Cyber-WISE: A Cyber-Physical Deep Wireless Indoor Positioning System and Digital Twin Approach.

Area of Science:

• Cyber-Physical Systems
• Wireless Indoor Localization
• Robotics

Background:

• Traditional wireless indoor localization relies on received signal strength (RSSI) fingerprinting.
• Existing methods require laborious Wi-Fi site surveys for radio map construction, leading to high costs and obsolescence issues.
• Challenges include adaptability, scalability, and suitability for hazardous or remote environments.

Purpose of the Study:

• To present a novel cyber-physical wireless indoor positioning system addressing data collection challenges in fingerprinting techniques.
• To develop an adaptable, scalable system for constructing radio maps in large and evolving environments.
• To reduce labor costs and improve the efficiency of indoor positioning through accurate ground-truth data collection.

Main Methods:

• Implementation of a cyber-physical system with a digital twin interface for informed decision-making.
• Autonomous radio map construction using a ground robot in real-time.
• Application of deep learning models (MLP, LSTM Model 1, LSTM Model 2) for RSSI-based indoor positioning.

Main Results:

• The system successfully generated a radio map autonomously in a challenging environment.
• RSSI-based indoor positioning using deep learning models achieved an average localization error of ≤2.16 m.
• LSTM Model 2 demonstrated superior performance with an average error of 1.55 m and 1.97 m, with over 81% of errors within 2 m.

Conclusions:

• The proposed cyber-physical system effectively leverages dynamic Wi-Fi RSS surveying and deep learning for accurate indoor localization.
• The approach offers practical applicability, suitability for real-world deployment, and improved human safety in hazardous locations.
• This method enhances efficiency and addresses radio map obsolescence, showing performance comparable to existing literature.