Reliability Cost Function

Quantifying Reliability Degradation in B-Type Control


Purpose of the Reliability Cost

In B-Type architecture, reliability is treated as a measurable quantity,
not an abstract design intention.

The reliability cost function provides:

The goal is not optimization, but bounded degradation.


Conceptual Definition

Let the system reliability be evaluated using a cost function:

\[J_{\text{rel}} \ge 0\]

where:

Adaptation is permitted only when the reliability cost remains below a predefined limit.


Normalized Metric Set

Primary contributors:

Response delay ratio: \(R_{\Delta t} = \frac{\Delta t}{\Delta t_0}\)

Gain compensation ratio: \(R_K = \frac{K}{K_0}\)

Amplitude ratio: \(R_A = \frac{A_{\text{out}}}{A_{\text{ref}}}\)

All metrics are dimensionless and centered around unity under nominal conditions.


Basic Reliability Cost Formulation

A weighted quadratic form:

\[J_{\text{rel}} = w_{\Delta t}(R_{\Delta t} - 1)^2 + w_K(R_K - 1)^2 + w_A(R_A - 1)^2\]

where:


Threshold-Based Decision Rule

\[J_{\text{rel}} \le J_{\text{rel}}^{\max}\]

If violated, the FSM transitions toward ADAPT_BLOCKED or SAFE_MODE.


Time-Accumulated Reliability Cost (Optional)

\[J_{\text{rel}}^{\text{acc}}(T) = \int_0^T J_{\text{rel}}(t)\,dt\]

This captures chronic reliability degradation over long horizons.


Summary

The reliability cost function:

In B-Type, adaptation is allowed
only when reliability cost remains bounded.