Collaborative Sensing for Heterogeneous Sensor Networks

Collaboration between low-quality sensor and high-quality sensor can achieve tradeoff between accuracy and energy efficiency in heterogeneous sensor networks (HSNs). Generally, HSNs are deeply integrated with dynamic physical environments. The monitored target's dynamics are the most important and common factors of the dynamic environments. Some important parameters (such as active opportunity, sampling frequency and sampling time) fail to adapt to the changes, which undermines the collaboration's performance when the state of the monitored target changes. Even the system performance is not up to user requirements or large amounts of energy are consumed. To solve this problem, we propose an adaptive collaboration scheme named EasiAC by the collaboration between magnetic and camera sensors. First, for the dynamics of the monitored target, EasiAC utilizes the magnetic sensors to predict the target's state via Bayesian filtering based method. Second, to achieve good performance of such collaboration, EasiAC adjusts the camera sensors' active opportunity, optimal sampling frequency and sampling time dynamically according to the estimated results from the magnetic sensors. Finally, a boosting based algorithm named BbTC is proposed to make classification for the target to achieve high accuracy (average classification accuracy is more than 98%). We evaluate EasiAC method through extensive simulations and real road environment experiments. The results demonstrate that EasiAC needs less energy consumption (saving 97% energy) than traditional solutions, while maintaining the performance at acceptable level (average image integration ratio is 90%) in the presence of target's dynamics.

關鍵字

Bayesian filtering ； dynamic environments ； collaboration ； sampling frequency ； target classification ； wireless sensor networks

國際替代計量

全文下載

主題瀏覽

Collaborative Sensing for Heterogeneous Sensor Networks

摘要

關鍵字

延伸閱讀

國際替代計量

本網站使用Cookies