This paper proposes the architecture of accident fall detection, positioning and rescue system that uses 3G networks. To perceive the fall detection algorithm, the angles acquired by the electronic compass and the waveform sequence of the tri-axial accelerometer on the smart phone are used as the system inputs. The obtained signals are used to produce an ordered feature sequence and then tested in consecutive way by the proposed cascade classifier for identification purpose. As soon as equivalent characteristic is confirmed by the classifier at the current state, it can continue to the state; contrarily the system will come back to the initial state and halt for the emergence of another feature sequence. When a fall accident event is diagnosed the victim’s position can be acquired by global positioning system (GPS) and forwarded to the rescue centre via 3G network so that immediate medical help can be provided.With the proposed cascaded classification architecture, the computational burden and power consumption issue on the smart phone system can be mollified. Moreover, known fall incident detection accuracy and reliability up to 92% on the sensitivity and 99. 75% on the specificity.