Input-to-State Stability: ISS Bounds

• Input-to-State Stability (ISS) bounds are explicit quantitative criteria that ensure robust system stability by bounding state decay and input gains.
• They are derived using Lyapunov functionals, comparison lemmas, and spectral decompositions to analyze both finite- and infinite-dimensional systems.
• ISS bounds offer practical insights for control design and disturbance attenuation in systems ranging from PDEs to switched and networked models.

Input-to-State Stability (ISS) bounds provide a fundamental quantitative framework for describing the robust stability of controlled dynamical and distributed parameter systems subject to exogenous inputs or disturbances. Central to ISS theory is the derivation of explicit bounds—typically of the so-called $\mathcal{KL}$–$\mathcal{K}_\infty$ form—that capture the decay of the state norm from initial data in the presence of bounded inputs, enabling direct analysis of the global stability and robustness of both finite- and infinite-dimensional (PDE or networked) systems.

1. Input-to-State Stability: Definition and Structure of Bounds

ISS formalizes the robustness of a dynamical system by specifying that its state norm is bounded at all times by a function decaying in time from the initial state, plus a function of the input's norm. For a system in a Banach (or Hilbert) space $X$, with solution $x(t)$ under input $u$, the ISS property is expressed as

$$\|x(t)\|_X \leq \beta(\|x_0\|_X, t) + \gamma(\|u\|_U), \quad \forall t \geq 0$$

where $\beta \in \mathcal{KL}$, $\gamma \in \mathcal{K}_\infty$, and the specific function class for the input norm depends on the application ($L^p$-norm, $\sup$-norm, etc.) (Mironchenko et al., 2019, Schwenninger, 2019). The function $\beta$ encodes exponential or polynomial convergence (typically $\beta(r, t)=M e^{-\omega t} r$ for exponentially stable systems), while $\gamma$—often linear—dictates the ultimate gain from persistent input.

Table 1: Typical ISS Comparison Functions

System Type	$\beta(r,t)$	$\gamma(s)$
Linear ODE/PDE	$M e^{-\omega t} r$	$C s$
Damped PDE	$C e^{-\sigma t} r$	$C s$
Burgers eq.	$C e^{-\eta t} r$	$C s$

2. Lyapunov Functionals and Derivation of ISS Bounds

The construction of ISS-type Lyapunov functionals is central to extracting explicit ISS bounds. The canonical methodology uses a Lyapunov function $V:X \to \mathbb{R}_+$ satisfying

$$\alpha_1(\|x\|) \le V(x) \le \alpha_2(\|x\|), \qquad \dot{V}(x,u) \le -\alpha_3(\|x\|) + \sigma(\|u\|)$$

where $\alpha_i \in \mathcal{K}_\infty$, $\sigma \in \mathcal{K}$ (Mironchenko et al., 2019, Xu, 13 May 2025). Employing a comparison lemma or Grönwall's inequality, one integrates the differential inequality to recover the standard ISS bound as above.

In infinite-dimensional systems, the technical construction of $V$ is often problem-dependent. For example, energy-based functionals or system-specific test functions (such as the free-boundary adapted function in (Xu, 13 May 2025)) are essential for nontrivial systems with PDE/ODE coupling or moving boundaries.

3. ISS Bounds for Infinite-Dimensional Systems and PDEs

ISS theory extends rigorously to linear and nonlinear PDEs, including parabolic and hyperbolic types. For boundary control of parabolic equations, e.g., (Schwenninger, 2019), the ISS property is granted under the stability of the underlying semigroup and admissibility of the (possibly unbounded) input operator. Lyapunov or semigroup-based estimates yield explicit ISS bounds, even with boundary disturbances: $$\|x(t)\|_X \leq M e^{-\omega t}\|x_0\|_X + C\|u\|_{L^p(0,t;U)}$$ with detailed $L^p$-dependence determined by operator trace properties.

For nonlinear PDEs or hybrid systems, ISS bounds typically leverage monotonicity properties, maximum principles, or comparison reductions (Mironchenko et al., 2018, Zheng et al., 2019), sometimes transforming the problem (e.g., to distributed input with homogeneous BCs) where standard Lyapunov or energy techniques become applicable.

4. Explicit ISS Bounds: Techniques and Recent Results

Across the literature, explicit ISS bounds are systematically derived using specialized Lyapunov constructions, modal (spectral or Riesz basis) decompositions, or monotonicity-based reductions. An archetype is provided by (Xu, 13 May 2025), which establishes for a Burgers fluid–particle system with a free boundary the explicit estimate

$$\|x(t)\|_X \le 4 e^{-\tfrac\eta2 t} \|x(0)\|_X + \sqrt{\tfrac{3}{2}} \|u\|_{L^2(0,\infty)}$$

where each term in the norm captures a different physical component, and all constants are specified in terms of the Lyapunov dissipation chain.

Similar explicitness is achieved in parabolic PDEs with boundary or mixed disturbances, often yielding

$$\|x(t)\| \le C_1 e^{-\lambda t} \|x_0\| + C_2 \sup_{s \le t} |d(s)|$$

for boundary disturbance $d(t)$ (Karafyllis et al., 2015), or even for coupled multi-equation systems with both domain and boundary inputs (Zheng et al., 2018), typically by ensuring all trace and cross-terms are absorbed through Young or Poincaré inequalities.

5. ISS Bounds for Switched, Hybrid, and Networked Systems

For impulsive, switched, and networked systems, ISS-type bounds must account for switching, jump counts, and interconnection topology (Mancilla-Aguilar et al., 2021, Mironchenko et al., 2019). The key conditions involve verifying exponential stability of the underlying nominal linear system and bounding the perturbation in an affine-in-state fashion: $$|x(t)| \leq K e^{-\lambda(t-s + n_{(s,t]})} |x(s)| + \gamma(\|u\|)$$ with $n_{(s,t]}$ counting discrete events. Sharp small-gain conditions and averaging over nonuniform jump sequences are used to guarantee global ISS without dwell-time restrictions in the strong case.

Composite ISS bounds for large-scale interconnections are constructed recursively, using superposition results and the “gain operator” $\Gamma$ governing subsystem interaction (Mironchenko et al., 2019).

6. Practical Implications and Applications

ISS bounds provide concrete criteria for robust performance, disturbance attenuation, and stabilization under actuator errors or measurement noise, including in boundary feedback-controlled PDEs (Karafyllis et al., 2015), sampled-data and sampled-measurement systems (Vallarella et al., 2018). The explicit gain constants and decay rates obtained in ISS bounds are instrumental for control design, especially in distributed or networked systems where performance guarantees must be traced through system interconnections.

For discretized or numerical schemes, such as finite-volume approximations for hyperbolic balance laws, preservation of ISS properties under discretization is established by transferring the quadratic Lyapunov dissipation chain directly to the discrete setting (Weldegiyorgis et al., 2020).

7. Outlook and Ongoing Research Directions

Current research emphasizes sharpening the tightness of ISS and iISS bounds, notably for systems with unbounded input operators, nonlinear or highly degenerate dynamics, and very large-scale or infinite network couplings (Mironchenko et al., 2019). Open problems include the construction of non-coercive or non-smooth ISS-Lyapunov functionals for PDEs outside analytic or sector class, optimization of constants or gain functions for practical control design, and extension of ISS theory to general classes of infinite-dimensional, switched, or delay systems, including rigorous numerical Lyapunov methods for verifying input-to-state stability bounds in complex models.