Counter-strike: accurate and robust identification of low-level radiation sources with crowd-sensing networks

Chaocan Xiang,Panlong Yang,Shucheng Xiao

DOI: https://doi.org/10.1007/s00779-016-0976-y

2016-01-01

Personal and Ubiquitous Computing

Abstract:The use of crowd-sensing networks is a promising and low-cost way for identifying low-level radiation sources, which is greatly important for the security protection of modern cities. However, it is challenging to identify radiation sources based on inaccurate crowd-sensing measurements with unknown sensor efficiency, due to uncontrollable nature of users. However, existing methods assume the sensor efficiency is available, while their identification accuracy tightly depends on identification threshold. To address these problems, we present Counter-Strike, an accurate and robust identification method. Specifically, we use truthful probability of sources for robust identification. And then, we propose an iterative truthful-source identification algorithm, alternately iterating between sensor efficiency estimation and truthful probability estimation, gradually improving the identification accuracy. The extensive simulations and theoretical analysis show that our method can converge into the maximum likelihood of crowd-sensing measurements, achieving much higher identification accuracy than the existing methods. Further, the identification threshold makes slight influence on the identification accuracy in our method, facilitating its practical use.