Remote sensing of elderlies' call for help is of great interest to caregivers, the elderly's family, and, of course, to the elderly who most often doesn't want to be bothered with a wearable device or cannot physically reach the panic button. According to the center for disease control and prevention (CDC), 2.5 million elderlies are treated in emergency rooms due to fall injuries. Personal emergency response systems (PERS), also known as panic buttons or medical alerts, are very common in the industry. Some of them require the injured person to operate, some are passive but have to be on the injured person body in the form of a necklace or a wrist band. We propose a single transmitter (TX), single receiver (RX), fully remote ultrasonic radar system for classifying a call for help signal, which only requires the person to wave one of her/his hands for 3 seconds. A simple, yet robust algorithm is employed to classify the movement extracted out of the phase of the received signal emitted from a 40 kHz continuous-wave (CW) ultrasonic transmitter. The proposed algorithm is highly immune to thermal noise and employs a novel real time, optimal Doppler estimation without the use of spectral analysis algorithms such as the Fourier transform.