The scope of the report is to discuss the problem of the exact identification of the dynamical system parameters in online mode. Main steps of solution of such problem is considered: parametrization (transformation of the dynamical system model into linear regression equation form) and estimation. Parameter identifiability conditions are discussed. Several different methods to solve this problem are described, analyzed in details and compared. The application of one of such methods to identify parameters of an unmanned underwater vehicle is demonstrated. Main unsolved problems are elucidated.