Multiprocessor Motion Control Pipeline Scheduling with Task Offsets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion control software experiences task activation jitter due to non-uniform intervals between wake-up times, leading to increased end-to-end delay and obstacles in shortening task cycles, which current solutions like PREEMPT_RT and dual operating systems require device driver corrections.
Innovation Solution
A method for scheduling a multiprocessor-based motion control software pipeline that calculates parameters for optimal periodic and input/output offsets, using a reference task to generate timers and set offsets for non-reference tasks, allowing for a three-step pipeline execution with optional delay between steps to minimize jitter and maximize resource availability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a PREEMPT_RT real-time patch or dual operating system structure is used to reduce task activation jitter, then task activation jitter is reduced, but device driver correction or re-creation is required
Solution Approach 1:
The patent introduces a task offset value as an intermediary parameter that mediates between the reference task and non-reference tasks. This offset value allows non-reference tasks to be scheduled at precise intervals relative to the reference task, reducing task activation jitter without requiring device driver corrections. The intermediary parameter decouples the timing synchronization requirement from the device driver layer.
Solution Approach 2:
The patent changes the scheduling parameter by introducing a configurable task offset value that can be adjusted to achieve precise task activation timing. By modifying the scheduling parameter (task offset) rather than the device driver structure, the system achieves real-time performance without the complexity of device driver re-creation.
2Device complexity
If task activation jitter is not reduced, then device driver correction is avoided, but end-to-end delay time increases and cycle shortening is obstructed
Solution Approach 1:
The patent performs preliminary action by calculating and setting the task offset value before task execution begins. The admission control program pre-calculates the optimal task offset based on the maximum allowable period and execution times, ensuring that tasks are activated at precise intervals from the start, thereby minimizing end-to-end delay without requiring device driver modifications.
Solution Approach 2:
The patent introduces dynamic scheduling by allowing the task offset value to be adjusted based on system conditions and performance requirements. This dynamic parameter enables the system to optimize task activation timing adaptively, reducing end-to-end delay while maintaining flexibility in the scheduling approach.
3Reliability
If a reference task and non-reference task structure is implemented with task offsets, then task activation jitter is minimized, but scheduling parameter calculation complexity increases
Solution Approach 1:
The patent implements self-service by enabling the admission control program to automatically calculate the task offset values based on provided parameters (maximum allowable period and execution times). The system serves itself by internally computing the optimal scheduling parameters without requiring external intervention or complex manual configuration, thereby minimizing task activation jitter while keeping the calculation process manageable.
Solution Approach 2:
The patent incorporates feedback mechanisms where the admission control program uses the maximum allowable period and execution time information to calculate and adjust task offset values. This feedback loop ensures that the scheduling parameters are optimized based on actual system performance data, achieving minimal task activation jitter through iterative parameter refinement.
4Productivity
If pipeline execution is implemented without delay time between steps, then productivity is improved, but task activation jitter increases
Solution Approach 1:
The patent changes the timing parameter by introducing a configurable delay time between pipeline steps that can be optimized to balance productivity and jitter reduction. By adjusting this parameter, the system can achieve the desired level of task activation regularity while maintaining high pipeline execution speed, resolving the contradiction between productivity and reliability.
Solution Approach 2:
The patent introduces dynamic delay adjustment between pipeline steps, allowing the system to adaptively control the timing between execution stages. This dynamic approach enables the system to maintain high productivity while minimizing task activation jitter by optimizing the delay parameters based on actual performance requirements.
Data Source
AI summary
According to the present invention, a method comprising: (a) calculating, by an admission control program, a parameter for scheduling a pipeline using a maximum allowable period and an execution time for each of N motion tasks, wherein the parameter includes an optimal periodic value, a task offset and an input/output offset for each of the N tasks; (b) generating, by a coordinator program, N motion tasks, wherein one of the N motion tasks is determined to be a reference task, and the other is determined to be a non-reference task; (c) allowing, by the reference task and the non-reference task, an operating system to generate a task wakeup timer and an input/output (I/O) initiating timer; and (d) setting, by the non-reference task, a task offset and an input/output offset for an own timer on the basis of the task wakeup timer of the reference task.