Topic domains:

Internet of Things, distributed analytics

Abstract:

Today's networking infrastructure is composed of numerous and heterogeneous devices that consume and produce large volume of data in real-time.

Extracting valuable information from this voluminous data is a challenging task that is currently undertaken by both academia and industry (Big-data analytics). As we experience the birth of the Internet of Things (IoT) where objects of various sizes and properties will be connected to the Internet, it will stress even more the needs of efficient analytics.

Nowadays, analytical computing is mostly done on a cloud-based backend or on high computational powered devices. However, we foresee that all these "smart" objects may be able to produce some computational operations closer to the edge (i.e. source of the data). Furthermore, these objects may have serious limitations (e.g. battery limited, computing power, latency, etc..) and may not be able to send a constant flow of information to the cloud for analysis.

The research question of this thesis topic is how to perform distributed analytics on a network of highly heteregeneous devices and to evaluate the potential benefits of the solution. For example, distributed analytics may increase security and privacy, reduce latencies or maximize the usage of available resources which are currently unused.

The thesis work will require to build a distributed network of IoT devices (e.g. phones, embedded devices, cloud, desktop) and demonstrate the feasibility of performing distributed analytics on such network infrastructure (most probably analytics of air quality data).