Human Occupancy Detection via Passive Cognitive Radio.

Area of Science:

• Electrical Engineering
• Computer Science
• Signal Processing

Background:

• Human occupancy detection (HOD) in enclosed spaces is challenging due to spectrum variations.
• Passive cognitive radio (CR) offers a novel approach for HOD but faces detection difficulties.

Purpose of the Study:

• To develop an advanced HOD system using passive CR for dynamic RF signal collection.
• To identify sensitive frequency bands for HOD using PCA and RFE-LR algorithms.
• To evaluate the performance of multiple machine learning classifiers for HOD.

Main Methods:

• Dynamically reconfiguring CR to collect passive RF signals.
• Applying Principal Component Analysis (PCA) and Recursive Feature Elimination with Logistic Regression (RFE-LR) for feature selection.
• Utilizing Support Vector Machine (SVM), k-Nearest Neighbors (KNN), Decision Tree (DT), and Linear SVM with Stochastic Gradient Descent (SGD) for classification.

Main Results:

• The proposed system accurately detects human subjects in both residential rooms and commercial vehicles.
• RFE-LR with SGD demonstrated superior performance, identifying key frequency bands efficiently.
• The adaptive spectrum sensing method ensures robust detection across diverse environments.

Conclusions:

• Passive CR is a viable technology for robust human occupancy detection.
• The developed system enhances CR efficiency in terms of speed and power consumption.
• The adaptive spectrum sensing approach improves HOD performance in various settings.