Research On Indoor Fingerprint Localization Algorithm Based On Shapelet

With the rapid development of the mobile Internet and the Internet of Things industry,wireless localization technology has played an important role in the fields of industry,medical care,home and security.Wi-Fi has become a hot spot in the field of indoor localization due to its wide coverage and low price.In Wi-Fi localization,compared to Received Signal Strength Indication(RSSI),Channel State Information(CSI),as fine-grained physical information,has a more sensitive environmental perception capability,therefore is more favored by many researchers.Based on the research of the existing CSI-based indoor fingerprint positioning technology,this thesis introduces the Shapelet algorithm to extract the location fingerprint information,and designs indoor fingerprint localization algorithm based on CSI.The main work is as follows:(1)Based on the differences shown by CSI in different positions,this thesis proposes a indoor fingerprint localization algorithm based on Shapelet decision tree.In the offline stage,the 3-? outlier processing method and Kalman filter are used to preprocess the amplitude of the CSI,which reduces the interference caused by the surrounding environment and people walking.Then combined with CSI data characteristics to improve the Shapelet algorithm,the CSI sub-sequences,which are most distinguishable from other grids as grid fingerprints,is extracted.Then we build a fingerprint database and a Shapelet decision tree.In the online stage,3-? outlier processing and Kalman filtering are performed on the CSI amplitude,and gridal localization is achieved through Shapelet decision tree.This algorithm fully considers the differences of the CSI sequences in each grid.This algorithm also considers the influence of the size of the grid on the localization in the experiment.The experimental results show that,compared with the existing localization algorithm,robustness is significantly improved.(2)In view of the high time complexity of the indoor fingerprint localization algorithm based on Shapelet decision tree and the limitation of classification algorithm,a fingerprint localization algorithm based on Shapelet transformation is further proposed.In the offline fingerprint construction phase,important point is introduced to solve the problems of large shapelet candidate set data,slow screening process,and large deviation of results due to emphasis on local fitting during the feature extraction process.The introduction of the Shapelet transformation algorithm realizes the separation of the feature extraction and classification process.In order to avoid repeated calculations,a filter mechanism is designed.In the online localization phase,the weight K-Nearest Neighborhood(WKNN)algorithm is used to estimate the target position.In order to obtain the best localization result,the value of K is determined.Experimental results show that compared with existing algorithms,the localization performance of this algorithm is further improved.Finally,this thesis uses the Intel 5300 network card to build an indoor localization prototype system,and conducts experiments in a variety of environments and different time periods to verify the feasibility of the fingerprint localization algorithm based on Shapelet.The thesis has 41 pictures,10 tables,and 78 references.