Several algorithms are developed for speed estimation and sensorless control of induction machines which among them, reduced-order based methods have proved acceptable performance. In this paper, a reduced-order speed estimation method, is introduced for sensorless control of DFIGs, in which, rotor currents and stator fluxes are used as state variables and dynamic of rotor currents error is applied to extract adaptation mechanism. Since current dynamic is faster than flux dynamic, proposed observer dynamic is improved. Dynamic stability of proposed method is demonstrated using Lyapunov stability criteria. The suggested speed estimation algorithm is confirmed utilizing a laboratory setup, in addition to simulations. Laboratory setup is based on a 16-bit DSPIC and practical results are extracted using a 250 watts induction machine. Both simulation and practical results show effective and stable operation of proposed method, especially at speed tracking.