[Paper Review] Multi-mode Sampling Period Selection for Embedded Real Time Control

Rajorshee Raha, Soumyajit Dey+2 more|arXiv (Cornell University)|Jun 29, 2015

Real-time simulation and control systems6 references4 citations

TL;DR

This paper proposes multi-mode sampling period selection for embedded real-time control, where controllers switch between a limited set of predefined sampling rates based on control-theoretic stability analysis. By precomputing schedules for each mode and using an automaton-based scheduler, the method achieves 30–50% reduction in ECU bandwidth with no loss in control performance, validated on ABS and ACC systems.

ABSTRACT

Recent studies have shown that adaptively regulating the sampling rate results in significant reduction in computational resources in embedded software based control. Selecting a uniform sampling rate for a control loop is robust, but overtly pessimistic for sharing processors among multiple control loops. Fine grained regulation of periodicity achieves better resource utilization, but is hard to implement online in a robust way. In this paper we propose multi-mode sampling period selection, derived from an offline control theoretic analysis of the system. We report significant gains in computational efficiency without trading off control performance.

Motivation & Objective

• Address the inefficiency of fixed sampling rates in multi-controller embedded systems sharing a single ECU.
• Overcome the practical challenges of fine-grained adaptive sampling by using coarse-grained, discrete sampling modes.
• Ensure robustness and determinism in real-time scheduling by precomputing schedules for each mode, accounting for non-deterministic execution factors.
• Enable dynamic switching between sampling modes based on control state, without degrading control performance.
• Demonstrate scalability and resource savings in both single-controller (ABS) and multi-controller (ACC-ABS) scenarios.

Proposed method

• Define a finite set of discrete sampling rates per control loop based on use-case and stability analysis.
• Use control-theoretic analysis (e.g., Bode plots, pole-zero analysis) to determine the minimum sampling rate required for stability under varying system states.
• Construct a finite-state automaton to model switching between sampling modes based on system state and performance requirements.
• Precompute exact task schedules for each global sampling state, considering timing non-determinism (e.g., variable execution times, message delays).
• Concatenate individual controller modes into a global system state to enable coordinated scheduling across multiple control loops.
• Integrate the automaton scheduler into the ECU to enable online, deterministic switching between sampling modes during runtime.

Experimental results

Research questions

• RQ1 Can a coarse-grained, multi-mode sampling strategy achieve significant computational resource savings without degrading control performance?
• RQ2 How can sampling rate switching be determined a priori to ensure stability and robustness in the presence of non-deterministic execution?
• RQ3 What is the optimal set of discrete sampling rates and switching criteria for a given control system, based on control-theoretic principles?
• RQ4 How does multi-mode sampling scale in systems with multiple controllers sharing a single ECU?
• RQ5 To what extent can ECU bandwidth be reduced while maintaining equivalent or better control performance compared to fixed-period sampling?

Key findings

• The multi-mode controller achieved nearly identical stopping distances (within 0.2%) compared to the fixed-period Matlab model across all road surfaces, confirming no performance degradation.
• In a panic braking scenario, the multi-mode ABS controller reduced ECU bandwidth usage by 30–50% compared to fixed-period sampling.
• In a general cruising scenario, the multi-mode controller operated in infrequent sampling modes for most of the time, significantly reducing average bandwidth demand.
• The method enabled effective ECU sharing between ACC and ABS controllers, with both systems maintaining satisfactory performance using a single ECU.
• Stability analysis confirmed that sampling periods as low as 0.1ms ensured stability, while 0.01s led to instability, validating the need for adaptive selection.
• The precomputed schedules ensured deterministic execution and robustness, avoiding the non-determinism issues of fine-grained adaptive sampling.