Low-complexity channel estimation via joint localization of angles and ranges for near-field massive MIMO systems

Abstract

In conventional communication systems, it is common to assume plane waves in far-field systems for the convenience of calculation. Through this assumption, when estimating channels and designing and analyzing communication systems, the results could be derived relatively simply. Electromagnetic waves, however, essentially take the form of spherical waves, and these plane wave assumptions no longer hold in near-field systems. In recent years, to support high data rate/large capacity in next-generation communication, channel estimation techniques based on massive multi-input multi-output (MIMO) systems are actively studied. In such a next generation communication system, the wavelength is short because of the high center frequency, while the size of the antenna aperture increases due to the large number of antennas, and the users/scatterers are on the wider near-field region. Therefore, in this case, the conventional MIMO system model based on the plane wave becomes inappropriate, and a new MIMO system model based on the spherical wave is required. This thesis presents an efficient channel estimation method for near-field massive MIMO systems. In order to estimate the channel parameters in terms of angles and ranges to the users/scatterers at the transmitter and receiver sides, we formulate the signal model for massive MIMO systems as a function of angle and range parameters. The signal model is then used to construct a variety of cumulant matrices, from which two-dimensional (2D) multiple signal classification (MUSIC) pseudo-spectrums for the angles and ranges are derived. By applying the proposed cumulant matrices to the 2D MUSIC technique, it is possible to localize the users/scatterers with respect to angles and ranges at the transmitter and receiver side, respectively, and the channel state information (CSI) can be estimated more accurately with the corresponding channel parameters. The effectiveness of the proposed algorithm is verified through the derived 2D MUSIC pseudo-spectrums and the simulations by comparing with the conventional channel estimation techniques.