You might have come across the phrase big data recently. Large quantities of data can provide valuable information in a range of situations, providing they can be effectively processed through analytics. This is no easy task. Not only are the datasets huge, they also contain data types that are generated and updated very quickly and may not always be accurate. Extracting value from such data requires statistical and analytical techniques, such as machine learning algorithms and complex network theory, which are often challenging to use.
Big data is a fast-evolving field. As new data sources emerge and new applications become plausible, novel approaches and technologies need to be developed to process results. Some of the most useful methods appear in the paper “Big sensor data applications in urban environments”, published in the Elsevier journal Big Data Research by Li-Minn Ang and Kah Phooi Seng of the School of Computing and Mathematics, Charles Sturt University, Australia.
Currently, big data research focuses primarily on ‘human-generated’ data. Companies like Google, Facebook and Twitter gather real-time data, such as emails, tweets, documents and photos, from their users to inform services like customer analytics, social network analytics and fraud detection.
However, big data is increasingly ‘machine-generated’ from networked sensor systems like the Internet of things and large-scale wireless sensor networks. Indeed, in the near future, the volume of data from machine-generated sources is set to outstrip that from human-generated ones.
In their paper, Ang and Seng survey the current state of affairs in big data derived from networked sensor systems. They illustrate their findings with results from studies that use sensor systems in urban environments to collect and analyse data. These can then offer solutions for challenges, such as air pollution, disaster management and intelligent transportation.
Ang and Seng go on to coin the term ‘big sensor data systems’ to highlight the emerging importance of big data derived from networked sensors. “In our view,” Ang explains, “the term ‘Internet of things’, with its usual definition of a large-scale network of connected objects for collecting and exchanging data, doesn’t adequately convey the importance of machine learning and analytics in deriving benefit from the data. Thus, we propose using the term ‘big sensor data’ to focus on the increasing importance of analytics in sensor-based systems.”
It is important to provide such a framework. Ang and Seng write that “researchers have yet to establish a systematic framework to address the challenges of designing and building big sensor data systems.” Indeed, Ang notes, one aim of the article is to motivate researchers to develop practical solutions in this field. These include sensor-social networks that work with both machine-generated and human-generated data, and (near) real-time systems such as earthquake early-warning systems, which enable us to respond immediately to threats.
Read Article free online until 25th May 2018
Ang, L-M., Seng, K. P.: “Big sensor data aplications in urban environments,” Big Data Research (2016)