The paper tackles a classic problem in statistics applied to a specific type of random physical system: a wave equation that is being jostled by noise. Wave equations describe how things like sound, light, or vibrations propagate through space over time. Here, the randomness comes from a particular kind of noise called fractional Brownian motion, which is more complex than simple white noise because it has memory, meaning past fluctuations influence future ones. The parameter being estimated controls how the system behaves, perhaps something like the speed of wave propagation or the strength of damping.
The challenge is that the noise is "infinite dimensional," meaning it acts across a continuous spatial domain rather than just at isolated points. This makes the mathematics considerably harder than standard settings in statistics or signal processing. The authors develop methods to estimate the unknown parameter by observing the system over time, and they work within a framework where the noise has a specific roughness characterized by something called the Hurst parameter. Depending on whether this roughness index is above or below a certain threshold, the statistical behavior of the estimators can be quite different.
The paper proves that the proposed estimators are consistent, meaning they converge to the true parameter value as more data is collected, and establishes their asymptotic distributions, which tell you how reliable the estimates are in practice. This kind of theoretical groundwork is important because stochastic partial differential equations like the wave equation are used to model real phenomena in physics, engineering, and finance, and understanding how to reliably infer their governing parameters from observations is both mathematically rich and practically useful.