there have been numerous studies involving research and development, for detecting falls exhibited by the elderly. Considering that the prevention of a falling elderly is much more complex to address and estimate, very little research has been done. In fact research is often strictly limited resourceful medical organizations that have specialized clinical tools. Human locomotion, particularly "Walking" is defined by sequences of cyclic and repeated gestures. The variability of such sequences can reveal information about drive failure and motor / motor-neuron disorders. Studying and exploiting the Cyclostationary (CS) properties of such sequences, offers a complementary way to quantify human locomotion and its changes with progressing aging and the development of diseases. This quantization may provide an insight into the neural function and the neural control of walking which would be altered by changes associated with aging and the presence of certain diseases. As part of the collaboration between LASPI and CHU Saint Etienne, we decided to focus on certain advanced signal processing theory and methods, to study very complex phenomena of human walking, which is often subject to numerous motor and / or motor-neurons malfunctions, such as in the case of the falling elderly population, that often has serious and severe consequences. Furthermore, this paper also examined the effects on walking in elderly subjects in three task conditions: (a) single task (MS) and (b) dual task: walking by performing a fluency task(MF) and (c) walking while backward counting (MD). Results show that the conditions of walking impacted the Cyclostationarity and its known indicator: the cyclic autocorrelation function. Such indicator also evolved between fallers and non-fallers and between the fallers who have history of falls and those who haven't.

No abstract