Real-time target tracking in passive wireless sensor networks using sensor fusion and Kalman filtering

Engineering notes

Key topics: autonomous driving, sensor fusion, control. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

This work presents a detailed study on passive wireless sensor network tracking for adaptive estimation and control. Existing models mainly use a fixed Kalman framework for data fusion under linear and time-invariant assumptions. These available approaches show limited efficiency when operating under nonlinear measurement dynamics and asynchronous sensor dropout. The research introduces an adaptive hybrid fusion method for handling tracking instability and high latency during network change. The process integrates predictive state extrapolation with dynamic gain adjustment using an Extended Kalman Filter structure. The challenge lies in the implementation of synchronization between distributed nodes and an accurate timing of communication cycles. The aim of the analysis is to enhance locating precision and stability subject to element absence. The simulation uses MATLAB and Simulink with predefined network topology and node distribution. The selected parameters include an RMSE, a latency, and a link quality representing dynamic response. The chosen parameter values reflect practical network ranges derived through sensitivity analysis on node density. The data are generated using iterative simulation + n to maintain a consistent update cycle. Data processing uses a filtering and normalization approach for eliminating an outlier interference. Evaluation metrics are based on RMSE accuracy, a latency response, and a throughput utilization. As can be noted from the results, an improvement has been achieved, and this has been reflected in the reduction of the RMSE from 0.64 to 0.41 and the latency from 290 to 180 ms after integrating the Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM)-based feature extraction module with the proposed adaptive sensor fusion and Extended Kalman Filter-based tracking framework. The findings confirm a higher network reliability with a stability ξ₈ increasing from 0.71 to 0.89, supporting a stronger dynamic response under a practical condition.

Links and sources

• PDF from original source

Comments