Analysis of the Use of the Kalman Filter for Signal Detection in a Direct Conversion Receiver in MIMO

In this paper, we consider an algorithm of detecting a signal with quadrature amplitude modulation (QAM) in a multiple inputs and multiple outputs (MIMO) communication system. The algorithm consists of two steps: soft-decision decoder using Kalman filtering and hard-decision decoder using criterion of the minimum distance between soft decision vectors and possible information symbols at each receiving antenna. The Kalman filter estimates symbols at the i-th time point during K iterations. The comparison of the noise immunity of QAM signal reception was performed using the Zero Forcing equalizer, the minimum mean square error (MMSE) algorithm, and the recurrent Kalman filtering. Signal processing takes place under conditions of Rayleigh fading multipath channel with Doppler spectrum spreading. Two cases were considered: the channel matrix is known precisely and the second when it is estimated by the least squares method (LS) using a first-order polynomial approximation for the elements of the channel matrix. The channel matrix contains not only signal distortions due to the multipath channel, but also distortions of the direct conversion receiver, such as amplitude and phase imbalance between signal in-phase and quadrature components, frequency offset due to non-ideal detection algorithm, as well as direct current offset (DC offset). In addition, the computational complexity of the detector based on the Kalman filtering is analyzed.