A LS-SVM based Measurement Points Classification Algorithm for Adjacent Targets in WSNs.

Area of Science:

• Computer Science
• Electrical Engineering
• Signal Processing

Background:

• Wireless Sensor Networks (WSNs) face challenges in multi-target tracking due to measurement origin uncertainty.
• Accurate target localization is crucial for various WSN applications.

Purpose of the Study:

• To propose a novel algorithm for precise multi-target tracking in WSNs.
• To address the issue of measurement origin uncertainty for adjacent targets.

Main Methods:

• Utilizing the Extended Kalman Filter (EKF) to predict classification lines for sampling points.
• Employing Least Squares Support Vector Machine (LS-SVM) to refine classification lines for accurate target center estimation.
• Integrating EKF with LS-SVM for iterative refinement of target location estimation.

Main Results:

• The proposed algorithm effectively classifies measurement points for adjacent targets.
• Simulations demonstrate the feasibility and accuracy of the LS-SVM and EKF integrated approach.
• Experimental results confirm the efficiency and effectiveness of the new tracking algorithm.

Conclusions:

• The novel LS-SVM and EKF-based algorithm significantly improves multi-target tracking precision in WSNs.
• This method provides a robust solution for handling measurement origin uncertainty in dense target environments.
• The validated approach offers enhanced accuracy and efficiency for WSN-based tracking systems.