How can time-sensitive data be processed most efficiently?
It is expected that data processing will increasingly occur in the cloud, but sending large volumes of data to the cloud, for example from Internet of Things (IoT) devices, creates challenges in processing that data in a timely way. One solution, to reduce delays for processing time-sensitive data, is to perform at least the preliminary data analysis, near to where the data is generated. This network edge (the "Fog"), between end devices and traditional cloud computing data centers, is where the physical world meets the internet. The initial analysis in the Fog could include cleaning out redundant or invalid information, and filtering and aggregating data.
Dr Farhad Mehdipour, a Senior Lecturer at Otago Polytechnic's Auckland International Campus, is leading a research team with colleagues at the University of Auckland, Unitec, and Western Sydney University, which has proposed a solution called Fog-engine. IoT devices would be deployed in a Fog at the edge of the cloud. Several Fog engines can be deployed together to create a peer-to-peer network in a smart system. The Fog-engine functions as a gateway to the cloud for a cluster of IoT devices. Fog-engine consists of modular Application Programming Interfaces (APIs). Software-wise, all Fog-engines utilize the same API, which is also available in the cloud to ensure vertical continuity for IoT developers.
One case study considered by the team was a smart home application including a heart rate monitoring and activity monitoring system. Using the Fog-engine the size of data transferred to the cloud could be reduced by 40%.