Mu Zhou

Wi-Fi/MARG Integration for Indoor Pedestrian Localization

With the wide deployment of Wi-Fi networks, Wi-Fi based indoor localization systems that are deployed without any special hardware have caught significant attention and have become a currently practical technology. At the same time, the Magnetic, Angular Rate, and Gravity (MARG) sensors installed in commercial mobile devices can achieve highly-accurate localization in short time. Based on this, we design a novel indoor localization system by using built-in MARG sensors and a Wi-Fi module. The innovative contributions of this paper include the enhanced Pedestrian Dead Reckoning (PDR) and Wi-Fi localization approaches, and an Extended Kalman Particle Filter (EKPF) based fusion algorithm. A new Wi-Fi/MARG indoor localization system, including an Android based mobile client, a Web page for remote control, and a location server, is developed for real-time indoor pedestrian localization. The extensive experimental results show that the proposed system is featured with better localization performance, with the average error 0.85 m, than the one achieved by using the Wi-Fi module or MARG sensors solely.

PILA: Sub-Meter Localization Using CSI from Commodity Wi-Fi Devices

The aim of this paper is to present a new indoor localization approach by employing the Angle-of-arrival (AOA) and Received Signal Strength (RSS) measurements in Wi-Fi network. To achieve this goal, we first collect the Channel State Information (CSI) by using the commodity Wi-Fi devices with our designed three antennas to estimate the AOA of Wi-Fi signal. Second, we propose a direct path identification algorithm to obtain the direct signal path for the sake of reducing the interference of multipath effect on the AOA estimation. Third, we construct a new objective function to solve the localization problem by integrating the AOA and RSS information. Although the localization problem is non-convex, we use the Second-order Cone Programming (SOCP) relaxation approach to transform it into a convex problem. Finally, the effectiveness of our approach is verified based on the prototype implementation by using the commodity Wi-Fi devices. The experimental results show that our approach can achieve the median error 0.7 m in the actual indoor environment.

Competitive agglomeration based KNN in indoor WLAN localization environment

In this paper, we proposed a novel localization algorithm in indoor Wireless Local Area Network (WLAN) environment. First of all, to conduct the Received Signal Strength (RSS) preprocessing, we eliminate the RSS outliers based on the density function of the difference of RSS. Second, to overcome the problem of the manual selection of the cluster number, as well as the number of the nearest neighbors in K nearest neighbor (KNN) algorithm, we propose to use the Competitive Agglomeration (CA) algorithm to achieve the localization. Third, the extensive experimental results conducted in an actual Nonline-of-sight (NLOS) indoor WLAN environment, as well as in a simulated Line-of-sight (LOS) environment prove that the proposed approach performs well in localization accuracy.

Indoor WLAN Collaborative Localization Algorithm Based on Geometric Figure Overlap

Most indoor localization methods only focus on the relationship between the user locations and environmental layout, while ignoring the relations among different user locations. Thus, we come up with an idea of collaboration to reduce the impact of noise on localization performance. First of all, according to mutual information between the target user and collaborative ones, we construct the geometric figure for different user locations. Second, the candidate marker reference points with maximum overlap are selected for solving a specific localization problem. Finally, the extensive experiments conducted in both the indoor straight corridor and lab demonstrate the effectiveness of the proposed approach with the average localization error within 2 m.

Antenna Ports Detection Algorithm in LTE System Using the Repetition of the Reference Signal

In the Long Term Evolution (LTE) system, the conventional detection algorithm generally use 1, 2, or 4 transmitting antennas to decode the PBCH. Although the power detection algorithm based on the combination of the PBCH and Secondary Synchronization Sequence (SSS) is featured with low detection complexity, the performance of this algorithm may be much poor under the low Signal Noise Ratio (SNR) condition. In this paper, we perform the repetition of cell reference signal for different antenna ports to detect the number of antenna ports. Simulation results demonstrate that the proposed algorithm can reduce the detection complexity, as well as enhance the anti-noise capacity for antenna ports detection.

Carrier Phase Based Attitude Determination Using Tightly Coupled BDS/INS

In order to realize the attitude determination in high accuracy and stability of BeiDou Navigation Satellite System (BDS) receiver carrier, a BDS/INS tightly coupled attitude determination algorithm was proposed. First, the error model of BDS system was put forward. Then, an extended Kalman Filter System with the double differences carrier phase as the main observation and the error state equation of INS as the system state equation which can ensure the attitude in a high precision were designed. Last, a platform was set up for testing the effectiveness of the algorithm with a single-frequency BDS receiver and an inertial sensor. The results show that the algorithm can effectively improve the measurement accuracy and output frequency of the attitude.

A Novel Method to Generate Wi-Fi Fingerprint Database Based on MEMS

Fingerprint-based positioning in Wi-Fi environment has caught much attention recently. One key issue is about the radio map construction, which generally requires significant effort to collect enough Wi-Fi Received Signal Strength (RSS) measurements. Based on the observation that the Micro Electromechanical System (MEMS) can automatically calibrate the target locations without complex equipment, we propose an efficient radio map construction method based on the technology of multi-sensor. Different from the conventional methods, the proposed one first relies on the gait detection approach and quaternion-based extend Kalman filter algorithm to estimate the velocity and heading of the target. Second, the Pedestrian Dead Reckoning (PDR) algorithm is used to calculate the current location of the target in a real-time manner, and meanwhile the data from Wi-Fi module are collected to generate the fingerprint database. The experimental results show that the proposed method is effective in positioning accuracy and efficient by saving the time and energy.

A Method of Fast Synchronization Based on PSS in TD-LTE Cell Search

During the first step of cell searching in Time Division Long Term Evolution (TD-LTE) system, the symbol timing synchronization is the most import part, it can not only enhance the performance of anti-frequency-offset but also reduce the complexity. In this paper, we propose a novel Primary Synchronous Signal (PSS) timing synchronization algorithm. Based on the fast convolution and Overlap-save, the proposed algorithm achieves the joint estimation of the symbol timing synchronization and coarse frequency offset in frequency domain. The theoretical analysis and simulation results prove that the proposed algorithm can not only reduce the complexity, but also improve the performance of anti-frequency-offset for PSS timing synchronization in TD-LTE system.

Quadrilateral model based Radio Tomographic Imaging in random Wireless Sensor Network

Radio Tomographic Imaging (RTI) is a popular technique for locating the target by using the characteristic that the Received Signal Strength (RSS) varies dramatically as the signal is obstructed by the target in Wireless Sensor Network (WSN). The existed works mainly rely on the establishment of linear models to locate the target by using the weight matrix in an elliptical model. This paper first gives an indoor environment with randomly distributed wireless sensors, and then proposes a linear model in a quadrilateral relating the changes of RSS to locate the target. In our approach, the Minimum Mean Square Error (MMSE) is utilized to construct the weight matrix for the linear model. Furthermore, the concept of pictures superposition is considered to process the images to enable the highly-accurate localization. The experimental results show that based on the proposed approach, the localization error in the single target situation is lower than 0.2 m by using the superposition of only five quadrilaterals.

Indoor device-free passive localization for intrusion detection using multi-feature PNN

Indoor device-free passive localization is an emerging technique that can be used in a variety of fields, like the intrusion detection and smart homes, which does not require the target to carry any devices or participate actively during the localization. In this paper, we rely on the Probabilistic Neural Network (PNN) algorithm which has been widely used in pattern recognition in combination with the device-free passive localization technique to realize the intrusion detection. We utilize the variance of RSS to classify the different intrusion states. Due to the limitation of single-feature input in providing information for classifier, we propose the multi-feature PNN to improve the accuracy of intrusion detection, as well as area localization. Our experiments conducted in an actual indoor Wi-Fi environment shows that the multi-feature PNN can reach better performance than the PNN with a single-feature input. Finally, the proposed approach achieves higher accuracy compared to some exited device-free passive detection approaches, and our approach can locate the area which the intruder is really located at accurately.