SensE

SENSors on the Edge”

SensE is a joint research project between the Chair of Computer Arcitecture and Parallel Systems of Technische Universität München and IfTA Ingenieurbüro für Thermoakustik GmbH sponsered by the Bayerische Forschungsstiftung.

Context

The continuous availability of system-critical infrastructure is essential for today’s society. Energy and drinking water supply systems, mobility and communications infrastructure, and many others are therefore permanently monitored with sensors. Enormous volumes of data that are generated; analyzing it requires intelligent use of the available computing resources both on the site - close to the sensors (at the “Edge”) - and in High Performance Computing systems or in the “Cloud”.

Goal

The aim of SensE is to efficiently combine all available resources for sensor data processing from the edge to the Cloud in a collaborative fashion. Its particular emphasis is on scalability and adaptivity of the sensor processing system through the intelligent coupling of system components and data.

Approach

In SensE, we aim to implement this vision: small and energy-saving systems, on site, process and analyze the locally available sensor data and use it in local process control loops. At the same time, the locally generated analysis results and a reduced (intelligently-selected) subset of the data are forwarded to the Cloud or to a High Performance Computing system for more complex analyses and combine data and models from multiple systems.

Sense aims to investigate the coupling of the resources, models, and data, and through this integration, an optimized sensor processing system can be developed. This system can be used on the computer systems at the edge to improve the monitoring systems and optimize the local control loops.

Overview of SensE

Data is generated by sensors and preprocessed on site (left). In a central computing system (right), more computing capacity is available to perform more complex analyses. The results can then be transferred back and used for more precise analyses at the edge.