The current trend towards a diverse, largely mobile user computing environment brings new challenges to computing systems architecture. Currently, cloud computing technologies offload devices by moving software and computation into the infrastructure shifting away from the client-server computing paradigm. One advantage in doing so is that user devices can be increasingly resource constrained without loss of functionality. For example, battery life time can be increased by offloading tasks that are computationally complex and energy hungry.
However, centralised datacenter processing can be unsuitable for many tasks, for example where specialised hardware is needed or where latency is key. In such computing scenarios, the cloud resources need to be in close proximity with the user device. In addition, as more and more consumer devices become networked through wireless interfaces a possibility arise to locally construct computing platforms. For example, a video camera can be added as a camera interface to a smartphone so that the camera app can select it instead of the inbuilt camera, thus adding a more powerful and high quality hardware platform when needed and available.
In the boundless concept, we research fundamental aspects of building distributed computing platforms on a per-need basis. The computing platform resources are configured to match the needs of the application based on the availability of hardware and software resources in a distributed local cloud. Resources include both local computing resources such as consumer electronics devices as well as capillary network resources, e.g. GPU or CPU resources located in the access network.
For more information, contact Jörn Janneck
Shubhabrata Sen introducing the project: