On parameter estimation for cusp-type signals (Q1695754)

The authors consider the problem of parameter estimation by continuous time observations \((X_t,\; 0\leq t \leq T)\) of a signal \(S(\theta,t)\) in a white Gaussian noise (WGN) \(dX_t=S(\theta,t)dt+\epsilon dW_t\), \(X_0=0\). Here, \(S(\theta,t)\) is a known signal with unknown parameter \(\theta\) which must be estimated. It is supposed that the signal \(S(\theta,t)\) has a cusp-type singularity, i. e. at the vicinity of the point \(t=\theta,\) it has the representation \(S(\theta,t)\approx a|t-\theta|^k,\) where \(k\in(0,\frac{1}{2}).\) The properties of the maximum-likelihood and Bayesian estimators are described in the asymptotics of small noise. Special attention is paid to the problem of parameter estimation in the situation of misspecifiation in regularity, i.e., when the observed signal is supposed to have a cusp-type singularity, but it is really smooth. The rate and the asymptotic distribution of the maximum likelihood estimator in this situation are described.