Rotor speed is a key parameter of vector control techniques in induction motor drive systems. Speed measurement using sensors such as encoders can be affected by electrical noise, mechanical failure, or disconnection. This study proposes a method for diagnosing speed sensor faults based on the Mean Squared Error (MSE) between the measured speed and the estimated speed obtained from a voltage model based Model Reference Adaptive System (RF-MRAS) observer. When the MSE exceeds a predefined threshold, the system triggers a fault flag to a high level to indicate that a speed sensor fault may have occurred. Simulation results in MATLAB/Simulink confirm the method's effectiveness in accurately detecting faults and ensuring timely response when a sensor failure happens.