FSM Dataflow Mapping for Low-Power Multicore Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multicore system design fails to accurately account for dynamic factors such as QoS changes and power budget fluctuations, leading to inefficient power management and suboptimal performance in embedded systems, particularly in image processing for unmanned flying objects.
Innovation Solution
A dataflow optimization method and apparatus that creates a Finite State Machine (FSM) with multiple system states to adapt to dynamic factors, optimizing dataflow by mapping tasks to cores and scheduling them to minimize power consumption or maximize QoS levels based on QoS-Controlled-Power-Minimization (QCPM) or Power-Constrained-QoS-Maximization (PCQM) optimization problems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional mapping optimization is performed at design time based on static assumptions, then device complexity is reduced, but power consumption efficiency deteriorates due to inability to adapt to dynamic factors
Solution Approach 1:
The patent implements dynamic mapping optimization by introducing an FSM that transitions between different system states based on runtime conditions. The mapping configuration is no longer static but dynamically adjusted according to detected dynamic factors such as workload changes and power budget fluctuations, allowing the system to adapt its task-to-core mapping in real-time
Solution Approach 2:
The patent employs feedback mechanisms by continuously monitoring dynamic factors (workload characteristics, power budget) and using this information to adjust the mapping optimization. The FSM transitions between states based on detected conditions, and the mapping configuration is updated accordingly, creating a closed-loop control system that responds to runtime changes
2Adaptability or versatility
If the number of system states in FSM is increased to cover more dynamic scenarios, then adaptability improves, but device complexity increases
Solution Approach 1:
The patent segments the dynamic factor space into discrete, manageable states within the FSM. Rather than attempting to handle continuous variations, the system divides the operational space into distinct states (e.g., low workload/high power budget, high workload/low power budget), making the complexity tractable while maintaining adaptability to different conditions
Solution Approach 2:
The patent changes the parameters of the FSM by selectively adjusting the number and type of states based on the specific optimization problem (QCPM vs. PCQM). The system can configure different numbers of QoS levels or power budget levels depending on the problem type, allowing parameter optimization to balance adaptability and complexity
3Use of energy by moving object
If mapping optimization is performed for each possible execution scenario, then power consumption is minimized, but the difficulty of detecting and measuring all scenarios increases
Solution Approach 1:
The patent applies partial action by optimizing mapping for representative system states rather than attempting to enumerate and optimize for every possible execution scenario. The FSM captures the essential dynamic behavior through a有限的 set of states, providing sufficient optimization without the prohibitive complexity of complete scenario coverage
Solution Approach 2:
The patent creates a universal FSM framework that can handle multiple types of optimization problems (QCPM and PCQM) and various dynamic factor combinations through a single unified structure. This multi-functional approach allows the system to adapt to different scenarios without requiring separate detection and optimization mechanisms for each case
Data Source
AI summary
The present disclosure relates to a dataflow optimization method for low-power operation of a multicore system, the dataflow optimization method including: a step (a) of creating an FSM including a plurality of system states in consideration of dynamic factors that trigger a transition in system states for original dataflow; and a step (b) of optimizing the original dataflow through optimization of the created FSM.