A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer

Abstract

The present invention provides a method for dynamically sharing and reserving heterogeneous resources at the communication and perception layers. The method comprises: obtaining a raw data set through multi-dimensional resource perception and acquisition; constructing a synaesthesia heterogeneous resource state map based on the data set using a graph structure modeling method that comprehensively represents the system state; a core method involves making a decision based on the map and business task characteristics using a three-state control strategy. This decision is based on the task demand vector, resource matching degree, and pressure index, and determines the task scheduling mode as one of communication-dominated, perception-prioritized, or resource-sharing mode using preset decision rules; performing multi-task resource mapping, dynamic reservation, and migration based on this scheduling mode; and iteratively updating the scheduling strategy using a feedback optimization algorithm. By introducing an intelligent three-state control mechanism and a unified resource state map, the present invention achieves intelligent and efficient collaborative scheduling of heterogeneous resources.

Description

Technical Field

[0001] The present invention belongs to the technical field of heterogeneous resource scheduling, and in particular relates to a method for dynamically sharing and reserving heterogeneous resources between a communication layer and a perception layer. Background Art

[0002] With the deep integration of wireless communications and sensing technologies, interawareness integration has become a key technical direction for future information networks. Communication tasks such as high-definition video transmission and perception tasks such as environmental monitoring and target tracking require the sharing of heterogeneous resources such as spectrum, computing, and storage. Dynamically and efficiently coordinating these limited and diverse resources to meet the complex service requirements of multi-tasking concurrency, high reliability, and low latency is a core challenge facing current technological development and is key to improving overall system performance. Current heterogeneous perception resource scheduling solutions suffer from the following deficiencies:

[0003] (1) Resource modeling is not comprehensive. Usually, the relationship is defined based on a single physical indicator, which makes it difficult to quantify the complex coupling relationship between resources at multiple levels. The scheduling decisions made are prone to cause resource conflicts, reducing the overall resource utilization of the system.

[0004] (2) Existing scheduling strategies are mostly static and single-mode. When business demands change, the system cannot adjust resource allocation trends in a timely manner. For example, it still allocates resources conservatively when high-bandwidth communication is required, or fails to concentrate resources when precise sensing is required, which ultimately leads to a serious decline in service quality.

[0005] (3) When determining the specific resource allocation plan, simple polling or priority-based greedy algorithms are often used. The resource allocation plan is prone to fall into local optimality, resulting in low system operation efficiency and high operating costs.

[0006] Therefore, we need to develop a method for dynamic sharing and reservation of heterogeneous resources in the communication layer and the perception layer, which can realize intelligent scheduling and dynamic allocation of heterogeneous resources by constructing a unified resource status map and introducing a three-state task control mechanism. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for dynamic sharing and reservation of heterogeneous resources in the communication layer and the perception layer, so as to solve the problems mentioned in the above background technology, such as one-sided resource state modeling, poor adaptability of scheduling decision strategies, and inefficient resource allocation schemes in existing perception heterogeneous resource scheduling schemes.

[0008] To achieve the above objectives, the present invention provides a method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer, the method being as follows:

[0009] Step S1: Through multi-dimensional resource perception and collection, the original data set of synaesthesia heterogeneous resources is obtained, including communication layer resource data, perception layer resource data and synaesthesia resource coupling indicator data;

[0010] Step S2: Based on the original data set, a graph structure modeling method is used to obtain a synaesthesia heterogeneous resource state map;

[0011] Step S3: Based on the state map of the synaesthesia heterogeneous resources and the business task characteristics, a task scheduling mode is determined by a three-state control strategy; the task scheduling mode includes a communication-dominated mode, a perception-first mode, or a resource-sharing mode;

[0012] The three-state control strategy judgment includes: based on the business task characteristics and the synaesthesia heterogeneous resource state map, performing mode confirmation through a fuzzy decision algorithm; the fuzzy decision algorithm is based on the task requirement vector, the resource matching matrix and the resource pressure index, and completes the mode confirmation through the following judgment rules:

[0013] (1) If the preset communication delay requirement and reliability requirement are met at the same time, and the resource pressure index of the communication layer is less than a first preset threshold, the communication-dominant mode is determined;

[0014] (2) Otherwise, if the preset perception class energy consumption constraints and computing resource requirements are met at the same time, and the resource pressure index of the perception layer is less than the second preset threshold, the perception priority mode is determined;

[0015] (3) Otherwise, if the preset resource matching degree is met and the system's resource pressure index is within the first preset range, the resource sharing mode is determined;

[0016] (4) otherwise, dynamically adjusting the resource allocation strategy based on the resource pressure index and the synaesthesia resource coupling indicator data;

[0017] Step S4: Based on the task scheduling mode and the synaesthesia heterogeneous resource state map, an optimal resource allocation solution is obtained through a multi-task resource mapping algorithm;

[0018] Step S5: Based on the resource allocation plan, a resource scheduling execution result is obtained through a dynamic reservation and migration mechanism;

[0019] Step S6: Based on the resource scheduling execution result, updated scheduling policy parameters are obtained through a feedback optimization algorithm.

[0020] Based on the above-mentioned scheme, the construction process of the synaesthesia heterogeneous resource state map includes: performing data preprocessing on the original data set to obtain standardized resource data, performing node construction based on the standardized resource data to obtain a resource node set, performing association analysis on the resource node set to obtain a resource association edge set, and using the graph structure modeling method to construct a graph of the resource node set and the resource association edge set to obtain the synaesthesia heterogeneous resource state map.

[0021] Based on the above solution, during the node construction process, a resource state vector is defined for each resource node as follows: ,in, Represents a resource node The resource state vector, 、 Represents resource nodes respectively Available transmission bandwidth, communication link quality, sensor module energy consumption budget, pre-processing computing unit occupancy rate, resource conflict degree, sensor bandwidth sharing degree, and interference superposition index.

[0022] Based on the above solution, the association analysis is to determine the edge connection relationship by calculating the resource correlation matrix between resource nodes, including:

[0023] For any two resource nodes, a directed edge is established when one of the following conditions is met: there is a direct data transmission path; they share the same physical resource pool; there is a timing dependency in the execution of business tasks;

[0024] A weight is defined for the directed edge. The weight is obtained by weighted fusion of three indicators that characterize the relationship between two resource nodes. The expression is as follows: ,in Represents a resource node , resource nodes The cosine similarity of the resource state vector; Resource Node , resource nodes The transmission capacity between The business task priority of the directed edge; is the weight coefficient, satisfying .

[0025] Based on the above solution, the step of obtaining the task requirement vector and resource matching matrix includes: performing demand analysis on the business task characteristics to obtain the task requirement vector, performing resource availability analysis based on the task requirement vector and the synaesthesia heterogeneous resource state map to obtain the resource matching degree;

[0026] A priority evaluation is performed on the task requirement vector to obtain a task priority score, which is used as the numerical value of the business task priority.

[0027] Based on the above solution, the resource availability analysis process is as follows: the task demand vector is matched with the resource status vector to obtain the resource matching matrix. Each element in the matrix is ​​calculated according to the following formula:

[0028]

[0029] in, Represents a resource node and business tasks Resource matching, is an indicator function, which takes the value 1 when the condition is met, otherwise it is 0; is the first Components, representing resource nodes The k-th dimension resource attribute value; is the task requirement vector No. Components, representing business tasks For the first Required value of dimension resources; The number of resource dimensions.

[0030] Based on the above scheme, the priority evaluation is achieved through the multi-attribute decision-making method, and the calculation formula of the task priority score is: ,in For the The weight coefficient of each attribute; and The preset Minimum and maximum values ​​for dimension resource requirements.

[0031] Based on the above solution, the steps for obtaining the resource pressure index include:

[0032] Based on the synaesthesia heterogeneous resource state map, the current total resource state vector of the system is calculated using the following formula: ,in Resource Node The weight coefficient of is the total number of resource nodes in the graph, Resource Node The resource state vector;

[0033] Based on the synaesthesia heterogeneous resource state map, calculate the current resource pressure index of the system ,in A collection of currently active business tasks.

[0034] Based on the above solution, the multi-task resource mapping algorithm is a heuristic search algorithm based on graph matching theory. It constructs a task-resource bipartite graph to find the optimal matching solution as the resource allocation solution.

[0035] Based on the above solution, the method of finding the optimal matching solution includes:

[0036] Calculate the matching benefit of each edge in the bipartite graph. The calculation formula is:

[0037]

[0038] in, Improve the value of resource utilization; To allocate costs; is the potential conflict degree; is the balance coefficient;

[0039] Based on the matching benefits, the Hungarian algorithm is used to solve the maximum weight matching problem of the bipartite graph to obtain an initial allocation plan;

[0040] After optimizing the initial allocation plan through the conflict detection and resolution mechanism, the resource allocation plan is obtained.

[0041] The present invention has the following advantages and effects compared to the prior art:

[0042] (1) By adopting a graph structure modeling method, we define resource state vectors for resource nodes and edge weights that integrate cosine similarity, transmission capacity, and business task priority. This allows for a comprehensive and accurate representation of heterogeneous resources, constructs a resource state graph that reflects complex coupling relationships, and provides a reliable basis for subsequent refined scheduling.

[0043] (2) A three-state control strategy decision mechanism is proposed. Based on the task demand vector, resource matching matrix, and resource pressure index, it confirms and switches between the communication-dominant mode, the perception-priority mode, and the resource sharing mode through preset decision rules. The system can adopt the optimal resource allocation tendency according to different business scenarios, avoiding the limitations of a single fixed strategy.

[0044] (3) A multi-task resource mapping algorithm based on graph matching theory is adopted. By constructing a task-resource bipartite graph and solving the optimal allocation plan based on a matching benefit formula that integrates the resource utilization improvement value, allocation cost and potential conflict degree, the macro scheduling model is accurately mapped to the optimal resource allocation plan for each task, ensuring that the decision is executed efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0046] Figure 1 This is a flowchart of a method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to more clearly illustrate the purpose, technical solutions and advantages of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The example implementation methods can be implemented in various forms and should not be understood as being limited to the examples described herein. On the contrary, these implementation methods are provided to make the present invention more comprehensive and complete, and to fully convey the concepts of the example implementation methods to those skilled in the art.

[0048] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present invention. However, it will be appreciated by those skilled in the art that the technical solutions of the present invention can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present invention.

[0049] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0050] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0051] The present invention will be described in detail below with reference to specific embodiments:

[0052] As attached Figure 1As shown, embodiment 1 of the present invention provides a method for dynamically sharing and reserving heterogeneous resources of the communication layer and the perception layer. The specific steps of the method are as follows:

[0053] Step S1: Obtain the original data set of synaesthesia heterogeneous resources through multi-dimensional resource perception and collection;

[0054] Step S2: Based on the original data set, a graph structure modeling method is used to obtain a synaesthesia heterogeneous resource state map;

[0055] Step S3: Based on the state map of the heterogeneous resources and the business task characteristics, a task scheduling mode is obtained through a three-state control strategy judgment;

[0056] Step S4: Based on the task scheduling mode and the synaesthesia heterogeneous resource state map, an optimal resource allocation solution is obtained through a multi-task resource mapping algorithm;

[0057] Step S5: Based on the resource allocation plan, a resource scheduling execution result is obtained through a dynamic reservation and migration mechanism;

[0058] Step S6: Based on the resource scheduling execution result, updated scheduling policy parameters are obtained through a feedback optimization algorithm.

[0059] Specifically, the multi-dimensional resource perception and collection in step S1 refers to real-time monitoring and data collection of various heterogeneous resources in the communication layer and the perception layer; including:

[0060] (1) Regularly scan all communication links through the network monitoring module to obtain real-time bandwidth occupancy and link quality parameters;

[0061] (2) Read the working status of each sensor device through the sensor management module, including the current sampling rate, energy consumption level and computing load;

[0062] (3) The resource coupling relationship between the communication layer and the perception layer is calculated through the resource monitoring agent, and resource coupling indicators such as conflict degree and sharing degree are generated;

[0063] (4) All collected data are aligned according to timestamps to form a raw data set of synaesthesia heterogeneous resources in a unified format.

[0064] Specifically, the original data set includes three parts: communication layer resource data, perception layer resource data, and interaural resource coupling index data. The communication layer resource data includes available transmission bandwidth, communication link quality, relay node occupancy rate, and modulation scheduling queue length; the perception layer resource data includes sensor module energy consumption budget, sensor sampling rate upper limit, and preprocessing calculation unit occupancy rate; and the interaural resource coupling index data includes resource conflict degree, sensor bandwidth sharing degree, and interference superposition index.

[0065] Preferably, the graph structure modeling method described in step S2 is specifically as follows: abstracting each resource module as a node in the graph structure, abstracting the dynamic coupling relationship or migration path between resource modules as an edge in the graph structure, and assigning a multi-dimensional resource state vector to each node to construct a directed weighted graph model that can characterize the complex association relationship between resource modules.

[0066] Preferably, the process of constructing the synaesthesia heterogeneous resource state map includes: performing data preprocessing on the original data set to obtain standardized resource data, performing node construction based on the standardized resource data to obtain a resource node set, performing association analysis on the resource node set to obtain a resource association edge set, and using the graph structure modeling method to construct a graph of the resource node set and the resource association edge set to obtain the synaesthesia heterogeneous resource state map.

[0067] Specifically, data preprocessing includes data cleaning, outlier removal, and normalization. For transmission bandwidth, the Min-Max normalization method is used to map it to the [0, 1] interval. For link quality, Z-score normalization is used to eliminate dimensionality effects. For occupancy indicators, their original percentage form is maintained.

[0068] Specifically, during the node construction process, a resource state vector is defined for each resource node as follows: ,in, Represents a resource node The resource state vector, 、 Represents resource nodes respectively Available transmission bandwidth, communication link quality, sensor module energy consumption budget, pre-processing computing unit occupancy rate, resource conflict degree, sensor bandwidth sharing degree, and interference superposition index.

[0069] Specifically, the association analysis is to determine the edge connection relationship by calculating the resource correlation matrix between resource nodes, including:

[0070] For any two resource nodes, a directed edge is established when one of the following conditions is met: there is a direct data transmission path; they share the same physical resource pool; there is a timing dependency in the execution of business tasks;

[0071] A weight is defined for the directed edge. The weight is obtained by weighted fusion of three indicators that characterize the relationship between two resource nodes. The expression is as follows: ,in Represents a resource node , resource nodes The cosine similarity of the resource state vector; Resource Node , resource nodes The transmission capacity between The business task priority of the directed edge; is the weight coefficient, satisfying ;

[0072] Finally, all the directed edges with defined weights constitute the resource association edge set.

[0073] Preferably, the three-state control strategy judgment in step S3 is based on the business task characteristics and the synaesthesia heterogeneous resource state map, and the mode is confirmed by a fuzzy decision algorithm to obtain the task scheduling mode; the task scheduling mode includes three modes: communication-dominated mode, perception-priority mode or resource sharing mode.

[0074] Preferably, the step of obtaining the task requirement vector and the resource matching matrix includes: performing demand analysis on the business task characteristics to obtain the task requirement vector, performing resource availability analysis based on the task requirement vector and the synaesthesia heterogeneous resource state map to obtain the resource matching degree;

[0075] Performing a priority evaluation on the task requirement vector to obtain a task priority score, and using the score as the numerical value of the business task priority;

[0076] Specifically, the task requirement vector is defined as: ,in, For business tasks bandwidth requirements, energy consumption constraints, computing resource requirements, latency requirements, and reliability requirements.

[0077] Specifically, the resource availability analysis process is: matching the task demand vector with the resource state vector to obtain the following resource matching matrix, where each element is calculated according to the following formula:

[0078]

[0079] in, Represents a resource node and business tasks Resource matching, is an indicator function, which takes the value 1 when the condition is met, otherwise it is 0; is the first Components, representing resource nodes The k-th dimension resource attribute value; is the task requirement vector No. Components (5 components in total), representing business tasks For the first Required value of dimension resources; The number of resource dimensions.

[0080] Specifically, the priority evaluation is implemented through a multi-attribute decision-making method, and the calculation formula of the task priority score is: ,in For the The weight coefficient of each attribute; and The preset Minimum and maximum values ​​for dimension resource requirements.

[0081] Preferably, the step of obtaining the resource pressure index includes:

[0082] Based on the synaesthesia heterogeneous resource state map, the current total resource state vector of the calculation system is calculated as follows: ,in Resource Node The weight coefficient of is the total number of resource nodes in the graph;

[0083] Based on the synaesthesia heterogeneous resource state map, calculate the current resource pressure index of the system ,in A collection of currently active business tasks;

[0084] Preferably, the fuzzy decision algorithm is based on the task requirement vector, resource matching matrix and resource pressure index, and completes the pattern confirmation of the business task through the following decision rules, wherein: is the preset threshold parameter; and These are the resource pressure indexes for the communication layer and the perception layer:

[0085] (1) If the preset communication delay requirement and reliability requirement are met at the same time, and the resource pressure index of the communication layer is less than the first preset threshold, it is determined to be the communication dominant mode; in this embodiment, the preset communication delay requirement and reliability requirement can be specifically reflected as follows 、 ; The first preset threshold can preferably be set to 0.7;

[0086] (2) Otherwise, if the preset perception class energy consumption constraint and computing resource requirement are satisfied at the same time, and the resource pressure index of the perception layer is less than the second preset threshold, the perception priority mode is determined; in this embodiment, the preset perception class energy consumption constraint and computing resource requirement can be specifically embodied as 、 ; The second preset threshold can preferably be set to 0.8;

[0087] (3) Otherwise, if the preset resource matching degree is met and the resource pressure index of the system is within the first preset range, the resource sharing mode is determined; in this embodiment, the preset resource matching degree can be specifically embodied as , the first preset range can preferably be set to ;

[0088] (4) Otherwise, based on the resource pressure index and the synaesthesia resource coupling index data, dynamically adjust the resource allocation strategy, specifically including:

[0089] When the resource pressure index of the communication layer is greater than the first preset threshold, the following communication resource release strategy is implemented: the sampling frequency of the perception layer is reduced to 70% of the original frequency, and the released bandwidth resources are preferentially allocated to business tasks in the communication-dominated mode;

[0090] When the resource pressure index of the perception layer is greater than the second preset threshold, the following perception resource optimization strategy is executed: suspend the data transmission of non-critical business tasks, and reallocate the released computing resources to business tasks in perception priority mode; the non-critical business tasks are business tasks other than critical business tasks, and the critical business tasks include: high-priority business tasks with a task priority score greater than or equal to 0.8, real-time business tasks with a delay requirement of less than or equal to 100ms, high-reliability business tasks with a reliability requirement greater than or equal to 0.95, and emergency scheduling tasks specified by the system administrator.

[0091] When the resource conflict degree is greater than the third preset threshold, the time division multiplexing mechanism is started: according to the formula , dynamically adjust the allocation ratio of communication time slots and sensing time slots, where Determined according to the current business task type ratio, 、 They are total time slot, communication time slot and perception time slot respectively; the third preset threshold is preferably ;

[0092] Furthermore, the system executes corresponding resource allocation strategies for business tasks based on the task scheduling mode, and the resource allocation strategies include:

[0093] (1) If the business task The task scheduling mode is the communication-dominated mode:

[0094] From all available resource nodes, give priority to The resource node is the business task Serve;

[0095] In the selected resource node At least 80% of the available transmission bandwidth should be reserved for business tasks. ;

[0096] Adjust other business tasks based on the interference superposition index to reduce their frequency to the lowest sampling rate;

[0097] (2) If the business task The task scheduling mode is the perception priority mode:

[0098] Based on the sensor bandwidth sharing, 50% of the communication bandwidth is released to support business tasks , and adjusting the routing strategy to reduce the number of transmission hops according to the weights of the directed edges in the synaesthesia heterogeneous resource state graph;

[0099] (3) If the business task The task scheduling mode is the resource sharing mode:

[0100] Based on the resource conflict degree, a time division multiplexing mechanism is used to dynamically allocate communication and sensing time slots, thereby minimizing the overall resource conflict degree of the entire resource pool. To achieve flexible and efficient sharing of resources.

[0101] Preferably, the multi-task resource mapping algorithm in step S4 is a heuristic search algorithm based on graph matching theory, which constructs a task-resource bipartite graph to find the optimal matching solution as the resource allocation solution, specifically including:

[0102] Build a task-resource bipartite graph , where T is the set of business task nodes, R is the set of resource nodes, and E is the set of feasible allocation edges;

[0103] Calculate the matching benefit of each edge in the bipartite graph. The calculation formula is:

[0104]

[0105] in, Improve the value of resource utilization; To allocate costs; is the potential conflict degree; is the balance coefficient.

[0106] Based on the matching benefits, the Hungarian algorithm is used to solve the maximum weight matching problem of the bipartite graph to obtain an initial allocation plan;

[0107] To ensure the feasibility of resource allocation, the resource allocation plan is obtained after optimizing the initial allocation plan through a conflict detection and resolution mechanism.

[0108] Specifically, the dynamic reservation and migration mechanism in step S5 includes three core functional modules: soft reservation marking, resource allocation snapshot, and seamless migration.

[0109] Specifically, the process of obtaining the resource scheduling execution result includes:

[0110] Soft reservation marking is performed on key business tasks: soft reservation marking is applied to resource nodes required for key business tasks in the synaesthesia heterogeneous resource state map to prevent low-priority business tasks from preempting them;

[0111] Periodically save resource allocation snapshots ,in 、 、 They are the resource state vector set, allocation matrix, and business task state set at time t respectively;

[0112] When a business task execution anomaly is detected at time t, the following migration process is triggered: select the valid snapshot closest to the current time point from the snapshot library; calculate the difference between the resource allocation snapshot at time t and the valid snapshot, using the following formula: ,in is the effective snapshot; based on the differences, a migration path is generated to ensure the continuity of business tasks; the system performs resource reallocation, updates abnormal business tasks to the newly planned resource path based on the migration path, and updates the synaesthesia heterogeneous resource state map in real time to generate the resource scheduling execution result.

[0113] Specifically, the feedback optimization algorithm adopts an adaptive parameter adjustment method based on reinforcement learning to continuously optimize system performance by analyzing historical scheduling effects.

[0114] Specifically, in step S6, the process of obtaining the updated scheduling policy parameters includes:

[0115] Collect performance feedback metrics, including task completion latency, system total energy consumption deviation, and resource conflict rate;

[0116] Based on the performance feedback indicators, the following optimization objective function is constructed:

[0117]

[0118] Among them, S is the set of scheduling policy parameters; is the task completion delay of business task j, ΔE is the total energy consumption deviation is the resource conflict rate, For business tasks The importance weight of N is the total number of business tasks; 、 is the penalty factor;

[0119] Based on the optimization objective function, the policy gradient algorithm is used to update the scheduling policy parameters. The calculation formula is as follows:

[0120]

[0121] in, is the scheduling policy parameter obtained in the tth iteration, For The scheduling strategy parameters obtained after updating; η is the learning rate; The gradient of the optimization objective function with respect to the scheduling policy parameters;

[0122] The updated scheduling policy parameters are subjected to constraint checks to ensure that their values ​​are within a reasonable range before being applied to the next round of scheduling cycles.

[0123] In this embodiment, by cyclically executing the above six steps, a unified synaesthesia resource state map is constructed, and a three-state task control mechanism is introduced to achieve intelligent collaborative scheduling of communication and perception heterogeneous resources, effectively improving the overall performance of the multi-task system.

[0124] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present invention are indicated by the claims. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and that various modifications and changes can be made without departing from its scope. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer, characterized in that: include: Step S1: Through multi-dimensional resource perception and collection, the original data set of synaesthesia heterogeneous resources is obtained, including communication layer resource data, perception layer resource data and synaesthesia resource coupling indicator data; Step S2: Based on the original data set, a graph structure modeling method is used to obtain a synaesthesia heterogeneous resource state map; Step S3: Based on the state map of the heterogeneous resources and the business task characteristics, a task scheduling mode is obtained through a three-state control strategy judgment; The task scheduling modes include communication-dominated mode, perception-first mode or resource-sharing mode; The three-state control strategy judgment includes: based on the business task characteristics and the synaesthesia heterogeneous resource state map, performing mode confirmation through a fuzzy decision algorithm; the fuzzy decision algorithm is based on the task requirement vector, the resource matching matrix and the resource pressure index, and completes the mode confirmation through the following judgment rules: a. If the preset communication class delay requirements and reliability requirements are met at the same time, and the resource pressure index of the communication layer is less than the first preset threshold, it is determined to be the communication-dominant mode; b. Otherwise, if the preset perception class energy consumption constraints and computing resource requirements are met at the same time, and the resource pressure index of the perception layer is less than the second preset threshold, the perception priority mode is determined; c. Otherwise, if the preset resource matching degree is met and the system's resource pressure index is within the first preset range, the resource sharing mode is determined; d. Otherwise, dynamically adjust the resource allocation strategy based on the resource pressure index and the synaesthesia resource coupling index data; Step S4: Based on the task scheduling mode and the synaesthesia heterogeneous resource state map, an optimal resource allocation solution is obtained through a multi-task resource mapping algorithm; Step S5: Based on the resource allocation plan, a resource scheduling execution result is obtained through a dynamic reservation and migration mechanism; Step S6: Based on the resource scheduling execution result, updated scheduling policy parameters are obtained through a feedback optimization algorithm.

2. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 1, characterized in that: The process of constructing the synaesthesia heterogeneous resource state graph includes: performing data preprocessing on the original data set to obtain standardized resource data, performing node construction based on the standardized resource data to obtain a resource node set, performing association analysis on the resource node set to obtain a resource association edge set, and using the graph structure modeling method to construct a graph of the resource node set and the resource association edge set to obtain the synaesthesia heterogeneous resource state graph.

3. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 2, characterized in that: During the node construction process, a resource state vector is defined for each resource node as follows: ,in, Represents a resource node The resource state vector, 、 Represents resource nodes respectively Available transmission bandwidth, communication link quality, sensor module energy consumption budget, pre-processing computing unit occupancy rate, resource conflict degree, sensor bandwidth sharing degree, and interference superposition index.

4. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 3, characterized in that: The association analysis is to determine the edge connection relationship by calculating the resource correlation matrix between resource nodes, including: For any two resource nodes, a directed edge is established when one of the following conditions is met: there is a direct data transmission path; they share the same physical resource pool; there is a timing dependency in the execution of business tasks; A weight is defined for the directed edge. The weight is obtained by weighted fusion of three indicators that characterize the relationship between two resource nodes. The expression is as follows: ,in Represents a resource node , resource nodes The cosine similarity of the resource state vector; Resource Node , resource nodes The transmission capacity between The business task priority of the directed edge; is the weight coefficient, satisfying .

5. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 4, characterized in that: The step of obtaining the task requirement vector and resource matching matrix includes: performing demand analysis on the business task characteristics to obtain the task requirement vector, performing resource availability analysis based on the task requirement vector and the synaesthesia heterogeneous resource state map to obtain the resource matching degree; A priority evaluation is performed on the task requirement vector to obtain a task priority score, which is used as the numerical value of the business task priority.

6. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 5, characterized in that: The resource availability analysis process is as follows: matching the task demand vector with the resource state vector to obtain the resource matching matrix. Each element in the matrix is ​​calculated according to the following formula: ,in, Represents a resource node and business tasks Resource matching, is an indicator function, which takes the value 1 when the condition is met, otherwise it is 0; is the first Components, representing resource nodes The k-th dimension resource attribute value; is the task requirement vector No. Components, representing business tasks For the first Required value of dimension resources; The number of resource dimensions.

7. A method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 6, characterized in that: The priority evaluation is achieved through a multi-attribute decision-making method, and the calculation formula of the task priority score is: ,in For the The weight coefficient of each attribute; and The preset Minimum and maximum values ​​for dimension resource requirements.

8. The method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 6, characterized in that: The steps of obtaining the resource pressure index include: Based on the synaesthesia heterogeneous resource state map, the current total resource state vector of the system is calculated using the following formula: ,in Resource Node The weight coefficient of is the total number of resource nodes in the graph, Resource Node The resource state vector; Based on the synaesthesia heterogeneous resource state map, calculate the current resource pressure index of the system ,in A collection of currently active business tasks.

9. The method for dynamically sharing and reserving heterogeneous resources between the communication layer and the perception layer according to claim 1, characterized in that: The multi-task resource mapping algorithm is a heuristic search algorithm based on graph matching theory. It constructs a task-resource bipartite graph to find the optimal matching solution as the resource allocation solution.

10. A method for dynamic sharing and reservation of heterogeneous resources between the communication layer and the perception layer according to claim 9, characterized in that: The searching for the optimal matching solution includes: Calculate the matching benefit of each edge in the bipartite graph. The calculation formula is: ,in, Improve the value of resource utilization; To allocate costs; is the potential conflict degree; is the balance coefficient; Based on the matching benefits, the Hungarian algorithm is used to solve the maximum weight matching problem of the bipartite graph to obtain an initial allocation plan; After optimizing the initial allocation plan through the conflict detection and resolution mechanism, the resource allocation plan is obtained.

Images