Study and development of a BLE dual proximity motion detection device

This thesis describes a device for preventing loss or theft of an item, using distance and motion detection. The concept from the user point of view is a device that allows to be moved by the owner only. The connection between the device and the owner is established by an app on the owner’s phone. The device is aware of its distance from the owner (i.e. from the owner’s phone). When in proximity, the motion detection algorithm is disabled. When the user is far away or absent, the motion detection algorithm ensures the security of the item. The minimum configuration of the device comprises: a BLE transmitter configured to transmit a short range signal, a motion processing unit to compute 3-axisaccelerations and angular velocities, a central processing unit, a buzzer to alert the user. The device can be improved adding a second BLE long range signal, allowing for a more sophisticated detection of the different use cases. While the theoretical study will cover this latter case, the actual implementation will focus on a device with one BLE signal only. The motion detection algorithm is in charge of calculating the movements of the item with respect to the surrounding environment. The local environment can be moving as well with respect to the earth frame (e.g. a car, a train etc).In proper terms, the environment where the item is placed can be non-inertial. The algorithm at its core computes the standard deviation on a moving-window sample of accelerations. The standard deviation is then used to characterize the environment and set the thresholds accordingly. As an improvement, a deep learning technique can be implemented to substitute the threshold-logic and make the device reactivity more dynamic.