Object Material Sensing And Detection Technology Based On CSI Statistical Characteristics Of Wi-Fi Signal

In recent years,mobile Internet and Internet of Things technologies have been rapidly developed,which has also increased people's demand for multiple types of context-aware technologies based on wireless communication systems.Among them,how to achieve high-precision and robust sensing and detection for the material of an object without contact under complex channel environment conditions has become a hot issue in the field of communication sensing technology.With the widespread deployment of Wi-Fi infrastructure,it is easy to obtain Channel State Information(CSI),which can reflect the characteristics of current channel state changes in a fine-grained manner,making CSI-based item material sensing detection a may.However,the existing CSI-based object material perception and detection technology still has many key difficulties to be solved:first,the existing CSI-based object material statistical feature modeling method is difficult to efficiently distinguish the material of the object;Secondly,existing perceptual identification algorithms have disadvantages such as low detection accuracy and poor robustness under complex indoor channel conditions.In response to the above problems,this thesis conducts an in-depth study on the material sensing and detection technology based on the CSI statistical characteristics of Wi-Fi signals.The specific research content is as follows.First of all,this thesis introduces the research background of object material perception and detection technology,and analyzes the difficulties faced by the existing Wi-Fi signal-based object material perception and detection technology.It mainly focus on the two key technologies of statistical feature modeling and perceptual identification algorithm of material materials.Secondly,in terms of the statistical feature characterization and modeling methods of wireless signals oriented to the material of objects,the existing CSI statistical feature modeling has the problem that the statistical characteristics obtained are relatively single,and thus cannot effectively distinguish different item materials.This thesis proposes a CSI perception statistical feature modeling method for object material characterization to extract statistical features.First,the amplitude and phase of the original CSI data are preprocessed separately to solve the problem of large fluctuations and phase shifts of the CSI data caused by the interference of environmental noise and hardware equipment;Then,according to the different effects of multipath on each subcarrier,it is proposed to use the variance information of CSI data to select subcarriers that are less affected by multipath propagation;Finally,through the designed sub-carrier selection strategy,in-depth analysis of the difference of different material materials,so as to extract the statistical characteristics that can efficiently distinguish different material materials.Finally,in terms of the item material perception and identification algorithm,in view of the low efficiency of the existing algorithm classifier and the limited detection accuracy caused by only considering the scene of a single receiving antenna,this thesis proposes a multi-antenna joint judgment algorithm for item material perception and identification.The algorithm first selects an effective classifier,and then uses the detection result of each receiving antenna of the classifier to combine with the majority voting algorithm to get the final judgment result,which realizes the high-precision detection of the material of the object.At the same time,this thesis also designs and builds an experimental platform for object material detection based on the CSI statistical characteristics of Wi-Fi signals.After experimental verification,the proposed perceptual identification algorithm has better detection accuracy and robustness than existing algorithms.