2D Direction of Arrival Estimation using Channel State Information for UCA

Abstract

This paper presents technique of two dimensional direction of arrival (DOA) estimation of a source using uniform circular array (UCA) and channel state information (CSI). In orthogonal frequency division multiplexing (OFDM) system signal is simultaneously transmitted over subcarriers with different frequencies. CSI is a value that describes complex gain/attenuation and phase shift introduced by wireless channel/environment on every subcarrier. Noting the fact that CSI values among antennas and subcarriers include phase shift due to DOA and Time-of-Flight (ToF), a novel technique of virtually extending the number of antennas by estimating the DOA and ToF at the same time is described. Simulation results show that accurate DOA estimation can be achieved even with small number of antennas. Introduction Direction of arrival estimation (DOA) problem arouse in the middle of last century in radar and sonar systems, which are classical applications of sensor array signal processing. The main task of DOA estimation is to extract azimuth and elevation angle informations from the received signal field. Wireless technologies developed very fast during the last few decades, and nowadays millions of people are using wireless equipment in daily life. To meet the increasing requirements of data traffic more complex modulation techniques and smart antenna structures were developed, and it became possible to use array signal processing techniques in wireless communication for mobile/wireless user localization. Multiple Input Multiple Output (MIMO) is a smart antenna technology that is already used in many communication systems. ArrayTrack[1] is an indoor localization system that exploits MIMO in Wi-Fi to track users at a very fine granularity in real time indoors. In ArrayTrack Access Points (AP) overhear the transmission and compute DOA based on the received frame. But, it doesn’t exploit multiple carrier property of orthogonal frequency division multiplexing (OFDM), which is implemented in widely used 802.11W-Fi and Long-Term Evolution (LTE) standards. In [2], a novel approach called FILA explores frequency diversity of the subcarriers and leverages channel state information (CSI) values to build a propagation model for positioning with fingerprinting, while in [3] a novel location signature CSI-MIMO incorporates CSI and MIMO of each subcarrier for fingerprinting. In [4] system called SpotFi utilizes CSI information reported by commodity Wi-Fi card to simultaneously estimate DOA and Time-of-Flight (ToF) of different sources. But all these methods provide only one dimensional angle information, while nowadays DOA estimation in 2D, i.e. azimuth and elevation angles, is expected due to accuracy requirements. This paper describes method of two dimensional DOA estimation using uniform circular arrays. In OFDM system data is sent on multiple carriers simultaneously, and channel state information value Joint International Information Technology, Mechanical and Electronic Engineering Conference (JIMEC 2016) © 2016. The authors Published by Atlantis Press 119 that represent complex attenuation and phase shift of different subcarriers. CSI among antennas include phase shift due to angle of incoming plane wave, so it is possible to extract azimuth and elevation angle informations. And CSI among different subcarriers include phase shift due to ToF, so by simultaneously extracting angles and ToF informations from CSI it is possible to virtually extend the number of antennas in UCA. System Model and Problem Formulation It is assumed that the array is circular withN antennas, andL narrowband source signals are impinging on it from the far-field. The geometry of the array is shown in Fig. 1. All array elements are assumed to be identical, omnidirectional, and uniformly distributed over the circumference of a circle of radius R in the xy plane. A spherical coordinate system is used to represent azimuth and elevation angles of incoming signal waves, with the origin being located at the center of the array. Elevation angle is measured down from the z axis, and azimuth angle is measured counterclockwise from the x axis. The nth antenna array is displaced from the x axis by an angle γn = 2πn/N,n = 0, 1, . . . , N − 1. Fig. 1: UCA geometry under consideration Due to different positions on the xy plane, complex envelope of incoming signal at the origin and at nth element’s position has phase difference ψn = e jk0Rsinθcos(φ−γn)[5], where k0 = 2π/λ is a wavenumber of a plane wave propagating in space with speed λ, θ and φ are elevation and azimuth angles respectively. The array steering vector for the lth path can be written as a(θl, φl) =  e0ll ejk0Rsinθlcos(φl−γ1) .. ejk0Rsinθlcos(φl−γN−1)  (1) and there are as many steering vectors as the number of sources. The output signal vector x of antenna array, obtained by superposition of signals from all the paths is written as