Efficient DV-HOP Localization for Wireless Cyber-Physical Social Sensing System: A Correntropy-Based Neural Network...

Area of Science:

• Computer Science
• Wireless Sensor Networks
• Machine Learning

Background:

• Wireless Sensor Networks (WSNs) are increasingly integrated into cyber-physical social sensing (CPSS).
• Accurate localization of sensor nodes is crucial for WSN performance but challenging due to environmental noise and limitations of existing algorithms like RSSI and DV-HOP.
• Existing methods like RSSI are sensitive to noise, while DV-HOP struggles with insufficient anchor nodes.

Purpose of the Study:

• To develop a robust and accurate node localization scheme for WSNs.
• To enhance distance estimation by combining RSSI and DV-HOP.
• To improve localization accuracy in noisy environments using a novel neural network approach.

Main Methods:

• A new neural network (NN)-based localization scheme, RHOP-ELM-RCC, is proposed.
• It utilizes both Received Signal Strength Indication (RSSI) and Distance Vector Hop (DV-HOP) for distance evaluation.
• An Extreme Learning Machine (ELM) with a regularized correntropy criterion (RCC) is employed for robust localization, replacing Mean Square Error (MSE) estimation with RCC and Least Square Estimation (LSE) with half-quadratic optimization.

Main Results:

• The proposed RHOP-ELM-RCC scheme effectively combines RSSI and DV-HOP for improved distance estimation accuracy.
• The use of RCC in ELM enhances robustness against environmental noise and outliers compared to MSE.
• Simulation results demonstrate that the proposed scheme significantly outperforms traditional localization schemes.

Conclusions:

• The RHOP-ELM-RCC scheme offers a cost-effective and accurate solution for WSN node localization.
• The integration of RSSI, DV-HOP, and RCC-based ELM provides superior performance in challenging environments.
• This approach represents a significant advancement in WSN localization techniques for CPSS applications.