Deep Q Network for Selective Device Invoking in a Multi-Device Wakeup Environment

A deep learning model that successfully learns to select a device from a Multi-Device Wakeup environment, takes input from IoT devices where sensors record and transfer data to reinforcement learning models. The deep reinforcement neural network is trained with a different variant of deep Q-learning, whose input is recorded by the sensors in various devices and further converted into required numerical values, and whose output is an action function that will be executed by the user-intended device. The proposed methodology is applied in an imaginary world where there are no devices and sensors to record the inputs; instead, there are some random numerical values, which can be used by the model to learn and process the output. It is found that this methodology outperforms all previous approaches and gives an efficient and accurate result.