CN116339955B - Local optimization method and device for computing communication framework and computer equipment - Google Patents
Local optimization method and device for computing communication framework and computer equipment Download PDFInfo
- Publication number
- CN116339955B CN116339955B CN202310595581.7A CN202310595581A CN116339955B CN 116339955 B CN116339955 B CN 116339955B CN 202310595581 A CN202310595581 A CN 202310595581A CN 116339955 B CN116339955 B CN 116339955B
- Authority
- CN
- China
- Prior art keywords
- task
- agent
- local
- agents
- conflict
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 79
- 238000005457 optimization Methods 0.000 title claims abstract description 78
- 238000004891 communication Methods 0.000 title claims abstract description 73
- 238000012790 confirmation Methods 0.000 claims abstract description 9
- 230000008030 elimination Effects 0.000 claims abstract description 8
- 238000003379 elimination reaction Methods 0.000 claims abstract description 8
- 239000003795 chemical substances by application Substances 0.000 claims description 261
- 230000006870 function Effects 0.000 claims description 51
- 238000005070 sampling Methods 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 9
- 238000003860 storage Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000002035 prolonged effect Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 208000033830 Hot Flashes Diseases 0.000 description 1
- 206010060800 Hot flush Diseases 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012953 risk communication Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to a local optimization method, a device and computer equipment for a computing communication framework. The method comprises the following steps: and constructing a scoring function of tasks related to execution positions of all the agents, calculating tasks to be executed by adopting a local optimization method according to the scoring function to obtain conflict-free scheduling, calculating the tasks to be executed by adopting the local optimization method for all the agents to obtain conflict-free task scheduling, and transmitting the conflict-free task scheduling to other agents in a communication mode if the existence of conflict is detected when global convergence confirmation is carried out, sending a schedule to the other agents, and eliminating task conflict generated in a global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme. The application can reduce the communication times and improve the dispatching generation speed.
Description
Technical Field
The present application relates to the field of agent scheduling technologies, and in particular, to a local optimization method, apparatus, and computer device for a computing and communication framework.
Background
Multi-agent scheduling is not something new, but the recent swarm hot flashes activated it again. The bee colony is an important application form of the multi-agent theory, and endows the multi-agent theory with new characteristics of flexibility, locality, autonomy and the like, so that the bee colony has high engineering value and can be expressed as a group of vehicles, unmanned aerial vehicles, robots or other intelligent agents with individual dynamic states. Depending on individual cost and functionality, clusters may take group action (low cost, single function) or may take distributed precise scheduling (high cost, full function). For the former, a clustering algorithm based on threshold response can realize aggregation, dispersion and other actions. In the latter case, a multi-agent scheduling algorithm is required to achieve precise coordination of assembly, scheduling, etc.
Although scheduling is still a process of matching "agent-task" pairs to time series. The bee colony is mostly composed of low-cost agents, and has higher requirements on calculation, communication and storage, so that the bee colony needs to be considered in practical application. Among these influencing factors, the miniaturization, low power consumption and high performance of the computing units are endlessly layered due to the rapid development of chips, such as the Injeida, AMD or Intel series products, and the contradiction is less prominent. Although communication technologies such as ad hoc networks have made great progress, they remain a more challenging factor due to environmental impact, particularly in complex terrain shielding or electromagnetic environments. Communication between agents is inevitably affected by the environment, resulting in varying degrees of error, packet loss, delay and even interruption, thereby affecting the generation time of the schedule, even success or failure. Therefore, how to reduce the dependence on risk communication, and to enhance the robustness and timeliness of scheduling generation is a problem to be considered.
The traditional method solves the scheduling problem through two stages of calculation and communication, the scheduling obtained through the local agent calculation in the calculation stage, the conflict is resolved through the arrangement of the local agent calculation in the communication stage and the adjacent agent communication, and the two processes are iterated, so that a scheduling scheme is obtained, the former stage focuses on improving the scheduling efficiency and the latter stage focuses on improving the optimizing performance in terms of the two stages, and therefore the problems that the total scheduling generation time is short and the communication time is short are needed to be solved.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a network multi-agent scheduling computing communication system.
A method of local optimization of a computing communication framework, the method comprising:
constructing a scoring function of each agent execution position-related task;
according to the scoring function, calculating the task to be executed by adopting a local optimization method to obtain conflict-free scheduling; in the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through the scores of scoring functions corresponding to the execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task related selection strategy;
performing task calculation to be executed on each agent through a local optimization method to obtain conflict-free task scheduling, and transmitting the conflict-free task scheduling to other agents through a communication mode if conflict is detected to exist when global convergence confirmation is performed;
and sending the schedule to other agents, and eliminating task conflict generated in the global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme.
In one embodiment, the scoring function is expressed as:
；
wherein ,is intelligent body->Execution task->Score of->Representing intelligent agent->Is used for the time schedule of (a),representing intelligent agent->According to the schedule->Execution task->Return of (1)>Is a task->Discount factor of return, < >>Is intelligent body->Along the schedule->Or the position of the agent reaches the task->Estimated time of position->For tasks->Is a primary task report of (1).
In one embodiment, the optimization problem of task scheduling by the agent is:
；
wherein n represents the number of agents, m represents the number of tasks,is a binary variable, when->When in use, intelligent agent->Execution task->When->At the time, represent agent->Do not perform task->，/>Is intelligent body->Execution task->Is a score of (2); constraint conditions of the optimization problem are as follows:
；
；
；
；
wherein ,representing intelligent agent->In task scheduling->Execution task->Is a time of (a) to be used.
In one embodiment, the local optimization includes: task switching and local sampling;
increasing a local scope score by the task exchange; the local scope score refers to the score of all agents in the local scope to execute tasks;
and obtaining task scores of other agents in a local range through the local sampling, so as to obtain conflict-free task scheduling through local estimation.
In one embodiment, the task exchange process includes:
setting a task exchange strategy as follows:
；
；
；
wherein ,is intelligent body->Select exchange->Policy of (2), local fraction->Is intelligent body->And intelligent agent->Sum of scores of->Is to add intelligent agent->Task of (1)>And intelligent agent->Task of (1)>Exchange (I)>Is the local score after the exchange.
In one embodiment, the local sampling process is:
constructing a local auction according to the current scheduling, task information, position information and scoring functions of all the agents in the local scope, wherein the local auction comprises the following steps:
；
wherein ,representing the local estimation function of agent i on other agents within local range, +.>Local auction function for other agent k at current time t +.>Representing task->Task information of->Indicating the location information of agent k.
In one embodiment, calculating distance information between the agent and the task to be executed according to the scoring function;
and determining the position information of the intelligent agent according to at least three pieces of distance information.
In one embodiment, the display agent is the current agent solving the winner vector of other agents in the current iteration from the local memory and solving other agents having task conflicts with the current agent; the implicit agent is an agent that may conflict if no task conflict is currently displayed.
A local optimization apparatus for computing a communication framework, the apparatus comprising:
the scoring function construction module is used for constructing scoring functions of tasks related to execution positions of the intelligent agents;
the local optimization module is used for calculating the task to be executed by adopting a local optimization method according to the scoring function to obtain conflict-free scheduling; in the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through the scores of scoring functions corresponding to the execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task related selection strategy;
the scheduling module is used for carrying out task calculation to be executed on each agent through a local optimization method to obtain conflict-free task scheduling, and when global convergence confirmation is carried out, if conflict is detected, the conflict-free task scheduling is transmitted to other agents through a communication mode;
and the solving module is used for sending the timetable to other agents, and eliminating task conflict generated in the global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme.
In one embodiment, a computer device includes a memory storing a computer program and a processor implementing the steps of the method when executing the computer program:
constructing a scoring function of each agent execution position-related task;
according to the scoring function, calculating the task to be executed by adopting a local optimization method to obtain conflict-free scheduling; in the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through the scores of scoring functions corresponding to the execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task related selection strategy;
performing task calculation to be executed on each agent through a local optimization method to obtain conflict-free task scheduling, and transmitting the conflict-free task scheduling to other agents through a communication mode if conflict is detected to exist when global convergence confirmation is performed;
and sending the schedule to other agents, and eliminating task conflict generated in the global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme.
The local optimization method, the device and the computer equipment of the computing communication framework reduce the communication times by properly increasing the calculated amount under the thought of proposing computing communication, thereby maintaining or even reducing the scheduling generation time. A new scoring function is designed that is more optimal and provides support for inferring the location of other agents. Under the above, a local optimization method is provided, so that the optimality and timeliness of scheduling are further improved. The method of task exchange and task sampling is utilized to explore a better scheduling method, and the problem of local optimum caused by individual greedy is avoided. The conflict among the agents is solved in advance through local estimation, the iteration times are reduced, and meanwhile, the calculated amount is spent on key agents by adopting the proposed task related agent selection strategy.
Drawings
FIG. 1 is a diagram of a communication computing framework in contrast to a traditional market framework in one embodiment, wherein (a) is a traditional market-based framework diagram and (b) is a communication computing framework diagram of the present application;
FIG. 2 is a schematic flow diagram of a method of local optimization of a computing communication framework in one embodiment;
FIG. 3 is a block diagram of a method of local optimization of a computing communication block in one embodiment;
FIG. 4 is an internal schematic diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
As shown in fig. 1, fig. 1 (a) is a schematic diagram of a conventional market-based framework, and fig. 1 (b) is a schematic diagram of a communication computing framework of the present application. As can be seen from the conventional market-based framework of fig. 1 (a), the total schedule time includes a planning time, a communication time, and a waiting time, and the calculation time and the communication time are included in the planning time, the communication time, and the waiting time. In the computing exchange communication framework, the conflict resolution strategy of the computing exchange communication is introduced, and the total incremental planning time is prolonged, so that the communication time is greatly reduced, and the scheduling efficiency is improved.
In one embodiment, as shown in fig. 2, a local optimization method for a computing communication framework is provided, as follows:
step 202, a scoring function of each agent performing a location-related task is constructed.
The scoring function is the basis for the agent's selection task, driving the generation of the agent's schedule.
And 204, calculating the tasks to be executed by adopting a local optimization method according to the scoring function to obtain conflict-free scheduling.
In the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through scores of scoring functions corresponding to execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task-related selection strategy.
In the step, although a certain calculated amount is increased through local optimization, the communication times can be greatly reduced, and along with the great expansion of the calculation resources, the communication frame is replaced through the calculation of the application, so that the scheduling generation time is ensured to be short, and the communication times can be reduced.
And 206, performing task calculation to be executed on each agent through a local optimization method to obtain conflict-free task scheduling, and transmitting the conflict-free task scheduling to other agents through a communication mode if the existence of conflict is detected when global convergence confirmation is performed.
Through local optimization, the conflict between tasks is intended to be resolved, so that conflict-free task scheduling is obtained, and global convergence is detected again.
And step 208, sending the schedule to other agents, and eliminating task conflict generated in the global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme.
In the local optimization method of the computing communication framework, the communication times are reduced by properly increasing the calculated amount under the thought of proposing computing communication, so that the scheduling generation time is maintained or even reduced. A new scoring function is designed that is more optimal and provides support for inferring the location of other agents. Under the above, a local optimization method is provided, so that the optimality and timeliness of scheduling are further improved. The method of task exchange and task sampling is utilized to explore a better scheduling method, and the problem of local optimum caused by individual greedy is avoided. The conflict among the agents is solved in advance through local estimation, the iteration times are reduced, and meanwhile, the calculated amount is spent on key agents by adopting the proposed task related agent selection strategy.
For the scheduling calculation of network multi-agent, the method generally comprises two stages, namely a calculation stage and a communication stage, wherein the agents pursue the common goal of the maximization of the score in the calculation stage, but the scores of the agents are all upper bound due to the limited capability of the agents. In the communication stage, each agent follows the consistency rule obtained by the highest bidding, namely, the task belongs to the agent with the highest score, and the task conflict can be resolved through the score. Thus, market-based methods must converge as long as the problem of deadlock due to fractional oscillations of tasks does not occur. There are three main reasons for influencing convergence speed: the first is the number of task conflicts, which relates to the distribution of agents or tasks and the performance of the algorithm, e.g., individual greedy algorithms are prone to task conflicts; secondly, the monotonicity of the task score, if the score is not monotonic, the task preferred by the agent can be lost due to low score, the task needs to be selected again for iteration, and even the task selection loop is trapped in the dead office; third, after the task score is updated, deleted or added, once the task score is updated, the attribute of the task may change, and further iteration is needed to resolve the task conflict.
In summary, the core of improving the scheduling efficiency of network multi-agent is how to accelerate convergence, i.e. reduce the iteration times. Therefore, in the application, the task conflict is reduced by increasing the calculated amount of other agents, so that iteration is reduced, the calculated time is prolonged, but the iteration number is reduced, so that the communication round is reduced, the calculated amount is increased mainly because of the local optimization including other agents, the communication times are expected to be reduced, and the total scheduling generation time is kept unchanged or even reduced.
In one embodiment, as shown in fig. 3, the computing and communication system of the present application, similar to the conventional computing, retains two iterative stages, in the computing stage, the agent rejects tasks that do not belong to itself according to the task attribute after the conflict resolution, then calculates the task without the conflict through local optimization, and adds it to the agent schedule. Next, convergence confirmation is performed, if a conflict exists, the schedule is transmitted to other agents, the conflict is resolved in the communication stage, and otherwise, the final schedule is obtained. In the communication stage, the agent sends its own schedule to other agents in the form of task winner, task winning bid and time stamp, receives the schedule of other agents, and then resolves task conflict according to the consistency rule. A common consistency rule is that the highest priced agent gets the task higher than the other agents.
In one embodiment, the scoring function is expressed as:
；
wherein ,is intelligent body->Execution task->Score of->Representing intelligent agent->Is used for the time schedule of (a),representing intelligent agent->According to the schedule->Execution task->Return of (1)>Is a task->Discount factor of return, < >>Is intelligent body->Along the schedule->Or the position of the agent reaches the task->Estimated time of position->For tasks->Is a primary task report of (1).
In one embodiment, the optimization problem of the agent for task scheduling is:
；
wherein n represents the number of agents, m represents the number of tasks,is a binary variable, when->When in use, intelligent agent->Execution task->When->At the time, represent agent->Do not perform task->Constraint conditions of the optimization problem are as follows:
；
；
；
；
wherein ,representing intelligent agent->In task scheduling->Execution task->Is a time of (a) to be used.
Specifically, the multi-agent scheduling is that n agents allocate m tasks to form respective scheduling. Each agentThe scheduling pi of (1) is to allocate the execution sequence and time of the tasks, and meet the constraints of the start time and expiration date of the tasks. Can be generally expressed as a constraint optimization problem. The above optimization problem is related to parameters->Improvement is made to introduce tasks->For intelligent agent->The impact of the schedules of (a) drives agents to explore new tasks based on previous tasks, regardless of whether the new tasks are in the vicinity of other agents, which can lead to blind outward exploration. Unlike common scoring functions, the latter term is introduced to drive the agent to select tasks near its location so that the agent is kept from competing tasks near other agents as much as possible, which helps reduce collisions between agents. Each agent tries to select a task closer to itself, so that it is possible to avoid that the agent closer to the task waits for other agents at a distance to execute the task with time limitation, resulting in task failure. In addition, introduced intoDistance information between the agent and the task is also implied.
In one embodiment, the local optimization includes: task exchange and local sampling, wherein the local range score is increased through the task exchange; the local scope score refers to the score of all agents in the local scope to perform tasks; and obtaining task scores of other intelligent agents in a local range through local sampling, and performing local estimation to obtain conflict-free task scheduling.
Specifically, local optimization refers to that an agent considers scheduling of other agents within a local scope, rather than just considering its own scheduling as in individual optimization. This allows agents to avoid possible collisions with other agents within local scope in advance and to attempt to evade individual optimization of the local optima trapped. The local optimization mainly comprises two steps: the first step is task exchange, which improves the scheduling performance by exchanging assigned tasks between agents; and secondly, local sampling is carried out to sample various feasible schedules, and the optimal schedule is selected, so that the scheduling performance is further improved. In both steps, task conflicts will be resolved as the scheduling of other agents is calculated.
In one embodiment, the task exchange process includes:
setting a task exchange strategy as follows:
；
；
；
wherein ,is intelligent body->Select exchange->Policy of (2), local fraction->Is intelligent body->And intelligent agent->Sum of scores of->Is to add intelligent agent->Task of (1)>And intelligent agent->Task of (1)>Exchange (I)>Is the local score after the exchange.
Specifically, in the task exchange step, the local scope is two agents that exchange tasks, and in the local sampling, the local scope is a plurality of agents that are related to tasks. Thus, locally optimized performance is assessed by local scores, i.e. the sum of the scores of all agents within a specified local range. Local scoreThe method comprises the following steps:
；
wherein ,is all agents within local scope, differing in the task exchange and local sampling phases.
For task exchange process, agentTry and other agents->Tasks are exchanged to improve local optimization, which can be manifested by a local score of two agent compositions. Since there may be many possible exchanges, the optimal exchange is chosen to maximize the local score increase.
Specifically, since task exchange requires a large amount of traversal, a heuristic algorithm may be used to solve the task exchange policy, so as to improve the computing efficiency.
In one embodiment, the process of local sampling is:
constructing a local auction according to the current scheduling, task information, position information and scoring functions of all the agents in the local scope, wherein the local auction comprises the following steps:
；
wherein ,representing the local estimation function of agent i on other agents within local range, +.>Local auction function for other agent k at current time t +.>Representing task->Task information of->Indicating the location information of agent k.
Specifically, in each sample, the agentCalculate all relevant agents +.>The task selection and the result progress of the intelligent agent are estimated, so that task conflict with other intelligent agents is avoided in advance, and communication required between the intelligent agents is reduced. The process of estimating other agent scheduling is local estimation +.>Based mainly on all agents in local area +.>All tasks->Information of->Intelligent agent->Position->And scoring function->To construct a local auction.
Specifically, the specific steps of local sampling are as follows:
1. task lattice loss
Task lattice lossMainly prevent intelligent body->Select task->. Optimal disqualifying task->Is the task of maximizing the estimated local score. If->If 0, then there is no non-limiting task:
；
；
2. unqualified task candidates
The disqualified task consists of optional disqualified tasks, which are all agents in a local scopeRelated tasks of (a)：
；
Wherein, the intelligent agentExplicit task candidates of->Consisting of tasks with higher bid than opponents, implicit task candidates->Consists of potentially conflicting tasks:
；
；
as illustrated below, schedules for agent-1 and agent-2 are shown in state-1. Through task exchange, the agent gets better scheduling (state-2). In state-2, agent-3 obtains two samples by local sampling, where the agent of sample-1 knows in advance that agent-2 will obtain task by evaluating task-G and task-H is obtained by discarding task-G. In sample-2, agent-3 actively disqualifies task-H, and lets agent-2 select task-I farther away, with a higher local score.
In one embodiment, to implement the above described local optimization process, it is also necessary to infer the location of other agents. These locations are typically obtained by communication. However, since the communication protocol of the conventional method does not include the location of the agent, it is necessary to add an additional communication protocol and traffic. To maintain the low flow advantage of the traditional approach, the present application attempts to infer the location of other agents from historical data received from the communication without adding a new communication.
In the application, the agentTask->Calculated from the scores, agent +.>To task->Is the estimated arrival time of agent +.>Task->Is associated with the intelligent agent +.>The ratio of the speeds of (2) and thus the agent +.>To task->Can be derived from the inverse of the scoring function:
；
wherein all agents are aware of the taskPosition->Value->And discount factor->Intelligent agent->Cruise speed of (2)Is known.
Computing agentTo task->Distance of->After that, intelligent body->Is at task->Is the center of a circle and the distanceIs a circle of radius. From the three circles defining an intersection, it can be seen that the agent is obtained +.>Only three different tasks are needed for the position of (2)>。
In another embodiment, the computation time will be greatly increased due to the computation of too many other agents. Thus, a suitably increased amount of computation should be used for "competitive" agents that may compete for the task. The core idea of selecting these competing agents is task correlation. Therefore, it is also necessary to specify a corresponding task related agent selection policy.
1. Task related agent
Task related agentInterested in the same task, there may be conflicts, with explicit agents +.>And implicit agent->The composition is as follows:
；
2. display agent
An explicit agent is an agent that currently has an explicit task conflict. Intelligent bodyFrom local memory->Find the current iteration +.>Other intelligent agent->Winner vector->Then all and intelligent agent are determined>Scheduling of->Agent with task conflict->：
；
3. Implicit agent
Implicit agents refer to agents that may conflict in the future in the event that there is no explicit task conflict at the present time. Various strategies for exploring implicit agents may be consideredMainly uses other intelligent agents +.>Winner bid +.>To infer schedule->Or position->：
；
Strategies for implicit agents are explored for a limited time. Current scheduling from agentInitially, tasks under dispatch are explored in sequence. The main exploration is task->Is central, to task->Is>Is the nearest potential agent within a circle of radius:
；
it should be understood that, although the steps in the flowchart of fig. 2 are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in fig. 2 may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor do the order in which the sub-steps or stages are performed necessarily performed in sequence, but may be performed alternately or alternately with at least a portion of the sub-steps or stages of other steps or other steps.
In one embodiment, as shown in fig. 3, there is provided a local optimization device for a computing communication framework, including:
a scoring function construction module 302, configured to construct a scoring function of each agent execution position-related task;
the local optimization module 304 is configured to calculate, according to the scoring function, a task to be executed by adopting a local optimization method, so as to obtain a conflict-free schedule; in the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through the scores of scoring functions corresponding to the execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task related selection strategy;
the scheduling module 306 is configured to perform task calculation to be performed on each agent by using a local optimization method, obtain a conflict-free task schedule, and transmit the conflict-free task schedule to other agents in a communication manner if a conflict is detected when global convergence confirmation is performed;
and the solving module 308 is configured to send the schedule to other agents, and eliminate task conflicts generated in the global convergence stage by adopting a consistency elimination principle, so as to obtain a network multi-agent scheduling scheme.
In one embodiment, the scoring function is expressed as:
；
wherein ,is intelligent body->Execution task->Score of->Representing intelligent agent->Is used for the time schedule of (a),representing intelligent agent->According to the schedule->Execution task->Return of (1)>Is a task->Discount factor of return, < >>Is intelligent body->Along the schedule->Or the position of the agent reaches the task->Estimated time of position->For tasks->Is a primary task report of (1).
In one embodiment, the optimization problem of task scheduling by the agent is:
；
wherein n represents the number of agents, m represents the number of tasks,is a binary variable, when->When in use, intelligent agent->Execution task->When->At the time, represent agent->Do not perform task->，/>Is intelligent body->Execution task->Is a score of (2); constraint conditions of the optimization problem are as follows:
；
；
；
；
wherein ,representing intelligent agent->In task scheduling->Execution task->Is a time of (a) to be used.
In one embodiment, the local optimization includes: task switching and local sampling;
increasing a local scope score by the task exchange; the local scope score refers to the score of all agents in the local scope to execute tasks;
and obtaining task scores of other agents in a local range through the local sampling, so as to obtain conflict-free task scheduling through local estimation.
In one embodiment, the task exchange process includes:
setting a task exchange strategy as follows:
；/>
；
；
wherein ,is intelligent body->Select exchange->Policy of (2), local fraction->Is intelligent body->And intelligent agent->Sum of scores of->Is to add intelligent agent->Task of (1)>And intelligent agent->Task of (1)>Exchange (I)>Is the local score after the exchange.
In one embodiment, the local sampling process is:
constructing a local auction according to the current scheduling, task information, position information and scoring functions of all the agents in the local scope, wherein the local auction comprises the following steps:
；
wherein ,representing the local estimation function of agent i on other agents within local range, +.>Local auction function for other agent k at current time t +.>Representing task->Task information of->Indicating the location information of agent k.
In one embodiment, the determining the location information of the agent includes: calculating distance information of the intelligent agent and the task to be executed according to the scoring function; and determining the position information of the intelligent agent according to at least three pieces of distance information.
In one embodiment, the task related selection policy includes: displaying the agent and the implicit agent; the display agent is the current agent to solve the winner vector of other agents in the current iteration from the local memory and to solve other agents having task conflict with the current agent; the implicit agent is an agent that may conflict if no task conflict is currently displayed.
In one embodiment, a computer device is provided, which may be a terminal, and the internal structure of which may be as shown in fig. 4. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a local optimization method of a computing communication framework. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
It will be appreciated by persons skilled in the art that the architecture shown in fig. 4 is merely a block diagram of some of the architecture relevant to the present inventive arrangements and is not limiting as to the computer device to which the present inventive arrangements are applicable, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In an embodiment a computer device is provided comprising a memory storing a computer program and a processor implementing the steps of the method of the above embodiments when the computer program is executed.
In one embodiment, a computer readable storage medium is provided, on which a computer program is stored which, when executed by a processor, implements the steps of the method of the above embodiments.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples illustrate only a few embodiments of the application, which are described in detail and are not to be construed as limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of protection of the present application is to be determined by the appended claims.
Claims (8)
1. A method for local optimization of a computing communication framework, the method comprising:
constructing a scoring function of each agent execution position-related task;
according to the scoring function, calculating the task to be executed by adopting a local optimization method to obtain conflict-free scheduling; in the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through the scores of scoring functions corresponding to the execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task related selection strategy;
performing task calculation to be executed on each agent through a local optimization method to obtain conflict-free task scheduling, and transmitting the conflict-free task scheduling to other agents through a communication mode if conflict is detected to exist when global convergence confirmation is performed;
transmitting a schedule to other agents, and eliminating task conflict generated in a global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme;
the scoring function is expressed as:
；
wherein ,is intelligent body->Execution task->Score of->Representing intelligent agent->Is used for the time schedule of (a),representing intelligent agent->According to the schedule->Execution task->Return of (1)>Is a task->Discount factor of return, < >>Is intelligent body->Along the schedule->Or the position of the agent reaches the task->Estimated time of position->For tasks->Is a primary task return of (1);
the task related selection strategy comprises the following steps: displaying the agent and the implicit agent;
the display agent is the current agent to solve the winner vector of other agents in the current iteration from the local memory and to solve other agents having task conflict with the current agent;
the implicit agent is an agent that may conflict if no task conflict is currently displayed.
2. The method for optimizing a local communication framework according to claim 1, wherein the optimization problem of task scheduling performed by the agent is:
；
wherein n represents the number of agents, m represents the number of tasks,is a binary variable, when->When in use, intelligent agent->Execution task->When->At the time, represent agent->Do not perform task->，/>Is intelligent body->Execution task->Is a score of (2); constraint conditions of the optimization problem are as follows:
；
；
；
；
wherein ,representing intelligent agent->In task scheduling->Execution task->Is a time of (a) to be used.
3. The method of local optimization of a computing communication framework of claim 1, wherein the local optimization comprises: task switching and local sampling;
increasing a local scope score by the task exchange; the local scope score refers to the score of all agents in the local scope to execute tasks;
and obtaining task scores of other agents in a local range through the local sampling, so as to obtain conflict-free task scheduling through local estimation.
4. A method of local optimization of a computing communication framework as claimed in claim 3, wherein the task exchange process comprises:
setting a task exchange strategy as follows:
；
；
；
wherein ,is intelligent body->Select exchange->Policy of (2), local fraction->Is intelligent body->And intelligent agent->Is used to determine the sum of the scores of the (c),is to add intelligent agent->Task of (1)>And intelligent agent->Task of (1)>Exchange (I)>Is the local score after the exchange.
5. A method of local optimization of a computational communication framework according to claim 3, wherein the process of local sampling is:
constructing a local auction according to the current scheduling, task information, position information and scoring functions of all the agents in the local scope, wherein the local auction comprises the following steps:
；
wherein ,representing the local estimation function of agent i on other agents within local range, +.>Local auction function for other agent k at current time t +.>Representing task->Task information of->Indicating the location information of agent k.
6. The method for local optimization of a computing communication framework according to claim 1, wherein determining location information of an agent comprises:
calculating distance information of the intelligent agent and the task to be executed according to the scoring function;
and determining the position information of the intelligent agent according to at least three pieces of distance information.
7. A local optimization device for a computing communication framework, the device comprising:
the scoring function construction module is used for constructing scoring functions of tasks related to execution positions of the intelligent agents;
the local optimization module is used for calculating the task to be executed by adopting a local optimization method according to the scoring function to obtain conflict-free scheduling; in the local optimization method, all the agents interact in a communication mode, the position information of the agents is determined through the scores of scoring functions corresponding to the execution tasks of other agents stored by the agents, and the local agents contained in the local optimization method are selected through a preset task related selection strategy;
the scheduling module is used for carrying out task calculation to be executed on each agent through a local optimization method to obtain conflict-free task scheduling, and when global convergence confirmation is carried out, if conflict is detected, the conflict-free task scheduling is transmitted to other agents through a communication mode;
the solution module is used for sending the timetable to other agents, and eliminating task conflict generated in the global convergence stage by adopting a consistency elimination principle to obtain a network multi-agent scheduling scheme;
the scoring function is expressed as:
；
wherein ,is intelligent body->Execution task->Score of->Representing intelligent agent->Schedule of->Representing intelligent agent->According to the schedule->Execution task->Return of (1)>Is a task->Discount factor of return, < >>Is intelligent body->Along the schedule->Or the position of the agent reaches the task->Estimated time of position->For tasks->Is a primary task return of (1);
the task related selection strategy comprises the following steps: displaying the agent and the implicit agent;
the display agent is the current agent to solve the winner vector of other agents in the current iteration from the local memory and to solve other agents having task conflict with the current agent;
the implicit agent is an agent that may conflict if no task conflict is currently displayed.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 6 when the computer program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310595581.7A CN116339955B (en) | 2023-05-25 | 2023-05-25 | Local optimization method and device for computing communication framework and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310595581.7A CN116339955B (en) | 2023-05-25 | 2023-05-25 | Local optimization method and device for computing communication framework and computer equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116339955A CN116339955A (en) | 2023-06-27 |
| CN116339955B true CN116339955B (en) | 2023-08-11 |
Family
ID=86889764
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310595581.7A Active CN116339955B (en) | 2023-05-25 | 2023-05-25 | Local optimization method and device for computing communication framework and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116339955B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6490611B1 (en) * | 1999-01-28 | 2002-12-03 | Mitsubishi Electric Research Laboratories, Inc. | User level scheduling of inter-communicating real-time tasks |
| CN107122857A (en) * | 2017-04-26 | 2017-09-01 | 南京航空航天大学 | Workshop multiple target collaboration Optimization Scheduling based on multiple agent |
| CN107479380A (en) * | 2017-08-25 | 2017-12-15 | 东北大学 | Multi-Agent coordination control method based on evolutionary game theory |
| CN108304937A (en) * | 2018-01-30 | 2018-07-20 | 中国计量大学 | A kind of intelligent body electric business agreement based on population metathesis reaction algorithm |
| WO2019154944A1 (en) * | 2018-02-08 | 2019-08-15 | Prowler.Io Limited | Distributed machine learning system |
| CN111586696A (en) * | 2020-04-29 | 2020-08-25 | 重庆邮电大学 | Resource allocation and unloading decision method based on multi-agent architecture reinforcement learning |
| US11269683B1 (en) * | 2020-07-27 | 2022-03-08 | United States Of America As Represented By The Secretary Of The Navy | Agent conflict resolution |
| CN114819273A (en) * | 2022-03-22 | 2022-07-29 | 上海航天壹亘智能科技有限公司 | Workshop scheduling method based on combination of multi-Agent global optimization and local optimization |
| WO2022193534A1 (en) * | 2021-03-17 | 2022-09-22 | 北京交通大学 | Service orchestration system and method based on intent driving in intelligent fusion identification network |
| CN115794341A (en) * | 2022-11-16 | 2023-03-14 | 中国平安财产保险股份有限公司 | Task scheduling method, device, equipment and storage medium based on artificial intelligence |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3270536B1 (en) * | 2016-07-14 | 2019-03-06 | Huawei Technologies Co., Ltd. | Sdn controller and method for task scheduling, resource provisioning and service providing |
| WO2019134254A1 (en) * | 2018-01-02 | 2019-07-11 | 上海交通大学 | Real-time economic dispatch calculation method using distributed neural network |
| US20220413455A1 (en) * | 2020-11-13 | 2022-12-29 | Zhejiang University | Adaptive-learning intelligent scheduling unified computing frame and system for industrial personalized customized production |
-
2023
- 2023-05-25 CN CN202310595581.7A patent/CN116339955B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6490611B1 (en) * | 1999-01-28 | 2002-12-03 | Mitsubishi Electric Research Laboratories, Inc. | User level scheduling of inter-communicating real-time tasks |
| CN107122857A (en) * | 2017-04-26 | 2017-09-01 | 南京航空航天大学 | Workshop multiple target collaboration Optimization Scheduling based on multiple agent |
| CN107479380A (en) * | 2017-08-25 | 2017-12-15 | 东北大学 | Multi-Agent coordination control method based on evolutionary game theory |
| CN108304937A (en) * | 2018-01-30 | 2018-07-20 | 中国计量大学 | A kind of intelligent body electric business agreement based on population metathesis reaction algorithm |
| WO2019154944A1 (en) * | 2018-02-08 | 2019-08-15 | Prowler.Io Limited | Distributed machine learning system |
| CN111586696A (en) * | 2020-04-29 | 2020-08-25 | 重庆邮电大学 | Resource allocation and unloading decision method based on multi-agent architecture reinforcement learning |
| US11269683B1 (en) * | 2020-07-27 | 2022-03-08 | United States Of America As Represented By The Secretary Of The Navy | Agent conflict resolution |
| WO2022193534A1 (en) * | 2021-03-17 | 2022-09-22 | 北京交通大学 | Service orchestration system and method based on intent driving in intelligent fusion identification network |
| CN114819273A (en) * | 2022-03-22 | 2022-07-29 | 上海航天壹亘智能科技有限公司 | Workshop scheduling method based on combination of multi-Agent global optimization and local optimization |
| CN115794341A (en) * | 2022-11-16 | 2023-03-14 | 中国平安财产保险股份有限公司 | Task scheduling method, device, equipment and storage medium based on artificial intelligence |
Non-Patent Citations (1)
| Title |
|---|
| 基于安全智能管理Agent的虚拟企业结构;黄蓓等;《机械与电子》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116339955A (en) | 2023-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lei et al. | Deep reinforcement learning for autonomous internet of things: Model, applications and challenges | |
| Dai et al. | Task co-offloading for D2D-assisted mobile edge computing in industrial internet of things | |
| Kong et al. | A belief propagation-based method for task allocation in open and dynamic cloud environments | |
| Gao et al. | Com-DDPG: Task offloading based on multiagent reinforcement learning for information-communication-enhanced mobile edge computing in the internet of vehicles | |
| Zhang et al. | Balancing energy consumption and reputation gain of UAV scheduling in edge computing | |
| CN114340016A (en) | Power grid edge calculation unloading distribution method and system | |
| Li et al. | Task offloading scheme based on improved contract net protocol and beetle antennae search algorithm in fog computing networks | |
| CN112672382B (en) | Hybrid collaborative computing unloading method and device, electronic equipment and storage medium | |
| Liu et al. | Energy-efficient joint computation offloading and resource allocation strategy for ISAC-aided 6G V2X networks | |
| Sacco et al. | A self-learning strategy for task offloading in UAV networks | |
| Liu et al. | A QoS-guaranteed online user data deployment method in edge cloud computing environment | |
| Wu et al. | Distributed reinforcement learning algorithm of operator service slice competition prediction based on zero-sum markov game | |
| Aslan | Mathematical model and a variable neighborhood search algorithm for mixed-model robotic two-sided assembly line balancing problems with sequence-dependent setup times | |
| Yang et al. | Multi-agent reinforcement learning based file caching strategy in mobile edge computing | |
| CN116339955B (en) | Local optimization method and device for computing communication framework and computer equipment | |
| Khani et al. | Approximate Q-learning-based (AQL) network slicing in mobile edge-cloud for delay-sensitive services | |
| Malandrino et al. | Efficient distributed DNNs in the mobile-edge-cloud continuum | |
| Oh et al. | A few good agents: multi-agent social learning | |
| Sun et al. | Subset selection for hybrid task scheduling with general cost constraints | |
| Pillac et al. | A fast re-optimization approach for dynamic vehicle routing | |
| Kim | New bargaining game based computation offloading scheme for flying ad-hoc networks | |
| CN116634026A (en) | Network multi-agent scheduling calculation exchange communication system, method and computer equipment | |
| Chen et al. | A variable neighbourhood search algorithm for solving dynamic vehicle routing problem under Industry 4.0 | |
| Wang et al. | An Efficient Hybrid Graph Network Model for Traveling Salesman Problem with Drone | |
| Li et al. | Energy–latency tradeoffs edge server selection and DQN-based resource allocation schemes in MEC |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |