative value of the indicator G i The calculation method is expressed by the following formula (1):

[0023]

[0024] In formula (1), X i represents the index value of the i-th index;

[0025] G i Represents X i The corresponding relative value of the indicator;

[0026] The activity It is calculated by the following formula (2):

[0027]

[0028] In formula (2), K represents the number of indicators;

[0029] The logistic regression operation in step S26 is expressed by the following formula (3):

[0030]

[0031] In formula (3), μ represents the activity corresponding to each device end. The mean of

[0032] γ represents the activity corresponding to each device end The standard deviation of

[0033] Represents the logistic regression function.

[0034] Preferably, in step S27, the data encryption object of the first processing module of the edge end includes any one or more of a device unique identification number, an IoT card unique identification number, a resource package level, a device type, and a timestamp when the edge end receives the device-end data reporting request.

[0035] Preferably, in step S3, the method for calculating cloud resource pre-allocation rules in the central cloud comprises the following steps:

[0036] S31, after receiving the data reporting request forwarded by the second processing module at the edge, the third processing module in the central cloud temporarily stores the request body data and obtains the ciphertext from the request header;

[0037] S32, the first processing module in the central cloud decrypts the ciphertext to obtain the device unique identification number, IoT card unique identification number, resource package level, device type and the timestamp of the edge receiving the device end data reporting request carried in the ciphertext;

[0038] S33, the third processing module of the central cloud obtains the user information and required resource type bound to the Internet of Things card corresponding to the unique identification number of the Internet of Things card obtained by decryption, and the third processing module presets resource packages of different levels for each resource type;

[0039] S34, the third processing module of the central cloud traverses each resource type required by the device, and matches the resource package corresponding to each resource type;

[0040] S35, further matching the cloud resource pre-allocation rule corresponding to each matched resource package according to the relationship between the pre-bound resource packages and the cloud resource pre-allocation rule.

[0041] Preferably, in step S32, the central cloud verifies the decryption result while decrypting the ciphertext, and the verification method includes the steps of:

[0042] S321, determining whether the decrypted device unique identification number and the IoT card unique identification number belong to the same device,

[0043] If yes, the verification is determined to be successful and the process goes to step S322;

[0044] If not, the verification is determined to have failed, and resource allocation to the corresponding device is terminated;

[0045] S322, determine whether T0-T device <0∪T0-T device >5000 milliseconds, T0 represents the current timestamp, T device Represents the timestamp obtained by decryption;

[0046] If not, the verification is determined to be successful;

[0047] If so, the verification is determined to have failed, and resource allocation to the corresponding device is terminated.

[0048] The present invention also provides a cloud-edge collaborative self-sensing resource allocation system, which can implement the self-sensing resource allocation method, including an edge terminal and a central cloud.

[0049] The edge end is used to receive the basic records regularly issued by various platforms of the central cloud and store them after overwriting and updating, and save the device activity records of the device end, and perform header enhancement processing on the request header of the data reporting request initiated by each of the currently connected devices and forward it to the central cloud;

[0050] The central cloud is used to decrypt the encrypted data in the received request header, calculate the cloud resource pre-allocation rules, and schedule and allocate resources according to the cloud resource pre-allocation rules calculated for each device end.

[0051] The present invention adopts a plug-in design concept, and through cloud-edge collaboration, associates users, devices, and IoT cards from the cloud to the edge, and proposes a head-enhanced computing method and processing mode, which realizes the self-perception allocation of the rate, storage resources, and computing resources required for device data transmission at the lowest cost, and realizes the unified operation of cloud (cloud computing) and object (IoT) resources, and automatically schedules them on demand. Users do not need to pay attention to the docking method of objects and cloud resources, and do not need to understand how to evaluate the demand for objects and cloud resources required by the business, and resources are automatically allocated on demand. The present invention is of great significance to promoting the refined management and control of objects and cloud resources, improving resource utilization, reducing user costs, and deepening the comprehensive application scenarios of cloud computing, and on this basis, it realizes the integrated operation service of equipment and resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0053] Figure 1 It is a diagram of the implementation steps of the cloud-edge collaborative self-aware resource allocation method provided by an embodiment of the present invention;

[0054] Figure 2 It is a schematic diagram of resource allocation by a cloud-edge collaborative self-aware resource allocation system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0056] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on this patent; in order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0057] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "inner", "outer", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as a limitation on this patent. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0058] In the description of the present invention, unless otherwise clearly specified and limited, if the term "connection" or the like appears to indicate the connection relationship between components, the term should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two components or the interaction relationship between two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0059] The self-sensing resource allocation method of the cloud-edge system provided by the embodiment of the present invention is as follows: Figure 1 and Figure 2 As shown, the following steps are included:

[0060] S1, the first recording module at the edge receives basic records regularly issued by various platforms at the center cloud through a standard data interface. The basic records include: user unique ID, device type, device unique identification number, IoT card unique identification number, device initial resource allocation level (a large level number indicates a large amount of allocated resources and a high priority). After collecting the data, it is overwritten and updated and stored at the edge;

[0061] The first recording module at the edge also saves the device activity records at the device end, including device data reporting logs, device control command reception logs, device online and offline record logs, device server resource occupation records, and the proportion of device resources actually used to the allocated size. The fixed period for saving device activity records is set to N months, and the edge end deletes log data that exceeds the storage period;

[0062] Device data reporting is initiated by the device, transferred to the central cloud via the edge, and logged at the edge;

[0063] The device control command reception is initiated by the central cloud, forwarded to the corresponding device through the edge, and logged at the edge;

[0064] The device online and offline records are reported by the device and recorded on the edge. If subscribed, they are pushed to the central cloud.

[0065] The device's server resource usage records the server CPU, memory, storage space, and bandwidth occupied by all signaling communications between the device and the central cloud.

[0066] The actual usage of equipment resources, the size of allocated usage and their proportions are statistically fed back through the fourth processing module of the central cloud after the weekly call is completed;

[0067] S2, the device side initiates a data reporting request to the corresponding second processing module of the edge side according to business needs. The second processing module of the edge side regularly calculates indicators for all devices connected to the edge side, including indicator data preprocessing and characteristic value calculation. When the second processing module receives the data reporting request sent by the device side, it inserts the data in the agreed format into the request header without changing the data reporting request body, completes the header enhancement processing of the data reporting request, and forwards the reported data to the central cloud. The specific processing steps are as follows:

[0068] S21, the second processing module at the edge end periodically obtains the basic records of all device ends that initiate data reporting requests from the first recording module according to a preset fixed interval period M, and periodically obtains the device activity records of all device ends that initiate data reporting requests from the first recording module according to a fixed interval period of 2M;

[0069] S22, the second processing module calculates the index value for each device end according to the basic records and device activity records of each device end obtained, and the index values ​​to be calculated include the number of bytes uploaded by the device in the past N hours X1, the number of control instructions received by the device X2, the online time of the device in the past N hours X3, the proportion of the edge server CPU occupied by the device request X4, the proportion of the edge server memory occupied by the device request X5, the proportion of the edge server bandwidth occupied by the device request X6, the proportion of the edge server storage occupied by the device request X7, the initial resource allocation level of the device X8, and the proportion of the actual resource usage of the device in the past N hours to the allocation size X9;

[0070] S23, extract the maximum value Max of each type of indicator i 、Minimum value i ;

[0071] S24, normalize each indicator value of each device end to obtain the indicator relative value G corresponding to each indicator value i , the normalization method is expressed by the following formula (1):

[0072]

[0073] In formula (1), X i represents the index value of the i-th index;

[0074] G i Represents X i The corresponding relative value of the indicator;

[0075] S25, according to the relative value G of each indicator i and the corresponding preset weight ω i , calculate the activity of the device end of the feature value to be calculated Activity It is calculated by the following formula (2):

[0076]

[0077] In formula (2), K represents the number of indicators;

[0078] S26, the activity of each device connected to the edge Perform logistic regression calculation (the purpose of logistic regression calculation is to classify the activity and allocate different levels of resources to different types of equipment. The model is suitable for industry, with small classification calculation amount, fast speed and low storage. In addition, by taking the derivative, Smaller (smaller than all The average value of ), and the part with a smaller derivative (less than 0.25) after the operation is directly classified as the lowest level package; Larger (larger than all The average value of the calculated value is calculated, and the part with a smaller derivative (less than 0.25) is directly classified as the highest level package; the priority of devices with high frequency of resource use is guaranteed to the maximum extent, and the devices with low frequency of resource use are prevented from occupying resources. The logistic regression operation is expressed by the following formula (3):

[0079]

[0080] In formula (3), the activity corresponding to each device end is The mean of

[0081] γ represents the activity corresponding to each device end The standard deviation of

[0082] Represents the logistic regression function.

[0083] Determine the area where the derivative of the preset logistic regression function is larger (greater than 0.25) as the corresponding target definition domain ( When the reciprocal of is greater than 0.25, the corresponding scope);

[0084] The target definition domain is divided into n segments, corresponding to n schedulable resource package levels (different levels of resource packages have different allocations for data transmission rate, distance from cloud storage resource pool, cloud computing allocation computing power, and whether to accelerate. For example: cloud storage package level 1, the device uses 100G of cloud storage capacity per month, and the storage resource pool is randomly allocated; for example, cloud storage package level 2, the device can use 200G of cloud storage capacity per month, and the storage resource pool is closest to the device to ensure fast upload). The jth resource package level is recorded as L j , j∈[0,n-1], where n≥2 (if there is only one package level, no resource allocation is required, so n≥2), the ranges are:

[0085] Level 0: L1 level: …,L n-1 class: The activity is divided into n segments The boundary value corresponding to the i-th segment;

[0086] The larger n is, the higher the resource package level is and the better the performance is.

[0087] Resource packages include: transmission speed limit (speed limit for uploading data from the edge to the central cloud), cloud storage resource pool distance (the edge uploads data to the resource pool of the central cloud. The resource pool of the central cloud is set up with: Zhengzhou node, Shanghai node, Hangzhou node, Guangzhou node, etc. Which node to use, corresponding to the distance between the resource pool node and the edge, has a significant impact on the data upload speed), cloud computing allocation computing power, whether to accelerate and other attributes. The n schedulable resource package levels are pre-allocated when the system is initialized based on the operating scale of the Internet of Things and cloud computing.

[0088] It should be emphasized that through logistic regression calculation, the resource package level that needs to be added to the request header is obtained.

[0089] In addition, the derivative after logistic regression is used to determine devices with relatively low and relatively high activity. The relatively low activity is then directly classified into the lowest level resource package, and the relatively high activity is directly classified into the highest level resource package. This helps to improve the calculation speed of the resource package level corresponding to each device.

[0090] S27, inserting the ciphertext encrypted by the first processing module at the edge into the request header;

[0091] The original parameters for ciphertext encryption include: device unique identification number, IoT card unique identification number, resource package level, device type, and the timestamp when the edge receives the data reporting request from the device.

[0092] The first processing module at the edge end encrypts the original parameters using the existing RSA algorithm, and the specific encryption process is not described here.

[0093] The first processing module at the edge retains the public key to encrypt the original parameters, while the first processing module at the central cloud retains the private key to decrypt the encrypted data inserted in the request header.

[0094] S3, after the central cloud decrypts the encrypted data in the received request header, it calculates the cloud resource pre-allocation rules. The specific calculation process includes the following steps:

[0095] S31, after receiving the data reporting request forwarded by the second processing module at the edge, the third processing module in the central cloud temporarily stores the request body data and obtains the ciphertext from the request header;

[0096] S32, the first processing module in the central cloud (the first processing module in the central cloud and the edge end is the same) decrypts the ciphertext to obtain the device unique identification number, IoT card unique identification number, resource package level, device type and timestamp carried in the ciphertext;

[0097] Preferably, the central cloud verifies the decryption result while decrypting the ciphertext, and the verification method includes the steps of:

[0098] S321, determining whether the decrypted device unique identification number and the IoT card unique identification number belong to the same device,

[0099] If yes, the verification is determined to be successful and the process goes to step S322;

[0100] If not, the verification is determined to have failed, and resource allocation to the corresponding device is terminated;

[0101] S322, determine whether T0-T device <0∪T0-T device >5000 milliseconds, T0 represents the current timestamp, T device Represents the timestamp obtained by decryption;

[0102] If not, the verification is determined to be successful;

[0103] If so, the verification is determined to have failed, and resource allocation to the corresponding device is terminated.

[0104] The central cloud verifies the decryption result while decrypting the ciphertext, discarding the device with the wrong request, which helps to reduce the number of objects for resource allocation and ensure the correctness and rationality of resource allocation.

[0105] S33, the third processing module of the central cloud obtains the user information and required resource type bound to the IoT card corresponding to the unique identification number of the IoT card obtained by decryption, and the third processing module presets resource packages of different levels for each resource type;

[0106] S34, the third processing module of the central cloud traverses each resource type required by the device and matches the resource package corresponding to each resource type;

[0107] S35, according to the relationship between the pre-bound resource packages and the cloud resource pre-allocation rules, further matching the cloud resource pre-allocation rules corresponding to each matched resource package.

[0108] Cloud resource pre-allocation rules include resource allocation location, path, usage, computing power, resource type and other parameters. Resource types include object storage, cloud hard disk, cloud host, cloud container, Internet of Things or computing network bandwidth.

[0109] For example, the pre-allocation rules of object storage resource types are as follows: the distance between the object storage resource pool node and the corresponding edge end is inversely related to the resource package level (the edge node uploads data to the resource pool in the central cloud. The resource pools in the central cloud include: Zhengzhou node, Shanghai node, Hangzhou node, Guangzhou node, etc. Which node to use and the distance between the corresponding resource pool node and the edge end have a significant impact on the data upload speed. Therefore, the longer the distance, the higher the allocated resource package level); object storage data storage path = / edge end unique ID / user information unique ID / IoT card unique identification number / current date (yyyy-MM-dd); object storage data storage path and storage duration are positively related to the resource package level (the higher the resource package level, the longer the storage time), and will be automatically deleted if expired.

[0110] For example, the pre-allocation rules of cloud hard disk resource types are as follows: data storage path in cloud hard disk = edge unique ID / user information unique ID / IoT card unique identification number / current date (yyyy-MM-dd); the size of data and folders stored in cloud hard disk is positively correlated with the resource package level (the higher the resource package level, the larger the storage space).

[0111] The pre-allocation rules for cloud host and cloud container resource types are, for example: cloud host CPU speed and core count, memory size, GPU speed and core count, which are positively correlated with the resource package level. (The higher the resource package level, the faster the CPU speed and the more cores)

[0112] The pre-allocation rules of IoT or cloud computing resource types are, for example, bandwidth limits or alarm traffic values, which are positively correlated with the resource package levels.

[0113] S4, the central cloud performs resource scheduling and allocation according to the cloud resource pre-allocation rules calculated for each device end. Specifically, after the third processing module of the central cloud calculates the cloud resource pre-allocation rules, it calls the componentized resource scheduling capability of the fourth processing module of the central cloud. The fourth processing module includes various resource scheduling components, supports resource scheduling and allocation according to the cloud resource pre-allocation rules, and at the same time, through various resource scheduling components, regularly obtains resource usage and sends it to the first recording module of the corresponding edge end.

[0114] The specific methods of resource scheduling and allocation are:

[0115] After calculating the cloud resource pre-allocation rules, the third processing module of the central cloud calls the resource scheduling componentization capability of the fourth processing module; each resource scheduling component in the fourth processing module of the central cloud interacts with various resource open platforms to obtain temporary access authorizations from cloud computing, Internet of Things and other platforms, and allocates resource space (storage space), monitoring thresholds (bandwidth restrictions or alarm traffic thresholds, equivalent to transmission rate restrictions), paths (paths for storing data reported by devices. Generally, paths are created according to the user, device code, and time to which the device belongs, which facilitates statistics, queries, and data deletion), the size of computing resources occupied (for example: the size of computing resources occupied by subsequent computing and processing (image cutting and splicing) of the data uploaded by the device (video images), etc.), etc. within the current period M according to the cloud resource pre-allocation rules, to achieve self-perceiving classification of various resources.

[0116] The fourth processing module of the central cloud queries the actual usage of the resources allocated this time M time after the previous step ends (for example, if M = 1 hour, then the data reported by the device will be stored in the central cloud for 1 hour, and the central cloud will feedback the actual storage space used by the edge. This is because not every device report needs header enhancement processing and modification of allocated resources. After each header enhancement processing, the device may have uploaded data multiple times), and feedbacks to the first recording module of the corresponding edge to calculate the proportion of actual device resource usage to the allocated size.

[0117] like Figure 2 As shown, the present invention also provides a cloud-edge collaborative self-sensing resource allocation system, which can implement the above-mentioned self-sensing resource allocation method. The device includes an edge terminal and a central cloud.

[0118] The edge side is used to receive the basic records regularly issued by various platforms in the central cloud and store them after overwriting and updating. At the same time, it saves the device activity records on the device side, and performs header enhancement processing on the request header of the data reporting request initiated by each currently connected device side and forwards it to the central cloud;

[0119] The central cloud is used to decrypt the encrypted data in the received request header, calculate the cloud resource pre-allocation rules, and schedule and allocate resources according to the cloud resource pre-allocation rules calculated for each device end.

[0120] In summary, the present invention realizes reasonable resource allocation in an automated manner, effectively reduces labor costs, and realizes plug-in transformation of resource allocation and use.

[0121] It should be noted that the above specific implementations are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art should understand that various modifications, equivalent substitutions, changes, etc. can be made to the present invention. However, as long as these changes do not deviate from the spirit of the present invention, they should be within the scope of protection of the present invention. In addition, some terms used in the specification and claims of this application are not restrictive, but are only for the convenience of description.

Claims

1. A cloud-edge collaborative self-aware resource allocation method, characterized in that: Includes steps: S1, the edge receives the basic records regularly issued by various platforms in the center cloud and stores them after overwriting and updating, while saving the device activity records on the device side; S2, the edge end performs header enhancement processing on the request header of the data reporting request initiated by each of the currently connected device ends and forwards it to the central cloud; S3, the central cloud decrypts the encrypted data in the received request header and calculates the cloud resource pre-allocation rule; S4, the central cloud performs resource scheduling and allocation according to the cloud resource pre-allocation rules calculated for each of the device ends; In step S2, the method steps of the second processing module at the edge end performing header enhancement processing on the request header include: S21, obtaining the basic records of all the device ends that initiate data reporting requests from the first recording module of the edge end according to a preset fixed interval period M, and obtaining the device activity records of all the device ends that initiate data reporting requests from the first recording module according to a fixed interval period N; S22, calculating an indicator value for each of the device ends according to the acquired basic records and device activity records of each of the device ends; S23, extract the maximum value Max of each type of indicator i 、Minimum value i ; S24, normalize each indicator value of each device end to obtain the indicator relative value G corresponding to each indicator value i ; S25, according to the relative value G corresponding to each indicator i and the corresponding preset weight ω i , calculate the activity of the device end of the feature value to be calculated S26, the activity corresponding to each of the device terminals connected to the edge terminal Perform logistic regression operations; S27, inserting the ciphertext encrypted by the first processing module at the edge into the request header; In step S3, the method for calculating cloud resource pre-allocation rules in the central cloud includes the following steps: S31, after receiving the data reporting request forwarded by the second processing module at the edge, the third processing module in the central cloud temporarily stores the request body data and obtains the ciphertext from the request header; S32, the first processing module in the central cloud decrypts the ciphertext to obtain the device unique identification number, IoT card unique identification number, resource package level, device type and the timestamp of the edge receiving the device end data reporting request carried in the ciphertext; S33, the third processing module of the central cloud obtains the user information and required resource type bound to the Internet of Things card corresponding to the unique identification number of the Internet of Things card obtained by decryption, and the third processing module presets resource packages of different levels for each resource type; S34, the third processing module of the central cloud traverses each resource type required by the device, and matches the resource package corresponding to each resource type; S35, further matching the cloud resource pre-allocation rule corresponding to each matched resource package according to the relationship between the pre-bound resource packages and the cloud resource pre-allocation rule.

2. The cloud-edge collaborative self-aware resource allocation method according to claim 1 is characterized in that: The basic records stored in the first recording module at the edge end include any one or more of a user unique ID, a device type, a device unique identification number, an Internet of Things card unique identification number, and a device initial resource allocation level.

3. The cloud-edge collaborative self-aware resource allocation method according to claim 1 is characterized in that: The device activity record stored in the first recording module at the edge includes any one or more of a device data reporting log, a device control instruction receiving log, a device online and offline record log, a device server resource occupation record, and a ratio of actual device resource usage to allocated size.

4. The cloud-edge collaborative self-aware resource allocation method according to claim 1, characterized in that: N = 2M hours.

5. The cloud-edge collaborative self-aware resource allocation method according to claim 1 or 4, characterized in that: In step S22, the indicator values ​​to be calculated include any one or more of the number of bytes uploaded by the device in the past N hours X1, the number of control instructions received by the device X2, the online time of the device in the past N hours X3, the proportion of the edge server CPU occupied by the device requests X4, the proportion of the edge server memory occupied by the device requests X5, the proportion of the edge server bandwidth occupied by the device requests X6, the proportion of the edge server storage occupied by the device requests X7, the initial resource allocation level of the device X8, and the proportion of the actual resource usage of the device in the past N hours to the allocation size X9.

6. The cloud-edge collaborative self-aware resource allocation method according to claim 1, characterized in that: In step S24, the relative value of the index G i The calculation method is expressed by the following formula (1): In formula (1), X i represents the index value of the i-th index; G i Represents X i The corresponding relative value of the indicator; The activity It is calculated by the following formula (2): In formula (2), K represents the number of indicators; The logistic regression operation in step S26 is expressed by the following formula (3): In formula (3), μ represents the activity corresponding to each device end. The mean of γ represents the activity corresponding to each device end The standard deviation of Represents the logistic regression function.

7. The cloud-edge collaborative self-aware resource allocation method according to claim 1, characterized in that: In step S27, the data encryption object of the first processing module of the edge end includes any one or more of the device unique identification number, the IoT card unique identification number, the resource package level, the device type and the timestamp when the edge end receives the device-side data reporting request.

8. The cloud-edge collaborative self-aware resource allocation method according to claim 1, characterized in that: In step S32, the central cloud verifies the decryption result while decrypting the ciphertext, and the verification method includes the steps of: S321, determining whether the decrypted device unique identification number and the IoT card unique identification number belong to the same device, If yes, the verification is determined to be successful and the process goes to step S322; If not, the verification is determined to have failed, and resource allocation to the corresponding device is terminated; S322, determine whether T0-T device <0∪T0-T device >5000 milliseconds, T0 represents the current timestamp, T device Represents the timestamp obtained by decryption; If not, the verification is determined to be successful; If so, the verification is determined to have failed, and resource allocation to the corresponding device is terminated.

9. A cloud-edge collaborative self-sensing resource allocation system, which can implement the self-sensing resource allocation method as described in any one of claims 1 to 8, characterized in that: Including edge and central cloud, The edge end is used to receive the basic records regularly issued by various platforms of the central cloud and store them after overwriting and updating, and save the device activity records of the device end, and perform header enhancement processing on the request header of the data reporting request initiated by each of the currently connected devices and forward it to the central cloud; The central cloud is used to decrypt the encrypted data in the received request header, calculate the cloud resource pre-allocation rules, and schedule and allocate resources according to the cloud resource pre-allocation rules calculated for each device end.

Images