Manoeuvre Target Tracking in Wireless Sensor Networks Using Convolutional Bi-Directional Long Short-Term Memory...

Area of Science:

• Engineering
• Computer Science
• Signal Processing

Background:

• Traditional wireless sensor networks (WSNs) face significant positioning and tracking errors in motorised targets due to inherent noise interference.
• Accurate tracking of manoeuvre targets is crucial for various applications, including surveillance and autonomous systems.

Purpose of the Study:

• To develop an improved motorised target tracking method for WSNs that overcomes limitations of traditional approaches.
• To enhance the accuracy and robustness of target positioning and real-time tracking in dynamic environments.

Main Methods:

• A hybrid approach integrating a convolutional bi-directional long and short-term memory neural network (BiLSTM) with extended Kalman filtering (EKF).
• The BiLSTM is trained using simulated Received Signal Strength Indicator (RSSI) values and actual target data to provide accurate initial manoeuvre target states.
• EKF is employed for real-time target localization and trajectory tracking based on the refined initial states.

Main Results:

• Experimental simulations demonstrate the proposed algorithm's effectiveness in tracking both linear and non-linear manoeuvre targets.
• The method significantly reduces positioning and tracking errors compared to traditional WSN tracking techniques.
• Accurate real-time positioning and tracking information of the motorised targets is achieved.

Conclusions:

• The proposed convolutional BiLSTM and EKF integrated method offers a robust solution for motorised target tracking in WSNs.
• This advanced tracking technique shows superior performance in diverse manoeuvre scenarios, addressing critical limitations of existing systems.