Article Preview
Top1. Introduction
In recent years, multiple sensors data fusion as a rising cross-subject has been well studied in many literatures. Through fusing various data from several sensors, a unified result can be produced (Khaleghi & Khamis, 2013). Nowadays, data fusion is widely used in many fields, such as pattern recognition and classification, image processing, robotics, and expert systems (Shen & Liu, 2014), especially in the wireless sensor networks (Zhang & Lai, 2014). With the increasing application of target detection and classification, multiple sensors data fusion has emerged recently as an interesting technology (Zhang & Zhang, 2015). The raw data of each sensor are often harsh and imprecise, and cannot be used for target classification and decision making directly. As the further research of the community of multisensor data fusion in the past several years, decision fusion is developed which convert the raw data of each sensor into an individual decision. Fusion center collects the local decisions from sensors and combines them into a global one. By transmitting the decision instead of the raw data from an individual sensor, the energy and bandwidth consumption is reduced in WSNs (Zhang & Zhang, 2016). Compared with the detection and classification for a target using a single sensor, multisensor decision fusion achieves more accurate, comprehensive and reliable results. There are a number of theories which can be used during the fusion process, such as Bayesian fusion, evidential belief reasoning, fuzzy reasoning, rough set theory, and possibility theory (Khaleghi & Khamis, 2013; Zhang & Lai, 2014).