An In-Networking Double-Layered Data Reduction for Internet of Things (IoT)

Area of Science:

• Computer Science
• Data Science
• Network Engineering

Background:

• The Internet of Things (IoT) generates vast amounts of data, leading to network bandwidth, energy consumption, and cloud storage issues.
• Current data reduction methods often use single or two-tier approaches, which may not be optimal for diverse IoT data streams.

Purpose of the Study:

• To propose a novel, two-layer in-networking data filtering and fusion approach for effective IoT data reduction.
• To address the challenges of data redundancy and volume in IoT environments at the network edge.

Main Methods:

• Implemented a two-layer approach: a data filtering layer (using change detection and deviation from estimated values) and a data fusion layer (using minimum square error criterion).
• The approach can be adapted for single or two-tier architectures, processing data at the network edge before transmission.

Main Results:

• The proposed approach demonstrated significant efficiency in data reduction compared to existing methods like Least Mean Squares filter and Papageorgiou's (CLONE) method.
• Evaluation using a real-world dataset confirmed the approach's effectiveness in mitigating IoT data challenges.

Conclusions:

• The in-networking data filtering and fusion approach offers an efficient solution for IoT data reduction.
• This method helps alleviate network bandwidth, energy, and cloud storage constraints inherent in large-scale IoT deployments.