Distributed Processing and Analytics of IoT data in Edge Cloud

Abstract

Sensors of different kinds connect to the IoT network and generate a large number of data streams. We explore the possibility of performing stream processing at the network edge and an architecture to do so. This thesis work is based on a prototype solution developed by Nokia. The system operates close to the data sources and retrieves the data based on requests made by applications through the system. Processing the data close to the place where it is generated can save bandwidth and assist in decision making. This work proposes a processing component operating at the far edge. The applicability of the prototype solution given the proposed processing component was illustrated in three use cases. Those use cases involve analysis performed on values of Key Performance Indicators, data streams generated by air quality sensors called Sensordrones, and recognizing car license plates by an application of deep learning.