Machine Learning for Real-time Data Reduction in Cloud of Things

In the last few years, the number of Internet of Things devices (e.g., smart sensors) connected to the internet has increased significantly and is expected to exceed ten billion devices in the next few years. These devices generate massive amounts of data continuously that needs to be collected, stored, and analyzed for real-time monitoring and smarter decision making support in underlying systems. In this context, cloud computing provides significant potential to store and analyze data. However, several obstacles exist when applying cloud-based solutions to real-time data analysis, including network bandwidth size, energy consumption, cloud storage, and processing costs all due to overwhelming data generation. The paper aims to evaluate machine learning techniques (e.g., principal component analysis, independent component analysis, and singular value decomposition) to reduce unnecessary data in a network edge (e.g., Internet of Things gateways), minimizing bandwidth and energy consumption while avoiding high storage and processing costs through more efficient cloud analysis.