A REST Framework on IoT Streams using Apache Spark for Smart Cities

The proliferation in the sensory hardware and the generation of voluminous data in Internet of Things (IoT) necessitate the development of effective methodologies to process the data to create actionable knowledge. Existing works tend to train cognitive models on data and derive insights from it. The insights may or may not be in real-time. This proposed work illustrates a real-time streaming analytics framework to predict congestions on multivariate IoT data streams in a smart city scenario. With an increase in the connected devices and Machine to Machine (M2M) communications, there is also a rise in the volumes of data generated. To handle such voluminous data, the frameworks needs to address scalability and reliability aspects. In this research, we present an architecture with the help of open source components that simulate an IoT streaming scenario in a persistent way and a batch analytics engine that processes the data inputs in order to produce higher order, granular insights. For the proposed work, we have used unsupervised learning approaches to identify congestions in traffic scenarios in smart cities.