Metal Detection Based On Wireless Network

With the severe of domestic counter-terrorism, security is getting more and more attention from the government. The government has enforced patrols and security forces in public places. Security equipment such as security doors, hand-held detectors have been widely deployed in public places such as railway stations, airports. The main purpose for security equipment is to detect dangerous metal objects, such as guns, knives, and so on. The detection of metal objects is very important in security field. The ways to detect metal for these security equipments is mainly based on electromagnetic induction and X-rays. Detector based on the electromagnetic induction needs to be put close to target and the detector based on X-rays is usually very large, expensive, complex, harmful and the detecting distance is very short. For the above reasons, the current metal detection equipment cannot be large-scale deployment with a low cost and cover most the interior space. With the popular of Wi-Fi, most public places are already covered with Wi-Fi. Low-cost and large-scale deployment is possible if we can use Wi-Fi to detect metal. Since the dielectric constant and the reflection coefficient of the metal are different from nonmetal, the affection of Wi-Fi signal reflection caused by metal is different from nonmetal too. We use this theory to realize the detection of metal. But using Wi-Fi to detect metal will encounter some challenges. The factors such as the interference caused by multipath effects and other signal source, the shape, material, surface smoothness of the target and the angle of reflection will impact on detect result. However, we use CSI(channel state information) of Wi-Fi to get the feature of the communicate link. We implement the system by using a series of methods to eliminate most the interference and use machine learning algorithm to identify the metal. We implement this system in commercial products. We conduct several experiments and its average accuracy and false alarm is more than 90% and less than 10% in 1 meter detect distance.