Billions of connected devices send data every second. Sending everything to a distant cloud causes delays. This research speeds things up by processing data in “fog” nodes closer to the devices.
The problem
The Internet of Things depends on fast, reliable data delivery. Fog computing places small processing nodes between devices and the cloud to cut delays, but moving data efficiently across these spread-out networks remains hard.
What the research did
We proposed BDPS, a big data processing scheme built on Apache Spark’s resilient distributed datasets, designed for fog–IoT networks. It uses a depth-first-search approach to find efficient routes for data between nodes.
What we found
BDPS delivered data with lower latency and lower network overhead than the Bellman–Ford, Floyd–Warshall and Dijkstra algorithms and Apache Hadoop, supporting efficient data handling in decentralised networks.
Who benefits
- Smart cities running large sensor networks
- Network and cloud providers
- Industries using IoT for monitoring and automation
Limits and what comes next
Results came from simulation and testbed experiments; large real-world networks bring extra complexity such as device failures and changing traffic.
The team
This research was led by Rakib Hossen (Jagannath University) with Md Whaiduzzaman (QUT), Mohammed Nasir Uddin, Md. Jahidul Islam, me, Alistair Barros (QUT), Mehdi Sookhak (Texas A&M University) and Md. Julkar Nayeen Mahi.
Frequently asked questions
What is fog computing?
A layer of computing between devices and the cloud that processes data closer to where it is created.
What does BDPS do?
It moves and processes IoT data across fog nodes more efficiently, reducing delays.
Why use Apache Spark?
Spark is built for fast, distributed big-data processing, which suits large IoT networks.