Energy consumption minimisation at edge node using [Formula: see text] approach in predicting sensor parameters in...

Area of Science:

• Computer Science
• Electrical Engineering
• Environmental Science

Background:

• Wireless sensor nodes face energy sustainability challenges due to limited storage and battery power.
• Minimizing energy consumption involves predicting data from inactive sensors, but parameter selection impacts accuracy and complexity.
• Existing methods for selecting active sensor parameters are often inefficient and do not guarantee optimal solutions.

Purpose of the Study:

• To propose a novel cross-correlation-based parameter selection approach for wireless sensor networks.
• To ensure the selected parameter set is stable and Pareto-optimal for efficient sensor data prediction.
• To reduce computational complexity and enhance energy sustainability in sensor nodes.

Main Methods:

• Developed a cross-correlation-based parameter selection algorithm, denoted as [Formula: see text].
• Evaluated the [Formula: see text] approach using nine diverse, publicly available environmental datasets.
• Compared the performance against existing parameter selection methods in terms of speed and accuracy.

Main Results:

• The [Formula: see text] approach demonstrated faster selection of active sensor parameter subsets compared to existing methods.
• Simulations showed significant reductions in edge node energy consumption, ranging from [Formula: see text] to [Formula: see text], for predicting sleep sensor parameters.
• The selected parameter sets were found to be stable and Pareto-optimal across various datasets and sampling intervals.

Conclusions:

• The proposed [Formula: see text] method effectively addresses energy sustainability challenges in wireless sensor nodes.
• This approach offers a more efficient and accurate way to select sensor parameters for data prediction.
• The findings suggest a practical solution for optimizing resource-constrained edge computing environments.