Why does stochastic gravitational-wave background violate the weak-signal noise-dominance assumption?

# Why the Weak-Signal Assumption Breaks Down for Stochastic Gravitational Waves

The stochastic gravitational-wave background (SGWB)—a hum of overlapping signals from billions of distant sources—violates the weak-signal assumption because the combined signal can become *louder than the detector's instrumental noise*, invalidating the approximation that noise dominates [AG-2024.03-474]. This matters because standard detection pipelines are built on that assumption and can dramatically underestimate how strong the signal actually is.

The Core Problem

In the weak-signal limit, you assume that gravitational waves are so faint compared to detector noise that you can linearize your analysis—treat the signal as a small perturbation riding on top of noise. This simplifies the math enormously. But when many weak sources pile up coherently in a given frequency band, their combined amplitude can rival or exceed instrumental noise. Once that happens, the signal and noise are coupled in nonlinear ways, and the old formulas fail [AG-2024.03-474].

The practical consequence is striking: for space-borne detector networks like TianQin + LISA observing extragalactic double white dwarfs, the signal-to-noise ratio (SNR) estimated under the weak-signal assumption can be off by as much as an order of magnitude [AG-2024.03-474]. In other words, you might think a source is barely detectable when it's actually quite loud, or vice versa.

Why This Happens in Practice

Two things conspire:

1. The signal is unexpectedly large. Some candidate SGWB sources (like first-order phase transitions from the early universe, or certain binary populations) produce more gravitational wave power than standard models predict [AG-2024.01-368], pushing them into the strong-signal regime.

2. Shot noise amplifies nearby loud events. A few individually resolvable, bright sources can dominate the background in their frequency band, meaning you can't treat the overall background as a smooth, weak, isotropic hum—you have to account for discrete loud events whose stochasticity matters [AG-2025.09-387].

The workaround is to notch out the brightest individual events from your data before analyzing the remaining background [AG-2025.09-387, AG-2024.06-288], but that requires knowing which events to remove and accepting some sensitivity loss.

---

Note: The provided preprints address the *failure* of the weak-signal assumption but don't deeply explain the mathematical mechanism of that failure; a more technical source on nonlinear signal processing in GW analysis would round out the picture.