Apparently, we have been observing significant demand for mobile devices during recent years in various fields. In this regard, one of the attentive spheres is the analyzing of diverse human activities and actions through mobile phone. Therefore, if a mobile phone is able to recognize and identify user’s current situation spontaneously, then it can transmit appropriate data to the opposite part. Moreover, this paper investigates a new Multi-Sensor module construction scheme and its related peripherals for data acquisition which to be used as a human activity recognition system. To do so, data was collected by our new multi-sensor module which is located in three different static positions, such as in pants pocket, in hand and in a jacket pocket as well, We have been conducted six daily activities including walking, running, standing, ascending stairs, descending stairs and riding on bus to carry on our experiment. Afterwards, the features have been extracted from raw sensor data using mean and variance. We have used Multi-Layer Neural Network to classify different activities. Ultimately, based on the consequence of experiment, we managed to achieve 97.8% of accuracy in four activities such as walking, running, standing and riding on bus.