Adaptive Embedded Scheduling With Neuromorphic Fault Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing adaptive embedded systems face challenges in generating schedules that collectively address mode changes and various types of faults, leading to delays or interference, particularly in safety-critical real-time systems, without excessive power consumption or processing time.
Innovation Solution
A method utilizing a multi-model system with spiked neural networks in a neuromorphic accelerator to generate schedules dynamically, adapting to mode changes and faults by explicitly modeling system adaptivity, enabling real-time optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If static configurations are used to meet task timing requirements, then scheduling is simplified, but the system cannot adapt to mode changes or faults
Solution Approach 1:
The system transitions from static scheduling configurations to dynamic adaptive scheduling that automatically adjusts task schedules in response to mode changes and faults. The scheduling system continuously monitors system state and generates updated schedules without requiring full reconfiguration, enabling real-time adaptation while maintaining manageable complexity through incremental adjustments.
Solution Approach 2:
The system changes scheduling parameters dynamically based on operational modes and fault conditions. Different scheduling parameters (such as task priorities, time slots, and resource allocation) are adjusted according to the current system state, allowing the same hardware configuration to support multiple operational modes with appropriate scheduling adjustments.
2Productivity
If multiple cores run tasks in parallel to improve performance, then processing capacity increases, but scheduling problems and resource interference worsen
Solution Approach 1:
The system divides tasks and resources into segments that can be independently scheduled on different cores. By segmenting the task set and associating specific tasks with specific cores or time slots, the system reduces scheduling complexity while maintaining parallel processing capacity. Each core can execute its assigned tasks with minimal interference from other cores.
3Reliability
If the system adapts to faults by changing operation, then fault tolerance improves, but delays or interference occur
Solution Approach 1:
The system pre-configures alternative schedules and fault response strategies before faults occur. When a fault is detected, the system can immediately switch to pre-planned alternative schedules rather than computing new schedules in real-time, minimizing execution delays while maintaining fault tolerance. Critical tasks have pre-identified fallback options ready for immediate activation.
4Reliability
If comprehensive schedules addressing all mode configurations and faults are generated, then system reliability improves, but the time and power required for schedule generation increases
Solution Approach 1:
The system generates schedules partially rather than comprehensively for all possible scenarios. Instead of pre-computing schedules for every possible combination of modes and faults, the system generates schedules for the current operational mode and generates alternative schedules on-demand when mode changes or faults occur. This reduces the computational power and energy required for schedule generation while maintaining reliability through selective schedule generation.
Data Source
Figure 1~4
Figure 5
Figure 6
AI summary
A method of generating schedules for an adaptive embedded system, the method comprising: deriving task sets of all possible tasks to be performed by the embedded system; deriving sets of all possible hardware configurations of the embedded system; creating a multi-model system having a multi-model defining the adaptivity of the system for all possible tasks and all possible hardware and all combinations thereof, the adaptivity defining how the system can change operation responsive to a mode change requirement and/or occurrence of a fault; solving a scheduling problem for the models of the multi-model system in a neuromorphic accelerator implemented by spiked neural networks; and providing schedule instructions to the system, for performance of tasks, based on the solution.