Parameter identification in linear stochastic differential equations (Q1122222)

Consider the following system of one-dimensional stochastic equations: \[ dX_ t=aX_ tdt+dW_ t,\quad X_ 0=0,\quad dY_ t=X_ tdt+dV_ t,\quad Y_ 0=0, \] where \((W_ t)\) and \((V_ t)\) are independent one- dimensional Wiener processes and where \(Y_ t\) is observed at discrete moments \(t=kd\), where \(d>0\), \(k=1,2,3,...\) In the present paper an estimator of a is given (in case \(a<0)\). It is based on \(Y_{1d}\), \(Y_{2d}\), \(Y_{3d}\),... and it is shown that asymptotically the behavior of the system is the same whether one starts with a stationary solution or some initial state.