Principal component analysis of CSI for the robust wireless detection of passive targets

Recent standards for wireless networks introduced spatial and frequency diversity to improve the quality and the bandwidth of the wireless communications. The arising channel state information has been exploited for the wireless localization of device-free targets. In this work, a principal component analysis has been applied to the channel state information in order to extract a target-dependent feature suitable to enable the robust detection of passive targets in large indoor areas. The experimental validation performed with commercial WiFi access points pointed out a failure rate lower than 3 [%] in a real indoor test field.