A BLE based Adhoc network (Pseudo AODV protocol) of Super sensors for ubiquotous multi-purpose sensing
‘Smart appliances’ and Ubiquitous networks are on a rise in gaining popularity of the notion of connected smart devices. Such ‘smart appliances’ are able to detect events related to their usage, performance, and outputs; which is possible due to the inbuilt sensors that are shipped with almost every of such devices. Ubiquitous sensor networks, on the other hand, have helped such ‘smart appliances’ talk to each other, thereby performing actions together as a combined entity. This project presents an evolutionary step ahead in smart ubiquitous networks by presenting a novel approach towards sensing based on ‘super sensors’. Instead of embedding sensors into every single appliance, we could build a ‘smart environment’ that can detect any event happening related to the appliances in it. The approach presented involves the use of a single sensing unit that captures multiple sensor streams like Light, Heat, Temperature, Vibrations, Electromagnetic noise, etc and then uses machine learning to detect all possible events that could happen. This ‘super sensor’ will result in a ‘smart environment’, the range of which can be expanded on demand by involving Ad-hoc networks. The project developed uses TensorFlow to build machine learning model over data collected from Android/Android-things. The super sensors developed communicate using a pseudo-AODV protocol based on BLE. Such Ubiquitous smart sensor networks can potentially disrupt the notion of Internet of Things and the connected world by minimizing costs, enabling machine to machine communication, mobile sensing, low energy usage and a dynamic on-demand network topology.