Applications in the Internet of Things (IoT) require a lot of computing power. IoT sensors in a smart factory, for example, produce large amounts of data that need to be processed, analyzed and aggregated in a very timely manner. Cloud computing provides virtually unlimited computing capacity. But the data transfer between IoT devices and the cloud can experience delays that are critical for IOT applications operating in real-time. Edge computing addresses this problem by directly processing IoT sensor data at the source.

Due to the limited capacity of edge nodes, operators must decide wisely which data to process in edge nodes and which data to offload to the cloud. The choice depends on many factors: maximum tolerated latency, the current CPU utilization of the edge nodes, data protection requirements, energy consumption, and the cost of using the resources and transferring application components between the edge and the cloud.

A research team led by Prof. Dr. Klaus Pohl from the software technology institute paluno has worked with NOKIA to develop the FOGPART optimization algorithm. It helps to take these factors into account appropriately. FOGPART can be used online, i.e. during operation. This is important because the context of IoT applications is constantly changing: New IoT devices (edge nodes) join a network and others leave. This dynamic means that data protection requirements must be constantly reassured and secured. In addition, cloud resource workloads and prices fluctuate over the course of usage.

“The algorithm we developed in cooperation with NOKIA ensures that essential data protection requirements are satisfied across the cloud edge continuum,” says Prof. Pohl. “At the same time, it helps to make optimal use of the existing IT resources of edge nodes.” The researchers are currently working on an extension to FOGPART that, among other thing, considers the energy consumption when allocating data between edge and cloud.

Current Publication:

Zoltan Mann, Andreas Metzger, Johannes Prade, Robert Seidl and Klaus Pohl: Cost-optimized, data-protection-aware offloading between an edge data center and the cloud.IEEE Transactions on Services Computing, 2022, 1-1.   [DOI]

Further Information

The work was carried out as part of the EU research project FogProtect (H2020 no. 871525). FogProtect addresses end-to-end data protection across the computing continuum, i.e., from the cloud to geographically distributed fog or edge nodes to end devices. More information: https://sse.uni-due.de/en/research/projects/fogprotect