On the asymptotic efficiency of the least squares estimates of regression coefficients

A process $y(t)=x(t)+aA(t)$ is considered, where $x(t)$ is a weakly stationary process with continuous parameter $t$ and $A(t)$ is a nonrandom function. Sufficient conditions for the least-squares estimate of $a$ to be asymptotically efficient ($1^\circ$–$5^\circ$) are obtained.