Research On Indoor Localization Algorithm Using Channel State Information And Received Signal Strength Indicator Measurements

With the development of the information age,indoor localization has received more and more attentions.Wi-Fi network has been used for indoor localization widely,because of its wide coverage,and low deployment costs.Wi-Fi positioning research mainly includes Received Signal Strength Indication(RSSI)and Channel State Information(CSI).RSSI can be obtained from the Media Access Control(MAC)layer.As coarse-grained physical information,it is described as the sum of the amplitude of multiple path signals.Because of its universality,it is widely used for indoor positioning.Moreover,CSI can be obtained from the Physical Layer(PHY).As fine-grained physical information,it is composed of several subcarriers and each subcarrier is represented by amplitude and phase information.Therefore,CSI is more sensitive to environmental perception and has a better robustness for indoor positioning.Combing the RSSI and CSI measurements as the fingerprint,in this paper,a new indoor algorithm is proposed which can effectively mitigate the multipath effect on positioning accuracy.The main work of the paper includes:(1)The theoretical knowledge of CSI and RSSI measurements are studied and an experimental platform is built based on localization system.First some fingerprint based on localization approaches and techniques are described in detail.Then a hardware experimental platform for CSI and RSSI measurements is built.The main equipment required for the experiment is: Inter5300 wireless network card,TP-Link router,desktop computer.The operating system is Ubuntu 14.04,its kernel version is 3.13.0-24-generic,and the data acquisition software is CSI-Tool.Finally,this platform is used to complete data collection.(2)A novel localization algorithm using RSSI measurements and CSI measurement is proposed by the Support Vector Machine(SVM)method.In the off-line phase,with the obtained CSI measurement,the Principal Component Analysis(PCA)pre-processing is utilized for dimension reduction at first.Then,the six statistical features of the CSI measurements are extracted.Next,the RSSI measurements,the extracted feature of CSI measurements and the reference positions form the training data set.At last,the SVM technique is proposed for regression learning and obtain the position based on regression function.In the on-line phase,after PCA preprocessing and feature extraction of CSI measurements,the final position can be estimated straightly with the regression function.Since the measurements in medium access control layer and physical layer are used for localization,it can lead to better performance of the proposed algorithm.Moreover,both the CSI and RSSI can be obtained simultaneously.Thus,the cost of hardware is low.The experiment results are shown that the proposed algorithm can offer more accurate localization result than other existing algorithms.(3)A novel localization algorithm using RSSI measurements and CSI measurement is proposed by the extreme learning machine ELM-AdaBoost method.In the off-line phase,we first form the fingerprint training dataset that consists of both nine statistical information of CSI and RSSI.To resolve the performance variation issue of ELM because of its random weights,we formulate an AdaBoost algorithm to combine many weak classifiers and produce a strong classifier.The weak ELM classifiers are generated by using the iteratively updated distribution of the training data.The final classifier(stronger)is obtained weighted majority voting.In the on-line phase,after the feature extraction of CSI measurements,the final position can be estimated straightly with the classification function.It can reduce the data storage and computational needs of existing schemes,and to maximize localization performance.The experiment results are shown that the proposed algorithm has better performance than other existing algorithms,with respect to localization results,storage space and complexity.