Multi-mode Sampling Period Selection for Embedded Real Time Control

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.

💡 Research Summary

The paper addresses a fundamental inefficiency in embedded real‑time control systems: the widespread practice of fixing a single sampling period for each control loop. While a uniform period guarantees deterministic behavior and simplifies static scheduling, it is overly conservative when several control loops share a common electronic control unit (ECU). Adaptive sampling, which varies the period according to the plant’s state, can dramatically reduce computational load, but fine‑grained online adaptation is difficult to implement robustly because of nondeterministic timing effects such as message latency, execution‑time jitter, and path‑dependent software delays.

To bridge the gap between the robustness of a fixed period and the efficiency of fully adaptive sampling, the authors propose a “multi‑mode sampling” strategy. The central idea is to partition the operating space of the controlled system into a small number of representative modes, each associated with a pre‑computed, constant sampling period that guarantees closed‑loop stability for all states belonging to that mode. The mode set and the switching logic are derived offline using classical discrete‑time control theory.

The theoretical foundation rests on the relationship between the sampling interval T and the pole‑zero configuration of the discrete‑time transfer function C(z). Changing T modifies the location of poles in the z‑plane; stability is assured when all poles lie inside the unit circle. The authors employ unit‑circle analysis in the z‑domain, complemented by alternative criteria (Lyapunov, Nyquist, Routh‑Hurwitz, Bode) where appropriate, to determine the maximal admissible T for each operating region.

A concrete case study is the anti‑lock braking system (ABS) of a passenger vehicle. The vehicle dynamics are modeled by a quarter‑car representation, with wheel slip λ, longitudinal speed Vx, braking torque Mb, and a piecewise‑linear friction‑versus‑slip curve µ(λ). Linearization around operating points yields an affine state‑space model, which is then controlled by a discrete PID law. The PID gains (Kp, Ki, Kd) are fixed; only the sampling period T varies.

Three sampling modes are defined:

• N0 mode – low to medium speeds (0–85 km/h) and low/mild/medium brake‑pedal pressure. Stability analysis shows that a 2 ms period keeps all poles inside the unit circle for slip ratios up to 0.65.
• N1 mode – higher speeds (85–140 km/h) with medium or high pedal pressure. A 1.5 ms period is the largest that still guarantees stability.
• E mode (Emergency) – a universal 1 ms period that is safe for any speed, pressure, or slip condition; it is used when the system detects a sudden change or an unsafe state.

For each mode the authors plot three‑dimensional surfaces (speed, slip, pole magnitude) to visualize the stability region. Guard conditions for mode transitions are expressed as logical predicates on vehicle speed (v) and brake‑pedal pressure (bpp), e.g., τN0 = (v ∈