Real-Time Multi-Frame Task Scheduling Feasibility Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing real-time scheduling methods struggle to guarantee the feasibility of multi-frame software tasks with variable execution times, deadlines, and guard times, especially in scenarios without periodic patterns, leading to potential overloads and incorrect scheduler operation.
Innovation Solution
A dynamic priority scheduling method that calculates ratios of execution times to deadlines for each task, compares these across all tasks, and adjusts processor frequency or task parameters to ensure real-time execution feasibility, using the Earliest Deadline First (EDF) scheduling policy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a set of multi-frame tasks with variable execution times and deadlines is scheduled on a processor, then the system can handle complex real-time applications, but it becomes difficult to guarantee real-time feasibility and detect scheduling conflicts
Solution Approach 1:
The patent applies preliminary action by performing a feasibility test during the design phase before actual real-time execution. The test checks whether the sum of execution times of all tasks with the same period constraint is less than or equal to the processor period, allowing potential scheduling conflicts to be detected and resolved before deployment, thus guaranteeing real-time feasibility without compromising the ability to handle complex applications
Solution Approach 2:
The patent implements feedback by using the results of the feasibility test to provide information about scheduling compatibility. If the test fails (sum of execution times exceeds processor period), the system provides feedback that the task set cannot be scheduled in real-time, allowing designers to modify task parameters or add more processors. This feedback mechanism ensures reliable real-time execution while maintaining adaptability to different application requirements
2Productivity
If multiple software tasks execute concurrently on a single processor by sharing resources temporally, then real-time execution is possible, but scheduling conflicts become undetectable in complex applications
Solution Approach 1:
The patent applies parameter changes by transforming the scheduling problem into a mathematical feasibility condition. By changing the perspective from temporal scheduling to parameter validation (checking if sum of execution times ≤ processor period), the system can easily detect scheduling conflicts in complex applications without compromising real-time execution capability. This parameter-based approach simplifies the detection of what would otherwise be difficult to measure scheduling issues
3Loss of time
If the execution time of tasks is reduced to meet deadlines, then real-time constraints are satisfied, but the processor may become overloaded
Solution Approach 1:
The patent applies preliminary anti-action by performing a feasibility test that prevents processor overload before it occurs. The test checks whether the sum of execution times of all tasks with the same period constraint is less than or equal to the processor period. If this condition is not met, the system preemptively identifies the overload risk and provides feedback to modify task parameters or add resources, thus preventing both deadline violations and processor overload rather than reacting after the problem occurs
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for the real-time ordering of an application consisting of a plurality m of software tasks executing at least one processing over a plurality N of consecutive data frames, each of said tasks i being defined, for each of said frames j, at least by an execution time (I), an execution deadline (II), and a holding time (III) relative to the following frame j+1, said holding time (III) being greater than or equal to said deadline (II), wherein according to the method the following steps are performed: for each task i, calculating the ratios (IV) and then searching, over the set of said frames j, for the maximum of said ratios max (IV); comparing the sum (V) to the number p of processors operating in parallel and over which the total calculation load of the real-time application is distributed; if said sum (V) is lower than or equal to the number p of processors, concluding the feasibility of the real-time execution of the set of said software tasks distributed over said processors.