Distributed Computing (offloading) with RIOT

Development archive
1

I have been thinking and reading about the research challenges in IoT-specific operating systems.

One idea that I have encountered is to perform the computation outside the device on a more powerful machine. Let’s say a face recognition algorithm cannot be performed on a wearable device and the wearable device needs to send (offload) this computation to a nearby machine.

This can be done at the application level but also at the OS level. Let’s say a remote fork API hides the complexity of creating remote process from the application programmer.

I was wondering if anyone has ever worked on this subject and if the community thinks it is an interesting problem to work on at all… If not, would you have any suggestions for the research challenges in this domain?

Thanks.

2

Hi Arash,

just from the challenges perspective I know that in the working group of Prof. Dr.-Ing. Karl-Ragmar Riemschneider at the Hamburg University of Applied Sciences they do complex signal processing on one powerful processor, based on distributed sensor data from vehicle battery cells.

Best Peter

3

Hi everyone :^)

4

Hello,

I don’t like very much the use-case of face recognition, however I guess opening up an API to OpenCV is better done in application space.

It is not particularly about OpenCV APIs. The face recognition was only an example. It is more about any computation which is heavy to be performed locally.