Cloud platform computing power scheduling methods and systems for multiple terminals

Abstract

This invention discloses a cloud platform computing power scheduling method and system for multiple terminals. Based on the historical usage records of cloud platform computing resources, the computing power scheduling cycle of the cloud platform is determined; the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle are obtained to determine feasible resource transfer strategies; based on the feasible resource transfer strategies and the connection relationships of all terminals, several terminal clusters are defined; based on the task progress status of the terminal clusters, the computing power resource scheduling links of the terminal clusters are determined; based on the computing power requirements of several terminal clusters, the cluster computing power allocation strategy of the cloud platform is determined; and based on the operating status of the computing power resource scheduling links, the computing power resource scheduling status of the terminal clusters is adjusted. By constructing a multi-level computing power resource scheduling mechanism between the cloud platform and terminals, and between different terminals, it ensures that the cloud platform can simultaneously meet the needs of multiple terminals in a single process of allocating computing power resources, thereby improving the efficiency and accuracy of computing power resource scheduling.

Description

Technical Field

[0001] This invention relates to the field of cloud platform operation technology, and in particular to a cloud platform computing power scheduling method and system for multiple terminals. Background Technology

[0002] As the top-level server of distributed networks such as the Internet of Things (IoT), cloud platforms can interact with terminals connected to the network. Terminals connected to the network, as relatively independent operating units, can execute corresponding tasks. Considering the limited computing resources of their own, when performing complex and large-scale tasks, they typically need to initiate a computing resource scheduling request to the cloud platform. The cloud platform will then request and allocate appropriate computing resources to the terminal. Current computing resource scheduling uses a one-to-one approach, where the cloud platform directly allocates computing resources to the terminal, and the cloud platform directly reclaims the computing resources after the terminal has finished using them, thus ensuring accurate allocation of computing resources to the terminal and avoiding prolonged occupation of computing resources. This approach only achieves single-level scheduling of computing resources from the cloud platform to the terminal. Considering that the cloud platform spends time allocating and reclaiming computing resources, this approach reduces the efficiency of computing resource scheduling and increases the processing load on the cloud platform for scheduling and allocating computing resources. Therefore, how to achieve multi-level scheduling of computing resources between cloud platforms and terminals, as well as between different terminals, and ensure that the cloud platform can meet the needs of multiple terminals simultaneously in one process of allocating computing resources to the outside world, is of great significance for improving the efficiency of computing resource scheduling and improving the reliability and speed of the entire system. Summary of the Invention

[0003] Given that existing computing resource scheduling is limited to the cloud platform and terminals, the cloud platform cannot simultaneously meet the computing resource scheduling needs of multiple terminals during a single allocation of computing resources. This results in excessive time wasted by the cloud platform in allocating and reclaiming computing resources, reducing the efficiency of computing resource scheduling and the overall reliability and speed of the system. In view of the above problems, this invention is proposed to provide a cloud platform computing resource scheduling method for multiple terminals that overcomes or at least partially solves the above problems, comprising:

[0004] Based on the historical usage records of the cloud platform's computing resources, determine the computing power scheduling cycle of the cloud platform; obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle, and determine feasible resource transfer strategies;

[0005] Based on the feasible resource transfer strategy and the connection relationship of all terminals, several terminal clusters are defined; based on the task process status of the terminal clusters, the computing power resource scheduling link of the terminal clusters is determined.

[0006] Based on the computing power requirements of the aforementioned terminal clusters, the cloud platform's cluster computing power allocation strategy is determined; based on the operational status of the computing power resource scheduling link, the computing power resource scheduling status of the terminal clusters is adjusted.

[0007] Optionally, based on the historical usage records of the cloud platform's computing resources, the computing power scheduling cycle of the cloud platform is determined; the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle are obtained, and feasible resource transfer strategies are determined, including:

[0008] Extract computing resource lending and recycling records from the cloud platform's work logs. Based on these records, determine the time from lending to recycling of computing resources for several types of terminals, thereby determining the cloud platform's computing power scheduling cycle.

[0009] The active tasks to be executed within the computing power scheduling cycle are determined from the task logs of each terminal connected to the cloud platform. Based on the computing power resource requirements and execution duration of the processes under each active task of each terminal, the feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined, thereby generating a feasible resource transfer strategy for all terminals.

[0010] Optionally, based on the feasible resource transfer strategy and the connection relationships of all terminals, several terminal clusters are defined; based on the task process status of the terminal clusters, the computing resource scheduling links of the terminal clusters are determined, including:

[0011] Obtain the connection topology between all terminals within the corresponding network of the cloud platform; based on the resource transfer feasibility strategy and the connection topology, divide all terminals into several terminal clusters;

[0012] Based on the execution time limit and the amount of computing resources required for each task process under each terminal in the terminal cluster, the computing resource scheduling link of the terminal cluster is determined; wherein the computing resource scheduling link refers to the transmission path that simultaneously satisfies the task process execution requirements of each terminal during the relay process of the same computing resource in all terminals in the terminal cluster.

[0013] Optionally, based on the computing power requirements of the plurality of terminal clusters, a cluster computing power allocation strategy for the cloud platform is determined; and based on the operating status of the computing power resource scheduling link, the computing power resource scheduling status of the terminal clusters is adjusted, including:

[0014] Based on the computing power acquisition requests initiated by the plurality of terminal clusters to the cloud platform and the amount of available computing power resources of the cloud platform, the cluster computing power allocation strategy of the cloud platform is determined; wherein the computing power acquisition request includes the expected amount of computing power resources to be acquired and the request initiation time; the cluster computing power allocation strategy includes the amount of computing power resources allocated by the cloud platform to the plurality of terminal clusters and the allocation time.

[0015] Based on the utilization status of computing resources by each terminal within the computing resource scheduling link, the idle computing resources held by the terminal are determined; based on the accumulated amount of idle computing resources, the transfer progress of the idle computing resources to the next terminal within the computing resource scheduling link is adjusted.

[0016] As one aspect of the present invention, embodiments of the present invention also provide a cloud platform computing power scheduling system for multiple terminals, comprising:

[0017] The scheduling cycle determination module is used to determine the computing power scheduling cycle of the cloud platform based on the historical usage records of the computing power resources of the cloud platform.

[0018] The feasible transfer strategy determination module is used to obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle and determine the feasible resource transfer strategy.

[0019] The cluster delineation module is used to delineate several terminal clusters based on the feasible resource transfer strategy and the connection relationship of all terminals;

[0020] The scheduling link determination module is used to determine the computing resource scheduling link of the terminal cluster based on the task process status of the terminal cluster.

[0021] The computing power allocation strategy determination module is used to determine the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of the plurality of terminal clusters.

[0022] The scheduling adjustment module is used to adjust the computing resource scheduling status of the terminal cluster according to the operating status of the computing resource scheduling link.

[0023] Optionally, the scheduling cycle determination module is used to determine the computing power scheduling cycle of the cloud platform based on the historical usage records of the cloud platform's computing power resources, including:

[0024] Extract computing resource lending and recycling records from the cloud platform's work logs. Based on these records, determine the time from lending to recycling of computing resources for several types of terminals, thereby determining the cloud platform's computing power scheduling cycle.

[0025] The feasible transfer strategy determination module is used to obtain the task plans of all terminals accessing the cloud platform within the computing power scheduling cycle, and determine feasible resource transfer strategies, including:

[0026] The active tasks to be executed within the computing power scheduling cycle are determined from the task logs of each terminal connected to the cloud platform. Based on the computing power resource requirements and execution duration of the processes under each active task of each terminal, the feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined, thereby generating a feasible resource transfer strategy for all terminals.

[0027] Optionally, the cluster delineation module is used to delineate several terminal clusters based on the resource transfer feasibility strategy and the connection relationships of all terminals, including:

[0028] Obtain the connection topology between all terminals within the corresponding network of the cloud platform; based on the resource transfer feasibility strategy and the connection topology, divide all terminals into several terminal clusters;

[0029] The scheduling link determination module is used to determine the computing resource scheduling link of the terminal cluster based on the task process status of the terminal cluster, including:

[0030] Based on the execution time limit and the amount of computing resources required for each task process under each terminal in the terminal cluster, the computing resource scheduling link of the terminal cluster is determined; wherein the computing resource scheduling link refers to the transmission path that simultaneously satisfies the task process execution requirements of each terminal during the relay process of the same computing resource in all terminals in the terminal cluster.

[0031] Optionally, the computing power allocation strategy determination module is used to determine the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of the plurality of terminal clusters, including:

[0032] Based on the computing power acquisition requests initiated by the plurality of terminal clusters to the cloud platform and the amount of available computing power resources of the cloud platform, the cluster computing power allocation strategy of the cloud platform is determined; wherein the computing power acquisition request includes the expected amount of computing power resources to be acquired and the request initiation time; the cluster computing power allocation strategy includes the amount of computing power resources allocated by the cloud platform to the plurality of terminal clusters and the allocation time.

[0033] The scheduling adjustment module is used to adjust the computing resource scheduling status of the terminal cluster according to the operating status of the computing resource scheduling link, including:

[0034] Based on the utilization status of computing resources by each terminal within the computing resource scheduling link, the idle computing resources held by the terminal are determined; based on the accumulated amount of idle computing resources, the transfer progress of the idle computing resources to the next terminal within the computing resource scheduling link is adjusted.

[0035] The beneficial effects of the above-mentioned technical solutions provided in the embodiments of the present invention include at least the following:

[0036] This invention provides a cloud platform computing power scheduling method and system for multiple terminals. Based on the historical usage records of cloud platform computing resources, the method determines the computing power scheduling cycle of the cloud platform; obtains the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle, and determines feasible resource transfer strategies; based on the feasible resource transfer strategies and the connection relationships of all terminals, it delineates several terminal clusters; based on the task progress status of the terminal clusters, it determines the computing power resource scheduling link of the terminal clusters; based on the computing power requirements of several terminal clusters, it determines the cluster computing power allocation strategy of the cloud platform; and based on the operating status of the computing power resource scheduling link, it adjusts the computing power resource scheduling status of the terminal clusters. By constructing a multi-level computing power resource scheduling mechanism between the cloud platform and terminals, and between different terminals, it ensures that the cloud platform can simultaneously meet the needs of multiple terminals in a single process of allocating computing power resources, thereby improving the efficiency and accuracy of computing power resource scheduling.

[0037] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0038] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0039] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0040] Figure 1 This is a flowchart illustrating the cloud platform computing power scheduling method for multiple terminals provided in this embodiment of the invention.

[0041] Figure 2 This is a schematic diagram of the structure of a cloud platform computing power scheduling system for multiple terminals provided in an embodiment of the present invention. Detailed Implementation

[0042] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0043] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," "outer," "far," "near," "front," and "rear," etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the accompanying drawings and 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, and therefore should not be construed as a limitation of this invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0044] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0045] Please see Figure 1 As shown, an embodiment of this application provides a cloud platform computing power scheduling method for multiple terminals. This cloud platform computing power scheduling method for multiple terminals includes:

[0046] Based on the historical usage records of the cloud platform's computing resources, determine the cloud platform's computing power scheduling cycle; obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle, and determine feasible resource transfer strategies;

[0047] Based on feasible resource transfer strategies and the connection relationships of all terminals, several terminal clusters are defined; based on the task progress status of the terminal clusters, the computing power resource scheduling links of the terminal clusters are determined.

[0048] Based on the computing power requirements of several terminal clusters, determine the cluster computing power allocation strategy of the cloud platform; based on the operating status of the computing power resource scheduling link, adjust the computing power resource scheduling status of the terminal clusters.

[0049] The beneficial effects of the above embodiments are that the cloud platform computing power scheduling method for multiple terminals, by constructing a multi-level computing power resource scheduling between the cloud platform and the terminal as well as between different terminals, ensures that the needs of multiple terminals are met simultaneously in one process of the cloud platform allocating computing power resources to the outside, thereby improving the efficiency and accuracy of computing power resource scheduling.

[0050] In another embodiment, the computing power scheduling cycle of the cloud platform is determined based on the historical usage records of the cloud platform's computing power resources; the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle are obtained, and feasible resource transfer strategies are determined, including:

[0051] Extract computing resource lending and recovery records from the cloud platform's work logs. Based on these records, determine the time from lending to recovery of computing resources for several types of terminals, thereby determining the cloud platform's computing power scheduling cycle.

[0052] The active tasks to be executed within the computing power scheduling cycle are determined from the task logs of all terminals connected to the cloud platform. Based on the computing power resource requirements and execution duration of the processes under the active tasks of each terminal, the feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined, thereby generating feasible resource transfer strategies for all terminals.

[0053] The beneficial effects of the above embodiments are that, as the top-level server within the network, the cloud platform itself needs to perform corresponding computing tasks to maintain the normal and stable operation of the entire network. This means the cloud platform itself also requires a certain amount of computing resources to maintain its own operation. Only by ensuring that it retains sufficient computing resources to maintain its own operation can the cloud platform allocate surplus computing resources to terminals. To ensure the cloud platform continuously allocates computing resources to terminals connected to the network, it will reclaim the computing resources allocated to terminals, achieving sustainable and cyclical use of cloud platform computing resources. To prevent computing resources allocated by the cloud platform from being occupied by individual terminals for extended periods, the cloud platform will set a maximum allowed time for external allocation of computing resources (i.e., the computing resource allocation cycle), preventing computing resources from being occupied for extended periods and unable to enter the cyclical use phase. Considering that the computing resources of the cloud platform need to be scheduled between the cloud platform and the terminal, as well as between different terminals, the above computing power scheduling cycle needs to fully take into account the usage time of different terminals for the allocated computing resources. The computing resource lending and recycling records are extracted from the cloud platform's work logs. Based on the computing resource lending and recycling records, the time from lending to recycling of computing resources for each type of terminal is determined. The usage time of the allocated computing resources for different terminals is comprehensively and accurately counted. Then, the average time from lending to recycling of computing resources for each type of terminal is used as the above computing power scheduling cycle.

[0054] It is understandable that the aforementioned computing power scheduling cycle is a unit of time during which the cloud platform schedules and allocates computing power resources to all terminals connected to the network. That is, the cloud platform schedules and allocates its own computing power resources at the beginning of the aforementioned computing power scheduling cycle and reclaims the computing power resources scheduled and allocated at the end of the aforementioned computing power scheduling cycle. To ensure that computing resources are allocated to all terminals to support their task processing within the aforementioned computing power scheduling cycle, active tasks to be executed within the computing power scheduling cycle are determined from the task logs of each terminal connected to the cloud platform. These active tasks refer to the tasks being processed by the terminals within the aforementioned computing power scheduling cycle. Based on the computing power resource requirements (i.e., the amount of computing power resources required by the processes under each active task) and execution duration (i.e., the execution duration of the processes under the active task) of each terminal, a feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined. The feasible time interval for transferring computing power resources refers to the duration corresponding to each process during the execution of active tasks by the terminal, during which computing power resources are transferred to meet the running requirements of the corresponding process. The amount of resources to be transferred refers to the amount of computing power resources required by each process during the execution of active tasks by the terminal. Based on the feasible time interval and amount of resources to be transferred, a feasible resource transfer strategy for all terminals is generated, providing a basis for subsequently setting the cloud platform's computing power resource scheduling and allocation for terminals.

[0055] In another embodiment, based on feasible resource transfer strategies and the connectivity relationships of all terminals, several terminal clusters are defined; based on the task progress status of the terminal clusters, the computing resource scheduling links of the terminal clusters are determined, including:

[0056] Obtain the connection topology of all terminals within the corresponding network of the cloud platform; based on the feasible resource transfer strategy and the connection topology, divide all terminals into several terminal clusters;

[0057] Based on the execution time limit and the amount of computing resources required for each task process under each terminal in the terminal cluster, the computing resource scheduling link of the terminal cluster is determined; where the computing resource scheduling link refers to the transmission path that simultaneously meets the task process execution requirements of each terminal during the transfer of the same computing resource among all terminals in the terminal cluster.

[0058] The beneficial effect of the above embodiments is that the scheduling and allocation of computing resources between different terminals means that the same batch of computing resources from the cloud platform can be sequentially transferred between several terminals, and each of the aforementioned batch of computing resources can meet the task process execution requirements of each terminal. To ensure the effective transfer of computing resources between different terminals, the connection topology between all terminals within the corresponding network of the cloud platform is first obtained, thereby dividing all terminals into several terminal clusters. It is understood that all terminals within each cluster have the same or similar feasible resource transfer strategies, and the length of the connection links between them is less than a preset path length threshold. Then, based on the execution time limit and the amount of computing resources required for each task process under each terminal within the cluster, the transfer path corresponding to simultaneously meeting the task process execution requirements of each terminal during the transfer process between all terminals within the cluster is determined. Subsequently, the cloud platform can schedule and allocate computing resources on a per-terminal-cluster basis, ensuring that the computing resources allocated to each terminal cluster can meet the computing power requirements for the execution of all terminal task processes within it.

[0059] In another embodiment, a cloud platform cluster computing power allocation strategy is determined based on the computing power requirements of several terminal clusters; the computing power resource scheduling status of the terminal clusters is adjusted according to the operating status of the computing power resource scheduling link, including:

[0060] Based on the computing power acquisition requests initiated by several terminal clusters to the cloud platform and the amount of available computing power resources on the cloud platform, the cluster computing power allocation strategy of the cloud platform is determined; wherein the computing power acquisition request includes the expected amount of computing power resources to be acquired and the request initiation time; the cluster computing power allocation strategy includes the amount of computing power resources allocated by the cloud platform to several terminal clusters and the allocation time.

[0061] Based on the utilization status of computing resources by each terminal within the computing resource scheduling link, determine the idle computing resources held by the terminal; based on the accumulated amount of idle computing resources, adjust the transfer progress of idle computing resources to the next terminal within the computing resource scheduling link.

[0062] The beneficial effects of the above embodiments are that different terminal clusters have different numbers of terminals and different amounts of computing resources required for each terminal to execute its tasks. This results in different computing resource demands from different terminal clusters to the cloud platform. Considering the limited computing resources of the cloud platform itself, in order to enable the cloud platform to accommodate the computing resource needs of different terminal clusters, the amount and timing of computing resource allocation to several terminal clusters are determined based on the computing power acquisition requests initiated by several terminal clusters and the available computing power resources of the cloud platform. This allows the cloud platform to schedule and allocate computing resources to multiple terminal clusters in a time-division manner, effectively supporting the task execution of each terminal cluster. Furthermore, based on the utilization status of computing resources by each terminal within the computing resource scheduling link (i.e., the actual utilization of the computing resources allocated by each terminal), the idle computing resources held by each terminal are determined. During the relay and transfer of the same batch of computing resources within the computing resource scheduling link, not all resources can be fully utilized by each terminal; instead, idle computing resources gradually form as each terminal's task execution progresses. When the accumulated amount of idle computing resources reaches a preset threshold, the idle computing resources are transferred to the next terminal in the computing resource scheduling link to provide computing power support for the next terminal to execute its task process.

[0063] In another embodiment, the above-mentioned determination of the cloud platform's cluster computing power allocation strategy based on the computing power requirements of several terminal clusters, and adjustment of the computing power resource scheduling status of the terminal clusters based on the operating status of the computing power resource scheduling link, can also be implemented as follows:

[0064] The process of determining the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of the aforementioned terminal clusters can be implemented through the following sub-steps:

[0065] Step S51, in the The start time of each computing power scheduling cycle is based on the terminal cluster. The total available computing power of the cloud platform and the number of terminal clusters. The performance credit score will be used to determine the upcoming performance credit score in the next period. Initially, the computing power scheduling cycle is directed towards the terminal cluster. The amount of computing resources allocated.

[0066] Specifically, in the first The start time of each computing power scheduling cycle is for the terminal cluster. Read the stored data for the first time Performance credit score per computing power scheduling cycle (This value is based on the terminal cluster) The performance during the (t-2)th computing power scheduling cycle (calculated and stored at the start of the (t-1)th computing power scheduling cycle); based on the current total available computing power of the cloud platform. The performance credit value of each terminal cluster is determined according to the following formula (1) in the [missing information - likely a specific formula or formula]. Initially, the computing power scheduling cycle is directed towards the terminal cluster. Allocated computing resources :

[0067] (1)

[0068] in, For the upcoming in the Initially, the computing power scheduling cycle is directed towards the terminal cluster. The amount of computing resources allocated; This represents the total available computing power on the current cloud platform. For the already stored terminal cluster For the first The performance credit value for each computing power scheduling cycle; K is the total number of terminal clusters.

[0069] Formula (1) achieves fair allocation based on credit value proportions, according to the principle that higher efficiency, higher credit, and more allocation. It dynamically divides the total computing power to each cluster, which is the direct output of the scheduling strategy.

[0070] Step S52: According to the terminal cluster In the Determine the resource utilization status of the terminal cluster within each computing power scheduling cycle. For the first The performance credit value.

[0071] Specifically, obtain the terminal cluster In the The resource utilization status within each computing power scheduling cycle; based on the resource utilization status, the terminal cluster is determined according to the following recursive formula (2). No. Performance credit score :

[0072] (2)

[0073] in, This is a personalized, adjustable parameter for the credit update rate; the default value is [preset value]. It is preferable to set it to a value between 0.1 and 0.3 to control the smoothness of the credit score; For terminal clusters In the Demand satisfaction rate over -1 computing power scheduling cycle ,in For terminal clusters In the The actual computing power consumed in each computing power scheduling cycle For terminal clusters The reported number of its -Computing power requirement for one computing power scheduling cycle; To predict reward weights, a personalized adjustable parameter is used, which is a preset value, 0 ≤ β ≤ 1, preferably 0.2 to 0.5; For terminal clusters The reward for prediction accuracy in the (t-1)th computing power scheduling cycle, ,in, To optimize the terminal cluster during the (t-2)th computing power scheduling cycle The predicted value of computing resource demand in the (t-1)th computing power scheduling cycle.

[0074] In one embodiment, the specific implementation of adjusting the computing resource scheduling state of the terminal cluster according to the operating state of the computing resource scheduling link is to adjust the computing resource scheduling state of the terminal cluster at the beginning of the t-th computing resource scheduling cycle (e.g., immediately after the completion of the computing resource quantity determination process). The execution process allocated within terminal cluster k includes the following sub-steps:

[0075] Step S61: At the beginning of the t-th computing power scheduling cycle, the determined computing power scheduling cycle will be scheduled to begin in the t-th cycle. Each computing power scheduling cycle is directed to the terminal cluster The allocated computing resources are distributed to the terminal cluster. The first terminal of the computing power resource scheduling link; in the terminal cluster In the computing resource scheduling link, according to the terminal cluster The computing power resource scheduling link includes the idle computing power of upstream terminals, the computing power demand of downstream terminals for pending tasks, and the terminal cluster. The link transmission coefficient in the t-th computing power adjustment cycle determines the amount of computing power resources that need to be transmitted from the upstream terminal to its direct downstream terminal, and controls the transmission of computing power resources from the upstream terminal to its direct downstream terminal according to this amount of computing power resources.

[0076] Specifically, at the beginning of the t-th computing power scheduling cycle, As initial computing resources allocated to the terminal cluster The first terminal of the computing power resource scheduling link;

[0077] Let the terminal number be... Then its initial idle computing power Set as ;

[0078] In the terminal cluster In the computing power resource scheduling link, the amount of resources that need to be allocated from the upstream terminal is determined according to the following formula (3). to its direct downstream terminals The amount of computing resources transferred ,according to Control upstream terminals to its direct downstream terminals Transfer of computing resources:

[0079] (3)

[0080] in, To need to get from upstream terminals Transmitted to downstream terminals The amount of computing resources; For the decision-making moment of computing power transmission, the upstream terminal Idle computing power; For the decision-making moment of computing power transmission, downstream terminals The amount of computing resources required for the next process to be executed in the task queue; For terminal clusters The link transmission coefficient in the t-th computing power adjustment cycle is dimensionless and is a customizable adjustable parameter, typically ranging from [value range missing]. ; The initial value can be set as a preset constant based on system deployment experience or historical data. In a preferred embodiment, the initial value can be set as a fixed value between 0.5 and 0.8.

[0081] The computing power allocation method (steps S51-52) and the cluster internal link transmission method (step S61) based on performance credit value provided in this embodiment of the invention can bring the following beneficial effects:

[0082] By quantifying the historical performance and predictive accuracy of each terminal cluster using performance credit scores, the cloud platform can prioritize the needs of efficient and reliable clusters when allocating computing power. The dynamic updating mechanism of credit scores can sustainably incentivize clusters to optimize task planning and demand forecasting, thereby improving the overall fairness and efficiency of cloud platform computing resource allocation.

[0083] The cloud platform only needs to allocate computing power to the first terminal of the cluster once at the beginning of the scheduling cycle. This computing power can then be automatically passed on to the downstream terminals according to the link transmission rules and their real-time needs. This method enables the same computing power resource to serve multiple terminals within the same cluster sequentially in time, directly achieving the core objective of this invention: "to meet the needs of multiple terminals simultaneously in a single allocation process."

[0084] The decision-making process for transferring computing power between terminals within the cluster is autonomously completed locally according to preset rules, without the cloud platform needing to intervene in the real-time scheduling of each terminal. This hierarchical scheduling mechanism reduces the number of decisions and communication overhead for the cloud platform, allowing it to focus more on macro-level resource coordination, thereby enhancing the system's scalability and overall operational efficiency as the number of terminals increases.

[0085] Furthermore, the above method may also include step S62, upstream terminal. to its direct downstream terminals After the computing power resource transfer is completed, the upstream terminal is updated according to the following formula (4). Idle computing power status:

[0086]

[0087] in, For the updated idle computing power of upstream terminal p, for Idle computing power before transmission is initiated, .

[0088] Furthermore, the above method may also include: at the start of the (t+1)th computing power scheduling cycle or at the end of the tth computing power scheduling cycle, transferring the link coefficients. Optimization is performed to obtain the link transmission coefficient of terminal cluster k in the (t+1)th computing power scheduling cycle, specifically including the following sub-steps:

[0089] Step S71: Within the t-th computing power scheduling cycle, at the preset sampling time, the terminal cluster... The computing power requirements of the pending tasks at the first-end terminals of the computing power resource scheduling link are monitored; based on the sampled computing power requirements of the pending tasks at the first-end terminals, the average computing power requirement of the first-end terminals in the t-th computing power scheduling cycle is determined, and this average computing power requirement is used as the load characteristic value of terminal cluster k in the t-th computing power scheduling cycle. ,Right now ,in To sample at the preset time The collected computing power requirements of the tasks to be processed at the first-end terminal. This represents the total number of samples taken.

[0090] Step S72: Based on the load characteristic value of terminal cluster k in the t-th computing power scheduling cycle. Based on the average load characteristics of all terminal clusters in the t-th computing power scheduling cycle, the link transmission coefficient in the t-th computing power scheduling cycle is optimized to obtain the terminal cluster... The link transmission coefficient in the (t+1)th computing power adjustment cycle.

[0091] Preferably, the terminal cluster is determined according to the following formula. Link transmission coefficient in the (t+1)th computing power adjustment cycle:

[0092]

[0093] in, Indicates terminal cluster The link transmission coefficient in the (t+1)th computing power adjustment cycle; Let be the average load characteristic value of all terminal clusters in the t-th computing power scheduling cycle; To adjust sensitivity, a personalized adjustable parameter is provided. This parameter is dimensionless, represents a preset value, and its range is [range missing]. For example, 0.05 or 0.1 is used to control the amplitude of a single adjustment and avoid oscillation; The preset lower and upper limits for the security side of the link transmission coefficient are dimensionless; for example, they can be set as follows: , Ensure that the coefficients are within a reasonable and effective range; For the clipping function, when The function value is ,when The function value is Otherwise, the function value is .

[0094] The load characteristic value calculation and link transmission coefficient dynamic optimization method (steps S71-72) provided in this embodiment of the invention further enhances the adaptability and resource flow efficiency of the computing power scheduling system. This method dynamically adjusts the internal computing power transmission coefficient by monitoring the actual load of the terminal cluster. This intelligent adjustment mechanism based on real-time load allows the flow rhythm of computing power within the cluster to adaptively match the actual demand changes of each terminal, thereby optimizing the flow efficiency and timeliness of computing power resources and improving the system's flexibility and overall response speed in response to dynamic loads. This method is an important supplement to the aforementioned credit allocation and link transmission mechanisms, together forming a complete, adaptive, multi-level computing power scheduling system from macro-level allocation fairness to micro-level flow efficiency.

[0095] Please see Figure 2 As shown, an embodiment of this application provides a cloud platform computing power scheduling system for multiple terminals. This cloud platform computing power scheduling system for multiple terminals includes:

[0096] The scheduling cycle determination module is used to determine the computing power scheduling cycle of the cloud platform based on the historical usage records of the cloud platform's computing power resources.

[0097] The feasible transfer strategy determination module is used to obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle and determine the feasible resource transfer strategy.

[0098] The cluster delineation module is used to delineate several terminal clusters based on feasible resource transfer strategies and the connection relationships of all terminals;

[0099] The scheduling link determination module is used to determine the computing resource scheduling link of the terminal cluster based on the task process status of the terminal cluster.

[0100] The computing power allocation strategy determination module is used to determine the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of several terminal clusters.

[0101] The scheduling adjustment module is used to adjust the computing resource scheduling status of the terminal cluster based on the running status of the computing resource scheduling link.

[0102] The beneficial effects of the above embodiments are that the cloud platform computing power scheduling system for multiple terminals, by constructing a multi-level computing power resource scheduling between the cloud platform and the terminal as well as between different terminals, ensures that the needs of multiple terminals are met simultaneously in one process of the cloud platform allocating computing power resources to the outside, thereby improving the efficiency and accuracy of computing power resource scheduling.

[0103] In another embodiment, the scheduling cycle determination module is used to determine the computing power scheduling cycle of the cloud platform based on the historical usage records of the cloud platform's computing power resources, including:

[0104] Extract computing resource lending and recovery records from the cloud platform's work logs. Based on these records, determine the time from lending to recovery of computing resources for several types of terminals, thereby determining the cloud platform's computing power scheduling cycle.

[0105] The feasible transfer strategy determination module is used to obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle, and determine the feasible resource transfer strategies, including:

[0106] The active tasks to be executed within the computing power scheduling cycle are determined from the task logs of all terminals connected to the cloud platform. Based on the computing power resource requirements and execution duration of the processes under the active tasks of each terminal, the feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined, thereby generating feasible resource transfer strategies for all terminals.

[0107] In another embodiment, the connection topology between all terminals within the corresponding network of the cloud platform is obtained; based on the feasible resource transfer strategy and the connection topology, all terminals are divided into several terminal clusters.

[0108] The scheduling link determination module is used to determine the computing resource scheduling link of the terminal cluster based on the task process status of the terminal cluster, including:

[0109] Based on the execution time limit and the amount of computing resources required for each task process under each terminal in the terminal cluster, the computing resource scheduling link of the terminal cluster is determined; where the computing resource scheduling link refers to the transmission path that simultaneously meets the task process execution requirements of each terminal during the transfer of the same computing resource among all terminals in the terminal cluster.

[0110] In another embodiment, the computing power allocation strategy determination module is used to determine the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of several terminal clusters, including:

[0111] Based on the computing power acquisition requests initiated by several terminal clusters to the cloud platform and the amount of available computing power resources on the cloud platform, the cluster computing power allocation strategy of the cloud platform is determined; wherein the computing power acquisition request includes the expected amount of computing power resources to be acquired and the request initiation time; the cluster computing power allocation strategy includes the amount of computing power resources allocated by the cloud platform to several terminal clusters and the allocation time.

[0112] The scheduling adjustment module is used to adjust the computing resource scheduling status of the terminal cluster based on the operating status of the computing resource scheduling link, including:

[0113] Based on the utilization status of computing resources by each terminal within the computing resource scheduling link, determine the idle computing resources held by the terminal; based on the accumulated amount of idle computing resources, adjust the transfer progress of idle computing resources to the next terminal within the computing resource scheduling link.

[0114] The operation and effect of the cloud platform computing power scheduling system for multiple terminals of the present invention are consistent with the above-mentioned cloud platform computing power scheduling method for multiple terminals, and the description of the cloud platform computing power scheduling system for multiple terminals will not be repeated here.

[0115] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. This disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims. Thus, if these modifications and variations of the invention fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.

Claims

1. A cloud platform computing power scheduling method for multiple terminals, characterized in that, include: The computing power scheduling cycle of the cloud platform is determined based on the historical usage records of the cloud platform's computing power resources. Obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle, and determine feasible resource transfer strategies; Based on the feasible resource transfer strategy and the connection relationship of all terminals, several terminal clusters are defined; based on the task process status of the terminal clusters, the computing power resource scheduling link of the terminal clusters is determined. Based on the computing power requirements of the aforementioned terminal clusters, determine the cluster computing power allocation strategy of the cloud platform; The computing resource scheduling status of the terminal cluster is adjusted according to the operating status of the computing resource scheduling link.

2. The cloud platform computing power scheduling method for multiple terminals as described in claim 1, characterized in that: The computing power scheduling cycle of the cloud platform is determined based on the historical usage records of the cloud platform's computing power resources. Obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle, and determine feasible resource transfer strategies, including: Extract computing resource lending and recycling records from the cloud platform's work logs. Based on these records, determine the time from lending to recycling of computing resources for several types of terminals, thereby determining the cloud platform's computing power scheduling cycle. The active tasks to be executed within the computing power scheduling cycle are determined from the task logs of each terminal connected to the cloud platform. Based on the computing power resource requirements and execution duration of the processes under each active task of each terminal, the feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined, thereby generating a feasible resource transfer strategy for all terminals.

3. The cloud platform computing power scheduling method for multiple terminals as described in claim 1, characterized in that: Based on the feasible resource transfer strategy and the connection relationships of all terminals, several terminal clusters are defined; based on the task process status of the terminal clusters, the computing resource scheduling links of the terminal clusters are determined, including: Obtain the connection topology between all terminals within the corresponding network of the cloud platform; based on the resource transfer feasibility strategy and the connection topology, divide all terminals into several terminal clusters; Based on the execution time limit and the amount of computing resources required for each task process under each terminal in the terminal cluster, the computing resource scheduling link of the terminal cluster is determined; wherein the computing resource scheduling link refers to the transmission path that simultaneously satisfies the task process execution requirements of each terminal during the relay process of the same computing resource in all terminals in the terminal cluster.

4. The cloud platform computing power scheduling method for multiple terminals as described in claim 1, characterized in that: Based on the computing power requirements of the aforementioned terminal clusters, determine the cluster computing power allocation strategy of the cloud platform; Adjusting the computing resource scheduling status of the terminal cluster based on the operating status of the computing resource scheduling link includes: Based on the computing power acquisition requests initiated by the plurality of terminal clusters to the cloud platform and the amount of available computing power resources of the cloud platform, the cluster computing power allocation strategy of the cloud platform is determined; wherein the computing power acquisition request includes the expected amount of computing power resources to be acquired and the request initiation time; the cluster computing power allocation strategy includes the amount of computing power resources allocated by the cloud platform to the plurality of terminal clusters and the allocation time. Based on the utilization status of computing resources by each terminal within the computing resource scheduling link, the idle computing resources held by the terminal are determined; based on the accumulated amount of idle computing resources, the transfer progress of the idle computing resources to the next terminal within the computing resource scheduling link is adjusted.

5. A cloud platform computing power scheduling system for multiple terminals, characterized in that, include: The scheduling cycle determination module is used to determine the computing power scheduling cycle of the cloud platform based on the historical usage records of the computing power resources of the cloud platform. The feasible transfer strategy determination module is used to obtain the task plans of all terminals connected to the cloud platform within the computing power scheduling cycle and determine the feasible resource transfer strategy. The cluster delineation module is used to delineate several terminal clusters based on the feasible resource transfer strategy and the connection relationship of all terminals; The scheduling link determination module is used to determine the computing resource scheduling link of the terminal cluster based on the task process status of the terminal cluster. The computing power allocation strategy determination module is used to determine the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of the plurality of terminal clusters. The scheduling adjustment module is used to adjust the computing resource scheduling status of the terminal cluster according to the operating status of the computing resource scheduling link.

6. The cloud platform computing power scheduling system for multiple terminals as described in claim 5, characterized in that: The scheduling cycle determination module is used to determine the computing power scheduling cycle of the cloud platform based on the historical usage records of the cloud platform's computing power resources, including: Extract computing resource lending and recycling records from the cloud platform's work logs. Based on these records, determine the time from lending to recycling of computing resources for several types of terminals, thereby determining the cloud platform's computing power scheduling cycle. The feasible transfer strategy determination module is used to obtain the task plans of all terminals accessing the cloud platform within the computing power scheduling cycle, and determine feasible resource transfer strategies, including: The active tasks to be executed within the computing power scheduling cycle are determined from the task logs of each terminal connected to the cloud platform. Based on the computing power resource requirements and execution duration of the processes under each active task of each terminal, the feasible time interval and amount of computing power resources to be transferred during the execution of active tasks by different terminals are determined, thereby generating a feasible resource transfer strategy for all terminals.

7. The cloud platform computing power scheduling system for multiple terminals as described in claim 5, characterized in that: The cluster delineation module is used to delineate several terminal clusters based on the feasible resource transfer strategy and the connection relationships of all terminals, including: Obtain the connection topology between all terminals within the corresponding network of the cloud platform; based on the resource transfer feasibility strategy and the connection topology, divide all terminals into several terminal clusters; The scheduling link determination module is used to determine the computing resource scheduling link of the terminal cluster based on the task process status of the terminal cluster, including: Based on the execution time limit and the amount of computing resources required for each task process under each terminal in the terminal cluster, the computing resource scheduling link of the terminal cluster is determined; wherein the computing resource scheduling link refers to the transmission path that simultaneously satisfies the task process execution requirements of each terminal during the relay process of the same computing resource in all terminals in the terminal cluster.

8. The cloud platform computing power scheduling system for multiple terminals as described in claim 5, characterized in that: The computing power allocation strategy determination module is used to determine the cluster computing power allocation strategy of the cloud platform based on the computing power requirements of the plurality of terminal clusters, including: Based on the computing power acquisition requests initiated by the plurality of terminal clusters to the cloud platform and the amount of available computing power resources of the cloud platform, the cluster computing power allocation strategy of the cloud platform is determined; wherein the computing power acquisition request includes the expected amount of computing power resources to be acquired and the request initiation time; the cluster computing power allocation strategy includes the amount of computing power resources allocated by the cloud platform to the plurality of terminal clusters and the allocation time. The scheduling adjustment module is used to adjust the computing resource scheduling status of the terminal cluster according to the operating status of the computing resource scheduling link, including: Based on the utilization status of computing resources by each terminal within the computing resource scheduling link, the idle computing resources held by the terminal are determined; based on the accumulated amount of idle computing resources, the transfer progress of the idle computing resources to the next terminal within the computing resource scheduling link is adjusted.

Images