Summary: Presently, the explosive growth of the portable Internet of Things (IoT) products presents a major challenge to the safety of wearable IoT devices by collecting significant amounts of sensitive information. In the area of protection and access control, the gyro sensor-based shift recognition is seen as a new technology that is emerging and achieved excellent performance at certain speeds. Therefore, a survey suggested that both illness and behaviour can impact gait habits, the gait criteria for clinical applications cannot be used by clinicians without awareness of the activities. Hence in this paper, the acceleration- gait cycle adaptive optimization technique (AGAO) hybridized with a non-linear model of gait recognition based on matrix–vector parameter estimation (MVP) has been proposed to modify the approach for producing a corresponding threshold that can effectively reduce cognitive-related problems. The detection of gait phases is taken into account in more accurate and difficult situations where the subject goes through mental tasks. Compared with adaptive gait cycle extraction (AGC) and linear discriminant analysis (LDA), AGAO -MVP reduces the duration of the algorithm by incorporating data from surface EMG sensors into the IoT system. Experimental results indicate that AGAO -MVP is more accurate in terms of performance accuracy and shows that cognitive task limits between the pose phase and the swing cycle become more dynamic.