The identification of elderly activities through intelligent sensors in a smart home can effectively monitor the abnormal movements of residents in everyday life without adding supplementary loads caused by wearable sensors. To meet the question of societal stake where the need for a discrete and ambulatory follow-up is required by the gerontologists, we propose the development of a new inside surveillance system based on multi-sensor detection. Here, we propose a new mechanism for locating and detecting the elderly activities in a smart home. Moreover, the goal of our mechanism is to improve the supervision of areas and locate people effectively within wireless sensor networks. The contribution of this work is threefold: first, two different technologies of detection combined using the fuzzy logic method are used to minimize the error during the detection process. Second, the number of messages diffused to the base station is reduced through dynamic clustering method. Third, an optimal method for the selection of moving sensors is proposed for the localisation phase. We discuss in depth the proposed monitoring algorithm performance in terms of energy consumption, execution time, and location error. Furthermore, experiment results and relevant performance comparisons with related works are presented.